Protein triggered ordering transitions in poly (L-lysine)-coated liquid crystal emulsion droplets

ABSTRACT Here, we report a simple and label-free methodology for real-time monitoring of adsorption of proteins such as bovine serum albumin (BSA), concanavalin A (ConA) (a lectin) and cathepsin D (CathD) (a tumour marker) on micrometer-sized poly (L-lysine) (PLL) functionalised liquid crystal (LC) droplets dispersed in aqueous phases. Earlier, we had demonstrated that PLL, a positively charged natural peptide, can induce homeotropic ordering of LCs at LC-aqueous interface, and thus PLL-adsorbed LC droplets showed radial director configuration. Herein, it was observed that subsequent non-specific adsorption of anionic proteins such as BSA, ConA and CathD can trigger a quick transition in director configuration of PLL-LC droplets (primarily dominated by electrostatic interactions) to pre-radial or bipolar, thus acting as a simple optical probe for detection of these proteins up to μg/mL of concentrations. Further, the detection limits for these proteins are found to vary (BSA<ConA<CathD) which follow the similar order as their anionic charges, thus suggesting the role of different binding affinities of protein-PLL in realising the director configuration of LC droplets. Overall, this study offers new pathways utilising ordering transition in LC droplets which will strengthen the principles to recognise biomolecular interactions for various interfacial and sensing applications. GRAPHICAL ABSTRACT


Introduction
Understanding protein interactions with surfaces (majorly driven by non-covalent forces) forms a basis to develop the fundamental principles of biological processes inside the cells and for various biomedical applications [1][2][3][4][5]. Considering the basic importance of cooperative electrostatic interactions between proteins and other biomolecules/membranes in the biological system, it becomes crucial to design stimuli-responsive interfaces for protein adsorption and principles to identify such interactions between proteins and the modified surfaces. Currently, there are numerous physicochemical methods available to detect and study protein interactions such as affinity chromatography, protein arrays, surface plasmon resonance, quartz crystal microbalance, FTIR spectroscopy, NMR spectroscopy and X-ray diffraction, but the costly instrumentation and complexity involved limit their widespread use in daily life [6].
Nematic liquid crystal (LC) confined within micrometre-sized droplets in aqueous media has been considered as the new class of functional materials for the broad range of sensing and interfacial applications [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24]. Due to their tunable optical properties, large surface areas and rich phases, LC droplets have shown tremendous potential as the optical biosensors. The detection principle lies within the change in director configurations of the LCs inside the droplets that can be optically observed with naked eye, eliminating the need of expensive and complex detection systems for signal transduction. For instance, LC microdroplets have been applied as a sensing tool for developing immunoassays [13], to detect glucose [18], bacteria and viruses [10], bacterial endotoxin at pgmL −1 concentration [15] and many more [17,[19][20][21][22][23][24]. Recently, we reported that adsorption of poly (L-lysine) (PLL) (a cationic peptide) can induce homeotropic orientational ordering of LCs at LC-aqueous interface. The homeotropic ordering is mainly due to the intermolecular hydrogen bonding between PLL and the 5CB LC [23]. In addition, the LC droplets prepared by alternate multilayer assembly of PLL and polystyrene sulphonate (PSS) (an anionic polyelectrolyte) with outermost being PLL showed radial director configuration of the LC. We had also observed that adsorption of Annexin V (an anionic protein) on (PLL-PSS)-coated LC droplets could trigger the radial to bipolar configuration transition of LC. The change in the configuration was proposed to be caused due to the perturbation of intermolecular interactions between PLL and the LC in the presence of Annexin V. In this context, PLLfunctionalised LC-aqueous interfaces can be of great utility as the director configuration of the LCs can allow the optical investigation of non-specific protein adsorption at those interfaces. In this paper, first, we aim to study how PLL-coated LC droplets can optically respond to the various anionic proteins (with varying electronegativities). Second, we sought to determine how their respective binding with PLL would affect the detection limits for those proteins at the interfaces. Therefore, we extend this tool to probe, in detail, the adsorption of different anionic proteins at PLLfunctionalised LC-aqueous interfaces by monitoring director configuration transitions in LC droplets.
