Mucus-responsive functionalized emulsions: design, synthesis and study of novel branched polymers as functional emulsifiers

Mucus lines the moist cavities throughout the body, acting as barrier by protecting the underlying cells against the external environment, but it also hinders the permeation of drugs and drug delivery systems. As the rate of diffusion is low, the development of a system which could increase retention time at the mucosal surface would prove beneficial. Here, we have designed a range of branched copolymers to act as functional mucus-responsive oil-in-water emulsifiers comprising the hydrophilic monomer oligo(ethylene glycol) methacrylate and a hydrophobic dodecyl initiator. The study aimed to investigate the importance of chain end functionality on successful emulsion formation, by systematically replacing a fraction of the hydrophobic chain ends with a secondary poly(ethylene glycol) based hydrophilic initiator in a mixed-initiation strategy; a decrease of up to 75 mole percent of hydrophobic chain ends within the branched polymer emulsifiers was shown to maintain comparative emulsion stability. These redundant chain ends allowed for functionality to be incorporated into the polymers via a xanthate based initiator containing a masked thiol group; thiol groups are known to have mucoadhesive character, due to their ability to form disulfide bonds with the cysteine rich areas of mucus. The mucoadhesive nature of emulsions stabilised by thiol-containing branched copolymers was compared to non-functional emulsions in the presence of a biosimilar mucosal substrate and enhanced adherence to the mucosal surface was observed. Importantly, droplet rupture and mucus triggered release of dye-containing oil was seen from previously highly-stable thiol-functional emulsions; this observation was not mirrored by non-functional emulsions where droplet integrity was maintained even in the presence of mucus.


Introduction
Mucous membranes create the moist exterior of various regions of the body including oral, gastrointestinal (GI), genital and ocular surfaces. For example, mucus is secreted by goblet cells directly onto epithelium cells within the GI tract, 1 forming a mobile outer layer, that is relatively quickly replaced, and a stationary gel-like layer, 2 that is adhered to the surface of the goblet cells. Mucosa varies across different anatomical sites with different life-cycles and thicknesses that depend primarily on its specic local function. [3][4][5] As a biologically-derived hydrogel 6 with dened viscoelasticity, rheology and mesh size, mucous membranes act to moderate the passage of exogenous material and prevent infection; in addition, the permeation of many pharmaceutical agents is also restricted in this way. Mucous membranes have, therefore, been a focus for drug delivery over many years and mucoadhesion and mucus permeability have been the subject of much research. 7 Structurally, mucus comprises a water-swollen network of gel-forming mucins which are cysteine-rich high molecular weight glycosylated proteins. Disulde bridges between cysteine residues contribute to the physical properties of the mucosal membrane; these sulfur-containing domains offer a chemical target for mucoadhesion and mucopermeation strategies. Mucoadhesion is generally regarded as progressing through a two-step process, 8 namely contact and consolidation. In situations where adhesion to mucus occurs within a predominantly liquid environment, the contact stage may be considered using conventional DLVO theory, as the resultant contradictory forces of attraction and repulsion must favour a physical interaction; a positive wetting interaction will also favour the contact stage. The material in contact with the mucus may be removed through various shear or stress modes and therefore consolidation of the interaction is required to generate a long-acting adhesion. In a mainly aqueous medium, this may be achieved in several ways 9 including entrapment, non-covalent intermolecular forces (e.g. electrostatic, van der Waals' and hydrogen bonding interactions), and the formation of covalent bonds through chemical reactions (e.g. disulde formation and Michael addition).
Thiol-functional materials and thiol-bearing polymers, oen called "thiomers", [10][11][12] have been studied specically to interact with cysteine residues within mucins, forming disulde bonds either through thiol/disulde exchange or thiol-thiol reactions within the mucus network; 13 evidence for these mechanisms has been established in several reports. [14][15][16][17] To enable drug delivery, thiol-functional polymers have been used to produce tablets, 17 polymeric micelles 18,19 and microparticles. 20 Recently, alkyl-modied carbomers (lightly crosslinked poly(acrylic acid)) were modied with 4-aminothiophenol, L-cysteine or D/L-homocysteine to form a novel class of emulsiers with the aim of generating mucoadhesive emulsions; 21 the emulsication of medium chain triglycerides was reported with varying stabilities, especially if the thiomers used were able to oxidise or crosslink prior to emulsi-cation. These emulsions were also shown to adhere to porcine buccal mucosa, hence demonstrating the concept of mucoadhesive oil-in-water (O/W) emulsions.
In recent work, we have shown that branched vinyl polymers made by conventional and controlled radical polymerisation are able to act is highly efficient emulsiers. [22][23][24][25] Branching was achieved by the copolymerisation of the macromonomer oligo(ethylene glycol methacrylate), (OEGMA) with the divinyl monomer ethylene glycol dimethacrylate (EGDMA), leading to a large number of hydrophilic primary polymer chains bridged by a small number of branch points. Through the utilization of either hydrophobic initiators (controlled radical polymerisation) or hydrophobic chain transfer agents (conventional radical synthesis), amphiphilicity can be introduced into the hydrophilic water-soluble branched polymer at the end of each conjoined primary chain, Fig. 1B. The resulting high molecular weight branched copolymers interact with oil droplet surfaces through multiple attachment sites, providing a contrast to linear polymer surfactants which suffer from single dynamic interactions, and making them more akin to Pickering emulsions, Fig. 1C, as evidenced by their remarkable stability and resistance to dilution. 24 Herein, we aimed to investigate the impact of decreasing the number of hydrophobic chain-ends within the branched copolymer stabiliser as a strategy for introducing thiol functionality to emulsion droplets and controllably inducing mucoadhesion. Within this model study we monitored emulsion storage stability over prolonged periods and the triggering of demulsication on contact with biosimilar mucus surfaces. To the best of our knowledge, this represents the rst demonstration of a triggered mucus-responsive emulsion system enabled by novel branched copolymer emulsiers.

