Introduction

An important chemotherapy drug, routinely used as anticancer medicine, is 5-fluorouracil (5-FU) [1]. This drug is known to improve the survival rate in various cancer patients [1]. Cancers such as stomach, esophageal, colon, breast, pancreatic, and cervical are all treated with 5-FU [2]. This drug is usually administered via injection into a vein but may also be used as a cream to treat potential cancer-causing skin conditions [3]. Nonetheless, improved delivery methods remain a research interest to both improve the response rate and to reduce the risk of acute liver injury. Nanoparticles (NPs) have shown great potential to be used as drug-delivery vehicles since these may enter a cell and deliver the payload drug directly to the affected region, instead of injecting the patient with the drug that will then affect both healthy and tumorous cells [4,5,6]. Iron-oxide (specifically magnetite Fe3O4) NPs have recently attracted great interest as a potential candidate for drug-delivery vehicles, since this paramagnetic NP may be directed via external magnetic fields to targeted areas in vivo. However, before a drug-delivery system can deliver its payload to the intended cell, it has to cross the cell membrane. Additionally, since an external force may be applied to direct the NPs (i.e., an external magnetic field), this drug-delivery system may unintentionally also interact with healthy cells (in close approximation of the tumorous cells). In some cases, the size of the external magnetic field may be large enough to force an interaction between the cell membrane and the drug-delivery system.

Thus, the interaction with the cell membrane is therefore of great importance since the first barrier to entry for the drug-delivery system is this lipid bilayer. The most numerous membrane lipid molecules are phospholipids which are known to be amphipathic [7]. Thus, whether the NP drug-delivery system is designed to be hydrophilic or hydrophobic (or both) at some point during its entry into the mammalian cell, through the cell membrane, it will interact chemically with the bilayer lipid, and it is therefore important to understand this interaction if the drug (5-FU) is to be delivered successfully.

The most fundamental function of lipid bilayers is to serve as a barrier between a cell’s interior and exterior. This is highly dependent on the continuity of the bilayer structure. Some significant biological processes are dependent on the partial breakdown of the lipid matrix. For example, the movement of small molecules across the membrane (like water), cell lysis, fusion, and even budding events. Therefore, the development of a water channel or pore is a significant event where the lipid matrix is partially destroyed [8]. Some proteins and peptides also have the capacity to form water channels and then regulate the membrane permeability and stability in a specific manner.

It is less well known that ions and protons also penetrate protein-free lipid bilayers at rates that are higher than those predicted by simple diffusion. Therefore, it has been proposed that thermal and mechanical fluctuations can cause water pores to form momentarily in protein-free bilayers [9,10,11]. Such pores could be the starting point for structural flaws connected to phase transitions, cell fusion, and lysis, as well as making it easier for polar molecules to passively cross membranes.

The formation and expansion of water pores in protein-free lipid bilayers have been extensively investigated using pipette aspiration and electroporation experiments. Melikov et al. [12] used model membranes, vesicles, and cells of various sizes and chemical compositions to study the phenomenon of membrane permeabilization under strong electric fields and/or mechanical stress [13, 14]. It is not entirely clear how pores form as a result of electrocompressive stress. But it has been shown consistently that a critical transmembrane potential exists where the membrane ruptures. It is thought that when the pore enlarges past a critical radius, irreversible membrane breakdown (rupture) takes place. Spontaneous resealing will happen when the radius falls below a critical level. It has been shown that this critical level is 0.7 nm [8]. Furthermore, experiments have demonstrated that by applying surface tension to the membrane, a water bridge (water defect) forms, and it is possible to control the evolution of these water defects [15]. It was discovered that 1–2.8 mN/m is the critical membrane tension for mechanical breakdown. For several seconds, large pores with a radius of 2–12 µm were stable at this tension. Porated membranes in general fail at lower tensions than their respective nonporous counterparts. Thus, the critical tension required to rupture a membrane strongly depends on the rate at which the applied tension is being loaded. The critical tension varies between 10 and 20 mN/m at high loading rates (25 mN/m per second), whereas it is 4 mN/m at low loading rates (0.07 mN/m per second).

