Introduction

Polyvinylpyrrolidone (PvP) has many applications [1, 2]. These arise from its implicit non-ionic polymer [3] properties, apparent inert behaviour and lack of activity towards biological organisms and material [2]. As a result, PvP has been widely used for stabilising nanomaterial dispersions [4, 5] especially where they have a medical use in applications like drug delivery [6]. Its stabilising property comes from the fact that it adsorbs on and coats nanomaterial and, being neutrally charged, minimises aggregation within the material dispersions [4, 5]. PvP consists of monomer units of N-vinylpyrrolidone (N-vP) [2] (Fig. 1). The tendency of N-vP to form non-covalent interactions with biological material will be small since the monomer units have one H- bond acceptor as the carbonyl bond and a log P value of 0.37. Log P is defined as the log octanol–water partition coefficient (log KOW) of a compound. This is a thermodynamic quantity relating to the partitioning of a compound between octanol and water. It has a standard use for defining the lipophilicity of pharmaceuticals and toxicants [7]. Commensurate with the relatively low log P value of the PvP monomer, PvP is hydrophilic and water soluble, and in aqueous solution, it remains in a coiled configuration [3]. The PvP polymer adsorbs on surfaces [8,9,10] and this property, and its interaction with other surfactants [3] and compounds in solution [11] has been variously studied in particular in relation to the adsorptive behaviour of polymers in general. In fact generally the adsorption of polymers on surfaces has attracted a lot of interest [12,13,14,15,16,17,18]. Many studies agree that polymer adsorption is entropically driven [19]. The main mechanism here is the release of bound water molecules from the solution polymer and the adsorbate surface during the adsorption process[19]. In the case of the phospholipid surface, this effect could be specially significant in view of the large amount of water molecules associated with the phospholipid polar groups [20] some of which will be displaced. This entropy gain will generally override the entropy loss when the polymer transforms from a 3D to a 2D environment on the adsorbate surface [21,22,23,24]. PvP presents an interesting case of polymer adsorption in that the possibility of only selected segments adsorbing leading to a looped or bunched configuration on adsorption can occur [17].

Fig. 1
figure 1

Structures of a N-vinylpyrrolidone (N-vP) and b polyvinylpyrolidone (PvP)

This communication reports a preliminary study on the interaction of PvP with supported phospholipid layers as a function of its chain length. The work has been initiated since many nanomaterial dispersions are coated with PvP [6, 8, 10, 16]. It is instructive therefore to have some knowledge of the effect of the PvP coating on the interaction of the nanomaterial with the lipid layer and indeed its interaction with biological targets. However, as a first step, it is necessary to know how the PvP itself interacts with lipid layers. It is especially important to do this since to date, there have been very few investigations on the interaction of PvP with lipid vesicles [25, 26] and lipid monolayers [27], and on the interaction of Ag nanomaterial coated with PvP with lipid monolayers [28].

Materials and methods

Five samples of PvP of molecular weights, 3.5, 10, 55, 360 and 1300 kD mole−1, were obtained from Sigma-Aldrich, and stock solutions were prepared for electrochemical analysis in 18.2 M.Ωcm Milli-Q water. Each PvP sample is characterised throughout the text by the number of monomer units obtained through dividing the molecular weight of PvP by that of N-PvP of 111.14 g mole−1 and expressed as PvPm, i.e. for the five PvP samples as 31, 90, 495, 3239 and 11,697, respectively. The electrolyte used in the electrochemical experiments was 0.0138 mol dm−3 NaCl and 0.00027 mol dm−3 KCl buffered at pH 7.4 with 0.00119 mol dm−3 phosphate (hereinafter in the text referred to as PBS). The PBS was of analytical grade and purchased from Sigma-Aldrich. The microfabricated platinum electrodes (Hg/Pt) used in the electrochemical assay [29, 30] were supplied by the Tyndall National Institute, Ireland. The dioleoyl phosphatidylcholine (DOPC) was obtained from Avanti Polar Lipids Alabaster, AL, USA and was > 99% pure. The DOPC dispersion for electrode coating was prepared by gently shaking DOPC with PBS to give a 0.25 µmole cm−3 dispersion. All other chemicals and reagents were of analytical grade and purchased from Sigma-Aldrich.

