Synthesis of novel dendritic molecules based on pyrroloanthracene units

Novel pyrroloanthracenes were efficiently synthesized starting from Diels-Alder adducts of anthracene. Some of these orthogonally bridged, rigid and bulky compounds were used as building blocks for the construction of monodisperse dendritic macromolecules. Molecular masses over 4 kDa could be obtained. Further propagation was problematic because of the particular stability of the 4-(1-pyrrolyl)benzyl cation causing the instability or difficulty of formation of the analogous benzyl ether linkage in the proposed dendritic structures.


Introduction
During the last decade, dendrimers have attracted a lot of attention because of their unique molecular architecture. 1Numerous applications have been foreseen for these molecules e.g. as complexing agents for small molecules and indeed this concept has been experimentally proved. 2 One possible strategy to increase the loading capacity of dendrimers is to incorporate bulky units in order to create large cavities in the macromolecule.The use of highly rigid building blocks will avoid collapse of the dendrimer in poor solvents.It has been demonstrated that Diels-Alder adducts of anthracene, which are highly rigid, orthogonally bridged structures, possess interesting properties to construct supramolecular entities with excellent complexing behaviour.Hydrocarbon dendrimers with triptycene monomeric units have been shown to form crystalline complexes with acetone. 3Cyclophanes consisting of Diels-Alder adducts of anthracene were found to display strong affinity towards certain tetraalkylammonium salts. 4Simple Diels-Alder adducts of anthracene were found to form clathrates with several solvents with a stoichiometry depending upon the nature of the solvent and the structure of the host. 5Fluorescent polymers in which similar entities had been incorporated were shown to be useful for the construction of sensors for dinitrotoluene which allows their use in detectors for land mines as this compound is a volatile impurity in trinitrotoluene. 6It was proved that the cavities introduced in the polymer films by the presence of the bulky anthracene adducts were essential to allow the analyte to penetrate in the structure and hence to induce a detectable change of the fluorescence properties of the polymers.Our group has been interested in the chemistry of anthracene for several years and we have shown that N-phenylmaleimidoanthracenes also have remarkable inclusion properties. 7

Results and Discussion
Taking into account the described interesting properties of orthogonally bridged Diels-Alder adducts of anthracene, we wished to explore the possibility to construct dendrimers based on these species as the AB 2 monomer.We chose to functionalize such adducts with ester (as protected A functionality) and phenol groups (as B functionality) allowing a deprotection strategy by reduction of the ester to alcohol and activation for coupling by well established strategies such as the Mitsunobu reaction or halogenation and subsequent Williamson ether synthesis.
In a first approach, we started from the Diels-Alder adduct 1a of anthracene and 2,5dimethoxy-2,5-dihydrofuran (2a) (Scheme 1).Although the commercial cis-trans mixture of the latter was used, only the cis adduct was isolated.Addition of 10 mol% of potassium carbonate was found to increase the yield significantly, most probably by avoiding the acid catalyzed decomposition of the dihydrofuran.The obtained adduct 1a could readily be condensed in the presence of a catalytic amount of p-TsOH with anilines bearing an electron withdrawing or neutral substituent yielding the pyrroloanthracenes 3a-h listed in scheme 1.However, the reaction was found to fail when strongly electron donating substituents are present on the aniline.This can be explained assuming the mechanism presented in scheme 2. Electron donating substituents will destabilize the supposed enamine intermediate 4 and hence disfavor the cyclization to the pyrrole.Our observation that aliphatic amines, in spite of their higher nucleophilicity, fail to react as well is consistent with this reasoning as is the observation that the yields are significantly higher when strong electron withdrawing substituents are present on the aniline.

Scheme 1
In order to extend the scope of this reaction towards anilines bearing electron donating substituents, we prepared the Diels-Alder adduct 1b of anthracene and 2,5-bisacetoxy-2,5dihydrofuran (2b).The latter compound was prepared via a literature procedure. 8Also in this case, addition of KOAc as a base dramatically improved the yield of the adduct although a rearrangement to lactone 5 could not be completely avoided as still 10% of this compound was formed.However, adduct 1b was found not to react with aromatic amines and hence this approach turned out to be useless.

Scheme 2
The ethyl ester 3c could readily be reduced with LiAlH 4 yielding the benzyl alcohol 3i which we intended for use as peripheral unit for our dendritic branches.In order to allow easy coupling under Williamson conditions we tried to convert the alcohol function of 3i into a leaving group.However, numerous experiments towards this goal failed: neither the chloride (by treatment of the alcohol with SOCl 2 ) nor the bromide (by treatment with CBr 4 /PPh 3 ), nor the trifluoroacetate (by treatment with trifluoroacetic anhydride/Et 3 N), nor the trichloroacetate (by treatment with trichloroacetyl chloride/Et 3 N) could be detected by TLC or mass spectrometry.We found the dichloroacetate stable enough to be detected by TLC but after work up of the reaction mixture, this compound had decomposed as well.Finally, the monochloroacetate 3j could be obtained, characterized and substituted in a test experiment with phenolate (Scheme 3).As could be expected, this substitution did not proceed on the benzyl position but on the chloromethyl residue yielding phenoxyacetate 3k.On the same principle, chloroacetate 3j could be coupled with 1,3,5tris(4-hydroxyphenyl)benzene and with 3,6-diphenyl-2,5-diketopyrrolopyrrole 6 9 affording the respective dendritic molecules 7 and 8 of generation 0.

