Using pH-sensitive lipid nanoparticles to deliver cisplatin for treating ovarian cancer resistant to cisplatin

Quantities of 4-(2-Hydroxyethyl) morpholine (>99% purity), N-(3-Dimethylaminopropyl)-N′-ethyl carbodiimide hydrochloride (EDC), dimethyl aminopyridine (DMAP), dichloromethane (DCM), hydroxybenzotriazole (HOBt), n-hexane, ethyl acetate, sodium sulphate, sodium chloride, deuterated chloroform, Pluronic F-127, and cisplatin (95% purity) were purchased from Sigma-Aldrich (Melbourne, Australia). Ultrapure water (Milli Q, 18.2 MΩ·cm) was used for all solution preparations. Monoolein (MO) and oleic acid (OA) were sourced from Nu-chek-Prep, Inc. (Elysian, MN, USA) (>99% purity).

Synthesis of aminolipids was carried out using the procedures described in [14]. Nuclear magnetic resonance (NMR) imaging (¹H) analysis confirmed the synthesised lipid structures (Figure S2). The chemical structures of 2-morpholinoethyl oleate (OE-Mo), N-(pyridin-4-ylmethyl)oleamide (OA-Py), and N-(2-(piperidin-1-yl)ethyl)oleamide (OA-Pi) are presented in Figure 1.

LNPs were prepared by a dry lipid film hydration method, adapted with minor changes from our previous work [15,16]. To prepare an empty nanoparticle (control without cisplatin), the dried lipid film was rehydrated with 1 mL of a Pluronic F-127 solution (2 mg F-127 in 1 mL of 0.9% NaCl solution). The resulting dispersion was then probe sonicated (Q Sonica) in pulse mode for 5 min to obtain a uniform suspension. Cisplatin-encapsulated LNPs were prepared by hydrating the dry lipids in an aqueous phase containing cisplatin. The aqueous phase was prepared by dissolving cisplatin (2 mg) in 1 mL of F-127 solution (2 mg in 1 mL of 0.9% NaCl). Table 1 represents the composition of each of the formulations prepared.

The hydrodynamic diameter, size distribution (reported as polydispersity index, PDI), and ζ-potential of the nanoparticles were characterised using a Zetasizer Nano ZS (Malvern Instruments, Malvern, UK). For size measurements, samples were diluted 1:10 in ultrapure water (MilliQ, Bayswater, VIC, Australia). ζ-potential was determined in 1.5 mM phosphate buffer (PBS) at different pH values, adjusted by addition of 0.01 M citric acid. Diluted samples (900 μL) were transferred to disposable cuvettes, and all measurements were conducted at 25 °C (n = 3) in accordance with the manufacturer’s instructions. Electrophoretic mobility was converted to ζ-potential by the instrument software using the Helmholtz–Smoluchowski equation [17].

Nanoparticle phase behaviour was examined at the small- and wide-angle X-ray scattering (SAXS/WAXS) beamline of the Australian Synchrotron (ANSTO), using procedures adapted from our earlier work [14,15,16,17,18]. X-rays of wavelength λ = 1.128 Å (11.0 keV) with a typical flux of ~10¹³ photons/s were employed. A sample–detector distance of 1.6 m was selected, yielding a q-range of 0.01–0.5 Å⁻¹. Two-dimensional scattering patterns were azimuthally integrated to generate one-dimensional intensity versus q profiles for phase assignment. Silver behenate served as the calibration standard. Each sample was exposed for 2 s, and no evidence of radiation-induced structural changes was observed.

The pH study was performed by mixing 40 μL of the nanoparticle with 40 μL of a buffer of the desired pH (citric acid/sodium phosphate system). The sample temperature was maintained at 25 °C using a circulating water bath during data collection.

Encapsulation efficiency (EE%) and drug loading (DL) were quantified after separation of non-encapsulated cisplatin using an ultrafiltration–centrifugation approach with Amicon^® centrifugal filters (Sigma-Aldrich (Melbourne, Australia)) (10 kDa molecular weight cut-off), as illustrated in Figure 2. Cisplatin-loaded LNP dispersions were transferred into the Amicon^® units and centrifuged at 5000× g for 15 min at 25 °C. The filtrate collected from the lower chamber contained the unencapsulated (free) drug; this fraction was brought back to the original volume prior to analysis.

