Introduction

Ganglioside GM1 is an important effector responsible for many important cellular functions [1, 2]. It is mainly resident in the lipid-ordered regions of the cell membrane rich in CerPCho and C called lipid rafts, whose environmental effects may have important pharmacological implications on the molecular functional properties [3–6]. In order to best replicate the ganglioside functionality, its inglobation in artificial BLM biomembranes composed by CerPCho and C is thus a necessary prerequisite. So far, GM1-enriched BLM have been fabricated in the form of unilamellar vesicles [7, 8], but there is very little in the literature reporting on the formation and behavior of GM1-rich CerPCho/C BLM fabricated on solid-supports. In particular, the solid-supported bilayer lipid membranes (ssBLM) morphological structure, and the ganglioside organization have been locally investigated via atomic force microscopy and fluorescence microscopy [9, 10], but important properties of GM1, such as the binding affinity, could not be achieved with such tools. Recently, the SPR technique was exploited to recognize the presence of GM1 in GM1/CerPCho/C domains demixed from GM1/DOPC/CerPCho/C biomembranes via the binding of the ganglioside with the β-subunit of the cholera toxin (CTB) [11], but the analysis was restricted to the evaluation of the GM1–CTB association rates. In this work we built artificial GM1/CerPCho/C regions above plasmonic transducers. As in [11], the presence of GM1 was detected via its association with CTB but, in addition to the association rates, here we also measured the dissociation constant of the complex CTB–GM1 and the consequent apparent K D, gaining insight into the functionality of this important ganglioside when resident in a physiological-like environment.

Materials and Methods

The ssBLM:GM1 were assembled by exploiting the fusion of small unilamellar vesicles (SUV) on a SiO2 interface, using the same procedure described in our recent paper on the fabrication of ssBLM:GM3 [12]. The plasmonic transducers were fabricated accordingly.

The stock solutions for the preparation of GM1/CerPCho/C mixtures were obtained in the same way as that used to obtain GM3/CerPCho/C mixtures described in [12]. The resulting ssBLM had the molar composition GM1 x CerPCho0.6 − x C0.4 with x = 0, 0.02, 0.05, 0.1, 0.2 corresponding to 2, 5, 10, 20 mol% GM1 concentration. In the tests reported here, five CTB solutions (4.1, 8.3, 12, 24.5, 35 nM) were generated starting from a stock solution of CTB obtained by dissolving 2 μg of CTB (Sigma Aldrich) in 2 mL HBS.

The SPR spectrometer is a home made instrument described in [12]. For the convenience of the readers we briefly list the measurements steps:

  1. 1.

    Injection of GM1 x CerPCho0.6 − x C0.4 SUV solution in the reaction cell and recording of a first SPR spectrum. The plasma angle θ a (corresponding to the minimum reflectivity) is obtained (Fig. 1 ESM).

  2. 2.

    Recording of the SUV fusion on SiO2 kinetic at the reflectivity setpoint of 0.3 until stabilization occurs.

  3. 3.

    Rinsing three times with HBS and recording of a new SPR spectrum. The plasma angle θ b is obtained.

  4. 4.

    Resetting the reflectivity setpoint to 0.3 and inject the CTB solution.

  5. 5.

    Recording of the CTB association kinetic until the steady state is reached.

  6. 6.

    Rinsing three times and recording of the SPR spectrum. The plasma angle θ c is obtained.

  7. 7.

    Repetition of these steps with GM1-free ssBLM and 35 nM CTB solution.

  8. 8.

    Elimination of the SUV solution from the cell and filling it with HBS, in order to record the CTB dissociation kinetics in the case of [GM1] = 10 mol%.

Results and Discussion

GM1-enriched ssBLM were successfully obtained via SUV fusion with homogeneity and stability similar to those documented in [12] for the fabrication of ssBLM:GM3.

Our SPR measurements indicate a ssBLM thickness of 6.3 nm, that is consistent with a partial filling with GM1 [10], but too close to that, 6.1 nm, measured for pure CerPCho ssBLM [13] to assume that it is a conclusive proof of the enrichment. Thus, the GM1 presence has been inferred more directly via binding tests performed in the presence of CTB, the specific ligand of GM1, put in contact with GM1-rich and GM1-free ssBLM. In the case of GM1-free ssBLM, no significant spectrum shift was found upon CTB addition, indicating that GM1 was selectively recognized. In all the tests, the unaltered shape of the SPR spectra, as visible in the example shown in Fig. 1 ESM, shows good stability and uniformity of the ssBLM in the various processing stages. The measurement of the GM1–CTB dissociation constant was carried out using samples of GM10.1CerPCho 0.5C0.4 ssBLM, and CTB solutions at the concentrations indicated in “Materials and Methods”. The shifts of Δθ CTB = θ c − θ b vs CTB concentration are reported in Fig. 1. The data were fitted to a Langmuir isotherm and the value of K D = (1.5 ± 0.5) nM was calculated for the apparent dissociation constant. This is, to our knowledge, the first measurement of the binding strength of the complex GM1–CTB when GM1 is hosted in a CerPCho/C ssBLM.

