Heterotrimetallic Double Cavity Cages: Syntheses and Selective Guest Binding.

A strategy for the generation of heterotrimetallic double cavity (DC) cages [Pdn Ptm L4 ]6+ (DC1: n=1, m=2; and DC2: n=2, m=1) is reported. The DC cages were generated by combining an inert platinum(II) tetrapyridylaldehyde complex with a suitably substituted pyridylamine and PdII ions. 1 H and DOSY nuclear magnetic resonance spectroscopy (NMR) and electrospray ionization mass spectrometry (ESIMS) data were consistent with the formation of the DC architectures. DC1 and DC2 were shown to interact with several different guest molecules. The structure of DC1, which features two identical cavities, binding two 2,6-diaminoanthraquinone (DAQ) guest molecules was determined by single-crystal X-ray crystallography. In addition, DC1 was shown to bind two molecules of 5-fluorouracil (5-FU) in a statistical (non-cooperative) manner. In contrast, DC2, which features two different cage cavities, was found to interact with two different guests, 5-FU and cisplatin, selectively.

Introduction

Nature exploits self‐assembled architectures (DNA, RNA, proteins and virus capsids) in a myriad of important biological processes. Inspired by this, metallosupramolecular chemists have developed a vast range of increasingly larger, self‐assembled metal–organic cages (MOCs) and a wide variety of potential applications of these systems have emerged. However, the majority of MOCs studied to date are of very high symmetry and there has been a push recently towards developing lower symmetry systems in order to attain enhanced functionally. In addition to targeting cages with lower symmetry, including heterobimetallic cage systems, there has also been a surge of interest in developing MOCs that feature more than one cavity, as the compartmentalization of molecular recognition events could potentially lead to new applications. Several multicavity MOCs have been developed but the most common examples are constructed using palladium(II) ions as the metal component. Double‐ [Pd3L4]6+, triple‐ [Pd4L4]8+ and even quadruple‐ [Pd6L6]12+ cavity systems have been generated. In addition to these systems there are several mechanically interlocked PdII‐based MOCs that also display multiple cavities. The molecular recognition properties of some of these systems have been examined and, in some cases, segregated binding of two different types of guests has been achieved.[ , , ] However, these impressive results have all been generated with homometallic systems.

Herein, building on our recent work that developed heterobimetallic [PdPtL4]4+ cages, we describe a method for the assembly of heterotrimetallic double cavity cages [Pd Pt L4]6+ (DC1: n=1, m=2; and DC2: n=2, m=1) (Figure 1). In addition, we report the host–guest chemistry of these systems with several different neutral guest molecules.

Figure 1

Cartoon representations of homometallic [Pd2(Ltripy)4]4+ (top left) and the related [Pd3L4]6+ double cavity (top right) cages. Bottom: A lower symmetry heterobimetallic [PdPtL4]4+ cage (left) along with heterotrimetallic [PdPt2L4]6+ (middle) and [Pd2PtL4]6+ double cavity (right) cages. Colors: palladium(II)=purple, platinum(II)=pink, grey and light grey=semi rigid linker ligands. Ltripy=2,6‐bis(pyridin‐3‐ylethynyl)pyridine.

Results and Discussion

We set out with the idea to further develop the conditions we used previously to generate a mono‐cavity [PdPtL4]4+ cage (MC), with the goal of synthesizing two new heterotrimetallic double cavity cages [Pd Pt L4]6+ (DC1: n=1, m=2 and DC2: n=2, m=1; Figure 1 and Scheme 1). The required substituted pyridylamine linker precursors (L1 and L2, Scheme 1) were synthesized in modest yields (36–41 %) using Pd‐catalyzed Sonogashira cross‐coupling methods (Supporting Information, Scheme S1 and S2).[ , , ]