Cisplatin content in the formulations before (W_i) and after (W_f) ultrafiltration/centrifugation was determined using a validated HPLC assay. Chromatographic separation was performed on an Agilent Zorbax SB-C18 column (Agilent, Santa Clara, CA, USA) (4.6 mm × 250 mm, 5 μm) equipped with an UV detector. The mobile phase consisted of acetonitrile and PBS (pH 5.0) at a 49:51 (v/v) ratio, delivered isocratically at 0.5 mL·min⁻¹. The column temperature was maintained at 30 °C, and cisplatin was monitored by UV absorbance (detector wavelength as per the validated method; e.g., 210 or 232 nm). For analysis, 10 μL of formulation was diluted with 990 μL of a 50:50 (v/v) mixture of buffer and acetonitrile, and 20 μL of this solution was injected. Drug concentrations were obtained from a calibration curve.

EE% and DL were then calculated from W_i and W_f using the following relationships:

where W_i is the initial amount of drug in the formulation and W_f is the amount of free drug in the filtrate after centrifugation.

All animal experiments were performed at Monash University according to approved animal ethics (animal ethics approval granted for in vivo experiment (MMCB/2020/01/Ho) and human ethics approval had also been granted for the use of human specimens in the experiments (Monash University Human Research Ethics Committee—MUHREC/25377, dated 14 July 2020). Patient-derived xenograft (PDX) models were used to evaluate the efficacy of the cisplatin-loaded LNPs. PDX models have been identified as better representations of tumour heterogeneity and tumour microenvironment with retention of cellular complexity, cytogenetics, and stromal architecture, and are powerful tools for determining cancer characteristics, developing new treatments, and predicting drug efficacy [19]. Cisplatin-refractory ovarian cancer tumours from the patient were implanted subcutaneously in immunosuppressed NOD.CgPrkdc^scid H2-Ab1^em1Mvw H2-K1^tm1Bpe H2-D1^tm1Bpe Il2rg^tm1Wjl/SzJ (NSG-(KbDb)^null (IA)^null) NSG MHC^null mice (depicted in Figure 3). The tumour volume (mm³) was determined with the following equation:

Once the tumour volume reached 180–300 mm³, treatments were administered at days 1, 8, and 18 at 4 mg/kg as an intraperitoneal injection. Tumour-bearing mice were randomised to cisplatin-loaded LNPs, treatments with cisplatin alone (cis-alone), and drug-free LNPs and buffer solutions (control). Mouse weight and tumour volume were measured twice a week for up to 120 days. The mice were culled once tumour volume reached 700 mm³ or if they showed signs of significant clinical deterioration.

The time to progressive disease (PD) was defined as the interval (in days) from treatment initiation to the point at which the mean tumour volume for a given treatment group exceeded the nadir value by more than 20%. The nadir was taken as the smallest mean tumour volume recorded after treatment commenced, or 180 mm³ if the minimum value was <180 mm³, as lesions below this size are difficult to measure reliably.

The time to harvest (TTH) was defined as the time (in days) from the beginning of treatment to the day of harvest at 700 mm³, and the median TTH was calculated and plotted using Kaplan–Meier curves (Graphpad Prism version 10). Tumour growth was monitored by measuring tumour volume three times per week and plotted as mean tumour volume ± SEM for each treatment group. Formal statistical comparison of treatment efficacy was performed using time to harvest (TTH), defined as the time required for tumours to reach the predetermined ethical endpoint requiring animal sacrifice. TTH was analysed using Kaplan–Meier survival curves, and significance between treatment groups was assessed using the log-rank (Mantel–Cox) test in GraphPad Prism (version 10). This endpoint was selected as it provides a robust and clinically relevant measure of tumour progression and treatment response in this model.

Paraffin-embedded kidney sections were evaluated semi-quantitatively for tubular injury. Tubular damage was assessed based on tubular epithelial cell loss, tubular necrosis, cellular debris accumulation, and tubular cast formation. The extent of injury was scored according to the percentage of affected tubules as follows: 0, normal; 1, <10–25%; 2, 26–50%; 3, 51–75%; and 4, >75% affected tubules.