To explain the emergence and development of these meta-stable pores, a variety of theoretical models have been put forth [16,17,18,19,20]. All of them are predicated on the notion that mechanical or electrocompressive stress causes lipid area defects. The lipid tails exposed to water are located in small, hydrophobic initial pores. The lipids then reorient, lining the water channel to create a hydrophilic pore. The approximate free energy of formation, E, of a cylindrical pore with radius, r, in such simplified models is given by

$$E\left(r\right)=2\pi r\gamma -\pi {r}^{2}\tau$$
(1)

where \(\gamma\) represents the line tension (opposing the pore formation) and \(\tau\) is the surface tension lowering the pore formation and expansion energy barrier [8].

The meta-stable hydrophilic pores can therefore be stabilized or destroyed in response to changes in the membrane’s surface tension. The critical radius for the formation of a hydrophilic pore is thought to be 0.3–0.5 nm based on experimental and theoretical research [8]. The ability to observe pore formation and closure has been demonstrated in simulation studies using simplified models [21]. In course-grained simulations, beads (that represent individual water molecules or three lipid carbon atoms) are used. The parameters of these simplified models are then related to specific chemical information of the system using the dissipative particle dynamics method. Using these simplified, coarse-grained models generally allows to produce qualitative results at the expense of quantitative information that is lost in the absence of the system’s atomistic details.

Hydrophilic pores have also been simulated at the atomic level. In these meta-stable intermediate structures, transient pores are seen during the spontaneous aggregation of lipids into bilayers [22]. These structures have relatively short lifetimes lasting only from 5 to 80 ns before they spontaneously reseal.

Here, we present extended, atomistic molecular dynamics (MD) simulations. We investigate the formation of pores under the working of a nanoparticle drug-carrier system where 5-FU was the chemotherapy molecule. This was done for polymer entrapped 5-FU molecules (i.e., PEGylated) and non-PEGylated systems in the presence of an external force (an external magnetic field) to assist with targeted drug delivery. Investigations are made into the mechanism of pore stability and expansion under stress. We also investigate the passive transport of water molecules through the pores and optimize the range of number of ligands that could be used to engineer such targeted drug-delivery systems. An earlier report by H. Leontiadou et al. [8] has also showed how molecular dynamic simulations may be used to investigate the formation of hydrophilic pores in lipid bilayers. In this work, we report on the formation of similar pores. However, here we introduce (for the first time and to the best of our knowledge) the presence of the Fe3O4 drug carrier loaded with 5-FU molecules and investigate the pore formation due to this nanocarrier and drug.

In our earlier work, we have investigated the interaction of PEGylated and non-PEGylated interaction of 5-FU with Fe3O4 nanoparticles of various sizes. We have also performed density functional theory (DFT) calculations on this drug, in that work. There we reported on the iso-electrostatic potential surfaces of the 5-FU molecule as well as on the frontier molecular orbitals, and the interested reader is referred to Harris [3] for a detailed DFT analysis.

Methods

Nanoparticle formation

Spherical magnetite (Fe3O4) nanoparticles were prepared with the same core diameter of 2.6 nm. This size was selected based on our earlier work on this system [3, 22]. A fixed number of 5-FU molecules, ranging from 1 to 100 in incremental step sizes of 5, were sequentially adsorbed onto the NP. This was done via a simulated annealing Monte Carlo scheme. Both a Universal as well as a Condensed-phase Optimized Molecular Potentials (COMPASS) (for atomistic simulations studies) forcefield was employed. The difference in output was negligible. PEGylated molecules were also produced, and the same procedure as above was applied to these systems to attach the PEGs with 5-FU to the NPs (for more details, refer to [3] and [22]).

The simulated annealing task in Adsorption Locator simulates a substrate loaded with an adsorbate or an adsorbate mixture of a fixed composition. A low energy adsorption site is identified by carrying out a Monte Carlo search of the configurational space of the substrate-adsorbate system as the temperature is slowly decreased. This process is repeated to identify further local energy minima. During the course of the simulation, adsorbate molecules are randomly rotated and translated around the substrate. The configuration that results from one of these steps is accepted or rejected according to the selection rules of the Metropolis Monte Carlo method. Adsorption Locator uses the Metropolis Monte Carlo method to search for adsorption configurations. In this method, only the positions and orientations of the sorbate molecules are sampled; each conformation is treated as a rigid body. The Metropolis method assumes that the adsorbate molecules do not have a high degree of torsional flexibility and ignores any internal degrees of freedom that the adsorbate components may possess on the substrate surface. Now, since a molecular system may be described by a small number of parameters (of relevance here: volume and temperature), the collection of molecular configurations, i.e., the ensemble, may be described by the distribution function \({\rho }_{m}.\) This represents the probability of each configuration m in the ensemble. Thus, the Monte Carlo method used here generates a chain of configurations (m, n, o, etc.) with the probability of transition from m to n represented by \({\pi }_{nm}.\) Now, if the configuration m is sampled with a frequency of \({\rho }_{m}\), then on average, \({\rho }_{m}{\pi }_{nm}\) of the configurations n are transformed to m. These fluxes must be the same to preserve the density; otherwise, there would be a net flow from m to n which would increase \({\rho }_{n}\) which means a different ensemble would be sampled. Therefore, the following balance must be obtained:

$${\rho }_{m}{\pi }_{mn}={\rho }_{n}{\pi }_{nm}$$
(2)

Now, the transformational step of the configuration occurs in two steps. At first, the trial configuration is generated with a probability of \({\alpha }_{mn}.\) Hereafter, the proposed configuration n is either accepted with a probability of Pmn or the original configuration, m, remains. It has the probability of (1 − Pmn). Thus, we can write the overall transitional probability as.

$${\pi }_{mn}={\alpha }_{mn}{P}_{mn}.$$
(3)

Consequently, the choice for the acceptance probability must satisfy.

$${P}_{mn}=min\left[1,\frac{{\alpha }_{nm}}{{\alpha }_{mn}}\frac{{\rho }_{n}}{{\rho }_{m}}\right].$$
(4)

Furthermore, since the number of particles and volume remain the same in these simulations, the NVT ensemble was chosen. Since we are investigating the formation of the hydropores in the phospholipid bilayer, no particle loss is anticipated during this process and the membrane volume is kept constant, similar to the conditions in the living organism.

Finally, the calculations were converged when the force on each ion was less than 0.001 eV/Å and a total energy convergence criterion of 10−5 eV was set.

Simulation of the membrane

A bilayer phospholipid (BLPL) membrane (dipalmitoylphosphatidylcholine (DPPC)) was manually sketched and geometrically optimized via a Universal forcefield. Figure 1 shows this membrane.

Fig. 1
figure 1

Bilayer phospholipid (BLPL) cellular membrane showing the hydrophilic head and hydrophobic tail. The “larger” hydrophilic headgroup particles are by no means coarse grain particles. The particles are just displayed with a larger particle size to clearly visually see the hydrophilic head and hydrophobic tail. However, it is still the same atomistic model on the right that was used

Full size image

This membrane was selected because it has been well studied experimentally and computationally. Hereafter, the NP systems (bare, i.e., no surfactants, PEGylated and non-PEGylated, respectively) were placed outside the cell, i.e., on top of the membrane, and allowed to interact with it via molecular dynamics (MD). The MD simulations were performed using the Accelrys Materials Studio package (version 7.0.100) Similar to before, a Universal and COMPASS forcefield was employed, respectively, in the context of an NVT ensemble. An external force with direction towards the inside of the cell, representing the external magnetic field, was applied onto the paramagnetic NP and set to − 1 mN/m. It is known that surface tensions between − 10 and − 38 mN/m will cause the pores to be stable (on a timescale of 160 ns) [8]. In this study, we need the pores to be unstable, i.e., to spontaneously reseal in the same timeframe.

Periodic boundary conditions were applied to the membrane, and the temperature was coupled to above the phase transition of DPPC (at 315 K) to 323 K. In all simulations, the pressure in the direction normal to the membrane surface (laterally outward) was either 0 osm or set to 0.285 osm. Since complex organisms, like mammals, maintain constant internal osm ≈ 0.285 osmol, matching that of 0.154 M NaCl [23]), this value for the lateral outward pressure was chosen.

Charges were assigned by the forcefield. These settings were applied to both Van der Waals and Coulomb forces, and the summation method was atom-based. The minimization method was selected as “smart,” which uses combinations of the steepest descent, conjugate gradient, and Newton methods. For the conjugate gradient method, the Fletcher-Reeves algorithm was used, and for the Newton method, the Broyden–Fletcher–Goldfarb–Shanno (BFGS) algorithm was used.

Molecular dynamics (MD) simulations were then employed on both the PEGylated and non-PEGylated systems with the internal cellular pressure set to both 0 and 0.285 osm, respectively. After the NPs penetrated the membranes and created the water bridge, 1000 water molecules were added to the system and allowed to interact with the pore under similar conditions with the same forcefield(s).