Results and discussion
Investigating chain-end criticality for branched copolymer emulsiers Polymerisation studies and copolymer characterisation. The synthesis of branched vinyl copolymers using the incorporation of low concentrations of divinyl monomers (less than 1 per propagating primary chain) has been the subject of several research reports. 26 The mixing of differing monovinyl monomer chemistries within the primary chains of the complex polymer architectures 27,28 has also been used to introduce main-chain functionality; 22,23 however, the joining together of polymer chains via this approach opens a unique opportunity to also control chain-end functionality, as demonstrated by our group in the formation of hyperbranched-polydendrons with varying combinations of dendritic and polyethylene glycol (PEG) chains. [29][30][31] As mentioned above, we have recently reported the formation and application of branched copolymers consisting of OEGMA (M n ¼ 300 g mol À1 ) and EGDMA, initiated by dodecyl a-bromoisobutyrate, 1 (Dod-Br, Scheme 1), under atom transfer radical polymerisation (ATRP) conditions, as highly efficient polymeric surfactants; the copolymers contain a hydrophobic dodecyl group at every chain-end. 24 We hypothesised that initiating this copolymerisation with mixed initiator feedstocks would allow the creation of branched copolymers with varying chain-end composition; however, the decrease in hydrophobic chain-ends may potentially impair the surfactant behaviour of the branched copolymers. To study this in detail, a series of branched copolymers with systematically varying numbers of hydrophilic and hydrophobic chain-ends was synthesised to establish the critical content of hydrophobic (lipophilic) chain-ends within the branched copolymer structures required for stable O/W emulsion formation.
As seen in Scheme 1A, co-initiation of the linear polymerisation of OEGMA with varying ratios of dodecyl-derived, 1, and PEG 17 -derived, 2 (PEG (17) -Br; starting material PEG-mono methyl ether 750 g mol À1 ), initiators leads to a mixed linear polymer product where a fraction of the polymer chains possess hydrophilic PEG chain-ends ([PEG (17)1.00 ]-p(OEGMA 50 )), and hence are exclusively hydrophilic, whilst the remaining fraction will carry a hydrophobic dodecyl chain-end ([Dod 1.00 ]p(OEGMA 50 )), and possess amphiphilic character. Conversely, the introduction of a low molar concentration of EGDMA, Scheme 1B, will combine the mixed linear population into a range of larger structures with an overall controlled average of PEG 17 and dodecyl chain-ends across the full polymer distribution. If a substantial number of hydrophobic dodecyl chainends could be replaced with hydrophilic PEG 17 chain-ends, whilst maintaining emulsier properties, an indication of chain-end redundancy would be derived and the potential for utilising the redundancy to introduce functional groups may become available.
A series of branched OEGMA/EGDMA copolymers were, therefore, synthesised via a mixed initiator, copper-catalysed ATRP strategy in an isopropyl alcohol (IPA)/water mixture (92.5/7.5 v/v%) at 40 C. All coinitiated copolymerisations targeted a number average degree of polymerisation (DP n ) of 50 monomer units ( ) were also synthesised, to allow comparison of synthesis conditions for each initiator and emulsication performance of mixed polymer samples as a control. Branched copolymerisations oen require extended reaction times, therefore each reaction was monitored by 1 H nuclear magnetic resonance (NMR) spectroscopy and terminated when high conversion ($99%) was established. Triple detection size exclusion chromatography (TD-SEC), using a dimethylformamide (DMF)/0.01 M LiBr eluent, was utilised to determine number average molecular weight (M n ), weight average molecular weight (M w ) and dispersity (Đ), Table 1.
The M n values of the linear polymer samples were higher than targeted, with Đ values indicating non-ideal initiator efficiencies and the inherent dispersity of the PEG 17 -derived macroinitiator. The branched copolymerisations showed relatively consistent M n values, in general, suggesting only a minor difference in the behaviour of the two initiators and comparable initiator efficiencies during the early stages of copolymerisation.
As expected, the M n and M w values were consistently higher for the branched copolymers than their linear analogues with correspondingly higher dispersities (ESI Fig. S7-S9 †). Within these broad distributions, a weight fraction analysis of the TD-SEC data indicated that between approximately 6-34 wt% of the various samples have molecular weights >500 kg mol À1 (therefore containing > 10 conjoined chains), ESI Fig. S10, S11 and Table S1. † As such, although there will be heterogeneity of the mixed initiator-derived chain-ends, a signicant proportion of branched copolymers will possess mixtures of PEG and Dod chain-ends even at the extremes of the 1 : 2 ratios targeted, Fig. 2.