The chemical structures of the synthesised ionisable lipids were confirmed by NMR analysis. The NMR analysis matched the analysis performed previously. Following the procedure explained in Section 2.3, LNPs were prepared using MO and three lipids: OE-Mo, OA-Py, and OA-Pi. Empty LNPs without cisplatin were named OE-Mo-NP, OA-Py-NP, and OA-Pi-NP, and cisplatin-loaded as Cis-OE-Mo-NP, Cis-OA-Py-NP, and Cis-OA-Pi-NP. The amount of cisplatin and Pluronic F-127 added to the nanoparticles (2 mg) was kept at 10 wt% to that of the total quantity of lipids in the LNPs. Table 1 shows the composition of the six LNPs.

The formulations were assessed for their particle size, PDI, drug loading (DL), and entrapment efficiency (EE%), and the results are presented in Table 2. All six formulations were well dispersed without visible phase separation or solid precipitations. The particle size of the LNPs prepared was 219–250 nm, while PDI was in the range of 0.15–0.22. These values are usually observed for non-lamellar LNPs [15]. Furthermore, cisplatin loading had a minimum effect on the particle size of the LNPs (Table 2).

The EE% and DL capacity of cisplatin-loaded LNPs Cis-OE-Mo-NP, Cis-OA-Py-NP, and Cis-OA-Pi-NP were also evaluated. As described in Section 2, the unencapsulated free drug was removed using the ultrafiltration and centrifugation method, and EE% was measured using HPLC. The calculated EE% for Cis-OE-Mo-NP, Cis-OA-Py-NP, and Cis-OA-Pi-NP was 62%, 59%, and 64%, respectively (Table 2), and there was no significant difference between their loading capacity. Compared to the study by Zhang et al. on similar MO-based cubosomes, our cis-loaded LNPs with MO and ionisable aminolipids possessed a drug concentration ~1.6–2 times higher [20]. Many factors influence the drug loading, including the lipid composition, the process of adding the drug and buffer, and the bilayer fluidity [21,22]. We have observed similar high drug loading of the water-soluble drug Fluconazole in our previous study [15]. The higher loading in this study may be due to aminolipids, which can act as ligands to replace chloride ions of cisplatin [23]. The higher loading in our studies may be due to the use of 0.9% NaCl solution, which will prevent the hydrolysis of cisplatin [Pt(NH₃)₂Cl₂] in water [24].

Synchrotron SAXS experiments were conducted with the cis-loaded LNPs to evaluate the effect of drug loading on the mesophase behaviour (Figure 4A). The phase behaviour of Cis-OE-Mo-NP at different pH values was very similar to that of the empty LNPs (OE-Mo-NP). Wherein, the Cis-OE-Mo-NP exhibited a H₂ phase at neutral pH values and transitioned into an Im3m cubic phase with less curvature when the pH was changed from 6 to 5. The phase behaviour of Cis-OA-Pi-NP varied slightly from the empty LNPs, OA-Pi-NP. At pH 7, the L₂ phase transitioned to an Im3m cubic phase instead of a Pn3m phase observed in the OA-Pi-NP.

On the other hand, Cis-OA-Py-NP differed from the control. The LNPs exhibited L₂ phase at pH 7, as compared to an H₂ phase for the drug-free OA-Py-NP. In the drug-free OA-Py-NP, a transition from the H₂ phase to a Pn3m cubic phase occurred when the pH changed from pH 6 to pH 5. In Cis-OA-Py-NP, the coexistence of two phases (L₂ + H₂) was observed at pH 5.0, and a complete transformation to a Pn3m cubic phase occurred at pH 4. Cisplatin [Pt(NH₃)₂Cl₂] contains an ionisable NH₃ group, and several past studies have shown that encapsulated ionisable drug molecules can influence the mesophase structure of LNPs [25,26]. A recent study by Azmi et al. has also seen cisplatin-induced morphological alterations in the phospholipid-containing phytantriol system, wherein the encapsulated cisplatin has been localised in the internal H₂ nanostructure, leading to an enlargement of its hydrophilic nanochannels [27]. The change can also be viewed from the ligand interaction of cisplatin with the aminolipid. The headgroup of OA-Py contains a pyridine moiety (Figure 1), and many studies in the past have shown that cisplatin can undergo ligand substitution with pyridine nitrogen atoms [23,28]. Cisplatin, being soluble in water, resides in the aqueous channels and can interact with the pyridine headgroup of OA-Py. An amount of OA-Py will participate in complexation with the drug, resulting in the phase transition pH (from L₂ to Q₂) occurring at one pH unit lower than the empty OA-Py-LNP.