In order to elucidate the protein-induced director configuration transition in PLL-modified LC droplets, three proteins: bovine serum albumin (BSA), concanavalin A (ConA) and cathepsin D (CathD) were studied which not only differ significantly in their respective electronegativities at physiological pH but also span a broad range of functions in biological systems [25][26][27][28][29][30][31]. We have chosen these three proteins because of their inherent anionic nature and hypothesised that possible formation of electrostatic complexes between PLL and those proteins may lead to an ordering transition of the LC at aqueous-LC interfaces. In this regard, the physiochemical properties such as isoelectric point (pI) of these proteins are relevant to predict the net charge on them at a particular pH (see Table 1).
For example, at physiological pH (7.4), all proteins are negatively charged with BSA being most anionic followed by ConA and CathD, respectively, as confirmed by zeta potential measurements (Table 1). Thus, the zeta potential values at pH 7.4 are in good agreement with their respective pIs [25,30,31]. As PLL being positively charged, we, therefore, sought to understand the fundamental insight into the formation of charged complexes in the presence of these anionic proteins and how they couple to the ordering of the LC at those interfaces. In addition, keeping in mind the wide ranges of functions these proteins serve, it becomes essential to understand the mechanism of protein adsorption at polymeric surfaces for various biomedical applications.
Literature survey few reports wherein LC confined within planar interfaces has been employed for detection of BSA at aqueous interfaces [32][33][34][35]. Recently, adsorption of BSA on polyelectrolyte-modified radial LC droplets was demonstrated to trigger the director configuration transitions to bipolar with a detection limit of 10 μg/mL of BSA [21]. However, the study lacks to offer the real-time monitoring of protein detection along with the higher observation time (30 min) limiting their application in biosensing. On the other hand, there is only one report concerning the detection of ConA utilising LC-aq planar interface where specific binding of ConA with saccharide leads to the reorientation of LCs at the interface [35]. Although their LC system is capable of detecting ConA up to 0.01 μg/mL, the longer response time (~2 h) and the ambiguous optical appearance of LCs at a lower concentration of ConA imply the difficulty in its quantification and detection at those interfaces. As far as CathD is concerned, no effort has been made yet to understand the adsorption of CathD on the orientational behaviour of decorated LC interfaces.
Herein, we observed that radial director configuration of PLL-LC could be triggered to bipolar/pre-radial due to the presence of these anionic proteins for up to 100 ng/mL of concentration and showed varying detection limits for the three studied proteins. The difference in detection limits and response times of these proteins can be attributed to the varying response of director configuration transition of LC due to the different electrostatic binding affinities of a particular protein with PLL which is largely determined by anionic charges present on that particular protein. Overall, the results would aid in understanding the interfacial electrostatic binding conduct of proteins at decorated LC-aq interfaces which would lead to the development of principles for various interfacial and biosensing applications.

Results and discussion
Optical imaging of interactions of PLL-modified 5CB droplets with BSA, ConA and CathD The adsorption of PLL on 5CB droplets stabilised the droplets in aqueous media and all the poly-dispersed PLL-5CB droplets exhibit radial director configuration under polarised microscope (POM) (Figure 1(a)) and show a point defect in centre when observed under bright field (BF), which is a result of homeotropic surface anchoring of 5CB at LC-aqueous interface [23]. It is important to note here that PLL-5CB droplets tend to settle down on the glass substrate and can have weaker non-covalent interactions with glass; however, this interaction does not affect the radial director configuration at least within the observation period (15 min). Next, we sought to observe the orientational response of PLL-5CB droplets towards the BSA at physiological pH 7.4. Interestingly, when PLL-5CB droplets were exposed to 1 mg/mL BSA solution (as shown in Figure 1(b,c)), the initial radial configuration of the droplets starts turning into bipolar/preradial in <2s. However, it took approximately 5 s for all the droplets to change their director configuration from radial to bipolar/pre-radial. The radial to bipolar/ pre-radial configuration transition can mainly be attributed to the adsorption of BSA at LC droplets surface due to ionic interactions between PLL and BSA. Figure 1(d) illustrates the schematic corresponding to the change in the director profile of PLL-coated LC droplets due to the interaction of BSA with PLL. In order to clearly elucidate the BSA-triggered director configuration transition in PLL-5CB droplets, fluorescence microscopy was carried out using FITC-labelled BSA. Epifluorescence microscopy image in Figure 2(a) shows the presence of strong fluorescence on the surface of the 5CB droplets which confirms the preferential adsorption of FITC-BSA on PLL-5CB droplet, while the corresponding BF image (Figure 2(b)) confirms the bipolar director configuration in the particular 5CB droplet.