Scheme 3
In order to overcome the instability problems preventing the conversion of the alcohol function of 3i into a leaving group, we prepared analogous pyrroloanthracenes starting from the Diels-Alder adduct 9a of anthracene and trans 1,2-dibenzoylethylene (Scheme 4). 10 Although we found that this compound required more drastic conditions to be transformed into the corresponding Nphenylpyrroloanthracenes 10a,c-e (refluxing xylene instead of toluene), this adduct also afforded the desired compounds when reacted with anilines bearing electron donating substituents.After reduction of the ester 10a to the benzyl alcohol 10b, we encountered the same problem to convert the alcohol function into a leaving group.
For the construction of a functional AB 2 monomer we prepared the Diels-Alder adduct 9b of 1,2-bis(4-methoxybenzoyl)ethylene and anthracene.This compound could readily be transformed into the bishydroxybenzyl alcohol 10i by cyclization with ethyl 4-aminobenzoate followed by demethylation of 10f with BBr 3 .The latter reaction resulted to some extent in hydrolysis of the ethyl ester.However, by working at -20°C this side reaction could be reduced to acceptable proportions.

Scheme 4
To avoid Williamson ether synthesis and hence the necessity of transforming the benzylic alcohol 3i into a leaving group, we coupled AB 2 monomer 10i with the peripheral unit 3i under Mitsunobu conditions (Scheme 5).The obtained G1-ester 11 was reduced to the G1-alcohol 12 with LiAlH 4 and converted into the bromoacetate 13.The latter compound allowed smooth coupling with the same core reagents as cited higher yielding the G1-dendrimers 14 and 15 (Scheme 6) which could be characterized by NMR spectroscopy.ES mass spectrometry allowed to confirm the molecular mass of 2762 and 4064 Da respectively.These macromolecules were found to degrade to a large extent when kept in solution in chloroform for a few weeks.Careful analysis of the degradation products showed that the benzyl acetate groups were hydrolyzed.This is consistent with the observed instability of the 4-(pyrrol-1-yl)benzyl moiety.Moreover, we found that even Mitsunobu coupling between G1-alcohol 12 and AB 2 monomer 10i failed.Two possible side reactions were experimentally confirmed.The first one yields the hydrazine derivative 16.It is well known that products of this type can be formed upon treatment of benzyl and allyl alcohols with DEAD and PPh 3 in the absence of a good nucleophile. 11It is evident, however, that there is no reason why our AB 2 monomer 10i would show a lack of reactivity.Therefore, we think the extreme activation of benzyl alcohol 12 is likely to facilitate the formation of 16 to a notable extent.A second side reaction results in cleavage of already formed benzyl ether linkages, which is detected by the presence of alcohol 3i in the reaction mixture.Most probably, the benzyl ether bonds are cleaved under the influence of the intermediate formed from DEAD and PPh 3 (Scheme 7).Again, this is a reaction for which excellent stabilization of the intermediate benzyl carbocation is a necessity.The fluorescence of the DPP core of dendrimer 14 was found to be almost totally quenched.This is in sharp contrast with the analogous G0-dendrimer 8 which was highly fluorescent.Most probably, backfolding of the highly electron rich 2,5-bis(4-benzyloxyphenyl)pyrrole units to the core region results in quenching of the fluorescence by electron transfer.Facing these problems mentioned above, we turned our attention towards the analogous AB 2 monomer 10j and peripheral unit 10k in which an extra methylene group is present (Scheme 4).The synthesis of these compounds was straightforward following the same strategy as described above.In this case, the removal of the methyl groups with BBr 3 was found to be cleaner as almost no hydrolysis of the ester occurred.Conversion of alcohol 10k into mesylate 10l was unproblematic.A test experiment in which mesylate 10l was coupled with 1,3,5-tris(4hydroxyphenyl)benzene under Williamson conditions showed that, although G0-dendrimer 17 could be obtained in reasonable yield, elimination giving rise to the alkene 18 would be a problem for further propagation towards higher generations (Scheme 8).Therefore, we coupled the alcohol 10k with our new AB 2 monomer 10j under Mitsunobu conditions which afforded G1-dendron 19 in reasonable yield (Scheme 9).This dendron could be fully characterized by NMR spectroscopy and MS mass spectrometry (m/z = 1586) and could be reduced cleanly with LiAlH 4 affording G1-alcohol 20.However, coupling of the latter alcohol 20 with AB 2 monomer 10j under Mitsunobu conditions failed, as only traces of the desired G2-ester 21 could be detected by mass spectrometry.