The ζ-potential values for Cis-OE-Mo-NP, Cis-OA-Pi-NP, and Cis-OA-Py-NP were measured at different pH using a Zetasizer instrument (as described in Section 2). The measured ζ-potential values are plotted against pH and are presented in Figure 4B. At neutral pH, the ζ-potential values of the Cis-OE-Mo-NP, Cis-OA-Pi-NP, and Cis-OA-Py-NP were 0.4 mV, 15.6 mV, and 1 mV, respectively. Interestingly, Cis-OA-Pi-NP had a high value of ζ-potential (15.6 mV) at neutral pH; this may be due to the apparent pKa of the added ionisable lipid in the LNP [26]. As the pH of the system decreases, the ζ-potential values increase for all the tested LNPs. The observed increase in ζ-potential values is due to the presence of ionisable aminolipids, which are weak bases, and proton-induced ionisation results in the rise in ζ-potential values.

The pH-responsive behaviour of the present LNPs can be interpreted using the critical packing parameter (CPP), defined as v/a0l, where v is the hydrophobic chain volume, a0 is the effective headgroup area, and l is the hydrophobic chain length. The aminolipids used in this study are weak bases whose headgroups contain a nitrogen atom; as pH decreases, these headgroups become increasingly protonated, depending on their apparent pKa within the lipid assembly. This protonation increases electrostatic repulsion between adjacent headgroups, expands the effective headgroup area, lowers the CPP, and consequently alters the preferred interfacial curvature. This observation resembles our previously developed antifungal drug Fluconazole-loaded LNPs [15]. These changes provide a mechanistic basis for the pH-dependent mesophase transitions observed in the SAXS experiments. Differences in the headgroup chemistry of OE-Mo, OA-Py, and OA-Pi are therefore expected to influence ionisation behaviours, internal nanostructures, and the retention/release characteristics of cisplatin. While direct release studies were not performed in the present study, the structural and ζ-potential data support a formulation-dependent pH-responsive mechanism.

To assess the in vivo efficacy of cisplatin-loaded LNPs in platinum-refractory ovarian tumours, tumour-bearing mice were randomised to empty LNPs, cisplatin-loaded LNPs, cisplatin alone, and saline injections. The treatment was conducted at a cisplatin dose of 4 mg per kg body weight. Empty LNPs were injected at the same lipid concentrations as cisplatin-loaded LNPs. Injections were given intraperitoneally on days 1, 8, and 18. Following treatment, the mice were monitored twice a week for tumour volume. The growth trajectories of the engrafted tumours over time have been presented on a line graph as mean of tumour volumes (mm³) ± SEM for all mice in their respective arms (Figure 5A–C). The present in vivo study was designed as an initial proof-of-concept evaluation using a single cisplatin dose across all formulations to enable direct comparison of formulation performance. However, this design does not allow a full assessment of dose-dependent efficacy and toxicity.

Cis-OA-Py-NP was the most effective therapy, demonstrating more than 60% reduction in the tumour volume compared to cis-alone (Figure 5C). Cis-OE-Mo-NP also showed in vivo efficacy in controlling tumour growth with 25–30% reduction in the tumour volume (Figure 5B). The reduction in tumour volume significantly prolonged the median TTH for the tumour-bearing mice group receiving the cis-loaded LNPs compared to the arms receiving cisplatin alone and PBS (as controls). These resulted in increases in the median TTH by 60% for the Cis-OE-Mo-NP arm and 100% for the Cis-OA-Py-NP arm, compared to the control arms (Figure 5E,F).

