5. 5
Abbreviations
Five supramolecularcages(C),fivesupramolecularhelicates(H) andfivecorrespondingligandsystems(L)
which have been synthesised in Lusby group labs are discussed in this report. Those made by other
workersare labelled alphabetically whilstthose made orhypothesized aspart of this writer’sprojectare
labellednumerically.The single exceptiontothis ishelicateH2
whichwassynthesizedbyanothermember
of the group.Fragments(F) have eitherbeenboughtorsynthesized.
FA
- 1-azido-2-(2-methoxyethoxy)ethane
LA
- 1,4-bis(2-(1-(2-(2-methoxyethoxy)ethyl)-1H-1,2,3-triazol-4-yl)pyridin-3-yl)benzene
CA
- [Co4(1,4-bis(2-(1-(2-(2-methoxyethoxy)ethyl)-1H-1,2,3-triazol-4-yl)pyridin-3-yl)benzene)6]12+
HA
- [Co2(1,4-bis(2-(1-(2-(2-methoxyethoxy)ethyl)-1H-1,2,3-triazol-4-yl)pyridin-3-yl)benzene)3]6+
7. 7
C∙12PF6 – supramolecularcage surroundedby12 PF6
-
counterions
Other abbreviations
CAN – (NH4)2Ce(NO3)6 - ceriumammoniumnitrate
n-ESI-MS – nano-electrospray ionization- mass spectroscopy
TBAF – tetrabutylammoniumflouride
1. Introduction
1.1 Overview
Supramolecular chemistry is the logical advancement from our now strong understanding of molecular
covalent bonding. Using intermolecular interactions, researchers have been able to consider discrete
compounds as a fresh set of Lego blocks. As the field’s collective grasp of the way the blocks interact,
cooperate and self-assemble has progressed, the number of new structures created has grown
exponentially.Advancements have allowedforthe synthesisof a variety(fig.I1)1
of large,elaborate and
functional assemblies.Initialapplicable examplesprove promisingbut aswiththe beginningsof covalent
chemistry we trail behind the natural world.1, 2
Biological systems already highlight the potential of
supramolecular structures for applications such as catalysis and photosynthesis.1, 2, 3
The current aim of
the field is to mimic and therebyunderstand these arrangements with the hope of someday surpassing
8. 8
them. Just as covalent chemistry has branched extensively, it is probable that successful utilization of
supramolecularspecieswill soonpermeate manyaspectsof society.
Fig. I1: Possible shapes of 2D and 3D supramolecular assemblies. Reproduced from reference 1.
Fig. I2: Fujita’s [Pd4(4,4’-bipyridine)4(en)4]8+
supramolecular square (1).11
1.1.1 Metallomacrocycles
Very often supramolecular structures incorporate metal centres into these systems.1
This is because we
can comprehensivelypredictthe strengthanddirectionalityof agivencentresbonding. Earlyexamplesof
9. 9
supramolecular chemistryused simple metal geometriesto create 2D architectures. There are a large
numberof macrocyclictriangles4,5
andsquares6,7
reported usingeitherlinearmetalunitsasedgesorlinear
ligands and metal vertices. Stang in particular has reported numerous examples of supramolecular
squareswhichuse cis-protectedPd(II)andPt(II)vertices8,9,10
.Fujitareportedthe firstexample of thistype
in 199011
; as with many examples in the field Fujita’s supramolecular square (1) was self-assembled.
Favourable M-L interactions between the 4,4’-bipyridine ligand and Pd(II) metal centres ((en)Pd(NO)2
precursor) resultedinthe formationof 1inethanol.The formationof 1isthermodynamicallydriven; 4,4’-
bipyridine is a better ligand for Pd than the NO precursor ligand and fully satisfied geometries between
4,4’-bipyridine andcis-protectedPd are onlypossible throughthe formation of the supramolecularsquare
1. The relative weaknessof the M-L bonding(whencomparedwithaverage covalentbondstrength)aids
thisprocess.The bonds weakenough tobreakif formedinan incorrectposition duringthe self-assembly
and then reform in a manner that builds towards the correct overall structure. As a result 1 was self-
assembled in 10 minutes. Whereas covalent synthesis of similarlysized structures would be a lengthy
processif at all feasible.
10. 10
Fig. I3: Fujita’s Pd squares 2 a/b/c in equilibrium with corresponding Pd triangles 2 a/b/c.12
Later Fujita noted that supramolecular assemblies can still form in the absence of fully satisfied
geometry.12
When longer ligands were used by varying the X position with linkers a, b and c the
corresponding squares 2a, 2b and 2c still formed but they were found to be in equilibrium with
corresponding supramolecular triangle (3a, 3b and 3c). It was thought that increasing the length and
therefore the flexibility of the ligand allowed for deviationfrom linearity. This in turn reduces the strain
on the Pd centre on formingthe triangle by retentionof near 90° bond angles. Thisway the equilibrium
was seen to shift towards 3 as the length of X or flexibility was increased on going from 2a to 2b to 2c.
Whilst 2 is thermodynamically favoured, 3 is a smaller more numerous species than 1 and is therefore
entropicallyfavoured. Multiplesupramolecularstructuresof varyingsizesare acommonfeature inthese
types of system. It is the flexibility of the ligand that creates this slight distortion of the geometry rules
and the resultingnewspecies.
11. 11
Fig.I4: Saalfrank’stetrahedral [Mg4(tetraethyl-2,3-dioxobutane-1,1,4,4-tetracarboxylato)6]4-
cage (4).13
The rest of the ligandshave beenomittedforclarity, this is the case for all following tetrahedral cages.
1.1.2 3D coordinationcomplexes
Flexible ligands gave rise to the serendipity that aided syntheses of the first 3D coordination based
supramolecular structures. Saalfrank and co-workers synthesized the first tetrahedral
metallosupramolecular cage 4 from a MgCl2 precursor and tetradentate ligand tetraethyl-2,3-
dioxobutane-1,1,4,4-tetracarboxylate.13
Fig. I5: Saalfrank’s tetrahedral [Fe4(tetraethyl-2,3-dioxobutane-1,1,4,4-tetracarboxylato)6] cage (5).14
The same workerslateruseda similarligandandaFe(II)Cl2 precursortosynthesisecoordinationcomplex
5.14
5 was able to encapsulate small cations (such as NH4
+
, Na+
, K+
and others) within the cavity of the
12. 12
neutral cage. Recognitionandencapsulationof thecationslistedstemsfromthe largesizeandframework
nature of 5. As the fieldcontainsmanyexampleswhichalsofitthisdescription,selective recognitionand
encapsulationare commonfunctionsof macrocyclesandcoordinationcomplexes.Oftenresearchgroups
have reported a new stable supramolecular species, then in the months or years after, publishedthe
functionof the cage as itsactivityis discovered.
1.2 Functional systems: catalysis
13. 13
As seen with Saalfrank’s cage it is the cavity which gives rise to host-guest functionality. When
encapsulatedbythe respective cage small moleculescanexhibitunusual chemistry.The reactivityof the
guest can be increased or decreased depending on the host-guest interactions, this feature has been
utilisedforapplicationsincatalysis.
1.2.1 Catalysisof the Nazarovcyclisation
Fig. I6: Raymond’s tetrahedral gallium cage [Ga4(N,N′-bis(2,3-dihydroxybenzoyl)-1,5-
diaminonaphthalene)6]12-
(6).15
Raymondandco-workerssynthesisedatetrahedral galliumcage 6whichwasable tocatalysethe reaction
of the dienol (7) tothe respective pentadiene (8) (fig.I7).15
Thisreaction,knownasthe Nazarovcyclisation
iscommonlyusedinorganicchemistry.The cage pre-organisedthe flexible substrate(8) e.gthe shape of
the cavity effectively held it in a more reactive conformation. As the rearrangement proceeded to the
intermediate, the molecule was evenbetter held by the cage. The shape of the intermediate was more
stable withinthe cavity,butmore importantlythe intermediatewasadiallyliccarbocation.Thispositively
charged species experienced an electrostatic attraction with the polyanionic cage. Stabilisation of the
intermediate rapidlyacceleratedthe rate of the reactionupto 2.1x106
timesthe uncatalysedreaction.
14. 14
Fig. I7: Nazarov cyclisation catalyzed by 6 and subsequent Diels-Alder reaction to form 9.15
The productof the reaction (8) was boundmore stronglythansubstrate 7 as the flexibilityof the guestis
removedbyformationof the cyclopentane ringsystem.Howevercatalyticturnovercannotoccurwithout
a vacant cavity, binding of the product caused inhibition of the cages function. This is common where
substrate andproduct are similarinstructure.Raymonddealtwiththisbyperformingasecondreaction,
addition of maleimide produced the Diels-Alder adduct of the product (fig. I7). This second product (9)
was now too large to act as a guest, the cavity was thereby vacated and catalytic turnover resumed.
Formationof the Diels-Alderadductcouldhoweverbe viewedasa negative asthe cyclopentadienewas
the desiredproduct,andinhibitionof the catalyticcycle wascausedbytrappingof thisproduct.
15. 15
1.2.2 Fujita’sPdbowl andcage
Fig I8: Fujitas Pd octahedron [Pd6(2,4,6-tris(pyridin-3-yl)-1,3,5-triazine)4]12+
(9) and Pd bowl [Pd6(2,4,6-
tris(pyridin-4-yl)-1,3,5-triazine)4]12+
(10).Reproducedfromref.16.
