Continuous-flow microreactor multi-step synthesis of an aminonaphthalene derivative as starting material for the preparation of novel anticancer agents

A multi-step synthesis of the aminonaphthalene derivate 1 as a key intermediate in the synthesis of the duocarmycin based prodrug 2 for a selective treatment of cancer in a microreactor is described. The conditions for the synthesis in the batch mode were adjusted for application in a microreactor and the results of both methods were compared showing that the transformations in the microreactor in most cases give similar or even better results with the advantage of a continuous-flow production, higher safety and faster reactions with an empirical accelerating factor of F = 3 – 10.


Introduction
A microreactor (MR) is a miniaturized reaction system fabricated using microtechnology and precision engineering, which generally contains a small reaction volume in the nano-to microliter range with characteristic microchannel dimensions between ten and several hundred micrometers. 1 The use of microreactors in the synthesis of chemical compounds is of high interest not only for research but also for production, as reactions in microreactors have several advantages over those in the usual equipment.When two or more different fluids are introduced into microreactors, a laminar flow will dominate.As a consequence from miniaturization, the molecular diffusion path is reduced and the diffusion-controlled mixing is significantly accelerated in a way that mixing goes to completion normally within milliseconds to a few seconds at the most. 2,3,5Moreover, due to a highly efficient heat transfer hot spots do not occur and thermally induced undesired side reactions are effectively suppressed.Thus, microreactors are especially useful for very fast and highly exothermic reactions. 3,4,5In addition, the continuous processing 6 has an inherent operational safety and an economical advantage, since the synthesis of large quantities of a product can be done by numbering-up the reaction vessels instead of using a conventional time-and cost-consuming scale-up process.9a On the other hand it should not be denied that microreactors still show several problems, 8,9 which hamper their use.Typical among these is the chemical clogging of the tiny microchannels, since many chemical transformations involve solid handling or the formation of precipitates.In addition, multi-step syntheses using microreactors have so far seldom been reported. 10n the development of the duocarmycin based prodrugs of type 2 for a selective treatment of cancer using the ADEPT approach (Scheme 1), 11 the aminonaphthalene derivate 1 serves as an important building block and its synthesis has been carried out according to a classical route (Scheme 2). 12Due to the need of large amounts of the prodrug 2 for clinical trials an efficient continuous synthesis of 1 was needed to allow the preparation of 2 in a kg-scale.We assumed that the use of a microreactor would open a reasonable way to achieve this aim with good reproducibility of product quality by precise control of reaction conditions in a continuous way.

OBn
Here we report on a multi-step synthesis of the aminonaphthalene derivate 1 in a microreactor (CYTOS ® 13 College, Figure 1).First we focused on the modification of the classical reaction conditions in order to allow the use of a microreactor by avoiding the employment and formation of solid material.At last the results using the batch mode and the microreactor were compared in respect to yield, reaction time and temperature.The following symbols are used, t: reaction time in the batch mode at a given reaction temperature; τ: residence time in the microreactor (residence volume / total flow rate); rt (room temperature): 22-25°C; T: reaction temperature in the batch mode (internal temperature) and in the microreactor (controlled by an external thermostat); F: empirical accelerating factor which characterizes in which extent the use of a microreactor can accelerate a chemical reaction, defined as F = t (reaction time) / τ (residence time).It should be noted that throughout the manuscript the given F values are only estimations, since an exact determination of the kinetics of the transformation using GC has not been performed.Furthermore, the given yields for the transformations in the microreactor refer to the results under steady-state conditions; thus, the production before that time is usually lost, which however is of no importance, if the total production time is much longer than the start-up time.For the synthesis of the tert-butyl ester 4 using the bromoacetyl bromide 3 as starting material the solvent diethylether employed in the batch mode was not suitable for the microreactor due to the formation of an ammonium salt as precipitate.14a THF proved also not to be appropriate due to a side reaction with 3. 15 Among other solvents investigated as CH 3 CN, DMSO, tBuOH and DMF, tBuOH and DMF showed an overall homogeneous reaction with moderate to high yields (Table 1, entry 2-4).The use of tBuOH as solvent afforded a yield of 81% in the batch mode within 20 h at room temperature, whereas in the microreactor with 34 min residence time at 25°C only 50% yield of 4 was obtained.It is possible to increase the yield by reducing the total flow rate to prolong the residence time, which however is less desirable due to a lower efficiency and limited flow rate control.The addition of catalytic amounts of DMAP as additive did not improve the conversion neither in the batch mode nor in the microreactor (entry 2 and 3); thus, under these conditions after 5.5 h 4 was formed in the batch mode in 43% yield.A better result was obtained employing DMF as solvent with 34 min residence time using catalytic amounts of DMAP and a reaction temperature of 35°C to give 4 in 66% yield.A comparison of the reaction rates under the different conditions revealed that the transformation in the microreactor proceeds with an accelerating factor of about 10.Table 1.Synthesis of 4 in the batch mode and in MR: [a] Isolated product by distillation.[b] Dry dichloromethane was used to dilute 3 (V/V=1:1).[c] Reaction without addition of DMAP