The final, geometry- and energy-optimized systems were calculated using a 15 ps dynamics time with 1 fs time steps in a NVT ensemble and using the Anderson thermostat. Atom-based electrostatics with a cubic spline truncation and cutoff distance of 11.5 Å were used with a buffer width of only 0.5 Å and spline width of 1 Å.

The binding energies between the surfactants (Es) and the NP (Enp) were determined after the total energy (Etotalsystem) was calculated as [3, 24,25,26,27,28,29]

$${E}_{b}=\left({E}_{total system}\right)-{E}_{nanocluster}-{E}_{surfactant}$$
(5)

The energies of the 5-FU and PEGylated 5-FU molecules were calculated by removing the NPs from the system and then performing a single point energy calculation. The energy of the NPs was calculated by the removal of the respective molecules and again performing a single point energy calculation.

Results and discussion

Binding free energy and hydrodynamic size

Figure 2 shows the calculated binding free energy (BE) for the bare, PEGylated (with 5-FU) and non-PEGylated (with only 5-FU molecules) systems. These results were obtained after the MD simulations were run. At first, no osmotic pressure was applied to the cell membrane (the bilayer phospholipid (BLPL)). This was done to establish a baseline for comparison to see what the effect of this pressure would be on the MD simulations. Thereafter, an osmotic pressure of 0.285 osm was applied (from the inside of the cell, outward onto the BLPL), and the simulations were repeated.

Fig. 2
figure 2

(Top) Binding free energy for different systems (bare, PEGylated, and non-PEGylated) with and without cell osmotic pressure. (Bottom) NPs with 40, 50, and 60 ligands respectively showing the different hydrodynamic sizes. The core NP diameter for all three of these was 2.6 nm

Full size image

It was observed that the bare systems with no molecules attached (neither PEGs nor 5-FUs) reacted (virtually) the same under both unpressurized and pressurized conditions with a BE of 3.13 × 10−13 eV and − 2.68 eV, respectively. Thus, it was established that virtually no interaction between the BLPL and bare NP exists without the membrane pressure and only a very small interaction when the bilayer is pressurized.

The non-PEGylated (with only 5-FU attached) systems all displayed an increased interaction with the BLPL (negative BE) compared to the bare system. When no osmotic pressure existed, a linear increase in BE was observed, whereas the presence of osmotic pressure increased the PLBL and NP interaction non-linearly. A similar non-linear increase in BE was observed for the NPs that were PEGylated, with the pressurized membrane system, at first, having a weaker interaction with the NP (i.e., lower BE values), and then a better interaction after 50 PEGs (with 5-FU molecules) were added. The reason for this is that at 50 ligands, the core 2.6 nm NP’s surface is saturated, with no more space left for additional molecules to bind. Consequently, any additional molecules can only physisorb onto the NP. Since these additional molecules are therefore not strongly bonded, they will easily be desorbed from the NP. Additionally, since most of the dangling bonds of the ligands on the NP will (after 50 ligands) interact with the additionally added ligands, less free energy is available to interact with the BLPL. Therefore, the gradient is observed to flatten out. The presence of osmotic pressures further reduces this interaction for the PEGylated system so that, at 100 ligands, the BE is almost the same as for the non-PEGylated system. Thus, to summarize, a maximum number of 50 PEGylated surfactants (with 5-FU on the PEGs) under a normal cell osmotic pressure of 0.285 osm on a 2.6 nm Fe3O4 NP will interact the best with a BLPL. Any additionally added PEGs (to stabilize the NP surface) will only reduce the interaction (free energy).

External force (magnetic field) applied

Hereafter, we ran simulations to observe what the effect of an external force (an externally applied magnetic field) would be on the cell membrane for similar NP systems as before. Since an externally applied magnetic field would be used in such drug-carrier systems to guide the NP to the target location after which it will unload its drug payload, we are interested in what effect this combined effect of paramagnetic NP and external magnetic field would have on the cell membrane. Figure 3 displays the resulting binding free energy under such circumstances. Like before, PEGylated and non-PEGylated systems were tested in 0 osm and 0.285 osm conditions.

