Weinreb Amidation as the Cornerstone of an Improved Synthetic Route to A-Ring-Modified Derivatives of Luotonin A

Weinreb amidation of ethyl 4-oxo-3,4-dihydroquinazoline-2-carboxylate with aromatic amines provides a significantly improved route to anilide-type key intermediates for the synthesis of the anticancer alkaloid, luotonin A, and new A-ring-modified derivatives thereof. This method has advantages concerning overall yield, brevity, and versatility with regard to the aromatic amine component, even if the latter has less favourable nucleophilicity, solubility and/or stability properties. This is demonstrated by the concise synthesis of a small library of luotonin A analogues, including a novel thiophene isostere of the alkaloid.


Introduction
Since the isolation of the alkaloid luotonin A (Figure 1) from the plant Peganum nigellastrum Bunge (Zygophyllaceae) in 1997 [1] and the discovery of its cytotoxic activity, which is based on stabilisation of the DNA-topoisomerase-I complex and is thus similar to that of the closely related natural product, camptothecin (CPT; Figure 1), there have been various reports describing the total synthesis and biological evaluation of luotonin A and compounds derived thereof (for reviews on luotonin A and related natural products, see refs. [2][3][4][5]). These synthetic approaches include the assembly of a pyrroloquinoline fragment (rings ABC) with an anthranilic acid synthon [6] as well as free-radical cyclisation reactions [7,8] or Pd-catalysed CC coupling reactions [9] as the key step, the OPEN ACCESS latter usually with final formation of ring C (for a comprehensive overview of all the synthetic strategies employed so far, cf. ref. [5]). As it has become quite evident that modification of ring A is highly interesting with respect to achieving enhanced cytotoxicity [5], synthetic approaches which utilise easily accessible and highly variable A-ring building blocks are of particular value. In this context, three different methods employing aniline or substituted anilines as such synthons have been developed. In Batey's approach [10], the key step is an intramolecular aza-Diels-Alder reaction of an imine-containing azadiene with a propargyl dienophile, mediated by dysprosium triflate (an application of the Povarov reaction). Closely related is a one-pot reaction recently published by Chu [11] which gives luotonin A or some derivatives thereof, albeit in modest yields, from isatoic anhydride, propargylamine, glyoxal and anilines in the presence of ytterbium triflate. In our hands, the cycloaddition approach reported by Yao [12] (Scheme 1) gave the highest overall yields and it worked most reliably in the final cycloaddition key step. Scheme 1. Yao's total synthesis of luotonin A [12]. This route is based on the generation of the requisite azadiene from an anilide structure 5 by treatment with Hendrickson's reagent [bis(triphenyl)oxodiphosphonium trifluoromethanesulfonate, prepared in situ from triphenylphosphine oxide and trifluoromethanesulfonic anhydride]. In Yao's protocol [12], the intermediate anilide 4 is prepared by saponification of ethyl 4-oxo-3,4-dihydroquinazoline-2-carboxylate (1) [13][14][15] with lithium hydroxide in aqueous tetrahydrofuran, followed by treatment of the carboxylic acid 2 with oxalyl chloride and subsequent reaction of the acid chloride 3 thus obtained with aniline/sodium bicarbonate in dichloromethane.

Results and Discussion
When we sought to employ this protocol for the preparation of a small library of new luotonin A derivatives bearing various substituents at ring A, we soon found out that the sequence of ester hydrolysis/acid chloride formation/aminolysis constitutes a serious bottleneck for two reasons: (a) the free carboxylic acid 2 is almost insoluble in most solvents, but it is highly sensitive towards decarboxylation, and special care has to be taken during the preparation and work-up of this intermediate; (b) the reaction of the very sparingly soluble acid chloride 3 with aromatic amines can also be problematic, especially if they have less favourable properties than aniline in terms of nucleophilicity, solubility, and/or stability. In order to overcome these limitations, we have now developed a highly efficient alternative route for the preparation of these key intermediates which enabled us to synthesize a series of A-ring modified luotonin A derivatives, including a novel thiophene analogue of the lead compound.
In a first attempt to circumvent the solubility problems mentioned above, we reversed the order of the two steps, N-alkylation (with propargyl bromide) and ester hydrolysis (Scheme 2). However, this approach met with failure: hydrolysis of the N-propargyl ester 6 with lithium hydroxide in aqueous tetrahydrofuran at room temperature or below always resulted in immediate decarboxylation of the initially formed carboxylic acid upon neutralisation (pH 7), leading to the 2-unsubstituted quinazolinone 7, which is known from literature [13][14][15].