Fig. 1
figure 1

Experimental shifts of the resonance angle (squares) vs CTB concentration, corresponding to GM1–CTB binding at [GM1] = 10 mol%. The line (red only in the online version) is the best fit of the data to a Langmuir isotherm, that corresponds to a dissociation constant K D = (1.5 ± 0.5) nM. The error bars (±0.1°) represent the reproducibility of the measurements of the angular minima

Full size image

Since this measurement has 33 % of uncertainty, we thought it is appropriate to measure K D also by recording the association and dissociation kinetics (Fig. 2a, c respectively). The predicted linear trend of the fast association rate constant k on with CTB concentration [14] is confirmed in Fig. 2b. Its linear best fit gives a value of k ass = (1.10 ± 0.04) × 105 M−1s−1, while the average k diss, as calculated by best fitting the kinetics of Fig. 2c, resulted in (2.8 ± 0.1) × 10−4 s−1. As a consequence, the value of (2.5 ± 0.2) nM was found for K D, in reasonable agreement with that found from the Langmuir isotherm fit.

Fig. 2
figure 2

a Experimental association kinetics of CTB to a ssBLM:GM1 (black traces) and their best fit (colored lines in the online version) at various concentrations in HBS (4.1, 8.3, 12, 35 nM). b Association rates and their experimental errors as calculated with the best fit procedure. The GM1 loading of the ssBLM is 10 mol%. Straight line linear best-fit, corresponding to an association constant k ass = (1.10 ± 0.04) × 105 M−1s−1. c Dissociation kinetics (black traces) obtained contacting the ssBLM:GM1 to pure HBS. The superimposed exponential best fit curves are colored only in the online version. The resulting dissociation constant k diss results (2.8 ± 0.1) × 10−4 s−1

Full size image

Considering a weighted average value of K D, namely 2.4 nM, it is worth noting that it is well within an order of magnitude of that (0.46 nM) measured for in vivo cells [15], which indicates a good mimetic action of the ganglioside incorporated into our ssBLM, at least for the GM1 recognition by CTB.

In addition to the 10 mol% GM1 concentration, the selective recruitment of CTB was demonstrated also for [GM1] = 2, 5, and 20 mol%, using a fixed CTB concentration (35 nM). The corresponding Δθ CTB shifts vs the GM1 loadings are reported in Fig. 3.

Fig. 3
figure 3

Experimental steady angular shifts (squares) recorded after the incubation of a 35 nM CTB solution with ssBLM:GM1 at various GM1 loadings. Vertical bars are the reproducibility errors. The line (red in the online version) is a guide for the eye

Full size image

At 10 mol% the angular shift Δθ CTB = 0.47° pertains to a thickness of 3.6 nm, lower than that (4 nm) expected for a complete coverage of CTB in the typical side-on configuration [16]. At 20 mol% it reaches 0.62°, corresponding to 4.8 nm thickness. According to [17] this result indicates that part of the CTB units binds to GM1 in the edge-on configuration with the pentameric ring orthogonal to the surface, and therefore these CTB units do not exploit all the five pockets available for binding. Although the reason for the occurrence of this kind of linkage is not clear, a possible explanation is the availability of a restricted number of separated GM1 headgroups above the ssBLM, indeed a possible result of their clustering due to the ganglioside density [18, 19]. This phenomenon, observed even at 2.1 mol% GM1 concentration in a POPC ssBLM, [19] is further favoured in CerPCho/C ssBLM as recently demonstrated by MD simulations [20].

Binding of CTB to a lower number of sites is responsible for the weakening of the bond strength [7, 19], that can manifest itself in the increase of k diss and/or a decrease of k ass. Actually, a decrease in k ass was accidentally found from the best exponential fit of the CTB association kinetics, reported in Fig. 2 ESM, that have been routinely recorded in order to monitor the stabilization of CTB adsorption on ssBLM:GM1. Indeed, the calculated k on for the GM10.1CerPCho0.5C0.4 ssBLM (2.1 × 10−3 s−1) is about two times higher than that (9.1 × 10−4 s−1) obtained for GM10.2CerPCho0.4C0.4 ssBLM, confirming the presence of a weaker bonding of GM1 to CTB for higher GM1 loadings.

In summary, we have fabricated homogeneous and stable GM1-rich CerPCho/C ssBLM on plasmonic transducers, which allowed us to monitor the dissociation of CTB from GM1 embedded in a physiological-like solid supported environment. The measured value of K D, 2.4 nM, is close to that found for in vivo cells, that indicates the potential of these systems as in vitro biomimicking platforms. In addition, whenever confirmed, the formation of GM1 clusters on these ssBLM will be a further benefit which could be fruitfully exploited to explore diseases, such as Alzheimer's disease, in which the clustering of GM1 plays a crucial key role [21].