Fig. 3
figure 3

Binding free energy for an increasing number of ligands on the NP

Full size image

Our first observation is that, for an external force up to 0.208 µN, the binding free energy is roughly the same as without the external force, i.e., the range of BE from 0 to − 3500 eV is roughly the same as in Fig. 2. Thus, the addition of this relatively small external force would not disrupt the experiment too much in that the main contributor to BLPL-NP interaction would still (mainly) be the NP and not the external magnetic field. The second observation is that the presence of the external force now shifts the non-PEGylated systems into positive BE values for ligand numbers less than 30. Thus, these NPs will no longer interact with the BLPL, and for ligand numbers higher than 30, this BE remains relatively low (below − 1000 eV), indicative of a weak interaction between lipid and NP. Therefore, it can be concluded that it would be better (disregarding the additional benefit of stabilization effects) to design PEGylated paramagnetic ultra-small NP drug carrier systems than non-PEGylated systems.

Furthermore, it is observed that the presence of the external force now shifts the PEGylated systems BE values in (virtually) the same range as the (non-realistic) non-pressurized cell-membrane system. Thus, this means that the presence of the external B-field effectively cancels the (opposite direction) osmotic pressure on the cell membrane and (incoming) NP system, so that the NP drug carrier system may indeed penetrate through the BLPL into the cell. Like before (Fig. 2), a nonlinear increase in the BE is observed, and at 100 ligands, the no-pressure and pressurized systems have the same free energy, indicative of this cancellation effect that the external force has on the osmotic pressure in the presence of the paramagnetic NP system.

Non-PEGylated and PEGylated NP interaction with BLPL

Figure 4a shows the geometrically optimized structures where no internal, cellular osmotic pressure was applied and the NPs were not PEGylated, with only the 5-FU molecule bonded. These geometries were computed in the presence of the external magnetic field. It can be observed that in all cases, the movement of the NP system through the BLPL causes the membrane to rotate with respect to the vertical axis, almost by 90°, thus closing in on itself. Additionally, since the NP system is not optimally stabilized (due to the absence of PEG molecules), a strong interaction between some of the membrane molecules and the NP system is observed. Thus, some of the PL molecules are dragged into the cell with the NP, during the formation of the protein corona. Figure 4b also shows the interaction when the osmotic pressure was turned on. This time, it is observed that the membrane stays intact much better than before. Again, some of the PL molecules are dragged into the cell via the NP for similar reasons as before. However, now the membrane is curved at the ends of the pore and the lipids are perturbed around it, like what is observed by Leontiadou et al. [8] in that it is shaped like an hourglass. At first, these pores are small and hydrophobic. Thus, the hydrophobic lipid tails will be exposed to the outside of the cell. Thereafter, the lipids reorient themselves, lining the pore opening with the headgroups to become a hydrophilic, hourglass-shaped water channel.

Fig. 4
figure 4

Non-PEGylated NPs, with B-field on. a No internal cellular pressure, b 0.285 osm applied. (The side view, the top part represents the outside of the cell, and the bottom part represents the inside of the cell)

Full size image

Figure 5a shows the optimized geometries for the PEGylated system, with no internal cellular pressure, but with the B-field on and (Fig. 5b) the same but with internal, cellular pressure present.

Fig. 5
figure 5

PEGylated NPs, with B-field on. a No internal cellular pressure, b 0.285 osm applied

Full size image

Similar observations are made as in Fig. 4, except that when from 50 to 100 ligands (in this case PEGylated ligands) are added to the NP, the membrane is destroyed. However, when the B-field is switched on, the membranes again become shaped like an hourglass. Therefore, it is important to know the pore size after the NP system has passed through the membrane to see if there is any change in this for the PEGylated versus non-PEGylated systems.

Pore size and BE for different number of ligands

Figure 6 shows the pore size and the binding free energy as a function of the number of ligands (PEGylated and non-PEGylated).

Fig. 6
figure 6

(Left) Pore size and BE vs number of ligands. (Right) Hourglass-shaped pore with lipid headgroups lining the pore

Full size image

Two regions are highlighted in Fig. 6: green (for the PEGylated system) and red (for the non-PEGylated system). From previous theoretical studies and experimental data, it is known that the critical size for which a hydrophilic pore will form is between 0.3 and 0.5 nm [8]. It is also known that the minimum size at which a pore can be stabilized is 0.7 nm [8]. Thus, the green area indicates the possible number of ligands required that would give a PEGylated system with non-stable pore sizes as well as negative binding free energies. This ranges from 10 to 28 ligands. In this range, the spontaneous resealing of the pores will occur. Additionally, the negative binding energy ensures the better stability of the NP-ligand complex. Thus, the NP system with this set of restrictions will be able to penetrate the membrane, without causing long-lasting damage to the membrane and would therefore be less toxic, i.e., it would not form a pore that remains open long enough for other more toxic molecules to enter into the cell.

