Polyethylenimine-CO₂ adduct templated CaCO₃ nanoparticles as anticancer drug carrier

Abstract

Background

Due to their porous structure and capability to degrade under acidic conditions, CaCO₃ nanoparticles in vaterite form can be used as carriers to effectively deliver drugs to low-pH sites such as tumors. The usually used intravenous administration requires long-term vaterite phase and colloidal stability for storage and blood circulation. While passive accumulation in tumors can be achieved via the enhanced permeation and retention effect, active accumulation requires reactive groups on vaterite nanoparticles to conjugate targeting molecules. Both requirements are hard to achieve in one simple and economical vaterite formulation. Herein, we used polyethylenimine (PEI)-based CO₂ adduct as both a CO₂ source and a template for vaterite mineralization to generate PEI-CO₂@CaCO₃ colloidal particles, with reactive amino groups from the PEI template.

Results

The obtained nanoparticles with a hydrodynamic diameter of 200–300 nm have a vaterite phase and colloidal stability in an aqueous solution for over 8 months. These nanoparticles could effectively load anticancer drug doxorubicin via coprecipitation and be surface-modified with polyethylene glycol (PEG) and folic acid for long-term blood circulation and tumor targeting purposes, respectively. After being endocytosed, the PEI-CO₂ adduct accelerates the dissolution of drug-loaded nanoparticles to generate CO₂ bubbles to break the lysosomes, leading to rapid doxorubicin delivery inside tumor cells. The degradation of PEI-CO₂ in the CaCO₃ nanoparticles could also release PEI and CO₂ and may contribute to the disruption of normal cellular functions. As a result, the drug-loaded PEI-CO₂@CaCO₃ nanoparticles strongly suppressed tumor growth in mice with HeLa tumor xenografts.

Conclusions

A new and effective vaterite drug carrier for anticancer therapy has been developed using PEI-CO₂ adduct as both a CO₂ source and vaterite template for CaCO₃ mineralization. This delivery system illustrates an application of CO₂ generation materials in drug delivery and has the potential for further development.

Introduction

As the most abundant biomineral found in skeletons of most invertebrates and eggshells of some vertebrates, calcium carbonate (CaCO₃) has been widely used in biomedical engineering, such as in fillers for bone implants (Unger et al. 2022) and as drug carriers for targeted delivery (Popova et al. 2021; Som et al. 2016). Calcite, aragonite, and vaterite are anhydrous polymorphs of calcium carbonate, whereas amorphous calcium carbonate (ACC), ikaite, and monohydrocalcite belong to the hydrated forms (Morse et al. 2007). Among them, vaterite with crystals of typical nano sizes attracts the most attention due to its spherical shape and porous structure, although it might recrystallize in water to non-porous calcite within several hours (Volodkin et al. 2004). The ACC is the least stable crystal form and can dissolute in water, despite its nanospherical structure (Zhao et al. 2015). The most stable form of CaCO₃ is calcite. The in vivo lysosomal acidic environment cannot effectively decompose it to release the encapsulated drugs (Wang et al. 2018), limiting its application for drug delivery systems. On the contrary, vaterite is a pH-sensitive porous carrier with tumor-triggered intracellular drug release (Wang et al. 2018; Min et al. 2015). In addition, vaterite with good biocompatibility can encapsulate some fragile biomolecules (Balabushevich et al. 2015; Binevski et al. 2019) to avoid degradation in vivo, achieve controlled release, and even reduce the immune response resulting from the payloads.

Compared with organic drug delivery systems, such as liposomes and synthetic copolymers, vaterite nanospheres are cost-efficient (Vikulina et al. 2020), safe (Trushina et al. 2022), and with an intrinsic degradation in the acidic tumor microenvironment (Som et al. 2016). Besides, they can release CO₂ bubbles in tumor tissues (pH 6.8–7.2) as a contrast agent for real-time ultrasound imaging (Min et al. 2015); and once endocytosed into lysosomes (pH 4.0–4.5), they also release CO₂ (Wang et al. 2018), potentially blasting lysosomes for the payload escape to the cytoplasm. Lysosomal escape is essential for antitumor drugs including doxorubicin (DOX), which then diffuses into the nuclei to disrupt DNA synthesis there (Agudelo et al. 2016).