Aqueous solution behaviour and emulsication studies
All branched copolymers showed high aqueous solubility (>5 wt%) and no observable solution differences across the samples with varying overall chain-end composition; lower critical solution temperature (LCST) studies exhibited a trend to lower values with increasing hydrophobic chain-end content, and branched [PEG 1.00 ]-p(OEGMA 50 -co-EGDMA 0.80 ) showed slightly lower values than its linear analogue; a relatively narrow range of transition temperatures was observed across the series (approximately 61-67 C), ESI, Fig S12. † The varying branched copolymer structures, Fig. 2, were also studied using static contact angle measurement of aqueous branched copolymer surfactant solutions (5 wt%). Aqueous solutions were placed onto a hydrophobic polytetrauoroethylene substrate to evaluate the presence and impact of the varying hydrophilic/hydrophobic chain-ends ( Fig. 3A(i)-(iv)). All solutions containing branched copolymers showed a decrease in contact angle from deionised water (125 AE 3.6 , Fig. 3A(i)), indicating the presence of the branched copolymer at the air/ water interface and an increase in hydrophobicity (ESI, Table S2 †). The lowest contact angle was observed for [Dod 1.00 ]p(OEGMA 50 -co-EGDMA 0.80 ) solutions (80 AE 2.2 ) and the introduction of PEG (17) -derived chain-ends increased the contact angle of the various solutions to range from approximately 99-105 ; the clear inference being that redundant hydrophobic chain-ends are present at the droplet surface for solutions containing [Dod 1.00 ]-p(OEGMA 50 -co-EGDMA 0.80 ) and these are somewhat removed and replaced by the more hydrophilic PEG 17 chain-ends in the solutions of copolymers containing mixed chain-end functionality. The contribution of the hydrophobic polymethacrylate backbone to the changes in observed contact angles must also not be ignored; however, this remains consistent within the polymer structures and may become more prominent with the changing chain-end chemistry.
Surface tension measurements were conducted to study the impact of the increasing number of hydrophilic PEG-derived chain-ends within the branched copolymer structures (ESI ,  Table S3, Fig. S13 †). Interestingly, changes to copolymer endgroup functionality across all materials had little impact on   their CMC behaviour. The intercept of the rst plateau of the surface tension curves were observed at concentrations of between 4.72 Â 10 À8 (AE8%) and 7.40 Â 10 À8 (AE2%) mg L À1 for linear polymers [Dod 1.00 ]-p(OEGMA 50 ) and PEG (17)1.00p(OEGMA 50 ) respectively ( Fig. 3B), suggesting a role for the hydrophobic methacrylate backbone in adsorption to the oil droplet surface. Whilst the slightly lower CMC obtained for [Dod 1.00 ]-p(OEGMA 50 ) could be attributed to the presence of the hydrophobic dodecyl chain-ends, comparable CMC values likely arise due to the low weight fraction of the polymer chains-ends (<2 wt% in all cases based on M n(GPC) ). This is supported by the consistent CMC values obtained for branched copolymers containing mixed chain-end functionality (ESI Fig. S14 †). All branched copolymer surfactants showed complex surface tension behaviour with increasing copolymer concentration; initial plateaus were followed by subsequent further decreases in surface tension at higher concentration. This behaviour is well reported for complex mixtures such as polymeric/small molecule surfactant combinations, and therefore is readily rationalised as being derived from the distribution of species within the statistically branched copolymer molecular weight distribution (ranging from linear polymers through to highly branched structures). CMC values obtained from branched copolymers containing mixed chain-end functionality ranged from 1.08 Â 10 À8 (AE16%) to 4.74 Â 10 À8 (AE2%) mg L À1 and showed no clear trend between chain-end composition and CMC values. This data indicates that, despite their contrasting chain-end compositions, branched copolymer emulsiers are expected to adopt similar conformations in aqueous solution and would exhibit similar responses to changes in concentration. Furthermore, this demonstrates that it would be possible to manipulate branched copolymer emulsifying properties with minimal impact on their aqueous solution behaviour.
Evaluation of the branched copolymers as emulsiers (5 mg mL À1 ) was conducted using high shear homogenisation of dodecane (1 : 1 v/v ratio) to create O/W emulsions, Fig. 4. Dodecane was selected as the disperse phase of the O/W emulsions due to its similarity to the dodecyl hydrophobic chain-end, introduced by the Dod-Br initiator, and the favourable interactions that should therefore occur with the resultant oil droplets. In the absence of a branched copolymer emulsier, stable emulsions could not be produced, with rapid demulsication and resulting phase separation within approximately two minutes of standing; similarly, all samples emulsied with linear polymers, or mixtures of linear polymers with varying hydrophilic/hydrophobic ratios of chain ends, demulsied relatively rapidly on standing for less than one week (ESI Fig. S15 †).
For all samples containing branched copolymer surfactants with various ratios of Dod and PEG (17) chain-ends, stable emulsions were formed. Creaming of the O/W emulsion was observed aer 24 hours equilibration due to density differences; however, no oil separation was apparent with emulsions characteristically white and opaque and stable for >3 years at ambient temperature ( Fig. 4A(i)-(iv)). Quantitative analysis by laser diffraction spectroscopy (ESI , Table S4, Fig. S16 †) showed highly comparable volume average mean diameters (D [4,3] ) ranging from 11.9-15.3 mm with a trend towards larger droplet diameters with decreasing Dod content within the branched copolymer solutions of identical concentration. Interestingly, the size distribution of O/W emulsion droplets stabilised by branched copolymer emulsiers resembled the smaller droplets This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 30463-30475 | 30467 within the bimodal size distribution generated by the equivalent mixed linear homopolymer system, described above (ESI Fig. S17 †).
The reduction of the number of hydrophobic chain-ends appears to be analogous to a reduction in small-molecule surfactant concentration, therefore a lower surface area may be stabilised and a smaller number of larger droplets result. As mentioned above, the dispersity of the co-initiated branched copolymer samples does also lead to the potential presence of a fraction of linear [PEG (17)1.00 ]-p(OEGMA 50 ) and some lightly branched materials having no Dod chain-ends. As may have been predicted, emulsions prepared using [PEG (17)1.00 ]p(OEGMA 50 -co-EGDMA 0.80 ) showed no long-term stability, with complete and rapid demulsication occurring within 72 hours (ESI Fig. S18 †). The role of the hydrophobic backbone in generating some stability to the emulsions cannot be ruled out; however, this material emphasises the critical importance of chain-end chemistry on the formation of stable emulsions.
Observation of the D [4,3] values over 40 days under ambient conditions for the emulsions stabilised with Dod-containing branched copolymers revealed very small changes in emulsion droplet sizes overall (Fig. 4B). The obtained D [4,3] values correlated well with those obtained by optical microscopy imaging (Fig. 4C(i)-(iv)), which show well-dened spherical droplets with no signs of coalescence across varying branched copolymer compositions, even aer several months of storage under ambient conditions. Laser diffraction was also used to assess long-term stability of the different O/W emulsions and minimal changes in the observed distributions and measured D [4,3] values were seen between samples studied 24 hours aer emulsication and aer storage under ambient conditions for 3 years (ESI, Fig. S19, Table S5 †). Dilution of the Dod-containing branched copolymer stabilised emulsions from the high concentration creamed layer through to 6.25% (v/v) also conrmed the unusual stability of the emulsions (Fig. 5) as reported previously for branched copolymer-stabilised O/W nanoemulsions. 24 This behaviour is remarkable given the probable presence of linear and low molecular weight species within the distribution bearing no hydrophobic chain-ends.