Scheme 1

Synthesis of the double cavity cages DC1 ([PdPt2L4](BF4)6) and DC2 [Pd2PtL4](BF4)6: DC1: i) Pt (2 eq.), L1 (4 eq.), and [Pd(CH3CN)4](BF4)2 (1 eq.), [D6]DMSO, RT, 5.5 h; DC2: ii) Pt (1 eq.), L2 (4 eq.), and [Pd(CH3CN)4](BF4)2 (2 eq.), [D6]DMSO, 50 °C, 10 h. The tube structures of DC1 and DC2 were generated using SPARTAN’16® (MMFF models). The molecular models indicated the cages are similar in size/length, the Pt−Pt distance for DC1 was 22.88 Å and the Pd−Pt distance 11.44 Å, while the end‐to‐end Pt−Pd distance for DC2 was 23.23 Å, the end‐to‐middle Pd−Pt distance 11.40 Å and the Pd−Pd distance 11.83 Å. Color: pink=platinum, purple=palladium, light blue=nitrogen, red=oxygen, grey=carbon, white=hydrogen.

We reasoned that combining the inert platinum(II) tetrapyridylaldehyde complex, Pt, with a suitably substituted pyridylamine and [Pd(CH3CN)4](BF4)2 in [D6]DMSO should then lead to the assembly of the two double cavity cages DC1 and DC2 (Scheme 1). Pt (2 eq.), [Pd(CH3CN)4](BF4)2 (1 eq.) and L1 (4 eq.) were combined in [D6]DMSO at room temperature (RT) and the reaction was monitored using 1H NMR spectroscopy (Scheme 1, Figures 1 and 2, and Supporting Information). After 5.5 h, the signals (Figures 2a, c) of the Pt complex and the diamine L1 had completely disappeared and were replaced by a new series of resonances (Figure 2b). Diagnostic of cage formation, the aldehyde and amine resonances of L1 and Pt, respectively, were replaced by a signal (Hi, δ=8.99 ppm) consistent with imine formation and the α‐pyridyl peaks (Ha and Hb) were shifted downfield, relative to L1, indicating coordination to the PdII ion. This behavior mirrored what was observed for the related heterobimetallic cage MC. However, the assembly of the DC1 system proceeded more slowly (5.5 h cf. 1 h).

Figure 2

Stacked partial 1H NMR spectra (400 MHz, [D6]DMSO, 298 K) of a) linker precursor L1, b) double cavity cage DC1, and c) platinum(II) tetrapyridylaldehyde complex Pt. The proton labels correspond to those shown in Scheme 1.

The second double cavity cage DC2 featuring two different cavities could be generated under similar conditions. Combining Pt (1 eq.), [Pd(CH3CN)4](BF4)2 (2 eq.) and L2 (4 eq.) in [D6]DMSO at RT slowly led to the formation of DC2 over 7 d. The reaction was repeated at 50 °C and after 10 h 1H NMR spectroscopy indicated the assembly of DC2 being complete with no resonances for the starting material observed and a new imine signal present at δ=8.96 ppm (Supporting Information). Although these NMR experiments provided no indication that any products other than the new cage systems DC1 or DC2 were formed in these reactions, the isolation and purification procedure adopted indicated small amounts of by‐products were also formed. Thus, addition of ethyl acetate to the DMSO solutions of the initially synthesized DC cages led to the precipitation of colorless or tan solids in high yields (87 or 95 %). Subsequent addition of acetonitrile to these solids resulted in selective dissolution of the cages DC1 or DC2, leaving small amounts of insoluble colorless by‐products behind in each case.

To obtain further insight into these assembly reactions we repeated the syntheses of DC1 and DC2 in the presence of an internal standard (tert‐butanol, Supporting Information). As with the initial assembly experiments the starting materials were completely consumed after 5.5 or 10 h, for DC1 and DC2, respectively, and only proton signals due to the cages could be seen. However, integration of the cage signals versus those of the tert‐butanol internal standard suggested that the cages were generated in 63 (DC1) and 79 % (DC2) yield, respectively. This indicated that there must be other species present, that are not observed in the NMR spectra, that account for the rest of the starting materials. Given that all the starting materials were consumed during the reaction we postulated that the by‐products could be oligomeric/polymeric materials with very broad NMR resonances. However, these by‐products can be removed using the method described above to obtain pure samples of the DC cages.