Apparatus and procedure

For the assay, the fabricated Hg/Pt electrode was contained in a flow cell consisting of a microfluidic flow cell containing the DOPC monolayer supported on a Hg sensing electrode, four automated bespoke syringe pumps enabling storage and transportation of fluids (electrolyte, test sample, phospholipid and water) into the flow cell, a field-programmable gate array (FPGA) data acquisition and control unit used to interface between software and hardware and an ACM Research Potentiostat for electrochemical measurements. A laptop was connected to control the screening platform, interfacing with syringe pumps and the FPGA control unit. The microfabricated electrode was prepared in advance by cleaning in a 1 mol dm−3 solution of NaOH in methanol, followed by HCl and Milli-Q water and then dried. Hg was manually deposited on the Pt disc of radius 0.480 mm to give a Hg/Pt electrode. The electrode was mounted as specified in Owen et al. [29]. Subsequently, all samples were deoxygenated (PvP solution, DOPC, electrolyte) with argon gas (Air Products) for a minimum of 30 min. Once purged, three syringes were filled with PvP sample (5 mL), DOPC dispersion (60 mL) and PBS (60 mL), respectively, and connected to tubing. Before any analysis, all tubing was flushed with deoxygenated PBS, and any bubbles were removed from the cell. Turning the potentiostat to run, the system was set to (i) clean with the electrochemical rejection of the previous used monolayer, (ii) deposit DOPC from dispersion, (iii) test the monolayer integrity in PBS and (iv) screen the sample solution as described previously in refs [29, 30] and in Table S1 in the SI. Upon single-sample completion, the sample tubing was flushed with PBS (5 mL). This was the analytical cycle for each sample. Samples were measured at increasing concentrations of one PvP sample, then switching to the next PvP sample. The sample syringe was replaced with every repeat of the same sample and between PvP sample solutions. All measurements were carried out in triplicate, and five PvP concentrations were screened for each PvP chain length sample. All fits to the data were carried out using the program IGOR Pro 9, and coefficients and their errors (SD%) for all fits are listed in Tables S2 and S3 in the SI.

Results and discussion

The system of a phospholipid monolayer adsorbed on a mercury electrode as a biomembrane model has been developed over four decades [31,32,33]. It has had fundamental biophysical applications for example in analysing ion channel [34], co-enzyme activity [35] and phospholipid behaviour in electric field [36,37,38]; however, its predominant practical implementation has been used in modelling the biomembrane activities of molecular [39,40,41] and nanoparticle species [41, 42]. Originally, a modified Langmuir–Blodgett technique was used for depositing the phospholipids on a hanging mercury drop electrode [32, 33]. Since these techniques were totally inappropriate for rapid and routine screening, the electrode was re-configured as a microfabricated Hg on Pt film electrode [29, 30, 43], and the phospholipid deposition was enabled from vesicles in a flow cell [29, 30]. In this study, a DOPC monolayer is deposited on the Hg electrode on the prepared Pt support and scanned at 40 Vs−1 from − 0.4 to − 1.2 V referred to throughout the text as rapid cyclic voltammetry (RCV) [29, 30, 43] (see Table S1 in the SI). The layers undergo potential-induced phase transitions characterised by two sharp capacitance current peaks (voltammetric), 1 and 2, respectively, as shown in Fig. 2 [29, 30, 43]. These two peaks correspond to the penetration of electrolyte into the layer and the reorganisation of the monolayer to form bilayer patches, respectively [36,37,38]. Changes in these capacitance peaks represent changes in the structure of the monolayer [29, 30, 43]. The interaction of the test substance with the monolayer selectively and systematically influences the capacitance-current potential profile [39,40,41,42]. An interaction of the test substance with the polar groups of the DOPC is reflected in a depression of the two peaks [39, 40] while an increase in the baseline of the capacitance current reflects the association of a polar compound with the apolar region of the DOPC layer and/or its disruption [38,39,40,41]. The reason for the latter effect is that the low value of baseline capacitance current is representative of the ordered DOPC layers on the electrode with the low dielectric apolar lipid tails adjacent to the electrode surface. When this low dielectric region is penetrated by a higher dielectric compound, the average dielectric constant of this region increases leading to an increase in the baseline capacitance current [39,40,41]. A potential shift in the capacitance current peaks indicates a change in the potential profile across the layer caused by the interaction of the compound with the layer [40, 44]. A monolayer disordering is shown as a broadening of the peaks [40, 41]. The screening results from this sensor platform have recently been shown to be related to biomembrane damage in in vitro cell cultures [45]. Other research groups have followed a similar approach, but not in rapid online screening format [46,47,48].