Entry
Solvent additive Batch mode yield [a] (t, T) MR yield [a] (τ, T) 3.0 eq tBuOH 81% (20 h, rt) 50% (34 min, 25°C) 3 [b] 3.0 eq tBuOH + 10-mol% DMAP 43% (5.5 h, rt) 50% (34 min, 25°C) 4 [b] DMF + 10-mol% DMAP 62% (19.5 h, rt) [c] 66% (34 min, 35°C) To prepare the phosphonosuccinate 9 and its anion 10 for the olefination of benzaldehyde 11 to give 5, sodium hydride is commonly used as base (Scheme 4). 16However, the intermediate 8 (M = Na) has a low solubility in THF and the by-product NaBr formed by the following nucleophilic substitution was almost insoluble in THF.On the other hand, LiBr dissolves well in THF, we therefore used the organic lithium reagents Li-HMDS, LDA and n-BuLi for the deprotonation.n-BuLi was most suitable to obtain an overall homogeneous reaction mixture and for that reason was employed in the microreactor (Table 2, entry 2-4 to give the desired compound 9 in a yield of 70% already after 24 min at 40°C, whereas in the batch mode only 57% of 9 after 36 h at room temperature were formed.Other approaches using phase transfer conditions (aqueous NaOH/TBABr) or the combination of LiCl/NEt  [a] MR yield (τ, T) [a] 1 NaH 91% (24 h, rt) ⎯ 2 Li-HMDS 87% [b] (12 h, rt) not performed 3 LDA 70% [b] (15 h, rt) not performed 4 n-BuLi 57% (36 h, rt) 70% (24 min, 40°C) In the following Wittig-Horner olefination 18 conventionally performed in two stages, 16 again we searched for an overall homogenous reaction.In the microreactor NaH and n-BuLi with THF as solvent could not be used, due to the formation of the in THF insoluble intermediate 10 (M = Na) and the by-product 12 (M = Li) (Scheme 4 and Table 3, entry 1 and 2).Using DMSO as solvent with n-BuLi allowed a transformation in the microreactor at elevated temperature, but the yield was much lower as in the batch mode.However, the olefination using NaOEt 19 in the microreactor gave a yield of 89% of the desired compound 5 with a residence time of 47 min at 25°C, while a shorter residence time afforded a mixture of 5 and the starting material (entry 4).
Compared to the batch mode the accelerating Factor F is about 6.It is worthwhile to note that 5 formed in the microreactor has an excellent purity and that the reaction proceeds with high stereoselectivity; thus, the Z-isomer of 5 could not be detected (NMR).The in situ generation of the intermediate 10 seems to be very beneficial, since another approach in the microreactor using a separate deprotonation with NaOEt under otherwise the same conditions as in entry 4 afforded only a mixture.Noteworthy, for reactions in the microreactor the collection period is important.The following different yields were obtained using different times to start the collection (conditions, entry 4): Collection from 1.0 τ over 21 min: 57%, collection from 1.5 τ over 20 min: 78%, collection from 1.9 τ over 30 min: 89%.Obviously, even after a time of 1.9 τ the steady state conditions have not been ensured, which resulted in an overall lower yield in the microreactor compared to the batch mode.Usually, a collection beginning time of 1.5 τ is suggested.13c,d