Scheme 2.
Hydrolysis of the ester 6, followed by spontaneous decarboxylation. Consequently, we envisaged direct transformation of the ester function into the corresponding anilide as a general and versatile solution of this problem. Weinreb's method [16,17] for amide synthesis from esters and (aliphatic or aromatic) amines by activation of the amine component with trimethylaluminium appeared to be an attractive option. Indeed, when the N-propargyl ester 6 was reacted with AlMe 3 -activated aniline in 1,2-dichloroethane solution (this solvent had been reported to be superior [18]), slow formation of the anilide was detected by TLC. However, the reaction could not be brought to completion even after prolonged refluxing and by using a larger excess of reagent. We assume steric hindrance of the ester group by the adjacent propargyl moiety to be responsible for this insufficient reactivity. Therefore, the N-3 unsubstituted ester 1 was chosen as a more promising substrate for Weinreb amidation. It turned out that this approach indeed gives excellent results: typically, 1 is completely consumed within 1-2 hours of refluxing, and the corresponding anilides 4 are obtained in >90% yield (Scheme 3). This applies even to sparingly soluble and/or electron-poor anilines like 4-nitroaniline or 4-aminobenzonitrile. After quenching of the amidation reactions with aqueous acid, the products 4 are typically isolated simply by filtration.
For the subsequent N-alkylation step with propargyl bromide, we again had to modify Yao's protocol [12] in order to circumvent solubility problems with some of the substituted anilides 4. Instead of the liquid/liquid two-phase method (toluene/water, tetrabutylammonium bromide as phasetransfer catalyst), we used dimethyl formamide/potassium carbonate as a reaction medium, which reliably gave the products (5) in yields between 54% and 86% (Scheme 3). Scheme 3. Two-step synthesis of the key intermediates 5 from the ester 1. With this optimised pathway to the 3-substituted quinazolinone-2-carboxylic acid anilides 5 now available, we prepared several of these cycloaddition educts and subjected them to the cycloaddition conditions [12] (triphenylphosphine oxide, triflic anhydride, dry dichloromethane as solvent; see Scheme 4). In most cases, the intramolecular aza-Diels-Alder reaction takes place very smoothly at room temperature within one hour. Thus, the known compounds luotonin A (8a) [1], 2-chloroluotonin A (8b) [19], and 4-methylluotonin A (8c) [11] as well as the new analogue 1,3-dimethoxyluotonin A (8d) were prepared in high yields (82-99%). When the aniline unit bears a strongly electronwithdrawing and solubility-decreasing group, as it is the case with the 4-nitroanilide 5e and the 4-cyanoanilide 5f, the cycloaddition proceeds somewhat less efficiently, affording the hitherto unknown 3-nitroluotonin A (8e) and 3-cyanoluotonin A (8f) in 69% and 54% yield, respectively, along with some unreacted educt [20].
In order to further explore the scope and limitations of this concise route to luotonin A derivatives, we briefly investigated its suitability for generating heterocyclic A-ring analogues, using 4-aminopyridine and 2-aminothiophene as building blocks. While the former is commercially available and undergoes Weinreb amidation with 1 very smoothly to afford the N-(4-pyridyl)amide 9 (Scheme 5), the latter amine is rather unstable and rapidly decomposes after isolation as the free base. We used a modification of Binder's method [21] for the preparation of 2-aminothiophene by deprotection of N-BOC-2-aminothiophene, employing a mixture of dichloromethane and trifluoroacetic acid at 0 °C.
Instead of attempting to isolate the free amine, we simply washed the organic solution with aqueous sodium carbonate and subsequently performed a solvent exchange from dichloromethane to 1,2-dichloroethane, keeping the temperature below 20 °C all the time. After drying, this solution of 2-aminothiophene is well-suited for use in the Weinreb amidation, thus affording the N-(2-thienyl)amide 11 in 95% yield (Scheme 5). lit. [12] Both of the N-(hetaryl)amides 9,11 were selectively alkylated at the quinazolinone nitrogen with propargyl bromide in analogy to the preparation of compounds 5. However, attempted intramolecular cycloaddition by treatment with Hendrickson's reagent showed that the pyridine derivative 10 does not react under these conditions: only unchanged starting material and triphenylphosphine oxide were recovered (a tiny TLC spot with an intense blue fluorescence indicated that traces of the 2-azaluotonin A might have been formed, though). On the other hand, we succeeded in the transformation of the thiophene intermediate 12 into the desired pentacyclic compound 13 in reasonable yield (55%), the latter product representing a novel thiophene isoster of luotonin A [22].
Preliminary in-vitro screening (using the Resazurin assay [23]) for antiproliferative activity towards six different human tumor cell lines showed that 13 is practically inactive, whereas 8d exhibits a slightly better activity profile than that of the lead compound, luotonin A [24]. More detailed biological investigations are intended.

Conclusions
By applying Weinreb's amidation method to the preparation of the anilide-type key intermediates, we achieved a significant improvement of Yao's synthetic route to luotonin A and A-ring modified derivatives thereof in terms of overall yields (8a: 65%, 8b: 66%, 8c: 47%, 8d: 50%, 8e: 30%, 8f: 33%; all isolated overall yields are based on commercially available anthranilamide as starting material), brevity and-most importantly-versatility with regard to the aromatic amine component, even if the latter has less favourable nucleophilicity, solubility and/or stability properties. This is demonstrated by the concise synthesis of a small library of luotonin A analogues, including a novel thiophene isoster of the alkaloid.