These findings were notable because the PDX tumours had been confirmed to be refractory to the platinum agent cisplatin, with treatment at the maximum tolerated dose (MTD) showing no impact on tumour growth. We hypothesised that achieving in vivo regression of these tumours would require a much higher intratumoral concentration of cisplatin to overcome the intrinsic tumour mechanisms of platinum resistance. Therefore, the in vivo tumour-killing efficacy of cis-loaded LNPs was likely owing to the ability of pH-responsive LNPs to deliver and concentrate cisplatin within the tumour micro-environment.

The Cis-OA-Pi-NP treatment arm was terminated early due to reported toxicities after the first injection. All the tumour-bearing mice randomised to this arm appeared to be extremely scruffy and developed respiratory distress within 1–2 days after the first dose. All mice were euthanised after 2 days post treatment. To evaluate toxicity, a full warm autopsy was performed on these mice with kidney, liver, pancreas, lung, heart and brain collected for histologic assessment. Gross kidney abnormalities were observed in mice receiving Cis-OA-Pi-NP. The histology results demonstrated that the kidneys of OA-Pi-NP receiving mice appeared normal (Figure 6A). In contrast, there were extensive signs of acute tubular necrosis (ATN) found in the kidneys of mice receiving Cis-OA-Pi-NP (Figure 6B).

The severe acute toxicity observed for Cis-OA-Pi-NP is likely to reflect multiple formulation-dependent factors rather than a single mechanism. One possible explanation is greater instability of this formulation under physiological conditions, which may promote premature drug leakage after administration. In addition, SAXS analysis showed that Cis-OA-Pi-NP exhibited a cubic internal structure at neutral pH, and this type of mesophase has been associated with more rapid escape of water-soluble cargo through interconnected aqueous channels. The comparatively higher positive ζ-potential of Cis-OA-Pi-NP at neutral pH may also alter biological interactions in vivo and contribute to its adverse toxicity profile. These observations suggest that future optimisation should focus on improving formulation stability, reducing premature release, and refining lipid composition before further in vivo evaluation

To further assess renal toxicity, kidney injury was evaluated semi-quantitatively using paraffin-embedded kidney sections. Tubular injury was scored based on tubular epithelial cell loss, tubular necrosis, cellular debris accumulation, and tubular cast formation. Scores were assigned according to the percentage of affected tubules: 0, normal; 1, <10–25%; 2, 26–50%; 3, 51–75%; and 4, >75%. Empty nanoparticle-treated mice showed minimal tubular injury, with a mean score of 0.4 (n = 3), whereas cisplatin-loaded nanoparticle-treated mice exhibited severe tubular injury, with a mean score of 3.9 (n = 3).

Clinically, cisplatin is known for causing renal damage. The main reason for Cis-OA-Pi-NP toxicity is attributed to the preferential accumulation of cisplatin in the kidneys. Cisplatin induces the production of reactive oxygen species (ROS), which leads to mitochondrial and non-mitochondrial pathways of apoptosis and necrosis [2]. ROS also cause acute inflammation and cell infiltration. Histological analysis also confirms a large amount of protein casts within these renal tissues, which can be caused by a significant reduction in blood flow to the kidney. The reduction in blood flow can be due to the plasma fluid entering the renal interstitium [29].

Another possible explanation for the renal toxicity of Cis-OA-Pi-NP is the potential sudden release of high-concentration cisplatin from the cubosomal formulation, and the higher positive charge associated with the LNP. As observed by the SAXS study, Cis-OA-Pi-NP exhibits cubic structure at pH 7, and several studies have shown that water-soluble drugs tend to leak out of cubosomes in a “burst release” manner when the cubosomes change to a cubic structure phase [30,31]. The cubic phase formation results in two aqueous channels that interdigitate throughout the internal nanoparticle matrix, and these aqueous channels may provide the route for the drug to escape [30]. In contrast, the H₂ phase maintains closed aqueous compartments, making it harder for the encapsulated drug to escape [31]. Cubosomes may also disintegrate under disintegration circumstances due to certain plasma interaction leading to a sudden release of the payload after injection [32].