Entry
Base / solvent Batch mode yield [a] (t, T) MR yield [b] (τ, T) NaOEt / EtOH 97% (5 h, rt) i) 80% (23.5 min, 25°C) ii) 89% (47 min, 25°C) In the following acid-catalyzed cleavage of the tert-butyl ester function in 5 the neat substrate 5 as solid needed to be dissolved in dichloromethane as additional solvent.By this way it could be separately pumped into the microreactor to react with TFA.In the batch mode the addition of dichloromethane had no influence on the yield.The transformation is a fast reaction and was completed in just 5 min at 34°C in the microreactor (Table 4), which is in accordance with data from the literature.The resulting acid 13 was used as crude product for the following Friedel-Crafts acylation; for this transformation the conventionally employed reagents Ac 2 O and KOAc or NaOAc 21 were not suitable for the application in the microreactor due to the poor solubility of the salts in organic solvents.Looking for modifications we realized that the transformation is baseaccelerated.In the absence of a base using only Ac 2 O the desired product was not formed.Therefore, NEt 3 was employed and proved to be more suitable than KOAc giving a yield of 86%-96% in the batch mode (Table 5).For the reaction in the microreactor the substrate 13 and Ac 2 O as well as NEt 3 were premixed at room temperature and then pumped into the preheated microreactor (130°C) by one inlet to give 6 with a yield of 100%, while at lower temperatures (about 100°C) only mixtures of the substrate and the product were observed (Table 5, entry 2 and 3).

Entry
Solvent Batch mode yield [a] (t, T) MR yield [a] (τ, T) The subsequent solvolysis of 6 was performed in the microreactor using NaOEt 22 in catalytic amounts instead of K 2 CO 3 as used in the batch mode.The best yield of almost 100% was achieved by using 30-mol% of NaOEt at 72°C for 23.5 min (Table 6, entry 3).With a smaller amount of NaOEt the reaction was incomplete under these conditions.However, it is also possible to obtain good results by using a lower flow rate with 20-mol% of NaOEt or less, while the reaction temperature can not be increased over 70°C, since EtOH would then form gas ARKAT USA, Inc. bubbles, which is not desirable using a microreactor.The accelerating factor in the microreactor was estimated as about 3.

Entry
Base in Ac 2 O Batch mode yield [a] (t, T) MR yield [a] (τ, T) The protection of the naphthol 14 as its benzyl ether 7 could conventionally be achieved using K 2 CO 3 as base in 89% yield (Table 7, entry 1).To allow a homogeneous reaction mixture LiOEt and (iPr) 2 NEt 23 were used as base, but under these conditions only an incomplete conversion was observed (entry 2 and 3), and with aqueous NaOH (>2.0 eq) a mixture of the undesired acid 15 24 und the desired product 16 was formed.Table 6.Synthesis of 14: [a] Crude product.[b] In addition, 10% of the starting material was found

Entry
Base in EtOH Batch mode yield [a] (t, T) MR yield [a] (τ, T) 87% [b] (72°C, 23.5 min) 3 30-mol% NaOEt ⎯ 100% (72°C, 23.5 min) However, by using aqueous NaOH (1.2 eq), the phase transfer catalyst TBABr and the waterinsoluble organic solvent 2-methyltetrahydrofuran (MTHF), 25 a liquid-liquid biphasic reaction in the micro-reactor gave the benzyl ether 7 in 72% yield within 31 min at 40°C.As also observed on other occasions, microreactors are especially suitable for liquid-liquid biphasic reactions.aq.NaOH (1.2 eq) 70% [d] (40°C, 2 h) 72% [d] (40°C, 31 min) For the hydrolysis of the ethyl ester 7 at first LiOH was used, which however formed a suspension and the reaction proceeded rather slowly (Table 8, entry 1).In contrast, using NaOH at elevated temperature the reaction was complete in the batch mode in a much shorter time (entry 2).Employing these biphasic conditions without the addition of a phase transfer catalyst in the microreactor, the acid 16 was obtained in 100% yield in 48 min at 68°C.  a] (t, T) MR yield [a] (τ, T) The obtained carboxylic acid 16 was directly used for the reaction with the Shioiri-Yamada reagent (DPPA) 27 in the presence of tert-butanol to give the desired final product 1 via a Curtius rearrangement.According to the proposed mechanism 27a the intermediate azide 17 would lead to the isocyanate 18 which further reacts with tert-butanol to give 1 (Scheme 8).It was expected that the nitrogen formation in the Curtius rearrangement would have a negative influence on the steady state situation in the microreactor.Thus, for a reproducible reaction we prepared the isocyanate 18 using 1,4-dioxane as solvent in the batch mode.After the gas evolution had finished, the resulting solution was introduced into the microreactor together with tert-butanol containing a little of 1,4-dioxane.However, even under these conditions the obtained yield of 52% of 1 was much lower than those in the batch mode (Table 9).This seems to be a result of the instability of the isocyanate 18; we first thought that the use of molecular sieves in the batch mode, which are clearly unsuitable for reactions in the micro reactor, has caused the difference; but a reaction in the batch mode without the addition of molecular sieves gave the same results.12b It is noteworthy that a homogenous solution could be collected at the outlet of the microreactor, while the formation of a suspension was observed in the batch mode.This phenomenon can be explained by the finer size of the particles obtained in the microreactor.