Thiol functional to branched copolymer emulsiers
The ability to introduce redundant chain-ends into the branched copolymer emulsiers, whilst maintaining efficient emulsication, offers the potential to add functional groups to the termini of a controlled number of primary chains across the polymer distribution without compromising performance of the emulsiers. Here, we have targeted thiol functionality, to facilitate mucus adhesion, through the use of a mixed initiation using 1 and a previously reported xanthate-functional ATRP initiator (Xan-Br), 3, Scheme 2.
As previously reported for analogous materials, the resulting [Xan y /Dod x ]-p(OEGMA 50 -co-EGDMA 0.80 ) branched copolymers, with varying ratios of Dod and xanthate chain ends, may be readily deprotected using n-butylamine to liberate the thiol functionality and create a series of functional emulsiers denoted as [SH y /Dod x ]-p(OEGMA 50 -co-EGDMA 0.80 ). The formation of linear polymers under identical conditions to those used above but initiated with 3, to yield [Xan 1.00 ]p(OEGMA 50 ), was investigated and shown to be comparable to the synthesis of [Dod 1.00 ]-p(OEGMA 50 ) and [PEG (17)1.00 ]p(OEGMA 50 ). The initiator efficiency of 3 does appear to be slightly higher than the other initiators (approximately 67% for 3; approximately 50% for 1) suggesting a slightly higher number of actual xanthate-bearing chains would be present in branched copolymer samples using the initiator than targeted through the nominal initiator molar ratios. The formation of a series of comparable branched copolymerisations utilising EGDMA and mixed 1 : 3 coinitiation achieved $99% monomer conversion in all cases and samples were analysed by TD-SEC to determine molecular weights and dispersities, Table 2 (ESI, Fig. S20 and S21 †).
To compare the xanthate-functional branched copolymers with those containing mixed Dod/PEG 17 chain-ends, the TD-SEC data was analysed to provide a weight fraction analysis, indicating that, again, between approximately 6-19 wt% of the species in branched copolymer samples have molecular weights >500 kg mol À1 which equates to >10 conjoined primary chains (ESI, Fig. S22 and Table S6 †). This was highly encouraging as the weight fractions of high molecular weight, highly branched structures are broadly similar to those shown to act as efficient emulsiers. Removal of the xanthate protecting group, and liberation of the thiol-functionality, was conducted for a selected series of functional copolymers, leading to the systematically-varying thiol-functional copolymers [ Fig. S23 †).