Fig. 2
figure 2

RCVs at 40Vs−1 of DOPC on microfabricated Hg/Pt electrode in PBS at pH 7.4 (black line) and with added a 5 and b 0.18 µmole dm3 of PvPm 31 and 3239, respectively (red line). Capacitance current peaks 1 and 2 labelled on (a)

The RCV plots in this study show a PvP interaction with the DOPC monolayer as a depression of the RCV capacitance current peaks (see Fig. 2). This peak depression directly corresponds to the presence of adsorbed species on the monolayer surface [42], and the extent of peak depression is linearly related to the coverage [42, 49]. In order to obtain a quantitative estimate of the effect of each compound on the DOPC layer, affinity constants (K2) and limits of detection (LoD) for PvP in PBS are estimated from the PvP calibration curves. The configuration of the RCV following the interaction displays exactly the same change in form with increasing PvP concentration irrespective of the PvP chain length. Significantly, with increasing PvP solution concentrations, the interactions showed the RCV capacitance peak currents as successively depressed with no potential shift, no peak broadening and no increase in the capacitance baseline current. The absence of an increase in the capacitance minimum current indicates that the PvP does not penetrate the DOPC layers with the interaction remaining at an adsorptive superficial level on the DOPC layer. Further evidence for this comes from the configuration of the calibration curves (capacitance peak current 1 suppression % versus PvP concentration) which are linear for lower PvPm values of 31 and 90 but Langmuirean [50] for higher PvPm values of 495, 3239 and 11,697 as shown in Figs. 3 and S1. This shows intuitively that PvP surface coverage becomes limiting at longer chain lengths.

Fig. 3
figure 3

Plots of % capacitance current peak 1 suppression versus solution concentration of PvP with LoD value indicated on horizontal axis a 90 and b 3239 PvPm, slope ‘c’ indicated by blue stippled line

The linear plots are fitted to Eq. (1):

$$Y=a+{\text{c}X}$$
(1)

The Langmuirean plots [50] are fitted to Eq. (2):

$$Y=a+\left[{\text{b}X}/\left(100+{\text{c}X}\right)\right]$$
(2)