For the non-PEGylated NP system (red area), the number of ligands ranges from 30 to 70. For less than 30 ligands, the binding free energy of the system would be positive. However, for the 30–70 range, the binding free energy is negative and therefore more stable. Additionally, in this range, the pore size is 0.7 nm or less and will therefore also spontaneously reseal. The binding energy for this system (0 to − 440 eV) is however not as negative as for the PEGylated system (− 474 to − 1771 eV), and therefore these are less stable. Thus, it may be easier for the non-PEGylated system to deliver the payload after membrane penetration than for the more stable PEGylated system.

Water interaction with pore

Non-PEGylated NPs

Figure 7 shows the geometry-optimized systems, with water molecules originally outside the cell, after the NP created the pores. For both the systems where no osmotic pressure was applied, the membrane was completely disfigured after NP penetration. Therefore, water easily penetrated through the membrane within 20 to 28 ns. For the more realistic systems with osmotic pressure, the following was observed. The membrane pore formed from the non-PEGylated system, with only 10 ligands, remained largely intact. However, the size of the pore (slightly above 0.3 nm), combined with the reorientation of the lipids to line and create a hydrophilic water channel, did not allow many water molecules to pass through the pore. These molecules became stuck inside the pore and could not penetrate it. However, the systems with 50 and 100 ligands did have a large enough pore size to allow for water molecules to travel into the cell. Now, if we include the limitations set by Fig. 6 on this, we can make two conclusions as follows. (i) If we allow for water to penetrate through the water pore after the NP has passed into the cell, then for the non-PEGylated system, the number of ligands that would create a stable enough NP system that creates a hydrophilic water pore that would spontaneously reseal would be between 50 and 70 ligands. For more than 70 ligands, the pore size would be too large to reseal. (ii) If we do not want any other molecules, including water, to enter the cell after the NP system has passed through the membrane, then 30–50 ligands should be used. For less than 30 ligands (Fig. 6), the pore size would be too small and water molecules will be captured in the pore.

Fig. 7
figure 7

In all these experiments, the B-field was on and 1000 water molecules were added to the system. Cyan molecules are water molecules. a 0 osm, non-PEGylated. b 0.285 osm, non-PEGylated. c 0 osm, PEGylated. d 0.285 osm, PEGylated

Full size image

PEGylated NPs

For the pores created by the PEGylated systems, the following is observed. For all the systems (10, 50, and 100 ligands), water penetrated the cell through the pore. However, for the system created with only 10 ligands, many of the water molecules became stuck in the pore (since it is hydrophilic). For the system created from 50 ligands, the same was observed but more water molecules now penetrated the cell. And, finally, for the pore created with the 100 ligand system, most of the water molecules penetrated the cell. Therefore, (i) since only pore sizes of between 0.3 and 0.7 nm will spontaneously reseal, the number of PEGylated ligands required would be between 10 and 30. Above 30 ligands, the pore size becomes too large to reseal. For 10 ligands and below, the pore size is too small, and its hydrophilic nature allows water to be captured into the membrane, instead of allowing it to pass through. Now, since the water passes through between 17 and 23 ns, spontaneous resealing would occur relatively quickly after the NP system has passed through the membrane.

Water inside cell

Figure 8 shows the percentage of water that permeates into the cell through the hydrophilic pore as a function of the number of ligands for the PEGylated and non-PEGylated systems. The absolute number of water molecules inside the cell was calculated by integrating the water density distribution across and below the pore (i.e., inside the cell). It is observed that for the PEGylated system, in the allowed range from 10 to 30 ligands, 33 to 35% of water molecules transverse through the pore. For the non-PEGylated system (in the range of 50 to 70 ligands), 39 to 50% of the water molecules enter the cell. Thus, (i) more water molecules enter the pore when the membrane is under tension, and (ii) more water molecules enter the pore when the NP that creates the pore is not-PEGylated, thus increasing the overall concentration of the hydrophilic 5-FU molecule in the system. Since the PEGylated NP system is more stable (with larger negative binding free energy values), it does not readily interact (in comparison with the non-PEGylated system) with the membrane. Consequently, the formation of the protein corona that forms during NP penetration is smaller than that of the non-PEGylated system. Thus, the pore’s hourglass structure lined with lipid headgroups remains more intact. Therefore, when water molecules permeate the cell through the pore, the pore structure is less affected. Now, since the average pore size of 0.55 nm is slightly larger than that of the non-PEGylated system (0.44 nm), two water molecules may enter the pore created by the PEGylated NP system at the same time, and only one molecule may enter at a time for the non-PEGylated system’s pores.