Conclusions
In conclusion, a nine-step synthesis of the aminonaphthalene 1 was accomplished using a microreactor as reaction vessel in a continuous-flow mode, allowing a large-scale production in an easier and safer way as in the batch mode.Moreover, an accelerating factor for the different reactions of F = 3−10 and mostly similar and sometimes even better yields using the microreactor were found.

Experimental Section
General Procedures.All reagents obtained from commercial sources were normally used without further purification.All solvents were dried by standard methods.Thin-layer chromatography (TLC) was performed on precoated silica gel plates (SIL G/UV 254 , Macherey-Nagel GmbH & Co. KG).Silica gel 60 (0.032−0.064 mm) (Merck) was used for column chromatography. 1 H-NMR spectra were recorded on Varian mercury-200 or Unity-300.All reactions in the batch mode were performed in normal laboratory glassware.As microreactor the CYTOS ® College System (MR, CPC Systems) was used consisting of one micromixer (core structure) with 2 mL residence volume and a residence unit with 45 mL volume.The flow rates of 0.4-20 mL/min were controlled by two integrated ISMATEC ceramic piston pumps.The temperature in the microreactor and the attached residence unit was regulated by an external thermostat (Lauda RKS-20 or Lauda Integral T1200).10b,13c : Before the reaction the MR was purged with the solvent used in the transformation (at least a 2-fold volume of the microreactor).The pumps A (for inlet A) and B (for inlet B) of the MR were calibrated to the desired flow rates.The residence time τ was defined by the equation: τ (min) = volume (mL) / total flow rate (mL/min).After preparation of the solution of the substrates (one or two) the transformation was started by pumping the solution(s) into the micromixer.13c The accuracy of the flow rate was verified by measuring the consumed volume of the solutions of the substrates after a given time period.
Procedure: Only the experiments with the best results in the batch mode and the MR are described.
(E)-2-Benzylidene-succinic acid 1-ethyl ester (13).Batch mode.A mixture of TFA/H 2 O (V/V = 9:1, total volume 62 mL) was added to a stirred solution of 5 (13.1 g, 45.1 mmol, 1.00 eq) in dichloromethane (40 mL) at rt and stirring was continued for 16 h.The solvent was evaporated in vacuo and for azeotropic removal of TFA the residue was taken up in toluene and concentrated in vacuo again (3 times with 25 mL).Then an aqueous saturated NaHCO 3 -solution (48 mL) was added carefully.After being stirred for 1 h the mixture was washed with pentane (2 × 30 mL) and acidified to pH 1 with 2 N HCl (36 mL).The resulting mixture was extracted with EtOAc (2 × 50 mL) and the combined organic lagers were washed with brine, dried (MgSO 4 ), filtered and concentrated in vacuo to afford 13 as a pale yellow solid (8.42 g, 80%).MR: Residence volume 2 mL, τ = 5 min, T = 34°C.Solution A, a solution of 5 (10.8 g, 37.2 mmol) in dichloromethane (total volume 35 mL) was introduced into inlet A at a flow rate of 0.2 mL/min.Solution B, a mixture of TFA/H 2 O (V/V = 9:1, total volume 45 mL) was simultaneously introduced into inlet B at a flow rate of 0.2 mL/min.The product solution corresponding to 23.6 mL of used solution A was collected over 130 min beginning from a run time of 6 τ.Work up and purification as described for the batch mode afforded 13 as a pale yellow solid (3.73 g, 82%).
The organic solution was dried (MgSO 4 ), filtered and concentrated in vacuo to furnish 14 as a brown solid (30.4-Benzyloxy-naphthalene-2-carboxylic acid ethyl ester (7).Batch mode.To a stirred solution of 14 (0.52 g, 2.4 mmol, 1.0 eq) in 2-methyltetrahydrofuran (MTHF, 10 mL) were added benzyl bromide (0.45 g, 2.6 mmol, 1.1 eq), tetrabutylammonium bromide TBABr (0.18 g, 0.56 mmol, 0.23 eq), water (9 mL) and an aqueous 2.5 M NaOH-solution (1.2 mL, 2.9 mmol, 1.2 eq).The mixture was heated to T internal = 40°C for 2 h.After cooling to room temperature the organic layer was separated, washed successively with 2 N HCl (5 mL), water (5 mL) and brine, dried (MgSO 4 ), filtered and concentrated in vacuo to afford a light brown-yellow solid (0.69 g), which was recrystallized from EtOH to give 7 as a light yellow solid (0.53 g, 70% 3 mmol, 1.19 eq) and DPPA (3.06 g, 11.1 mmol, 1.08 eq).The mixture was heated to reflux for 45 min with evolution of nitrogen to give the isocyanate 18 and afterwards cooled down to 40°C.Then mixture was treated with freshly distillated tert-butanol (192 mL) and heated to reflux (T internal = 85°C) using an oil bath (106°C) for 14 h.After being cooled to room temperature the reaction mixture was filtered to remove a white precipitate .The mother liquor was treated with EtOAc (120 mL), washed successively with 2 N HCl (2 × 70 mL), aqueous saturated NaHCO 3 -solution (2 × 40 mL), water (70 mL) and brine (60 mL).The organic solution was dried (MgSO 4 ), filtered and concentrated in vacuo to provide a yellow solid (3.03 g) which was taken up in diethylether.Some insoluble white solid was filtered off through a sintered-glass-fritted funnel (porosity 4).The resulting mother liquor was concentrated in vacuo and the residue was washed with EtOH to give 1 a light yellow solid (1.49 g, 53%