Y is the % current peak depression, X is the solution concentration of PvP and ‘a’ is the intercept due to some depression of the control current peak during the assay. The slope ‘c’ is shown on the calibration plots in Figs. 3a, b. The coefficients and their errors are displayed in Table S2 in the SI. Although the fits are good and the coefficients are reasonable, there are large errors on one of the calibration points of the 11,697 PvPm polymer. Since this communication is only focused on the sensor element response to the lower PvP solution concentrations as a molecular initiation event (MIE) [51], the response plots at higher PvP concentration are not of concern here and will be studied more comprehensively in further work. The LoD metric of the PvP affinity for the DOPC monolayer is the lowest significant solution concentration of PvP, which can structurally modify the DOPC layer and is estimated from 3 times the standard deviation of the DOPC capacitance peak 1 current corresponding to the PvP solution concentration on the appropriate calibration curve [39, 40]. The LoD error is taken from the error of the slope coefficient for the linear plot and the errors associated with both coefficients for the Langmuir plots, respectively. The relation of the LoD to the calibration curves is shown in Fig. 3. The metric K2 is defined as the slope ‘c’ of Eq. (1) in Fig. 3a and the slope ‘c’ of Eq. (2) in Fig. 3b both divided by 100 to give the fractional depression of the capacitance peak current 1 per unit PvP solution concentration. K2 is taken to represent the affinity constant [50] of PvP for the DOPC surface. This assumption is justified from the experimental data since log K2 linearly correlates with − log LoD for all polymer chain lengths (see Fig. 4a), and − log LoD has been an established measure of the affinity of a compound for the phospholipid layer [39, 40]. The empirical equation for the relation between log K2 and –log LoD can be extracted from the fit in Fig. 4a as:

Fig. 4
figure 4

a Plot of log K2 versus − log LoD values derived from % capacitance current peak 1 suppression versus solution PvP concentration calibrations and b plot of log K2 versus PvPm (red circles) fitted to Eq. (5) (black line), Eq. (5) terms: exp (1.7 × 10−4.PvPm) (red dashed line) and 3.58.PvPm0.037 (blue stippled line), versus PvPm

$${\text{log}}\,{K}_{2}=-0.91-1.03.{\mathrm{log\,LoD}}$$
(3)

The following can be derived from the linear relation between log K2 and − log LoD:

$${K}_{2}=1/\left(8.1{{\text{LoD}}}^{1.03}\right)$$
(4)

The relationship of K2 with the free energy of adsorption (ΔG) will ideally follow the formal equation, ΔG = − RT ln K2 [50] so log K2 is an effective metric relating to the adsorption energy and is used as such in this study.

Plots of log K2 versus PvPm are displayed in Fig. 4b and show an increase which is steepest for the shortest PvP chains and is curvilinear for the longer chains. The plot can be empirically fitted to the following:

$${\mathrm{log}}\,\mathrm{{}}K_{2}={\text{exp}}\left(1.7\times {10}^{-4}\times {{\text{PvP}}}_{{\text{m}}}\right)+3.58.{\text{Pv}}{{{\text{P}}}_{{\text{m}}}}^{0.037}$$
(5)

and similarly for − log LoD:

$$-{\mathrm{log\,LoD}}={\text{exp}}\left(1.6\times {10}^{-4}\times {{\text{PvP}}}_{{\text{m}}}\right)+3.91.{\text{Pv}}{{{\text{P}}}_{{\text{m}}}}^{0.037}$$
(6)

Interestingly, the LoD for a single unit of N-vinylpyrrolidone (N-vP) can be estimated from Eq. (6) as 12.3 µmole dm−3 which is reasonable for such a single-saturated ringed molecular structure as reported in previous publications [39, 40]. If each monomer unit contributes a linear increase to the affinity of the combined polymer for the DOPC, a constant value of log K2 per monomer number in the polymer would be expected. In fact if (log K2)/PvPm is plotted against PvPm (Fig. 5a), it exhibits a near reciprocal decrease which fits Eq. (5). Significantly, Eq. (5) fits the data better at higher PvPm than a reciprocal fit shown in the inset to Fig. 5a. This decrease in adsorption energy for individual units with increase in chain length can indicate two effects:

  1. (1)

    Only selected monomer units are adsorbing on the DOPC surface allowing sections of the chain to remain attached as loops [17] but not adsorbed.

  2. (2)

    The length of the chain affects the affinity of a monomer unit for the surface.