Fujitaand co-workersmanagedtoresolve asomewhat similarproblemwithadifferentcatalyticsystem.
The groupself-assembledaPd6L4 octahedron(9) aswellasthe Pd6L4 bowlshapedassembly fromaslightly
different ligand.16
The cage was able to bind Diels-Alder reactants 11 (9-hydroxymethylanthracene) and
12 (N-cyclohexylmaleimide) (fig.I9).Inthe absence of cage reactionbetween 11 and 12 has beenwidely
reported to produce the 9,10-bridged Diels-Alder adduct. Whenthe reactiontook place within the cage
the cavity pre-organised the molecules in a manner which changed the conventional selectivity. In this
arrangement12 wasclose inspace tothe terminusof 11,thisleadtobridgingatthe endof 11ratherthan
at the middle. The product of the reaction; the unusual 1,4-adduct (14) has conjugated π ring systems
originating from the starting material (11). 14 experienced evenstronger π-π stacking interactions with
the 3-tpt ligand that makes up the cage structure than 11. Binding of two substrates within 9 causes an
entropic destabilisation of the structure meaning replacement of product 14 in cavity by one of each
reactant(11 and12) wasveryunlikely.Againdue toproductinhibitioncatalyticturnoverwasnotachieved.
16. 16
Fig.I9: 9 catalysedDiels-Alderreaction between 11 and 12 forming unusual 1,4-adduct 14. Reproduced
from ref. 16.
With the bowl (10) this was not the case. A slightlydifferent dienophile substrate was used(15, fig.I10)
and without the pre-organisation afforded as with the cage, the conventional 9,10-adduct (16) was
formedfromreactionwith 11. The rate of this reactionhoweverwasgreatlyincreased,inabsence of 10
the yield of the reaction(fig.I10) was3%, butwith10 near-completeconversion(99% yield)wasachieved
in the same time. As withthe octahedron(9), 11 wasstabilisedasa substrate viaπ-π stacking.However
bridgingof 15 across the centre of the molecule removed linearity fromthe product(16) and diminished
the stacking interaction. This way the affinity between product and bowl was reduced, new substrate
moleculescouldreplace ittherebyachievingcatalyticturnover.
Fig.I10: 10 catalysedDiels-Alderreactionbetween 11 and 15 forming 9,10-adduct 16. Reproduced from
ref. 16.
17. 17
Fig. I11: Catalytic cycle of 9 during the reaction between 17 and 18 to form 19.
Octahedron 9 was foundinsteadtoachieve catalysisof a differentreaction.17
2-naphthaldehyde (17,fig.
I11) wasfoundto bindinthe hydrophobicpocketof 9. 17 couldthen undergonuceophilicadditionfrom
Meldrum’s acid (18, fig. I11) followed by dehydration to form the respective conjugated enone (19, fig.
I11). Stabilisationof the anionicintermediatewasachievedthroughstrongelectrostaticinteractionswith
3 Pd2+
centres around each portal. The Knoevenagel product (19) was found to be too large to fit in the
cavityand was thus dissociated.Thisway,thisworkersobservedhighyieldsforvariousderivativesusing
onlya catalyticamountof 9. Whenbowl 10 was usedinstead,the reaction proceeded witha17% yield as
the lackof a portal surroundedPd2+
centresmeant10 was unable to affordthe same stabilisationof the
intermediate.Thisgoesto showthat the interactionsnecessaryforsuccessful catalysisare veryspecific.
Stabilisationof the intermediate hasbeenseentobe most promising throughelectrostaticorπ-stacking
interactions.The shape of the moleculesmustthenchange sufficientlythatproductinhibitionisavoided.
18. 18
The difficulty of finding substrates which perform this binding-stabilisation-alteration sequence is the
mainreasonexamplesof supramolecularcatalysisare notsonumerous. The incrediblerange of enzymes
which have been tailored to each perform this sequence on a specific molecule is the reason we trail
nature inthis kindof catalysis.
1.2.3 stabilisationof P4
FigI12: Nitschke’stetrahedral Fe cage [Fe4(L)6]4-
(20),tetrahedral P4 (22) and host-guest complex (21).18
Supramolecular cages have also been reported to produce effects opposite to catalysis, where reactive
compoundshave beenindefinitelystabilised.Thiswasthe case whenNitschke andco-workers stabilised
molecules of the extremely air-sensitive and pyrophoric compound P4.18
Tetrahedral cage 20 was self-
assembledandwasabletobind 22withinthe cavity forminghost-guestcomplex21(fig.I12).Despite,the
bondingwithin22beingsignificantlyweakerthanwhenthecompoundisoxidised,themoleculeremained
stable indefinitely.Theportalsof 20were calculatedas1Å inradius,toosmalltoallow22orevendiatomic
oxygenthrough. AlthoughNitschkesuggeststhermal fluctuations create transientwideningof the portals
of 20 thusallowingO2 intothe cavityof the cage.Thisthenrules outprotectionof 22 (aspart of 21) from
oxidationsource asthe mechanismof stabilisation.Insteaditisproposedthatthe transitionstate of any
oxidative addition to the molecule creates a species too large to fit inside the cavity, the resulting high
19. 19
energy barrier produces indefinite stabilization of 22. Despite this, 22 could be easily replaced as guest
within21.Additionof benzenetothe solutiongavetwophases. 22wasthenreplaceddue toitspreference
for the organicphase workingincoordinationwiththe competitionof benzene topervadethe cavity (fig.
I13). Free white phosphorusin the organic phase regained its sensitivity to air and reacted over time to
form phosphoric acid, which then re-dissolved in the aqueous phase due to its polarity. Though the
applications are not immediately obvious the sequence demonstrates an incredible amount of control
overthe reactivityof an extremelysensitivespecies,affordedinpartbysupramolecularcage 20.
Fig. I13: Biphasic displacement of 22 from 21 by benzene and dissolution of free 22 in organic phase,
subsequentreaction of 22to give H3PO4 whichre-dissolvesinaqueousphase. Reproducedfromref.18.
20. 20
1.3 Functional systems: drug delivery
The field’s besteffortshave notyetachieved the aim of metallosupramoleculardrugdelivery. Examples
of successful drug vehicles have been reported in the form of surfactant based nano-particles;
micelles19,20,21
and liposomes22,23,24
. Anticancer drug doxorubicin has showed improved cytotoxivityin a
breast cancer MCF-7 cell assay as part of a micellular structure made up of poly(ethylene glycol)-
poly(aspartate hydrazide) copolymers.19
Whilst the same drug as part of liposomal therapy Myocet has
been approved for treatment of metastatic breast cancer.22
Liposomal vehicles however experience
problemswithinvitrostabilityandare rapidlyclearedby the reticuloendothelial system.25
Coordination
capsulestendtobe more stable thantheirorganiccounterpartsandexhibit the interlinkingpropertiesof
molecular recognition and encapsulation. These qualities suggest cages are well matched to the task of
drug delivery. Furthermore asthe space withinthe supramolecularframeworkhasproperties definedby
the cage, organicapolar ligands oftencreate hydrophobiccavities. Entropicallydrivenrepulsionof water
enablesbindingof hydrophobicmoleculessuch drugswithinthisspace.
1.3.1 Crowley’sPd“lanterns”
Selectivebindingof drugmoleculeshasthe potentialtoovercome many of the problemsassociatedwith
conventional delivery methods. Workers have alreadyreported examples of drug encapsulationsuch as
Crowley’s Pd2L4 “lantern” shaped cage (22, fig. I14) which was able to bind two molecules of the anti-
cancer drug cisplatin.26
21. 21
Fig.I14: Self-assemblyof Crowley’s[Pd2(2,6-bis(pyridin-3-ylethynyl)pyridine)4]4-
cage (22) from precursor
[Pd(CH3CN)4](BF4)2 andIL22
.Structure of anti-cancerdrugcisplatinisalsoshown(23).26
The hostguestinteractionbetween 22and23isstrengthenedby hydrogen-bondinginteractionsbetween
the NH3 groupsof 23 and the pyridyl nitrogenatomswithinthe IL22
struts. Thisinteractioncanbe seenin
the image of the host-guestcomplex(fig.I15) andisanexampleof functionof 22arisingfromaparticular
feature of the ligand IL22
.
Fig. I15: Encapsulation of 23 by 22 forming host-guest complex 24.
22. 22
This system however had problems with water solubility and therefore will not have the necessary
bioavailability to function as a drug delivery vehicle. Theoretically a similar water-soluble host-guest
systemcouldeffectivelyprotectthe drugwithinthe cavityof cage.Therebydefendingitfromdestruction
in the gut by acid or otherresistance mechanisms.Thisprotectionavoidsreductionof the effective dose
notedwhendrugsare deliveredviaconventionalmethods.Converselythe cage can protecthealthycells
from the cytotoxicityof some drugs.26
Deliveryof cisplatinwouldbe aprime example of this,the drug is
a major componentof solutionsadministeredduringchemotherapyasaresultthe harshside-effectsare
well known. Host-drug systems are able to target tumors, another feature that makes them extremely
well suited to the task of drug delivery. Cancerous cells exhibit poor drainage of macromolecules, this
leads to accumulation of large structures such as supramolecular cages within tumors.27
Disassembly of
the host-drug structure could then be triggered using some kind of stimulus and the drug would be
released. With Crowley’s cage competing ligands such as DMAP and Bu4NCl were used to initiate
disassembly,disassemblytriggersrequire considerationasare cage specific.