2 Scheme 1 .
Scheme 1. Compound 1 as building block of prodrug 2 for a selective treatment of cancer.

Table 7 .
Synthesis of 7: [a] Crude product.[b] Isolated product by column chromatography.In addition, 19% of the substrate 14 was reisolated.[c] In the batch mode 23-mol% TBABr and in MR 40-mol% TBABr were used.[d] Isolated product after recrystallization from ethanol

4-Acetoxy-naphthalene-2-carboxylic acid ethyl ester (6). Batch mode
. A solution of 13 (35.2g, 150 mmol, 1.00 eq) in Ac 2 O (770 mL) was stirred at rt for 16.5 h.Then NEt 3 (29.3mL, 210 mmol, 1.40 eq) was added and the mixture heated to reflux (T = 136°C, oil bath: 155°C) for 3 h.The solvent was evaporated in vacuo and the residue taken up in EtOAc (200 mL) and water (200 mL).After basification to pH = 9 with solid NaHCO 3 the aqueous layer was separated and extracted with EtOAc (100 mL).The combined organic layers were washed successively with an aqueous saturated NaHCO 3 -solution (2 × 150 mL), 2 N HCl (2 × 100 mL) and brine, dried (MgSO 4 ), filtered and concentrated in vacuo to yield 6 as a brown solid (37.3 g, 96%).MR: Residence volume 47 mL, τ = 47 min, T = 130°C.Solution A, a solution of 13 in Ac 2 O (total volume 190 mL) prepared by mixing a solution of 13 (8.24g, 35.2 mmol) in Ac 2 O (132 mL) with a solution of NEt 3 (4.98 g, 49.2 mmol, 1.40 eq) in Ac 2 O (50 mL), was introduced into inlet A at a flow rate of 1.0 mL/min.The product solution corresponding to 30 mL of used solution A was collected over 30 min beginning from a run time of 2 τ and over 30 min from the run time of 2.6 τ.Work up and purification as described for the batch mode yielded 6 as a brown solid (1.