Fig. 5
figure 5

a Plot of (log K2)/PvPm versus PvPm (red circles) fitted to rearranged Eq. (5) (black line) with inset of same plot at high PvPm showing reciprocal fit [(log K2)/PvPm = 3 × 10−4 + 4.1.PvPm−0.95] to plot as blue dashed line. b Plot of log (K2/PvPm) versus PvPm (red circles) fitted to Eq. (5) − log PvPm (black line) with terms: exp (1.7 × 10−4.PvPm) − log PvPm (red dashed line) and 3.58.PvPm0.037 − log PvPm (blue stippled line), versus PvPm

If the first factor is critical, it can lead to a near reciprocal decrease in log K2 with PvPm as the ‘looping’ of chains at the surface becomes more extensive proportionately with the chain length. In terms of the second factor, it is instructive to take a preliminary look at the thermodynamics of the adsorption in order to further understand the form of the log K2 vs PvPm relationship.

Eqs (5) and (6) show that the affinity of the PvP for the DOPC is divided into two terms. These are plotted out in Fig. 4b for log K2. These terms fit a model whereby a combination of entropic and enthalpic forces drive the polymer adsorption. The exponential term may be identified with the enthalpic contribution of this model showing an increase from a small value to a higher value at long chain lengths. The small value at short chain lengths is consistent with the low propensity of the PvP monomer unit to bind with the DOPC surface having only one H-bond acceptor group which would interact with the positively charged DOPC choline moiety and a low log P value of 0.37 which reflects a weak tendency of the molecule to partition into an organic phase such as lipid. The exponential increase in this term with chain length reflects the increasing tendency for the PvP chains to interact with each other on the DOPC surface as the chains become longer. Such an interaction is consistent with the increasing entanglement of ‘looped’ chains on the DOPC surface as the chains become longer and is commensurate with the finding that solution PvP has a propensity for self-assembly [52]. The power term in Eqs. (5) and (6) can represent the entropic contribution to the adsorption model due to dissociation of water molecules from PvP and from the DOPC surface [20] following PvP adsorption. This necessarily increases with increasing chain length but at longer chain lengths, it levels off due to the increasing relative significance of the entropy loss due to change in PvP configuration from 3 to 2 dimensions.

The PvP adsorption effects can be seen more clearly if the K2 value is divided by PvPm. Conceptually, this normalises the adsorption affinity constant to the PvP’s monomer segment instead of the polymer. Figure 5b shows a plot of log (K2/PvPm) as a function of PvPm. A steep decrease in log (K2/PvPm) with PvPm is observed followed by a curvilinear increase at longer chain lengths. Equation (5) − log PvPm fits the data and the separate terms minus log PvPm, respectively, have been plotted versus PvPm. Both terms show steep decreases at lower PvPm, but only the exponential term increases at longer chain lengths. These results fit the model whereby both the monomer normalised entropic and enthalpic contributions are significantly affected by the ‘looping’ of the polymer on the surface at shorter chain lengths. At longer chain lengths, the enthalpic contribution from the association of polymer on the DOPC surface becomes significant. The results and the model proposed in this work are consistent with the state-of-the-art findings on polymer adsorption where the adsorption is an interplay between entropic and enthalpic forces [53,54,55].

Conclusions

Aqueous PvP interacts with electrode-supported phospholipid layers of DOPC. This interaction is strongly dependent on chain length. For shorter chain lengths, the interaction is linearly related to PvP solution concentration whereas at longer chain lengths, the interaction is Langmuirean; however, the interaction voltammogram in all cases is representative of an adsorption process. Both the affinity constant, K2, and the limit of detection, LoD, are extracted from these plots, and log K2 and − log LoD are directly related to each other. Plots of log K2 and − log LoD versus the monomer segment number, PvPm, fit a two-term equation consisting of a power term and an exponential term. A plot of (log K2)/PvPm versus PvPm is near reciprocal showing that there is ‘looping’ of the chains on the DOPC surface during the adsorption process. The results fit a model whereby the adsorption is entropically driven at short chain lengths but at longer chain lengths, non-covalent interactions between the chains on the DOPC surface can assist in promoting adsorption.