1.3.2 Therrien’srutheniumprisms
Bruno Therrien and co-workers have self-assembled water-soluble prismatic ruthenium cage 27 from
precursor25 andligandsystem 26(fig.I16).28
The cage displayedthe novel behaviorof rather thanbinding
a drug,27 was able bindascaffoldwithinthe hydrophobiccavitytowhichthe anti-cancerdrugfloxuridine
(fig.I16) wastethered.
Fig. I16: Structure of Floxuridine (28).
23. 23
Fig. I17: Self-assembly of ruthernium prismatic cage from 25 and 26, subsequent encapsulation of
pyrene gave host-guest complex 27. Reproduced from ref. 28.
Molecules floxuridine were pre-tethered to the R position of the functionalized pyrene (fig. I17). 27
showedimprovedsolubilityandcytotoxicityoverfloxuridine alone,thusgivingademonstrationof host-
drug function.Increasedcytotoxicitytoovariancancer cellsisa positive result,althoughasystemwitha
cytotoxic guest and biologically inert vehicle would be preferable. This was not the case with the
rutheniumcage reportedasarene rutheniumcomplexes,whichare similartoeachmetal centre in 27,are
known anti-cancer agents.29
This system shows many of the desirable features of a functional
supramolecular drug deliveryvehicle althoughthe cyctotoxicity of 27 revives the issue of side-effects, it
isalso unclearwhetherthe effectivenessof 27 isdue to the hostor the guest.
Molecular recognition proves a double edged sword in regards to successful drug encapsulation by a
supramolecular species. Unique host-guest interactions are necessaryto ensure selective encapsulation
24. 24
of a givendrug,therefore designing asystemwhichissure tohave thenecessaryinteractionsisnottrivial.
The selectivityof cagesoftenresultsingoodencapsulationof asingle drugratherthanarange of different
molecularstructures. Although, the rate at which new examples are reportedissure toquicken asmore
cages are created.Whilstthe fieldisrapidlyexpanding,virtuallyall have a commonproblemof stability,
the weakbondswhichallowforself-assemblyalsoleadtorelativelyfragile structures.
25. 25
1.4 Lusby group previouswork
Recentworkoutof the Lusbygrouphaslookedtoincorporatethe firstrow transitionmetal-cobaltinself-
assembled supramolecular structure.30
Cobaltis smaller, cheaper and less toxic than the 2nd
and 3rd
row
transitionmetalsgenerallyemployedinmetallosupramolecularassemblies.Otherworkershave generally
looked to use the larger transition metals because of the relative structural stabilisation afforded by
substitutionallynon-labilecentres.31
1.4.1 Self-assemblyfollowedbyoxidation
Whensynthesizingthe tetrahedral supramolecularcagesCA
and CB
(fig.I18), the groupusedCo2+
centres
but were able to stabilise the structures in an alternative manner.30
CA
and CB
were self-assembledfrom
cobalt (II) precursor(CoPr, fig.I18) and linearpolyaromaticligandLA
or LB
(fig.I18). CA/B
was thenslowly
oxidisedusingCAN,whichconvertedall 4centresfromCo2+
to Co3+
.
Fig. I18: Synthesis of supramolecular cobalt cages CA
and CB
.
26. 26
Thisoxidation hada profoundeffectonthe cobaltd-orbital electronsandtherefore stabilityof the cage.
Oxidation converts the metal centres from high spin d7 to low spin d6 (fig. I19). The former has labile
bondingdue tothe highenergyunpairedelectronswhilst the latterisafilledhalf shell arrangementwith
bonding that is substitutionally inert. Octahedral geometry at the metal centres is retained during this
change.This self-assemblyfollowedbyoxidationisa novel wayof achievinga structure that was shown
viascramblingexperimentstobe constitutionallynon-dynamicupto70°C. Mixingof CA
and CB
solutionin
MeCN showednoexchange of ligandbetweenthe structures,despite the similarityof LA
and LB
.
Fig. I19: Molecular orbital diagram for cobalt d-orbitals before (high spin d7, left) and after oxidation
(low spin d6, right).
Whilst many workers have reported formation of a mixture of supramolecular species with the same
components.1,12
The recent research out of the Lusby group demonstrates control over the dynamic
equilibriumbetweenthe cages CA/B
andtheirrespective helicatespecies (HA/B
).
Fig. I20: Triple helicates HA
and HB
.
eg eg
t2g
t2g
27. 27
The group notedthatformationof tetrahedron(CA/B
,fig.I21) inthe self-assemblyreactionwaspromoted
by a concentrated reaction mixture followed by a slow oxidation. Conversely, selective synthesis of HA/B
(fig. I21) could also be achieved using the opposite conditions; a dilute self-assembly reaction mixture
followed by a fast oxidation gave helicates HA
and HB
pure. Fast addition of CAN should effectively trap
whicheverspeciesisinsolution initsgivenstate,byconversionto non-labile Co3+
.The contrastingreaction
conditionsthereforesuggest thatbeforeoxidationthe supramolecularequilibriumlieswell towardsHA/B
.
Conversely, slow-oxidation will give the structure time to rearrange and thereby form the tetrahedron.
Concentration has been shown in previous paper by other worker to promote formation of larger
supramolecular species over smaller structures due to Le Chatelier’s principle. As the concentration is
increasedthe equilibriumshiftstomaintainthe same numberof moleculesinsolution,this systemdoes
so by forming the entropically disfavoured CA/B
species. The opposite is true for dilute conditions during
HA/B
synthesis.
Fig. I21: Selective synthesis of cage (CA/B
) or helicate (HA/B
) species.
28. 28
1.4.2 LA/B
synthesisand counterionmetathesis
Synthesisof LA
andLB
wasachievedbyacoppercatalysedazide-alkyne“click”reaction(fig.I22).Usingthis
reactionthe central unit(F1c
) couldbe readilyconvertedto LA
or LB
dependingonthe azide fragmentused
(FA
/FB
). The PEG group of LA
gave a more water soluble cage (CA
) and the bulky adamantane group of LB
aidedcrystallisation of CB
allowingforx-raycrystallography todetermine CB
structure.
Fig. I22: Schemes for synthesis of LA
or LB
from F1c
and azides FA
or FB
.
The workerslaterfoundhoweverthatcounterionmetathesishadagreatereffectonsolubility thanligand
alteration. Conversion of the mixed counter ion species(MeCN:H2O, 2:1) produced by the self-assembly
to highly apolar PF6 counter ions resulted in a cage (CA
∙12PF6) with good solubilityin organic solvents
(MeCN, diisopropylether). Whilst conversionto all nitrate species CA
∙12NO3 was found to be make CA
water soluble up to 2.5 mM. As many of the bodies uptake mechanismsrely uponaqueousfluids,good
watersolubilityandtherefore bioavailability isessentialtoany potentialfunctionasadrugdeliveryvector.
29. 29
1.4.3 CA
guestbinding
Whilstthe cavities of HA
and HB
were too small to encapsulate molecules,CA
∙12NO3 exhibitedpromising
host-guest chemistry (fig. I24). The cage encapsulated triisopropylsilanol (TIPSOH) in slow-exchange (on
the NMR timescale).Thisstrongbindinginteraction (Ka = 1400 M-1
) wasshownby the presence of 4 sets
of peaks on the NMR spectrum (fig. I23), these corresponded to free CA
∙12NO3 (red peaks), bound
CA
∙12NO3 (blue peaks), free TIPSOH(largergreenpeak) andboundTIPSOH(smallergreenpeaks).
Fig I23: 1
H NMR spectra showing CA
∙12NO3 (a), CA
∙12NO3 with bound TIPSOH (b, slow exchange),
CA
∙12NO3 with bound nitrobenzene (c, fast exchange) and CA
∙12NO3 with bound chromone (d, fast
exchange). Reproduced from ref. 30)
A numberof othersboundwithinthe cavityinfast-exchangeonthe NMRtimescale,the mostinteresting
of which were 2-adamantanone (analogue of antivirals amantadine and rimantadine) and the
anticoagulant coumarin. Fast-exchange guests were confirmed by slight shifts and broadening of both
guest(greenpeaksinb,cand d fig.I23) and CA
∙12NO3 peaks(blue peaksb,c,dfig.I23).
31. 31
1.4.4 CA
stability
CA
∙12NO3 was acidic as a 2.5 mM D2O solution, the pH was measured at 2.5. Stability of CA
∙12NO3 was
observed to be a problem when the workers moved the pH to more basic conditions using sodium
phosphate buffersolutions.Whenthe stabilitywastestedat pH 6 the cage was observedtodecompose
into a cobalt solution and white LA
precipitate over a time period of a few days (fig.I25). D2O dilution
experiments showed conversion between the triazole proton (shown in green, fig. I25) and deuterium,
indicatingthatthe proton isweaklyboundto LA
and is therefore acidic.Neutralityof free LA/B
suggested
that this proton was acidic partly due to the highly charged Co3+
centres of CA
∙12NO3 removing electron
density from the coordinated triazole ring. The workers postulated that loss of the proton caused LA
to
become weaklynegativelycharged.Subsequentdonationof electrondensityontothe coordinatedcobalt
centre was thoughtto cause a trans-effectinteraction,wherethe oppositeCo-LA
bondwaslabilised.Any
biologicallyfunctional speciescannotexhibitthiskindof instabilityasinordertosuccessfullydeliverdrugs
the structure must be stable to a wide pHrange (approx.2-8).
Fig. I25: Possible scheme for buffer mediated decomposition of CA
∙12NO3.