In our previous report, we showed by PMIRRAS measurements that PLL could induce the homeotropic ordering in LC due to intermolecular hydrogen bonding between NH 3 + of PLL's side chains and CN of 5CB LC [23]. In the present case, we decorated LC droplets with a single layer of PLL which imparts radial director configuration of LC and can be attributed to the intermolecular interactions between PLL and LC as suggested earlier [23]. Here, we propose that the adsorption of anionic protein likely perturbs the intermolecular interactions between PLL and LC due to strong ionic interactions between protein and PLL and thus reorients the LCs in a bipolar/pre-radial configuration (as earlier shown in Figure 1(d)). Another plausible and least feasible mechanism could be that protein-PLL interaction can displace the PLL away from the droplets surface giving rise to the configurational transition. In that case, the LC droplets will be not stable and will coalesce within a few hours [9]. However, BSA-adsorbed PLL-LC droplets have been visually found to be stable against coalescence in aqueous solutions for a minimum of 3 days ( Figure S3), therefore minimising the possibility of this mechanism.
Next, we sought to observe the real-time examination of the PLL-5CB droplets while varying the concentration of BSA (C BSA ), keeping the volume of PLL-5CB droplets and observation time constant (̴ 5 μL and 15 min, respectively). We find that upon decreasing the concentration of BSA, the number of PLL-5CB droplets showing director configuration transition from radial to bipolar/preradial also decreases. For example, upon subsequent exposure to C BSA of 40 μg/mL, all the PLL-5CB droplets changed their director configuration to bipolar/predial (Figure 3(a,b)), whereas when C BSA was reduced to 20 and 10 μg/mL simultaneously, approximately 65% and 40% of the PLL-LC droplets changed to bipolar/preradial ( Figure 3(c-f)). In presence of C BSA ranging from 10 μg/mL to 1 μg/mL, a mixture of radial/pre-radial /bipolar exists with a decreasing number of pre-radial /bipolar droplets as the concentration decreases. Moreover, when PLL-5CB droplets were exposed to 0.5 μg/mL of BSA, PLL-5CB droplets did not show the director configuration transition from radial to bipolar or pre-radial within 15 min (Figure 3(g,h)).
Next, we sought to modulate the sensitivity of PLL-5CB droplets towards BSA by decreasing the volume of PLL-5CB emulsion from 5 to 1 μL ( Figure S4). It was observed that only 20 μg/mL of BSA is required now to induce the configuration transition in all the PLL-5CB droplets. Similarly, 10, 1 and 0.1 μg/mL of BSA are able to change approximately 70%, 55% and 35% of the total LC droplets to pre-radial/bipolar, respectively. For a concentration of BSA ranging from 0.1 to 0.01 μg/mL of BSA, LC droplets could still change the director configuration to bipolar/pre-radial but in a very small fraction of the droplets which are difficult to report. Below 0.01 μg/mL BSA, all the PLL-LC droplets remain radial. Therefore, the sensitivity of PLL-5CB droplets for the detection of BSA was found extremely modulated by reducing the volume of the emulsion. The difference in the detection limit of BSA upon varying emulsion volume can be contributed to the two factors here, first is the reduced volume of PLL-LC emulsion which results in the lesser number of droplets; therefore, the amount of BSA required to trigger the configurational transitions of PLL-5CB droplets is consequently decreased; second is that the reduced volume of emulsion itself decreases the dilution of BSA, thus ultimately raises the available amount of BSA to adsorb at the PLL-LC droplets.