The purified cages DC1 and DC2 were analyzed and characterized using 1H, and 1H DOSY NMR, HPLC and ESIMS (Supporting Information). The 1H NMR spectra of the acetonitrile‐soluble fractions of DC1 and DC2 in CD3CN were very similar to the spectra observed in [D6]DMSO. In both cases, the 1H DOSY NMR spectra (CD3CN, 298 K) showed that all the proton resonances within the individual samples had the same diffusion coefficients (D =4.50×10−10 m2 s−1 and D =4.55×10−10 m2 s−1), suggesting formation of a single product (Supporting Information, Figures S24 and S25, and Table S1).

Additionally, the diffusion coefficients for DC1 and DC2 were very nearly identical and consistent with the formation of two cage molecules of similar molecular size and shape. Furthermore, the diffusion coefficients of DC1 and DC2 were different to those found for L1, MC and Pt and consistent with the formation of the larger double cavity architectures (Supporting Information, Table S1). ESIMS data obtained from the acetonitrile solutions were also consistent with the formation of the cages DC1 and DC2. For example, the ESIMS data for DC1 featured several major isotopically resolved peaks observed at m/z=496.3036 [PdPt2(L1)4(Cl)]5+, 629.1223 [PdPt2L4(Cl)2]4+, 642.1310 [PdPt2L4(Cl)(BF4)]4+ and 885.1754 [PdPt2L4(Cl)(BF4)2]3+ consistent with the presence of the [PdPt2(L1)4]6+ cage (Supporting Information, Figure S16). HPLC (CH3CN, C18) chromatograms of DC1 and DC2 showed that the samples only contained one compound. Additionally, the retention times of the DC cages were different to the related MC system and the cage precursors L1 and Pt, providing further evidence for the formation of the new DC architectures (Supporting Information, Figure S28).

Having developed a robust method for the synthesis and purification of heterotrimetallic DCs, we next investigated the host–guest chemistry of the systems. We had previously shown that the related heterobimetallic cage (MC) would interact strongly with 2,6‐diaminoanthraquinone (DAQ). Therefore, we initially examined the host–guest chemistry of DC1 and DC2 with this guest molecule. The signals in the 1H NMR spectra ([D6]DMSO, 298 K) of host–guest mixtures were broad but clearly displayed large complexation‐induced shifts indicative of guest binding (Supporting Information, Figures S33 and S34). The NMR spectra obtained for the MC:DAQ host–guest system displayed slow exchange on the NMR time scale and suggested that the DC host–guest adducts behaved similarly. However, interpretation of the NMR data was complicated because the peaks were broadened and/or overlapped. Most likely this was caused by the presence of both 1 : 1 and 1 : 2 host–guest adducts. However, this was potentially also due to different respective orientations of the two bound guests relative to one another. The ESIMS data (Supporting Information Figure S35 and S36) obtained from the host–guest mixtures confirmed the formation of the host–guest complexes with spectra displaying peaks due to both 1 : 1 and 1 : 2 adducts, for example m/z=554.3263 [DC1 : DAQ(BF4)]5+ and 601.9412 [DC1 : 2DAQ(BF4)]5+ (Supporting Information). Ultimately, the molecular structure of the DC1 : 2DAQ host–guest complex was determined using single crystal X‐ray crystallography (Figure 3 and Supporting Information).

Figure 3

The molecular structure of the DC1 : 2DAQ host–guest complex determined using single crystal X‐ray crystallography. Selected distances [Å]: Pt1−Pd1 11.733(2), Pt1−Pt1′ 23.466(2). Color: pink=platinum, purple=palladium, light blue=nitrogen, red=oxygen, grey=carbon, white=hydrogen yellow=carbon atoms of the DAQ guest molecules. Solvent molecules and counterions are omitted for clarity. Only one orientation for each of the disordered DAQ guest molecules is shown.