Emulsication and mucus-responsive studies using thiolfunctional to branched copolymer stabilisers
The functional-emulsions initially envisaged for these studies require biologically/pharmacologically-relevant oil phases to be stabilised in water. As such, and building on our recent reports, 24,31 emulsions were generated using the naturally occurring polyunsaturated liquid hydrocarbon squalene using the thiol-functional branched copolymers. This more detailed study was limited to [SH 0.75 /Dod 0.25 ]-p(OEGMA 50 -co-EGDMA 0.80 ) due to its balance of high thiol content, a predicted efficient number of Dod chain-ends to maintain emulsication performance, a good balance of molecular weight, and a weight fraction of highly branched chains representing a high probability of thiol content (8 wt% $ 19 chains; 4 wt% $ 28 chains; and 2 wt% $ 37 conjoined chains) (ESI , Table S6 †).
Squalene emulsions were generated under identical conditions to dodecane emulsions described above and, as expected, this [SH 0.75 /Dod 0.25 ]-p(OEGMA 50 -co-EGDMA 0.80 ) branched copolymer acted as a highly efficient emulsier, generating approximately 16 mm oil droplets (D [4,3] ) within an aqueous continuous phase (ESI, Table S9 †). The potential for disulde bond formation on storage, leading to possible aggregation or demulsication, was studied by storing the thiol-functional emulsion under atmospheric conditions and ambient temperature for extended periods. Laser diffraction measurements taken each week for 4 weeks showed no meaningful variation or effect of storage and no visible separation of oil was observed (ESI Fig. S26 †). Squalene is a good solvent for a range of molecules including two dye molecules, the diazo dye Oil Red O and an anthraquinone Oil Blue A, which were studied as models of future therapeutic agents that may be incorporated into the emulsions (Fig. 6 & 7). This was achieved via addition of small amounts (0.1 wt% w.r.t. oil phase) of the respective dye into the This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 30463-30475 | 30469 emulsication process. Incorporation of the dye molecules resulted in slightly smaller emulsions (D [4,3] approximately 13 mm) and samples stabilised with either Dod 1.00 -p(OEGMA 50 -co-EGDMA 0.80 ) or [SH 0.75 /Dod 0.25 ]-p(OEGMA 50 -co-EGDMA 0.80 ) were generated to investigate the impact of thiol-functionality on the stable emulsions (ESI, Fig. S27 †).
Several biosimilar mucus formulations have been reported; a mimic of natural mucus, outlined by Boegh et al., 32 has been reported as being rheologically optimised to strongly resemble natural porcine intestinal mucus. The biosimilar mucus contains Carbopol 940 (a high molecular weight crosslinked poly(acrylic acid)), to provide a shear-thinning dominant elastic behaviour, whilst chemical similarity is obtained by incorporation of porcine mucin, bovine serum albumin and a lipid mixture of cholesterol, phosphatidylcholine and linoleic acid; the surfactant polysorbate 80 is also present. The biosimilar mucus was spread evenly across a glass microscope slide to provide a mucosal surface for investigation by optical   microscopy; each sample was viewed and monitored closely for 10 minutes aer the addition of the concentrated emulsions to the mucus substrate. Squalene emulsions stabilised with [Dod 1.00 ]-p(OEGMA 50 -co-EGDMA 0.80 ) (Fig. 6A(i)) were unaffected by the mucus surface, spreading evenly across the substrate and maintaining their ability to move throughout the study; ready identication of individual stable droplets was achieved via optical microscopy, with no observable aggregation or perturbation of the dispersed oil phase (Fig. 6A(ii)). This behaviour was in marked contrast to the emulsions stabilised with [SH 0.75 /Dod 0.25 ]-p(OEGMA 50 -co-EGDMA 0.80 ) (Fig. 6B); addition of this emulsion to the biosimilar mucus-covered glass slide led to instantly observable differences in spreading and aggregation. During the same 10 minute observation period, the formation of free oil aer droplet rupturing was also seen (Fig. 6B(ii)). A separate [SH 0.75 /Dod 0.25 ]-p(OEGMA 50 -co-EGDMA 0.80 )-stabilised squalene-in-water emulsion was prepared, containing Oil Red O (ESI Fig. S28 †). Exchange of Oil Blue A for Oil Red O had no impact on the mucus-responsive behaviour, with extensive demulsication occurring within 10 minutes of addition to a biosimilar mucosal surface (Fig. 7A(i) and (ii)).
The mucus-triggered demulsication was investigated further via simultaneous addition of the two dye-loaded, thiolfunctional emulsions to the same mucus substrate. As seen in previous experiments, the emulsions adhered to the mucus at the point of addition with a clear boundary between the coloured emulsions and limited mixing (Fig. 7B(i)). The resulting rupturing of the oil phases with each zone of coloured emulsion (Fig. 7B(ii)), led to a mobile liquid oil layer that quickly mixed (Fig. 7B(iii)), generating a purple colour, in the presence of distinct and static red and blue areas. This suggests opportunities for the isolation of incompatible oil-soluble chemistries, or combination products, with triggered demuslication and mixing on target mucosal surfaces.
The limited spreading, and observed lack of motion, of the [SH 0.75 /Dod 0.25 ]-p(OEGMA 50 -co-EGDMA 0.80 )-stabilised emulsion suggests rapid and strong mucus interactions, presumably via disulde bond formation between the cysteine-rich regions within the porcine mucin and the thiol functionality presented at the oil droplet surface. Chemical bond formation during the consolidation stage is characteristic of the so-called "second generation" mucoadhesive materials and disulde bond formation is the basis of most thiolated mucoadhesive macromolecules. This correlates well with the literature "contact/ consolidation" model of mucoadhesion but is surprising given the high stability of the thiol-functional droplets in the absence of mucus. We hypothesise that the thiols at the branched copolymer chain ends are unable to approach each other to form bridging disulde bonds upon storage, due to the steric repulsion of the branched p(OEGMA) chains at the oil-droplet/ water interface (Fig. 8A); intra-droplet reaction is also highly unlikely given the density of OEGMA side chains along the p(OEGMA) backbone. The rapid interaction with the mucus substrate appears to suggest that steric repulsion is not a major factor at this interface, and the thiols present in the cysteine-rich areas of the porcine mucus are able to bond readily with thiol-functional emulsion droplets (Fig. 8B).
The demulsication behaviour, and release of free oil, suggests that the formation of covalent bonds with the mucus substrate leads to either a removal of branched copolymer stabiliser from the oil/water interface, with subsequent destabilisation and agglomeration, or the modication of the contact angle at the mucus-emulsion interface and causing the droplet to rupture and spread resulting in agglomeration and release of the emulsied oil.