Motivated by the high sensitivity of director configuration in PLL-5CB droplets towards adsorption of BSA, next we sought to exploit the PLL-5CB droplets for detection of ConA. Since ConA overall possesses a negative charge at physiological pH, it can be predicted that ConA can also interact electrostatically with interfacial cationic PLL to trigger a configuration transition in LC droplets. As observed from POM and BF images in Figure 4(a,b), introduction of 1 mg/mL ConA on PLL-5CB droplets could also cause the LC director to move to the bipolar/ pre-radial configuration in PLL-5CB droplets in ̴ 5 s. However, it takes ̴ 20 s after addition of 1 mg/mL of Con A to induce the bipolar/pre-radial configuration in all the droplets which are slightly higher than the response time required by 1 mg/mL C BSA (̴ 5 s). This configuration transition in the presence of ConA can also be attributed to the non-specific adsorption of ConA on PLL-LC droplets via ionic interactions which are also evident through the fluorescence microscopy study ( Figure S5). Next, we sought to monitor the director  configuration of PLL-5CB droplets upon decreasing the concentration of ConA while keeping the emulsion volume 5 μL. The number of LC droplets exhibiting director configuration transition to bipolar/pre-radial droplets also decreases with a decrease in C ConA and radial, bipolar/pre-radial droplets coexist. For example, in the presence of 0.75 mg/mL ConA, approximately 70% of the LC droplets showed the director configuration transition to pre-radial or bipolar while remaining droplets retained radial (Figure 4(c,d)). Similarly, 0.5 mg/mL ConA could change the director configuration in only 45-50% of the droplets. For the concentration of ConA ranging from 0.5 to 0.1 mg/mL, only a small fraction of droplets (40% to 20%) showed bipolar/pre-radial configuration. Below 0.1 mg/mL of ConA, the PLL-LC droplets remained radial ( Figure S6). Next, the detection limit of PLL-LC droplets towards ConA was manipulated by decreasing volume of PLL-LC emulsion from 5 to 1 μL. It was realised that PLL-5CB droplets now can respond to lesser C ConA via director configurational transitions ( Figure S7). Now, 0.5 mg/mL of ConA could induce the transition in more than 95% of the total LC droplets. Upon varying the concentration of ConA to 0.1 mg/mL and 0.05 mg/mL could trigger the director configuration transitions to bipolar/pre-radial in approximately 70% and 55% of the droplets, respectively. Up to 0.01 mg/mL, there were some small fraction of bipolar/pre-radial droplets observed; however, it becomes  difficult to analyse the droplets with less transition percentage. Moreover, the PLL-5CB droplets could not respond to 0.5 μg/mL of ConA and remained radial even after 15 min of addition. Overall, it is evident that when emulsion volume is 5 μL, at least 0.500 mg/mL of ConA was necessary to trigger the radial to the bipolar/ pre-radial configuration in approximately 50% of the PLL-5CB droplets, while this value reduces to 0.05 mg/ mL when emulsion volume is reduced to 1 μL. These results show that similar to BSA, the detection limit of ConA can also be amended by varying the volume of LC emulsion. Inspired the exceptional ability of PLL-modified LC droplets to report the non-specific binding to anionic proteins BSA and ConA, we sought to explore the possible interfacial interactions between PLL and CathD via orientational transitions of PLL-5CB droplets. Zeta potential measurements reveal that CathD carries overall a negative charge at pH 7.4, yet small as compared to the other two studied proteins. Figure 5 shows the POM and corresponding BF images of PLL-5CB emulsion (5 μL) after exposure to a different concentration of CathD. Interestingly, CathD also induces radial to bipolar/preradial configuration transition of PLL-5CB droplets; however, it takes 10 s to initiate the configuration transition and requires a higher concentration of CathD (of 1.5 mg/mL) as compared to BSA and ConA to induce the configuration transition in all droplets. Careful observations reveal that it required at least 1.2 mg/mL of CathD to induce radial to bipolar/pre-radial transition in 100% of the droplets. Further decreasing the C CathD to 1 mg/mL and 0.5 mg/mL leads to the change in LC droplets from radial to bipolar in approximately only 90% and 50% of the total LC droplets, respectively. Upon reducing the volume of LC emulsion to 1 μL, the detection limit of CathD was found to be reduced to 0.25 mg/mL with a transition in approximately 50% of the LC droplets.