Cis-OA-Py-NP treatment resulted in significant mouse weight loss due to renal and lung damage. One of the long-term impacts of cisplatin treatment is severe cachexia, both in mice and humans [33]. This was observed in mice receiving cisplatin-loaded LNPs despite showing promising tumour growth inhibition. During a monitoring period of 40 days, the weight of the mice in all experiment arms were measured and presented in Figure 7A as the mean percentage change from the original weight. The results showed that tumour-bearing mice treated with Cis-OA-Py-NP exhibited significant body weight loss of >25% from baseline up to 30 days following treatment. The percentage weight losses in mice from the arms receiving buffer, cisplatin alone, or empty LNPs were not significant. This observation suggested that cisplatin LNPs can cause treatment-related toxicities leading to weight loss in these mice. In support of these findings, in vitro toxicity studies demonstrated that cisplatin-loaded nanoparticles were indeed more toxic to both A2780 and A2780cis (cisplatin-resistant) cells compared to free cisplatin and empty nanoparticles (Supplementary Figure S1 and Table S1). Notably, the inclusion of OA-Py appeared to further enhance the cytotoxicity of the nanoparticles.

Autopsies of Cis-OA-Py-NP-treated mice (Figure 7B) indicated gross damage to the mice kidneys and lungs after treatment, while the liver, spleen, pancreas and heart were spared. As a control, the autopsy results of the empty LNPs, OA-Py-NP, showed no gross damage to their kidneys and lungs (Figure 7C).

It is important to note that the empty control LNPs (OA-Pi-NP, OA-Py-NP, and OE-Mo-NP) do not have any anticancer activities or any toxicity effect. Therefore, the kidney damage seen for Cis-OA-Py-NP was likely related to increased concentration of cisplatin (a known renal toxic agent) in the kidney. The observed lung damage was also observed, possibly because of the vasculature and basement membrane of the lung are similar to those of the kidney, and toxicity is due to the higher concentration of drugs in the two organs. The observed kidney and lung toxicities are consistent with altered tissue exposure to cisplatin-loaded LNPs, but not definitive evidence of selective accumulation.

This higher accumulation in the lung may be attributed to the positive charge associated with the LNPs as explained by Cheng et al. in their recent study [34]. Cheng et al. developed a Selective Organ Targeting (SORT) strategy by adjusting the surface charge of LNP, which allowed LNP to accurately deliver the cargo molecule to the lungs, spleen and livers of mice when administered intravenously [34]. In the study, Cheng et al. added positively charged and negatively charged SORT molecules as a supplemental component to the traditional LNPs used for RNA delivery. An increase in the addition of positively charged SORT molecules using (2,3-dioleoyloxypropyl)trimethylammonium chloride (DOTAP) resulted in tissue-specific mRNA delivery shift from the liver to the spleen to the lungs. Adding a permanent positively charged SORT molecule completely changed the delivery from the liver to the lungs. On the other hand, adding permanently negatively charged SORT molecules meant that the majority of mRNA delivery was found in the spleen [34]. It is possible that due to the presence of the ionisable lipid, the LNPs are partially positive charge, leading to accumulation in these organs.

The differences observed between the three cisplatin-loaded formulations highlight a clear structure–property–performance relationship. OE-Mo-containing LNPs showed pH-responsive behaviour with relatively moderate in vivo efficacy and a more favourable tolerability profile. OA-Py-containing LNPs exhibited altered phase behaviour in the presence of cisplatin, potentially reflecting interactions between the pyridine headgroup and the drug, and this formulation showed the strongest anti-tumour activity but also substantial toxicity. In contrast, OA-Pi-containing LNPs displayed a higher neutral-pH ζ-potential and severe acute toxicity, suggesting that headgroup chemistry and resulting nanostructural behaviour strongly influence biological performance.

A study by Summers et al. demonstrated that mast cells mediate cisplatin nephrotoxicity and renal injury, and confirmed the therapeutic potential use of the mast cell stabiliser sodium chromoglycate for the prevention of cisplatin-induced acute kidney injury [35]. Administering sodium chromoglycate along with the cis-loaded LNPs may be a viable path to reduce the renal toxicity.

The original contributions presented in this study are included in the article/. Further inquiries can be directed to the corresponding author. Supplementary Materials