32. 32
1.5 Project aims
The aim of the project was to bypass the apparent instability of CA
∙12NO3 to neutral conditions by
synthesisinganewligandand cage systemdevoidof the acidicproton. Thiscalledfor a rethinkof the LA
designwithremoval of the functionalized triazoleringfromthe designand subsequentsubstitutionwith
an alternate heterocycle. Crucially, one devoid of an acidic proton. However many of the successful
featuresof LA
andLB
couldbe retained,givingthe brief forthe new ligand:
- Syntheticallyfeasible ligandsystem.
- Two N,N 5-memberedchelate ringsatthe endsof the ligandsystem.
- Central phenyl linkerunit.
- Newaromaticheterocycle attriazole position, devoidof acidicproton.
- Group available atendof ligandforpotential exo-functionalisation.
This brief lead to alteration of the LA
design to give the motifs shown in fig. I26. The fragments 2-[6-(5-
bromopyridin-2-yl)-1,2,4,5-tetrazin-3-yl]ethan-1-ol (pyridyl-tetrazine) and 2-(5-bromopyridin-2-yl)-1-
methyl-1,3-benzimidazole (pyridyl-benzimidazole) would theoretically be di-coupled with benzene-1,4-
diboronic acid to give the full ligands (L1
and L2
fig. I26) which could then be assembledwith a cobalt
precursor (using the same protocol for synthesisof CA/B
) into cage systems (C1
and C2
) that are stable at
neutral pH. The R positions show sitesfor potential functionalization. Whilst the structures of LA
and LB
contained PEG and adamantane groups for solubility and to facilitate crystallization respectively,
biological orfluorescentmarkerscouldalsobe added forbiomedical applications.
33. 33
Fig. I26: General structures of fragments used when designing ligands L1
and L2
.
34. 34
2. Experimental
2.1 General information
Reagentswere purchasedfromSigma-Aldrich,Alfa-AesarorVWR. All columnchromatographywas
performedusingstationaryphase GeduranSi60silicagel (particle size40-63) microns.Thin-layer
chromatography(TLC) was ran onsilica60 gel coated(0.2mm thick60F254. Merck,Germany) aluminium
plates,andthenobservedusingUV light(ChromatoVue CabinetmodelCC-10,UplandCA91786, USA).
Solventswere obtaineddryandpure fromsolventpurificationsystemmanufacturedby Innovative
Technology,Newburyport,MA,USA.Solventswere degassedbynitrogenpurgingorbysonication
(ultrasonicbath) undervacuumfollowedbyN2 backfill cyclesonSchlenk apparatus.
All 1
H and DOSYNMR experimentswere recordedonaBrukerAV500 instrumentata constant
temperature (298 K). The data was processed using Topspin 2.1 (Bruker) and MestReNova 6.0.3
(MestreLab Research). Chemical shifts are reported in parts per millionfrom low to high fieldand
are referenced against values for the residual solvent peaks. Coupling constants (J) are reported in
Hz. Standard abbreviations indicating multiplicityare used as follows:m = multiplet,t = triplet, d =
doublet, s = singlet.
36. 36
2.2.2 synthesisof F1b
(a) 2-[6-(5-bromopyridin-2-yl)-1,2,4,5-tetrazin-3-yl]ethan-1-ol,F1b
(i) 5-bromo-2-pyridinecarbonitrile (499mg,2.73 mmol), 3-hydroxyisopropionitrile (1.86mL,
27.3 mmol),hydrazine (3.43mL, 109 mmol) andethanol (5 mL) were addedto a two-necked
flaskandplacedundernitrogen atmosphere.The flaskwasheatedinanoil bathfor 24
hoursat 90°C behindablastshield. ReactionprogresswascheckedusingTLC.
(ii) The reactionmixture waswashedwithDCMand water,the organic phase wasthen
concentrated in vacuo.(ii) The orange protonatedproductwasoxidisedbystirringwith
sodiumnitrite (188mg, 2.73 mmol) inaceticacid (60 mL) for 30 minutesatroom
temperature.The reactionmixture waswashedwithDCMandwater,the organic phase was
concentrated in vacuo and purifiedviaflashcolumnchromatography.The title compound
was affordedasa brightpinkcrystalline solid.Yield=120mg(16%) 1
H NMR (500 MHz,
Chloroform-d)δ9.03 (dd, J = 2.3, 0.6 Hz, 1H, HE), 8.58 (dd, J= 8.4, 0.7 Hz, 1H, HC),8.16 (dd, J
= 8.4, 2.3 Hz, 1H, HD),4.36 (q, J = 5.9 Hz, 2H, HA), 3.72 (t, J = 5.8 Hz,2H, HB).
37. 37
2.2.3 attemptedsynthesisof L1
(b) 1,4-bis(2-[6-(pyridin-2-yl)-1,2,4,5-tetrazin-3-yl]ethan-1-ol)benzene,L1
(notsuccessfullysynthesised)
F1b
(25.2 mg, 88.6 µmol), benzene-1,4-diboronic acid (6.99 mg, 42.1 µmol), Na2CO3 (93.9 mg, 4.8 mmol)
and Pd(PPh3)4 (6.5 mg, 5.6 µmol) were added to a two-necked flasked and placed under nitrogen
atmosphere. THF (9 mL), EtOH (6 mL), and H2O (3 mL) were separatelydegassed before being added to
the two-neckedflask.The lightbrown reactionmixture wasleftheatinginanoil bath for16hoursat 80°C.
No productyieldwasobservedincrude 1
HNMR.
38. 38
2.2.4 attemptedsynthesisof L1
(2)
(c) 1,4-bis(2-[6-(pyridin-2-yl)-1,2,4,5-tetrazin-3-yl]ethan-1-ol)benzene,L1
(notsuccessfullysynthesised)
F1c
(100.1 mg, 0.3 mmol), 3-hydroxyisopropoionitrile (251.4 mg, 3.5 mmol) and hydrazine (0.4 mL, 14.1
mmol) were addedtoa microwave vial andplacedunderanitrogenatmosphere.EtOH(2 mL) was added
and the reaction mixture was heated for 12 hours at 90°C behind a blast shield. No product yield was
observedincrude 1
H NMR.
39. 39
2.3 synthesisof L2
2.3.1 General scheme forsynthesisof L2
Scheme E2: Reaction scheme for synthesis of 2-(5-bromopyridin-2-yl)-1-methyl-1,3-benzimidazole (F2b
)
from 5-bromo-2pyridinecarboxylic acid (F2a
) (route (d)) and 1,4-bis(6-(1-methyl-1,3-
benzimidazolyl)pyridin-3-yl)benzene (L2
) from F2b
(route (e)). Reaction conditions and yields: (d) F2a
, N-
methylphenylenediamine,borane-THF,toluene,72hours,110°C, 41%; (e) F2b
,benzene-1,4-diboronicacid,
Pd(PPh3)4,Na2CO3,THF,EtOH,H2O,72 hours,70°C, 48%.
40. 40
2.3.2 synthesisof F1b
(d) 2-(5-bromopyridin-2-yl)-1-methyl-1,3-benzimidazole,F2b
F2a
(3.27 g, 16.2 mmol) and toluene (81 mL) were added to a two-necked flask in an ice-water bath.
Borane-THF(518 µL, 5.41 mmol) wasaddeddropwise before the mixture stirredfor30 minutesat room
temperature.N-methylphenylenediamine(615µL, 5.41 mmol) wasaddedandthe reactionwasheatedat
110°C for 72 hours. ReactionprogresswascheckedusingTLC.
Reaction mixture was concentrated in vacuo and the product was purified by silica flash column
chromatography (1:1 DCM:hexane). The title compound was afforded as a white solid. Yield = 642 mg
(41%).m.p. X°C. 1
H NMR (500 MHz, Chloroform-d) δ8.75 (d, J = 2.3 Hz, 1H HH), 8.33 (d,J = 8.5 Hz, 1H HF),
7.97 (dd,J = 8.5, 2.4 Hz, 1H HG), 7.82 (d,J = 7.5 Hz, 1H HA), 7.44 (d,J = 7.6 Hz, 1H HD), 7.36 (td,J = 7.6, 1.3
Hz, 1H HC), 7.32 (td,J = 7.6, 1.3 Hz, 1H HB), 4.26 (s, 3H HE). HR-ESI-MS(m/z):288.014 (predicted[M+H]+
=
288.013) , 309.996 (predicted[M+Na]+
= 309.995).
41. 41
2.3.3 synthesisof L2
(e) 1,4-bis(6-(1-methyl-1,3-benzimidazolyl)pyridin-3-yl)benzene,L2
F2b
(642 mg,2.23 mmol),benzene-1,4-diboronicacid(175mg,1.06 mmol), Pd(PPh3)4 (61.3mg,53.1 µmol)
and Na2CO3 (562 mg,5.31 mmol) were addedto a two-neckedflask.THF (20 mL), H2O (10 mL) and EtOH
(16 mL) were individually degassed by purging and added to the reaction flask. The mixture was then
heatedat 80°C for72 hours. ReactionprogresswascheckedusingTLC.
The reaction formed a white precipitate which was filtered in vacuo and washed withreaction solvents.