Fig. 8
figure 8

Percentage of original (1000) water molecules that permeate into the cell through the hydrophilic pore

Full size image

Discussion

5-FU: bad membrane permeability

Conventional chemotherapies mainly destroy the proliferating bulk of tumor cells. However, they are not effective at destroying the cancer stem-like cells (CSCs). These may form new colonies leading to tumor relapse. While 5-FU is an effective chemotherapy drug with a broad spectrum of anti-cancer activity, it is hindered by shortcomings like short half-life, non-specific toxicity in all cells, and bad membrane permeability leading to side effects such as gastrointestinal, liver diseases, hair loss, etc. [30].

5-FU conversion to metabolites

The cytotoxic effect of 5-FU is induced by preventing cellular thymidylate synthase (TS). Thus, this inhibits DNA replication [31]. After 5-FU enters a cell, it will be converted to several active metabolites, such as fluorodeoxyuridine monophosphate (FdUMP), fluorodeoxyuridine triphosphate (FdUTP), and fluorouridine triphosphate (FUTP). Each of these metabolites prevents cell growth in a unique way. For example, FdUMP inhibits TS leading to indirect DNA damage by deoxynucleotide imbalances.

5-FU resistance

Furthermore, it is known that the primary factor leading to 5-FU resistance is the presence of cancer stem-like cells (CSCs) within a cancer cell niche. Targeted nano-formulation drugs may co-deliver anti-cancer drugs to target CSC-specific surface markers (such as CD44 and CD133) and signaling pathways to combat drug resistance [32]. Drug delivery with NP systems may improve 5-FU therapy. The effectiveness of 5-FU may also be enhanced by entrapping the molecule in polymer-based NPs (like PEG molecules) to reduce non-selective exposure and side effects. (The PEG molecule is a biocompatible, hydrophilic, synthetic polymer with widespread use in biomedical applications). This would also reduce the drug resistance and increase the circulation half-life.

Thus, it is important for the drug to (i) enter the cell through the membrane and then (ii) to damage the DNA before the cell is destroyed. To prevent drug resistance, this can be done as a (iii) targeting NP system.

It is well known that surfactants may affect membrane stability [20]. Thus, ligands may be used during the design of nanoscale drug-delivery systems to compensate for any detrimental effects that the nanoparticle may have on the stability of the membrane. This is important since damage caused to the membrane may not only lead to cell death but also allow other ions and molecules to enter the cell unhindered.

Through synthesis, surfactants can be changed, and nanoscaled systems can be designed arbitrarily. This provides useful pathways to control membrane stability. It is also known that larger quantities (i.e., concentrations of hydrophilic surfactants) may cause significantly larger perturbations for a given concentration [20]. Thus, an increase in concentration may also increase the pore size and stability of the water defect, as was observed in this study. However, if these pores have the critical size, it may spontaneously reseal. Thus, even though the nanocarrier may create a temporary water-bridge while penetrating through the membrane, this is short-lived and would only allow some water molecules to traverse into the cell for a few nanoseconds. This effect may be used to maintain the cellular osmotic pressure and stability, while the 5-FU is converted to the required metabolites to serve its purpose.

Conclusion

In this investigation, we reported on the interaction of a PEGylated and non-PEGylated Fe3O4 NP drug-delivery system, with 5-FU as the chemotherapy drug. This was done with particular focus on the interaction of this system with a DPPC phospholipid bilayer and the resulting pore formation. We observed the hourglass-shaped pores with hydrophilic lipid headgroups lining the pores. Furthermore, we showed that for PEGylated NPs, 10 to 28 ligands will allow for the formed pores to spontaneously reseal, whereas for the non-PEGylated NPs, 30 to 70 ligands will do the same.

We also showed that the PEGylated system in this range allows for 33 to 35% of water molecules to transverse through the pore. For the non-PEGylated system (in the range of 50 to 70 ligands), 39 to 50% of the water molecules entered the cell. This was the result of the less stable non-PEGylated NP system’s interaction with the pore prior to water penetration. This effect may be used to maintain the cellular osmotic pressure and stability, while the 5-FU is converted to the required metabolites to serve its purpose as a chemotherapeutic drug.