Conclusions
Mucoadhesive materials offer a considerable opportunity for the formulation and delivery of a range of compounds to target mucosal sites. Mucoadhesive emulsions are of particular interest as the ability to dissolve hydrophobic materials within the stabilised oil-droplets offers the potential to carry active molecules and deliver them to specic locations; antimicrobials, antivirals, hygiene agents and drug compounds are of clear interest but materials able to diagnose, label or indicate local environmental conditions also have applications. Here, we have shown a novel strategy for generating bespoke thiolfunctional, highly efficient emulsiers and, to the best of our knowledge, such materials have not been reported previously. The branched copolymer synthesis platform, copolymerising a low concentration of bifunctional monomer within a conventional ATRP polymerisation, allows not only the formation of high molecular weight materials, but also the introduction of amphiphilicity with controlled chain-end functionalisation; the This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 30463-30475 | 30471 demonstration of chain-end redundancy and thiol-functionality introduction offers new avenues for functional-emulsion research. Mucus-triggered oil release was not an initial target of this study but the ability to form such stable emulsions with rapid release is remarkable and requires additional research to understand both the limitations of the approach and its full opportunities. For simplicity, one branched copolymer emul-sier and one ratio of emulsier to oil, at one oil-droplet size, were selected for these initial mucus-substrate interaction studies, however, there are options to tailor the interactions through concentration of thiol-functional groups and oildroplet size; these studies are ongoing.