The crude product was purified by silica flash column chromatography (2% MeOH in DCM). The title
compound wasaffordedasawhite crystallinepowder.Yield=255 mg (48%). m.p.X°C.1
H NMR (500 MHz,
Chloroform-d) δ9.02 (dd,J = 2.3, 0.7 Hz, 2H HH), 8.54 (dd,J = 8.3, 0.7 Hz, 2H HF), 8.13 (dd,J = 8.3, 2.4 Hz,
2H HG), 7.86 (dd,J = 7.4, 1.4 Hz, 2H HA), 7.84 (s, 4H HI), 7.48 (dd,2H HD), 7.38 (td,J = 8.0, 7.6, 1.4 Hz, 2H
HC),7.34 (td, J= 7.6, 1.4 Hz,2H HB),4.36 (s,6H HE).HR-ESI-MS(m/z):493.211 (predicted[M+H]+
=493.214) ,
515.192 (predicted[M+Na]+
= 515.195).
42. 42
2.4 synthesis of C2
2.4.1 General scheme for synthesis of C2
Scheme 2: General scheme for synthesis of tetrahedra and helicate; C2
∙12NO3 from L2
(route (f)(i)),
C2
∙12PF6 from L2
(route (f)(ii)) and helicate species H2
∙6PF6 from L2
(route (g)). Reaction conditions and
yields: (f)(i) L2
(concentrated), Co(ClO4)2∙6H2O (concentrated), MeCN, CAN dropwise, AgNO3, 24 hours,
50°C, 38%; (f)(ii) L2
(concentrated), Co(ClO4)2∙6H2O (concentrated), MeCN, CAN dropwise, NH4PF6, 24
hours,50°C; (g) L2
(dilute),Co(ClO4)2∙6H2O(dilute), MeCN, fastadditionCAN,NH4PF6,24hours, 50°C,97%.
43. 43
2.4.2 Synthesisof C2
∙12NO3
(f)(i) [Co4(L6
2
)]12NO3,C2
∙12NO3
L2
(103 mg, 209 µmol) and Co(ClO4)2∙6H2O (51.6 mg, 139 µmol) were added to a 250 mL Schenk flask.
MeCN (16 mL) wasseparatelydegassedbeforebeingadded.The reactionmixturewaslefttoheatat50 °C
for24 hours.The reactionwaslefttocool toroomtemperature, CAN (84.3mg,153 µmol) inMeCN (6mL)
was then added drop-wise via syringe pump over 2 hours. The resulting orange precipitate was filtered
onto celite andelutedwith2:1 MeCN:H2O, CG-400 anion exchange resin(1.22 g, mol) was addedto the
resultingsolution.Thismixture wasleftstirringgentlyovernight.The resinwasthenremovedbyvacuum
filtrationontocelite,the solventmixturewasremovedbyconcentrationin vacuo andreplacedwithsolely
H2O (16 mL). To this solution AgNO3 (72.4 mg, 419 µmol) was added in an aluminiumfoil coveredflask,
this was left for 3 hours before being filtered onto celite to remove the silver precipitate present.
Successive cyclesof centrifugingfollowedby syringefilteringthenensuredcompleteeliminationof silver
precipitate.Theresultingsolutionwasthenconcentrated in vacuotogivethe titlecompoundas anorange
solid. Yield= 53.4 mg (39%). m.p.X°C. 1
H NMR (500 MHz, DeuteriumOxide) δ8.97 (d, J = 8.7 Hz, 1H HF),
8.82 (d, J = 8.6 Hz, 1H HG),7.98 (d,J = 8.7 Hz, 1H HD), 7.76 (s,1H HH), 7.63 (t,J = 7.9 Hz, 1H HC),7.38 (s,2H
HI), 7.12 (t, J = 8.0 Hz, 1H HB), 5.52 (d,J = 8.6 Hz,1H HA), 4.62 (s,3H HE).
44. 44
2.4.3 Synthesisof C2
∙12PF6
(f)(ii) [Co4(L6
2
)]12PF6,C2
∙12PF6
L2
(30.1 mg, 60.9 µmol) and Co(ClO4)2∙6H2O (14.8 mg, 40.6 µmol) were added to a 250 mL Schenk flask.
MeCN (5mL) wasseparatelydegassedbefore beingadded.The reactionmixture waslefttoheat at 50°C
for 24 hours. The reaction was left to cool to room temperature, CAN (33.8 mg, 44.6 µmol) in MeCN (2
mL) wasthenaddeddrop-wise viasyringepumpover2hours.
The resulting orange precipitate was filtered onto celite and eluted with 2:1 MeCN:H2O. Excess NH4PF6
(799 mg,4.90 mmol) wasthenaddedtogive aredprecipitatewhichwasremovedbyfiltrationontocelite
followedbyelutionwithMeCN.Thissolutionwasthenconcentrated in vacuo togive the title compound
as a brightred crystalline solid. 1
HNMR(500 MHz, Acetonitrile-d3) δ 8.86 (d, J = 8.7 Hz, 1H HF), 8.79 (dd,J
= 8.6, 1.8 Hz, 1H, HG), 7.93 (d, J = 8.6 Hz, 1H, HD), 7.59 (t, J = 8.2 Hz, 1H, HC), 7.56 (d, J = 1.6 Hz, 1H, HH),
7.46 (s, 2H, HI), 7.12 (t, J = 7.9 Hz, 1H, HB),5.33 (d,J = 8.6 Hz, 1H, HA), 4.52 (s,3H, HE).
45. 45
2.5 Experiments with C2∙12NO3/C2∙12PF6
2.5.1 Mass Spectrometry:
Mass spectrometry (n-ESI-MS) of C2
∙12NO3 was carried out using Waters nanoMate and SYNAPT G2
instruments. A spectrumwasproducedusinga10 µM cage solution.
2.5.2 X-rayCrystallography:
X-raycrystallographyqualitycrystalsof C2
∙12PF6 weregrownby slowdiffusionof diisopropyl etherintoan
acetonitrile cage solution. This produced a sample of large dark red plate shaped crystals of C2
∙12PF6.
These crystalswere submittedtothe Universityof Edinburgh crystal structure service.
2.5.3 ScramblingExperiments:
250µL of C2
∙12NO3 (2.5mM) and 250 µL CA
∙12NO3 (2.5 mM) were added to an NMR tube at room
temperature. The 1
H NMR spectrum of this mixture was taken after 25 minutes and then again after 48
hours.The sample wasthenheatedat 50°C then 60°C and finally70°Cover separate 24 hour periods, 1
H
NMR spectra were takenatregularintervals.
2.5.4 Buffersolutionexperiments:
The pH of a 2.5mM D2O solution of the C2
∙12NO3 was tested using pH paper. Stability tests were
performed by mixing with a NaxHxPO4 buffer solution. Buffer was added 1:1 (250 µL:250 µL) with the
2.5mM C2
∙12NO3 solution.1
H NMR were then takenat regular intervals.A 1:1 C2
∙12NO3 (2.5 mM to D2O
sample (250 µL:250 µL) was prepared andsubmitted for1
HNMR spectroscopy.
2.5.5 GuestBindingExperiments:
0.5mL of 2.5mM C2
∙12NO3 solutionswereaddedtoseparate NMRtubes,1molarequivalentof guestwas
thenaddedto this.The tube wasshakenvigorouslyandthensubmittedfor1
HNMR spectroscopy.
47. 47
Fig.R2: DOSY NMR (500 MHz, D2O) spectrumforC2
∙12NO3.
3.2 Mass Spectrometry
Fig. R3: Mass spectrum of C2
∙12NO3, peaks corresponding to cage are labelled according to fig R4.
P2
P1
P4
P3
P6
P7
P5
48. 48
Charge Mass Predictedmass Seeninspectrum
0 3932.83 - N
1+ 3870.84 3870.84 N
2+ 3808.85 1904.43 N
3+ 3746.87 1248.96 Y
4+ 3684.88 921.22 Y
5+ 3622.89 724.57 Y
6+ 3560.90 593.48 Y
7+ 3498.91 499.84 Y
8+ 3436.92 429.62 Y
9+ 3374.94 374.99 Y
10+ 3312.95 331.29 N
11+ 3250.96 295.54 N
12+ 3188.97 265.75 N
Table R1: Charge states of C2
∙12NO3with corresponding masses.
50. 50
Fig. R4: expansions of peaks corresponding to charge states labelled, red peaks modelled by isotopic
substitution have been superimposed, discrepancy due to calibration of n-ESI-MS.
3.3 Scrambling Experiments
Fig. R5: 1
H NMR spectra (500MHz, D2O) stack of CA
∙12NO3, C2
∙12NO3 and CA
∙12NO3+ C2
∙12NO3 mixture
after varying time and temperature.
CA
∙12NO3
C2
∙12NO3
C2
∙12NO3+ CA
∙12NO3 after25 mins
Plus48 hours
Plus24 hoursat 50°C
Plus24 hoursat 60°C
51. 51
3.4 Buffersolutionexperiments
The pH of a 2.5mM D2O solution of the C2
∙12NO3 was tested usingpH paper. The colour suggested that
the solutionwas aboutpH 6.
Fig.R6: 1
H NMR spectra (500MHz, D2O) stack of C2
∙12NO3 and NaxHxPO4 + C2
∙12NO3 mixture aftervarying
amounts of time.
C2
∙12NO3
NaxHxPO4 +C2
∙12NO3 after 25 minutes
Plus24 hours
Plus72 hours
Plus144 hours
52. 52
3.5 X-ray Crystallography
Crystal Data (returned from service): C250H231Co4F72N65P12, Mr = 6121.34, monoclinic, P21/c (No. 14), a =
18.6723(3) Å, b = 38.7772(8) Å, c = 38.7947(6) Å, = 92.0840(16)°
, = = 90°
, V = 28071.1(9) Å3
,T = 120.0
K, Z = 4, Z' ) = 3.447, 116463 reflections measured, 15031 unique (Rint = 0.0805) which were
usedinall calculations.The final wR2 was0.4577 (all data) and R1 was 0.1725 (I > 2(I)).