µm
Zeta potential measurements of PLL-5CB droplets upon adsorption of BSA, ConA, CathD Since orientational transitions in PLL-5CB droplets are prominently taking place through non-specific binding of anionic proteins and interfacial PLL residues, we carried out a series of experiments to realise the zeta potential measurements of PLL-5CB droplets and protein-adsorbed PLL-5CB droplets ( Figure 6). It is not surprising that zeta potential of PLL-5CB droplets is positive (+42 mV), mainly due to the adsorbed cationic lysine units of PLL at LC-aqueous interface. However, upon adsorption of BSA, ConA or CathD at PLL-5CB droplets, the zeta potential values become negative and drop to −17.5, −9.4 and −3.2 mV, respectively, confirming the adsorption of anionic proteins over cationic PLL-5CB droplets. We propose that more the anionic charge density offered by protein, elevated the possibility of ionic interactions between protein and cationic PLL residues and faster and significant the response of LC director configuration in LC droplets. We corroborated the trend between electronegativities of proteins and response of PLL-LC droplets by using two other proteins: human haemoglobin (pI is 6.87) and fibronectin (pI is 5.5-6.0) as both possess overall negative charge at pH 7.4. Although the study with fibronectin is under progress, we report the results with haemoglobin in Figure S8. It was found that 250 μg/mL of haemoglobin could trigger ordering transition in approximately 40-50% of the total PLL-LC droplets (volume of emulsion = 1 μL), while droplets remain radial when 100 μg/mL haemoglobin was added. The zeta potential of haemoglobin-coated PLL-LC droplets was −2.6 mV. The zeta potential value and the percentage transitions for haemoglobin are in good accordance with that of CathD.

Selectivity of PLL-5CB droplets for anionic proteins
Having established that PLL-modified LC droplets offer a stimuli-responsive interface that allows the adsorption of anionic proteins via non-specific binding to PLL, we sought to realise the relevance of ionic interactions between the two binding moieties and thus proposed that a cationic protein at pH 7.4 would not trigger the director configuration transition in PLL-5CB. As a proof of concept, we chose lysozyme, a single-chain polypeptide of 129 amino acids, as its isoelectric point is quite high (̴ 11.3) and therefore lysozyme possesses a net positive charge at physiological pH [36,37]. Figure 7 shows the POM images of PLL-5CB droplets before and after exposure to 1 mg/ mL lysozyme and apparently, the presence of lysozyme does not affect the director configuration of LC droplets suggesting that cationic PLL restricts the adsorption of cationic lysozyme at PLL-5CB droplets surface and thus retains the radial configuration in PLL-5CB droplets. Although additional non-covalent interactions between proteins and PLL residues cannot be ruled out, it is evident that ordering of LC droplets is largely influenced by attractive ionic interactions between anionic protein and cationic PLL. Therefore, we note that director configuration of PLL-5CB droplets is triggered selectively by the adsorption of anionic proteins and hence can be exploited to as an optical, label-free, fast and sensitive platform for the selective detection of anionic proteins over cationic proteins at physiological pH.
From the above results, it is clear that the ordering transition in PLL-LC droplets is not only predominantly triggered by negatively charged proteins but also differ significantly for the three studied proteins in terms of their low detection limits and response time. Such a difference in sensitivity of LC droplets towards various proteins could be attributed to the anionic density possessed by a protein molecule in aqueous solution as suggested from zeta potential measurements. Below, Table 2 summarises various sensitivity aspects associated with the PLL-5CB droplets for the detection of BSA, ConA and CathD.   10 30 a of protein-adsorbed PLL-LC droplets (mV); b of protein for >35% transition (μg/mL); c of LC to initiate the transition (radial to bipolar/pre-radial); d of LC for 100% transition (radial to pre-radial/bipolar) (1 mg/mL protein).

Conclusions
We have demonstrated the implication of director configurations of LC droplets to report the real-time adsorption of anionic proteins such as BSA in few seconds, primarily driven by electrostatic interactions between protein and cationic PLL, as confirmed by POM, bright field and fluorescence microscopy. Additionally, it is demonstrated that director configuration in PLL-LC droplets can respond to ConA, a lectin protein, and CathD, a tumour marker, within 1 min of response time and these proteins can be quantified with detection limits of 0.1, 50 and 250 μg/mL for BSA, ConA and CathD, respectively. Overall, we found that sensitivity and response time of PLL-5CB droplets differ for the three proteins and follow the similar order as the anionic charge density possessed by these proteins after adsorption on PLL-LC droplets as confirmed by zeta potential measurements. Director configuration transitions in PLL-5CB droplets are found more fast and responsive to BSA, the most anionic protein among three, followed by ConA and CathD (slightly lesser anionic than BSA). Overall, PLL-5CB droplets provide a responsive LC interface that enables a fast, label-free, quantitative platform to optically and selectively detect real-time adsorption of negatively charged proteins and hence can find promising application in biosensing and interfacial applications.