Characterisation
Nuclear magnetic resonance (NMR) spectroscopy was recorded using a Bruker DPX-400 spectrometer operating at 400 MHz for 1 H and 100 MHz 13 C nuclei respectively. Molecular weight data was obtained by triple detection size exclusion chromatography (TD-SEC) using a Malvern Viscotek SEC Max . The instrument was equipped with a GPC max VE2001 auto sampler, two Viscotek D6000 columns, a guard column and a triple detector array TDA305 (refractive index, light scattering and viscometer). TD-SEC was conducted at 60 C using a mobile phase of DMF containing 0.01 M lithium bromide (LiBr) at a ow rate of 1 mL min À1 . Emulsications were conducted using an IKA T25 ULTRA-TURRAX high-shear homogeniser. Oil-in-water (O/W) emulsions were characterised by laser diffraction using a Malvern Mastersizer 2000A equipped with a Hydro 2000 SM dispersion unit. Microscope images were obtained using a Leica DM4 B microscope tted with a CMOS camera. A Kibron Delta-8 surface tensiometer was used to measure the surface tension of aqueous polymer solutions at 20 C. The contact angles of aqueous polymer solution droplets rested on a Teon substrate were measured using A Kruss DSA100E drop shape analyser equipped with an advanced digital camera (Â20 magnication).

Polymer synthesislinear polymerisations
For the synthesis of linear p(OEGMA) samples, targeting DP n ¼ 50 monomer units, Dod-Br (0.027 g, 0.08 mmol, 1.00 equiv.), PEG 17 -Br (0.25 g, 0.34 mmol, 1.00 equiv.) or Xan-Br, (0.13 g, 0.30 mmol, 1.00 equiv.) along with 2,2 0 -bipyridyl (0.10 g, 0.64 mmol, 2.00 equiv.) and OEGMA (5.00 g, 16.0 mmol, 50.0 equiv.) were added to a 25 mL round bottom ask, equipped with a magnetic stirrer bar. The mixture was deoxygenated via N 2 bubbling for 30 minutes. A solvent system of IPA/H 2 O (92.5 : 7.5 v/v%, 4.39 : 0.36 mL, 55 wt% w.r.t to monomer, deoxygenated separately via N 2 purge) was added to the reaction vessel and the mixture was deoxygenated by N 2 bubbling for a further 5 minutes. Anisole (ca. 100 mL) was added to the reaction ask for use as an internal standard for determination of monomer conversion by 1 H NMR. This was calculated from the change in the intensity of OEGMA vinyl peaks during polymerisation, relative to those of the anisole aromatic 1 H resonances (ESI, Fig. S6 †). CuCl (0.032 g, 0.32 mmol, 1.00 equiv.) was added and the reaction vessel was then sealed. The ask was placed into a pre-heated oil bath (40 C) and le for 24 h aer which the polymerisation was stopped by cooling the ask to ambient temperature, exposure to oxygen and dilution with THF. A neutral alumina column was used to remove the copper catalyst, and the solvent was removed in vacuo. The crude polymer was puried by precipitation (twice) from acetone into cold petroleum ether (30-40 C). Residual solvent was removed in a vacuum oven at 40 C overnight. The puried polymers were characterised by 1 H NMR in CDCl 3 and TD-SEC using a DMF/LiBr (0.01 M) eluent.