Fig. R7: Structure of C2
∙12PF6 solved by X-ray crystallography.
54. 54
Fig. R9: Partial 1
H NMR (500MHz, D2O) spectra stack of host-guest species (red), G1 (green) and
C2
∙12NO3 (blue). D2O solvent peak and portion of aliphatic region omitted for clarity.
Chromone (G2)
Fig. R10: Partial 1
H NMR (500MHz, D2O) spectra stack of G2 (red), C2
∙12NO3 (green) and host-guest
species (blue). Aliphatic region omitted for clarity.
2-adamantanone (G3)
55. 55
Fig R11: Partial 1
H NMR (500MHz, D2O) spectra stack of G3 (red), C2
∙12NO3 (green) and host-guest
species (blue). Portion of aliphatic region and D2O solvent peak omitted for clarity.
Nitrobenzene (G4)
Fig. R12: Partial 1
H NMR (500MHz, D2O) spectra stack of G4 (red), C2
∙12NO3 (green) and host-guest
species (blue). Aliphatic region omitted for clarity.
Camphor(G5)
56. 56
Fig R13: Partial 1
H NMR (500MHz, D2O) spectra stack of G5 (red), C2
∙12NO3 (green) and host-guest
species (blue). D2O solvent peak omitted for clarity.
4. Discussion
4.1 Synthesisof L1
design
4.1.1 F1b
synthesis
The two synthetic routes to obtain L1
proved unsuccessful. The first route (a→b scheme E1) involved a
tetrazine formationreaction which gave F1b
from F1a
and Hi
PN in a small yield (14%) as was a statistical
distributionof products.Homo-couplingof F1a
andHi
PN likelygaveby-products3,6- BPa
and BPb
(fig.D1).
Fig. D1: Proposed products of tetrazine formation reaction.
4.1.2 L1
SynthesisfromF1b
The nextstepused a Suzuki coupling reactiontotryform L1
fromF1b
(fig.D2, full scheme 2.2.3).
57. 57
Fig. D2: Attempted L1
synthesis from F1b
.
Thisreactionwas triedusing Na2CO3 andTBAF as strongerand weakersourcesof base.Inboth casesthe
bright pink colour of the reaction mixture (caused by F1b
) was immediately changed to a brown colour.
Thiswasindicative of decompositionof F1b
,andwaslikelythe reasonforfailureof the reaction.Giventhe
ease of formationof the tetrazine ring,the conjugatedsystemcouldwellbe vulnerabletoattackfrom the
twosourcesof OH-
.The twocarbonatomswithin thissixmemberedringare likelytobe electrondeficient
through bondingwithtwoelectronegative nitrogenatoms,thiscouldleave themsusceptible to OH-
in a
mechanismthatwould breakthe ringsystem.
Fig. D3: attempted synthesis of L2
from Hi
PN and F1c
.
4.1.3 L1
synthesisfromF1c
The secondroute (fig.D3, full scheme 2.2.4) usedF1c
(previouslysynthesizedbyanothermemberof the
group)34
. It was thought that a reaction to that in fig. D1 would form tetrazine rings on each end of F1c
giving L1
. Instead a yellow precipitate was formed which was shown by 1
H NMR not to be the desired
product.Aswithfig. D1,carbonitrilegroupsare notspecificinformingtetrazinerings.Therefore itislikely
58. 58
that some polymerisationof F1c
hasoccurred (fig.D4).Thislikelypolymerizationcouldexplainthe failure
of the reaction.The syntheticinaccessibilityof the L1
designledtoeffortsbeingfocusedon L2
.
Fig. D4: Polymerization of 1,4-bis(6-ethynylpyridin-3yl)benzene.
4.2 synthesisof L2
design
4.2.1 F2b
andL2
synthesis
Duringworkon the seconddesign,F2b
andL2
were synthesizedcleanlyandinreasonable yields. The
reactionmechanismforsynthesisof F2b
isshowninfig.D5. The Suzuki couplingof F2b
tobenzene-1,4-
diboronicacidwasunaffectedbythe problemsnotedwith F1b
.Thissupportsthe reasoningthatthe
tetrazine ringsysteminF1b
isunstable to sourcesof OH-
.
Fig. D5: Reaction mechanism for synthesis of F2b
from N-methylphenylenediamine (29) and F2a
.
59. 59
As withCA/B
and HA/B
, an equilibriumbetweensupramolecular species ispresentinsolutionafterself-
assembly.Thiswasproventobe the case here as C2
was synthesizedusingaconcentratedreaction
mixture followedbyaslowoxidation.The helicate (H2
) correspondingto C2
was selectivelysynthesized
fromL2
byanothermemberof the group.34
The synthesis of H2
appliesthe same reaction conditionsas
the HA/B
syntheses. The conditionsforsynthesisof H2
and C2
are summarizedinscheme D6.Fast
oxidationisthoughttofreeze the equilibrium, the factthatthisgives H2
suggeststhatafterself-
assemblythe solutionismostlyH2
.Toform tetrahedra,single centresneedtobe oxidizedwhichare
thenable to rearrange intoa tetrahedral structures.Le Chatelier’sprinciple explainsthe effectof
concentration,asthe numberof moleculesinsolutionisincreasedthe systemseekstocorrectthis
change by formingthe specieswiththe lowerentropy(lessmolecules) e.g.the cage andvice versa.This
effectisthoughttobe negligiblethoughasthe equilibriumbeforeoxidationliessofarinfavourof the
helicate. The twocanbe distinguishedby characterisationusingNMRexperiments. MainlyDOSY,which
differentiatesbetweensize.
Fig. D6: Schemes for selective synthesis of supramolecular tetrahedron C2
or helicate H2
.
60. 60
4.3 Evidence for C2
4.3.1 1
H NMR experiments
Confirmation and characterisation of the C2
structure has been completed using several different
experimental techniques. 1
HNMR spectra(fig.D7) were usedto validate synthesisof F2b
andL2
.
Fig D7: 1
H NMR spectra of F2b
(bottom) and L2
(middle) and C2
∙12NO3 (top).
Shifting of the assigned peaksof L2
after self-assembly was indicative of a new supramolecular species,
although was not determinant of whether the species was H2
or C2
. The set of 7 aromatic peaks and 1
61. 61
aliphaticpeakwere seentoshifteitherupfieldordownfield,byvaryingdegrees.Particularlystrongshifts
upfield were observed for protons close to the nitrogen donor sites (HA and HG) due to withdrawal of
electrondensitybythe chargedcobaltcentresuponcoordination.
4.3.2 DOSY NMR Experiment
The new supramolecular species was confirmed as C2
rather than H2
by diffusion-ordered NMR
spectroscopy (DOSY) (fig. D8). The spectrum indicated presence of a single species with one diffusion
constantshownby the linearplotof couplings.
62. 62
Fig.D8: DOSY NMR (500 MHz, D2O) spectrumof C2
∙12NO3
The correspondingdiffusionconstantof 2.1x105
can be usedtocalculate the radiusof thisspeciesviathe
Einstein-Stokesrelationship(Eq. D1).Substitutingvaluesgivesaradiusvalue of 12.07Å.
α = kbT / 6πƞD
Eq. D1: Einstein-Stokes relationship between radius (α), diffusion constant (D), temperature (T) and
viscosity(ƞ).
Theoretical modellingof the cage usingSpartan software gave the structure shownin fig.D9. Takingthe
length of one ligand (which is an approximation of diameter) gave a value of 25.03 Å and therefore a
radius of 12.51 Å. This model agrees well withthe calculated hydrodynamic radius despite the cage not
beingspherical,the discrepancyof 0.44Å canbe easilyexplainedby the approximationusedand tumbling
of the cage in solution.Onlywhenleadingwiththe ligandedge will the tetrahedronhave the maximum
radius expected, other topologies will give an averaged and reducedvalue. Deviation from linearity due
to the slightflexibilityof the ligandalsocontributestoreduce the radius.
Fig. D9: Theoretical modelling of C2
using Spartan software.
25.03 Å
63. 63
4.3.3 Mass spectrometry
The mass of C2
∙12NO3 was determined using mass spectrometry. The masses of likely charge states,
producedbylossof subsequentcounterions, were calculatedbeforehandandcanbe seenintable R1. A
sample of 10 µM C2
∙12NO3 solution in D2O was submitted giving the spectrum seen in fig. D10.
Fig. D10: Mass spectrum of C2
∙12NO3.
64. 64
7 Peaks from the spectrum matched the predicted masses, these peaks have been expanded and
compared with peaks calculated using isotopic distribution modelling software (fig. R4, 3.2). The
expansionsmatch reasonablywell with the isotopic distribution peaks for each respective charge state.
The slightdiscrepancypresentisdue tocalibrationof the n-ESI-MSinstrument.Throughconfirmationof
the predictedcharge states,the non-chargedcage C2
∙12NO3 was showntobe 3932.83 Da.