Single crystals of the DC1 : 2DAQ host–guest adduct were grown by slow vapor diffusion of ethyl acetate into a DMSO solution containing a 1 : 2 host–guest mixture. The X‐ray structure confirmed the formation of the double cavity architecture with PdII coordinated to the central pyridyl moiety and PtII coordinated to the two pyridylimine ends. The crystallographically determined Pt1−Pt1′ 23.466(2) Å and Pt1−Pd1 11.733(2) Å distances correspond well with those determined by molecular modelling (Scheme 1 and Supporting Information). Unsurprisingly, the two cavities of DC1 were of a similar size to that of the analogous [PdPtL4]4+ monocavity cage host–guest adduct with Pd−Pt distances of ≈11.73 Å. Each cavity of the DC1 architecture contained a DAQ guest molecule and, as was observed previously with the monocavity architecture, hydrogen‐bonding interactions between the carbonyl groups of the guest and the acidic α‐pyridyl hydrogens of the cage stabilized the host–guest interaction (Figure 3 and Supporting Information).

Due to the complicated NMR spectra with overlapping peaks, we were not able to carry out a titration to determine the binding constants for the DC1‐DAQ interaction so we examined other guests in order to identify one that did not have this problem. When the known anti‐cancer drug 5‐fluorouracil (5‐FU) was added to either the MC or DC cages, small complexation‐induced shifts (Δδ=0.07–0.13 ppm) were observed for the endo α‐pyridyl hydrogen resonances that line the Pd−Pt cavities of the metallo‐architectures (Supporting Information). This suggested that 5‐FU was bound within the cage cavities. ESIMS data obtained for the cage:5‐FU mixtures displayed peaks consistent with the formation of 1 : 1 host–guest adducts in the cases of the MC and DC2 hosts. Congruent with the observations for the DC1:DAQ system, the ESIMS data for mixtures of 5‐FU and DC1 featured ions consistent with the presence of 1 : 1 and 1 : 2 host–guest complexes. Thus, the combined NMR and ESIMS data suggested that DC1 can form a 1 : 2 host–guest complex with 5‐FU, while the DC2 and MC systems can form only 1 : 1 adducts. A control experiment with [Pd2(Ltripy)4]4+ (where Ltripy=2,6‐bis(pyridin‐3‐ylethynyl)pyridine, see Figure 1 and the Supporting Information) indicated that 5‐FU did not bind with the cavity of that simple single cage system (Supporting Information, Figure S46), consistent with the observation that DC2 only forms a 1 : 1 host–guest adduct with 5‐FU.

The interaction of 5‐FU with each of the cage systems (MC, DC1 and DC2) was examined further using 1H NMR titrations (CD3CN, 298 K) and the corresponding data was curve‐fitted using bindfit to obtain the association constants (K, Supporting Information, Figures S38, S39, S41, S42, S43 and S44). For the MC:5‐FU system the 1 : 1 binding model (K 1=283±5 M−1) provided the best fit to the titration data. Similar results were obtained with the DC2 : 5‐FU system and the 1 : 1 association constant was determined to be K 1=210±3 M−1. 1 : 2 binding models provided the most reasonable fits to the DC1 : 5‐FU titration data (Supporting Information, Table S2). While all the 1 : 2 models provided similar association constants, the statistical model seemed the best giving K 1=1260±20 M−1 and K 2=315±5 M−1.

Enzymes with multiple guest binding sites often display allosteric behavior (either positive or negative cooperativity) with the binding of the first guest causing a conformation change that affects the interaction with the second. There is growing interest in these types of host–guest interactions with metallohosts.[ , ] Related metallomacrocyclic host systems that feature two identical guest binding sites have been shown to display positive cooperativity where the binding of the first guest causes a conformation change that preorganizes the second guest binding site, leading to an enhanced host–guest interaction. Presumably the observed lack of any cooperativity between the 5‐FU‐DC1 binding events reflects the rather rigid nature of the double cavity architecture. The cavities of DC1 are already preorganized for guest binding and the complexation of the first 5‐FU guest causes little to no conformational change in the host architecture leading to the statistical, non‐cooperative binding behavior.