Characterisationlower critical solution temperature measurements
For a typical lower critical solution temperature (LCST) measurement, an aqueous copolymer solution (5 mL, 1 wt%) was prepared in a round bottom ask (10 mL) equipped with a magnetic stirrer bar. The solution was studied through three heating/cooling cycles in an oil bath at a heating/cooling rate of approximately 1 C min À1 . The temperatures of copolymer solutions were monitored until the cloud point was observed, i.e. when the transparent solution turned opaque, indicating precipitation of the dissolved macromolecules. LCST values were quoted as the mean value (n ¼ 3 AE standard deviations).

Characterisationsurface tensiometry measurements
Aqueous copolymer solutions were prepared at a concentration of 30 wt% in deionised (DI) H 2 O. Serial dilutions of such copolymer solutions in DI H 2 O were used to yield 30 different copolymer concentrations ranging from 6.9 Â 10 À9 to 30 wt%. The use of a 96 well-plate allowed for high throughput analysis of samples; 50 mL of each polymeric surfactant (n ¼ 8) was used per well and measured against pure water as the control. Surface tension data was plotted against a log concentration of the surfactant. Critical micelle concentrations (CMC) were calculated by determining the intercept between the lines of best t obtained from the linear decreases in surface tension (slope) and the lower plateau area.

Characterisationcontact angle measurements
Aqueous copolymer solutions (5 wt%) were prepared and applied to a glass microscope slide covered with a polytetra-uoroethylene (PTFE) substrate, to create a hydrophobic surface. Aqueous copolymer solutions (5 mL) were applied to the slide and contact angles measured using the static sessile drop method (n ¼ 10 AE standard deviations).

Characterisationpreparation of biosimilar mucus
Carbopol® 940 (0.90 g, 0.90 w/v%) was dissolved in HEPES buffer solution (9.00 mL, 1.30 mM CaCl 2 , 1.00 mM MgSO 4 and 137 mM NaCl, pH 7.4) using magnetic stirring. A lipid mixture of phosphatidylcholine (0.018 g, 0.18% weight/volume (w/v)%), cholesterol (0.0036 g, 0.36 w/v%), polysorbate 80 (0.033 g, 4 : 1 ratio) and HEPES buffer solution (1 mL) was prepared and added to the Carbopol solution when approximately 90% of the Carbopol had dissolved. Mucin from porcine stomach (0.50 g, 5.00 w/v%) was added to the solution and the pH altered to 7.4 with the addition of NaOH (1 M), which was monitored using a pH probe. BSA (0.31 g, 3.10 w/v%) was added and the pH was once again adjusted to 7.4. Biosimilar mucus was stored overnight at 2-4 C before use and discarded if not used within a maximum of four days from the day of preparation.

Preparation and characterisation of O/W emulsions
O/W emulsions were prepared using a 1 : 1 v/v ratio of oil-: water, using either n-dodecane or squalene as the oil phase. Aqueous copolymer solutions (3 mL, 5 mg mL À1 ) and ndodecane or squalene (3 mL) were added to a glass vial (14 mL); the mixture was then homogenised using an IKA T 25 ULTRA-TURRAX over-head high-shear homogeniser at 24 000 rpm for 2 minutes. Emulsions were stored overnight in a sealed glass vial at ambient temperature before characterisation via laser diffraction. O/W emulsions were added dropwise to the dispersion unit containing approximately 100 mL deionised water (DI H 2 O) at a stirring rate of 1000 rpm at ambient temperature. The volume-average droplet diameters (D [4,3] ) are quoted as an average of $20 measurements, where D [4,3] ¼ P D i 4 N i / P D i 3 N i . For further stability studies, emulsions were stored in sealed glass vials at ambient temperature and were reassessed by laser diffraction and optical microscopy at the time intervals stated.

Mucosal trigged emulsion release studies
For mucosal triggered release studies emulsions were prepared as above with the inclusion of either Oil Red O or Oil Blue A (0.1 wt% with respect to the oil phase) as a hydrophobic drug mimic. In a typical mucosal triggered release study, biosimilar mucus was spread onto a glass microscope slide. Polymeric emulsier-stabilised O/W emulsions (100 mL) were applied to the centre of the glass slide. The slides were then imaged using optical microscopy both immediately aer application of the O/ W emulsion and again 10 minutes aer the addition.

Conflicts of interest
SEE and SPR are co-inventors of a recently led patent application that describes the work reported here.