4.3.4 X-raycrystallography
The connectivity of the components of C2
∙12NO3 was confirmed by x-ray crystallography. A service was
usedforthisdetermination,yieldingthe structure showninfig. D11.Growingcrystalsof qualitysufficient
for x-ray crystallography was a problem with CA
∙12NO3. Crystals of C2
∙12PF6 were used for this
investigation rather than C2
∙12NO3, as bulkier PF6 counter ions were more likelyto produce high quality
crystals.The bettersolubilityof C2
∙12PF6 thanC2
∙12NO3 in organicsolventsgave awiderrange of options
forrecrystallisation. The crystals whichwere grown byanothermemberof the group viaslow diffusionof
diisopropyl etherintoMeCN were darkred,plate shapedandlookedof good quality.
Fig. D11: Structure of C2
∙12PF6 solved by X-ray crystallography.
65. 65
4.3.5 Comparisonwith Spartanmodel
Throughthese experimentsthe size,mass,shape,and connectivity hasbeenconfirmed andshown tobe
in line with the modelled structure. Establishing within reasonable doubt the formation of a new
supramolecularspecies. The Spartanmodelof C2
whichhasminimisedstericinteractions,agreesverywell
withthe experimental evidence. The angle createdbycoordinationthroughpyridine andimidazolerings
is a good descriptor of the level of strain at the centres. Averaging all these bond angles for the real
structure (fromx-raycrystallography) givesanaverage coordinationangle(ACA)of 81.80°. Thiswas 0.97°
closerto the ideal 90° than the model (predictedACA 80.83°),a difference thatshowsthe model andthe
C2
structure are verysimilar.Electrostatics,betweenchargedmetalcentresandthe ligands,whichare not
taken into account by the model wouldexplain this small difference. Shorteningof the Co-L2
bonds due
to electrostaticinteractionswouldincreasethe coordinationangle.
Fig D12: Image of modelled cage showing coordination angles at one metal centre.
4.4 C2
stabilityexperiments
4.4.1 Scramblingexperiments
66. 66
Complete characterisationof system wasfollowedby testingof itsstability.Inthe Lusby group’sprevious
workthiswasdone byexaminingthe stabilityinscramblingandbuffersolutionexperiments. The stackof
1
H NMRspectraobtainedfrommixingof twocage solutions (scramblingexperiment)(fig. D13) shows very
little change in thepeaks(integralsorshifts) correspondingto C2
∙12NO3 orCA
∙12NO3 afterheatingat50 °C
and 60 °C.
Fig. D13: 1
H NMR spectra (500MHz, D2O) stack of CA
∙12NO3, C2
∙12NO3 and CA
∙12NO3+ C2
∙12NO3 mixture
after varying time and temperature.
In aqueoussolutiondecompositionintocomponentsis unlikelydue tothe hydrophobicnature of L2
and
LA
. Lack of change in these peaks shows that there is no ligandexchange between species (at 50 °C and
60 °C).The experimentisareflectionof themetal-ligandbonding,thoughisdependentonbothstructures.
In previous work CA
∙12PF6 exhibited ligand exchange with CB
∙12PF6 in acetonitrile but only at 70 °C. This
experiment highlighted that cooperative M-L bonding played a large part is stabilising CA
and CB
.30
The
similarityof LA
andLB
meansligandexchangeisfeasible andwouldbeexpectedtobe seenbythepresence
of newpeaks inthe 1
H NMR spectrum.ThoughL2
has a differentligandstructure,some exchange is still
67. 67
expected.The lackof this suggeststhatall 3structures (C2
,CA
andCB
) have similarconstitutionaldynamics
due to the cooperativityof the components.A recentpaperbyMike Ward and co-workerssuggeststhat
supramolecularassemblyisdrivenin partbythe hydrophobicityof the ligandcomponent.33
Aspartof the
cage,half the surface areaof L2
facesintowardsthe hydrophobiccavitytherebyachievingaloweroverall
amountof interactionwith water.Withthe scramblingexperimentperformedthe hydrophobicityof L2
is
such that this effect may play a part, contributing towards the relative stabilisation afforded by
cooperative structures as opposed to separate components. Future work monitoring the rate of self-
assemblyinwatervsacetonitrilewouldservetoconfirmordenythis.
4.4.2 Buffersolutionexperiments
CA
∙12NO3 has a pH of 2.5. Through D2O dilution experiments this acidity was attributed to the triazole
protonof LA
. The mainaimof thisproject was to bypassthisaciditywitha differentligandstructure.The
pH of C2
∙12NO3 wasshownto be about6 usingpH paper. A bufferexperimentwas conductedtotestthe
stabilityof the cage underconstantpH6conditions.Thisexperimentproduced thespectrastack(fig.D14)
from which C2
∙12NO3 can be seen to decompose over the course of roughly 9 days. The compound is
unstable to the buffer conditions (checked using a control experiment) and decomposes intothe ligand
as a white precipitate andacobaltsolution.
68. 68
Fig. D14: Scrambling experiment showing decomposition of C2
∙12NO3 over approx. 12 days.
The decomposition is roughlya factor of 10X slower than that of CA
∙12NO3. Puzzlingly, eventhough it is
devoid of the acidic proton C2
∙12NO3 is still unstable to neutral pH conditions. This evidence therefore
suggeststhatthe triazole protonisnotinvolvedinthe mechanismof decomposition. Instead anothersite
inthe structure mustbe interactwiththe buffersolution. A brandnew (butstill unpublishedatthe time
of writing) cage synthesised by workers in the Lusby group has again a slightly differing structure.35
CC
∙12NO3 has the same cobalt tetrahedral structure (fig.D15) bututilisesligand (LC
) ratherthanL1
or LA
.
Fig. D15: Structure of LC
and CC
.
69. 69
Thougha simplerliganddesign,itappearsthe cage isstable toa pH 8 sodiumphosphate buffersolution.
The major difference between C2
, CB
and CC
is that in the latter the ligands coordinate through two 6-
membered rings. In the former two, coordination is through 5-membered (imidazole/triazole) and a 6-
membered (pyridine) rings. From the refined crystal structure of C2
∙12PF6 (fig. R7, section 3.5) the
geometry atthe 4 metal centres isconsidered strained.Coordination througha5-memberedratherthan
6-memberedringleadstobondanglesthatare far off the ideal 90° foroctahedral geometry.
The average coordination angle was 81.38° for CB
∙12PF6 and 81.80° for C2
∙12PF6 (example centre shown
infig.D16). Thissmall difference of 0.42°couldexplainthe ten-folddifferenceinreactivitybetweenthem
althougha largerdifference isexpected.
Fig. D16: Cobalt centre taken from x-ray crystallography structure of C2
∙12PF6.
The ACA of centres in C2
∙12PF6 vary somewhat with the example given being one of the more strained
centres, although the differencesbetweencentresin CB
∙12PF6 are much greater. On closerinspectionof
the crystal structure forCB
∙12PF6 itcanbe seenthereare twotypesof centre withdifferentlevelsof strain.
70. 70
The first(type A) shown infigD17 hasan ACA of 81.94 ° whichisactuallycloserto the ideal 90 ° thanthe
ACA for C2
∙12PF6.
Fig. D17: Type A cobalt centre from x-ray crystallography structure of CB
∙12PF6.
The second type of centre in CB
∙12PF6 however is a single highly strained cobalt atom with an ACA of
79.56 ° (figD18). With the bondanglesbeing10.44 ° away from the ideal thissite wouldbe energetically
unfavourable.Thoughthisresultisfor CB
∙12PF6 and the buffersolutionexperiments were conductedon
CA
∙12NO3, comparisons can still be drawn as counter ion was shown not to perturb the cage structure.
The difference between LA
and LB
as part of CA/B
is an exterior functionalization (using a 2-PEG or
adamantane group) e.gone that probablydoesnotaffectthe coordinationangle.
71. 71
Fig. D18: Type B centre from x-ray crystallography structure of CB
∙12PF6.
Though CC
has not yet been submitted for x-ray crystallography due to crystallisation issues it is a
reasonable prediction thatthe ACA willbe 3-9° closerto the ideal 90 ° than for C2
∙12PF6.Confirmationof
this through x-ray crystallography could corroborate that there is a linear trend between ACA (as a
measure of geometry strain) and time for decompositionin a pH 6 sodium phosphate buffer solution
(measure of stability). The proposedjustificationforthisisthat more strainedgeometryleadsto poorer
overlap between the donating nitrogen orbital and the cobalt d-orbitals. This in turn could lead to the
highly charged cobalt metal centres being more electronically unsatisfied than the more
thermodynamicallysatisfiedcounterparts(e.g CC
). A potentialmechanismcouldpossibly be attackatthe
highly strainedand highlycharged centre by a nucleophile in solution such as OH-
which is formedfrom
phosphate in the equilibriumshowninEq.D2.Howeverfurtherexperimentsare necessarytoinvestigate
this.
HPO4
2−
+ H2O H2PO4
−
+ OH
Eq. D2: HPO4
2-
ionsandwaterproduce OH-
ionsand H2PO4
−
ionsinaqueousbuffersolution.
72. 72
In termsof an application indrugdelivery, instabilitytoneutral condition meansC2
∙12NO3 doesnotmeet
the criteriafor use withinbiological systems.Future workersare therefore more likelytofindsuccessin
pursuitof biologicallyfunctional systemsutilisingcageswithlessstrainedmetalgeometries.
4.5 Host-guestchemistry
4.5.1 Catalysis
In terms of catalysis, the instability of C2
∙12NO3 to OH-
limits its applications. Many base catalysed
reactions now become inaccessible as potentially cage catalysedsyntheses. However the huge range of
organic reactions proves an advantage and as described in the introduction,cages with these shapes of
cavity have shown promising catalytic activity. Therefore testing of the catalytic activityshouldproceed
usingsubstrateswithfavourable affinitiesforthe cage. Reactionswithanionicintermediateswhichmay
experience aColumbiceffect(withcationiccage),aswell asthose whichare a good fitfor the cavity(can
be constrictivelybound) shouldbe considered.