Finally, having demonstrated that mixtures of DC2 and 5‐FU only form a 1 : 1 host–guest complex (K 1=210±3 M−1) with 5‐FU bound within the Pd−Pt cavity of the architecture, we examined if the remaining empty Pd−Pd cavity could bind a second, different guest molecule. It is well established that [Pd2(Ltripy)4]4+ and related systems[ , , ] can bind two molecules of cisplatin (CP) in acetonitrile solution. Therefore, we used 1H NMR spectroscopy (CD3CN, 298 K) to study the segregated guest binding of 5‐FU and CP within the two different cavities of the DC2 cage system. The 1H NMR spectra (Figure 4, and Supporting Information) showed that the addition of 5‐FU (1 eq.) to DC2 caused a downfield shift of the endo α‐pyridyl cage protons Hj and Hs, consistent with the guest binding within the Pt−Pd cavity of the system. CP (2 eq.) was added to the mixture and this caused a shift and broadening of the endo cage protons Ha and Hi that line the Pd−Pd cavity of DC2. We also examined reversing the order of the guest addition (Supporting Information, Figure S48). The addition of 2 eq. of CP to DC2 caused a shift of the Ha and Hi cage proton resonances (again indicating preferential binding to the Pd−Pd cavity) and subsequent addition of 5‐FU (1 eq.) resulted in a final spectrum that was identical to that in Figure 4 (Supporting Information, Figure S48). This provides strong evidence for the selective, segregated guest binding of CP and 5‐FU within the two different cavities of DC2 with the selectivity determined by the different nature of the two cages, not the order of addition.

Figure 4

Stacked partial 1H NMR spectra (400 MHz, CD3CN, 298 K) of DC2 (top), DC2+5‐FU (1 eq.) (middle) and DC2+5‐FU (1 eq.) and CP (2 eq.) complex (bottom). The proton labels correspond to those shown in Scheme 1.

Conclusion

We have developed a method for the assembly of the heterotrimetallic double cavity cages, [PdPt2L4]6+ (DC1) and [Pd2PtL4]6+ (DC2). Combining an inert platinum(II) tetrapyridylaldehyde complex with suitably substituted pyridylamine linker units and PdII ions led to the assembly of the heterotrimetallic cages through reactions facilitated by the reversible nature of imine bond formation and the relatively labile Pd‐pyridyl bonds. 1H and 1H DOSY NMR, ESI‐mass spectra and HPLC data were all consistent with the formation of the double cages. The double cages DC1 and DC2 displayed cavities of similar sizes to those of related homometallic [Pd2L4]4+ and heterometallic cages, and the binding of a variety of different guest molecules within the double cage assemblies were studied. Heterotrimetallic [Pd Pt L4]6+ (DC1) features two identical cavities and forms a host–guest adduct with two DAQ guest molecules, as determined by X‐ray crystallography, and it also binds two molecules of the anticancer drug 5‐FU. In the latter case, the guest binding was statistical and the lack of any cooperativity between the guest binding sites was attributed to the rigid structure of the cage backbone which prevents extensive reorganization after binding the first 5‐FU molecule. DC2 features two different cage cavities and it was shown that the two different guests, 5‐FU and CP, could be bound selectively within the different cavities of the double cage architecture.

Cooperativity, is commonly exploited by enzymes. Ready access to these multicavity cage structures should enable the cooperativity of molecular recognition process within these systems to be studied in more detail potentially shedding new light on how subtle factors can alter non‐covalent interactions. In turn that may lead to new controllable, enzyme‐like multicomponent reactions and catalysis.

Additionally, the ability to organize different guests within segregated compartments of a single discrete metallosupramolecular structure could be exploited in a range of applications, including dual guest (drug) delivery and bio‐mimetic energy transfer processes.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.