Fig. D19: Guests which displayed encapsulation by C2
∙12NO3.
73. 73
4.5.2 Slow-exchange guests:TIPSOH
Some initial guestbindingexperimentshave beenperformedwith C2
∙12NO3.The guestsshowninfig.D19
were chosenaccordingto Rebek’ssuggestion thatmoleculeswhich occupy50-55% of the volume of the
cavity tend to bind well. Five of the molecules which behaved as guest for CA
∙12NO3 displayed signs of
encapsulation, these were the only 5 tested with C2
∙12NO3 due to time constraints. As with CA
∙12NO3,
TIPSOHwasseen tobindinslow-exchangeonthe NMRtimescale.The slightlyshiftedcage peaks (fig.D20)
whichare nowina newenvironmentduetothe guestpresentcanbe seen(redarrows) alongside a small
amountof free cage.Boundguestisalsopresent(yellowarrows) asisthe freeguestinlargeexcess(green
arrows). Sucha large shiftupfieldisobservedforthe guestasit iswell shieldedbythe cage.
Fig. D20: Partial 1
H NMR (500MHz, D2O) spectra stack of host-guest species, TIPSOH and C2
∙12NO3.
As there are two environments for each species, the spectrum is evidence for a host-guest binding
interaction. From the molar ratios of these peaks at equilibrium the association constant of TIPSOH
bindingwithinthe cage wascalculatedas280 M-1
viaeq.D3.
Ka = (1-χeq)[H0] / (χeq)[H0][Geq]
Eq. D3: Relationshipbetween associationconstant (Ka), % unbound cage (χeq), cage concentration ([H0])
and guest concentration ([Geq]).
74. 74
This binding occurs despite TIPSOH being 220 Å3
in volume and therefore much larger than the 55% of
cavity suggested. Spartan modelling of the encapsulation of TIPSOH by C2
∙12NO3 followed by energy
minimisation of steric interactions gives the image shown in fig. D21. It was found that after energy
minimisation the propyl groups of the guestaligned themselves with the portalsof C2
∙12NO3, this likely
necessaryforthe cavityto accommodate the size of TIPSOH.
Fig. D21: Spartan model showing host-guest complex C2
∙12NO3 TIPSOH.
The TIPSOH Ka forCA
∙12NO3 wasreported as1400 M-1
and veryrecentexperimentsinthe Lusbygrouplab
gave a value of 2800 M-1
for the CC
∙12NO3 cage. Such stark differences in this value between the three
cages was not expected. It is thought that strong encapsulation of TIPSOH occurs through constrictive
binding. With this view,the very slight differences in cavity size betweenthe cages are not sufficient to
explainthe orderof magnitude difference inassociationconstant. Furtherexperimentsare necessaryto
investigatethe cause of thisdiscrepancy,althoughonepossiblecause couldbe the relativeflexibilityof L2
vs LC
.
75. 75
4.5.3 Fast-exchange guests
For the rest of the guestsa differentmechanismof bindingwasthoughtto occur. Whilstshape and size
are still important, molecules with electron rich functional groups were observed to bind well within
CA
∙12NO3. Withthe 4 guests:chromone,2-adamantanone,nitrobenzene andcamphorthiswasobserved
to alsobe the case forC2
∙12NO3. These guestsboundinfast-exchange(onthe NMRtimescale),generally
a weakerinteractionthanslowexchange. Asthe host-guestNMRspectrum(fig.D22for chromone as an
example) shows a time-average of the fast equilibrium between bound and unbound cage species, an
NMR titration experiment is necessary to obtain an association constant. This process involves tracking
host-guestpeaks(via1
HNMR spectroscopy) asthe concentrationof guestisslowlyincreasedfrom0to 1
equivalent(of host).Thisexperimentwasnotperformeddue totime constraints.
Fig. D22: Partial 1
H NMR (500MHz, D2O) spectra stack of chromone, C2
∙12NO3 and host-guest species.
4.5.4 Fast-exchange guests:chromone
Spartan modelling of chromone encapsulated within C2
∙12NO3 gave the energy minimised (steric
interaction minimised only) host-guest structure shown in fig. D23. This suggests the guest molecule
occupiesthe space as shown. There were indicationswith CA
∙12NO3 that hydrogenbondingbetween the
pyridine proton facinginto the cavity (HH spectrumI) and an electron rich functional group of the guest
molecule helped tostrengthenbinding. Thismaybe the case here as the internal pyridine proton(HX) of
C2
∙12NO3 can be seen(fromx-raycrystallographydata) tobe ina similarorientation.The carbonyl group
76. 76
of chromone pointstowardsthe verticesof the cavity,e.g.the space whichthe internal pyridine proton
occupies.Itistherefore highlylikelythatasimilarhydrogen-bondinginteractionishelpingtostabilisethe
guestmolecule withinthe cavity.
Fig. D23: Spartan model showing host-guest complex C2
∙12NO3 chromone.
4.5.5 Fast-exchange guests:2-adamantanone nitrobenzene andcamphor
A similar interaction between this internal proton and the carbonyl group of 2-adamantanone and
camphor along with the nitro group of nitrobenzene is thought to encourage binding of these guests.
More time would enable NMR titrations of these guests, determination of association constants and
quantificationof the strength of binding. Comparison of Ka values with position of the internal pyridine
proton(slightlydifferentacrossthe 3cages) couldyieldaqualitative(possiblyevenquantitative withhost-
guestx-raycrystallographydata) insightintothe relationshipbetweenstructure andguest-bindingability.
77. 77
4.6 Summary and future work
Successful synthesisof C2
alone provesa positive result,aresultwhichconfirmsthe strengthof the self-
assembly followed by oxidation protocol. By the relatively simple act of oxidising the self-assembly
product, the interlinking problems of lability and stability that plague supramolecular chemistry are
somewhatovercome.Thisfeatureallowsfornow relativelyrapidsynthesisof tetrahedral cobalt cages.If
a suitable ligandsystemcan be designed and synthesised it is likelythat, via the self-assembly followed
by oxidationprotocol, acage can be obtainedpure. Thisisshownbythe rate at whichthe Lusby groupis
producingnewcages. Otherworkershave commonly reportedmixturesof supramolecularspecies,self-
assembly followed by oxidation allows for selective synthesis of either cage (CA
, CB
, CC
, C2
) or the
corresponding helicate (HA
, HB
, HC
, H2
) by variance of concentration and speed of oxidation. Since these
factors are now understood it would be interesting to examine the effect of a third variable. The
hydrophobiceffecthasbeenreportedtoaid supramolecularsynthesisof largerspecies(e.g. tetrahedra)
over smaller species (e.g. helicate), it would therefore be interesting to examine the self-assembly
reaction of C2
∙12NO3 (or CA
/CB
) in aqueous solution.33
An increase in the rate vs self-assembly in
acetonitrile would provide additional evidence for this theory. As would an increase in the relative
amountof cage vshelicate,duetothe reductionof hydrophobicinteractionaffordedbythelargerspecies.
Future work on C2
∙12NO3 would therefore include: NMR titrations of confirmed guests with the aim of
assessing the structure-function relationship,testing with substrates similar in structure to the Nazarov
substrate and otherunimolecularreagentswiththe aimof catalysis,andexaminationof the rate of self-
assembly whenperformed in water to assess whether the reaction is hydrophobically driven.This work
can continue hand-in-hand with the now relatively easy synthesis of new more stable cages which
currentlyinclude alternate linkerunitstothe central benzene aswell as6-memberedcoordinatingunits.
78. 78
Conclusion
During this project a new cobalt tetrahedral supramolecular cage (C2
) was synthesized from a cobalt (II)
precursor and ligandsystem L2
. The structure has been confirmedusingNMR,X-ray crystallography and
n-ESI-MSexperiments.Initial guestbindingshows behaviorsimilartoa previously synthesized cage (CA
),
encapsulationof TIPSOHoccurredinslow-exchangewhilstcamphor,2-adamantanone,nitrobenzene and
chromone were boundinfast-exchange. Stabilitystudiesshowedthatthoughdecompositionof C2
∙12NO3
(in a pH 6 sodium phosphate buffer solution) is slower than for CA
∙12NO3, though the structure is still
unstable tonear-neutralconditions.Comparisonwithamore stable newlysynthesizedcage (CA
) suggests
that strained geometry (around cobalt centres in C2
∙12PF6) due to coordination through 5- membered
imidazole rings is the cause of this instability.Though this feature limits the use of the systemas a drug
delivery vector, the cage’s ability to catalyse unimolecular reactions is still to be tested. The result
demonstrates the strength of the self-assemblyfollowedby oxidationprotocol when synthesizingthese
types of cage, but also shows the necessity for ideal cobalt geometry when designing for functionally
stable assemblies.
79. 79
Acknowledgements
I would like to thank P.J. Lusby for use of his laboratory and chemicals and for answering all questions
duringthe course of the project, howeverbasictheymayhave been.
I wouldlike tothankfellowprojectstudentsR.StewartandU. Mitreviciute fortheir companyaround the
laband for theirsharingof glassware.
Finally I would like to thank the PhD students in the group T. Sooksawat,D. August, M. Edwards and M.
Burke withoutwhom completionof the projectwouldhave beenimpossible.