However, the drawbacks of vaterite particles as drug carriers are obvious. Bare vaterite particles lack phase and colloidal stability; this stability guarantees prolonged blood circulation prior to accumulation in tumors. Passive accumulation is due to the presence of leaky vessels in tumors, with pores from 100 nm to 2 μm in diameter (Kalyane et al. 2019), highlighting the suitable particle sizes of nanomedicines (e.g., several hundred nanometers). Too small particles tend to be cleared by the reticuloendothelial system in the liver and the spleen. In some cases, active accumulation is achieved by grafting with specific ligands (e.g., folic acid (Shi et al. 2012)) that recognize tumor cells. This, however, requires reactive groups that are absent on bare vaterite particles.

Some bioactive molecules (Yang et al. 2016; Dong et al. 2016; Xu et al. 2022; Kim et al. 2013) and synthetic polymers (Popova et al. 2021; Min et al. 2015) have been used to template vaterite mineralization. Vaterite particles with nano range sizes, as well as long-term phase and colloidal stability (Popova et al. 2021; Xu et al. 2022; Kim et al. 2013), have been achieved in some formulations. To add reactivity, vaterite particles have been coated with specifically designed lipid layers with reactive groups to conjugate with targeting molecules (Dong et al. 2016; Xu et al. 2022; Kim et al. 2013), which complicates the preparation process and largely increases the material cost. Until now, a simple and economical synthesis to achieve submicron size, long-term colloidal stability, as well as reactivity in one vaterite formulation remains a challenge.

As one of the most extensively studied polycationic materials, polyethylenimine (PEI) plays an important role in gene delivery therapy (Kircheis et al. 2001; Buscail et al. 2015; Pandey and Sawant 2016; Cao et al. 2021). Moreover, PEI contains CO₂-philic amino groups, making it an attractive CO₂ absorbent that has been extensively investigated for CO₂ capture (Shen et al. 2017). In this study, using the advantage that PEI absorbs CO₂, we used PEI-CO₂ adduct as a CO₂ source for CaCO₃ mineralization, leading to a mesoporous particulate form of CaCO₃ with a high surface area under mild conditions. The porous structure is perfect for the loading of both small and large biomolecules (Volodkin et al. 2004). The PEI as a template can maintain vaterite colloidal particles in an aqueous solution for over 8 months. Besides the high stability, the amino groups of PEI chains on vaterite particles provide reactive sites for further potential functionalization. Considering that a long circulation in the blood is necessary for tumor cellular internalization (Sun et al. 2014, 2017), poly(ethylene glycol) (PEG) was chosen to graft onto the vaterite nanoparticles through dynamic chemical bonds, making the particles “stealthy” in the blood (pH 7.4) but “activated” for cell internalization in acidic tumor tissues due to the departure of PEG chains (Chen et al. 2018). Doxorubicin (DOX) was selected as a model drug to be loaded into the vaterite CaCO₃ particles by the coprecipitation method. The strategy of tumor-triggered intracellular drug release is shown in Fig. 1, with folic acid (FA) as a tumor-targeting molecule (Shi et al. 2012).

Fig. 1

Schematic design of tumor-triggered intracellular drug release of DOX/PEI-CO₂@CaCO₃-PEG-FA

Experimental section

Materials

Polyethylenimine (PEI, M_n = 10 k Da) was purchased from Gobekie New Materials Co., Ltd. (Shanghai, China). Calcium hydroxide (95%), 4-hydroxy benzaldehyde (98%), and p-toluenesulfonyl chloride (99%) were purchased from Chron Chemical Co., Ltd. (Chengdu, Sichuan, China). Fluorescein 5-isothiocyanate (FITC, 95%), doxorubicin hydrochloride (DOX, 98%), methoxy-polyethylene glycol (PEG, M_n = 1,900), and folic acid (FA, ≥ 97%) were purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Fluo-4 AM was purchased from Beyotime Biotechnology Co., Ltd. (Shanghai, China), and Lyso-Tracker Red was purchased from Solarbio Life Sciences Co., Ltd. (Beijing, China). The L929 and HeLa cells were donated by the West China School of Pharmacy, Sichuan University, China. The BALB/c-Nude mice (SPF) were purchased from Gempharmatech Biological Co., Ltd. (Jiangsu, China). Dulbecco's modified Eagle's medium (DMEM) was obtained from Thermo Fisher Scientific (China) Co., Ltd. (Shanghai, China). Fetal bovine serum (FBS), 2-[4-(2-Hydroxyethyl)piperazin-1-yl]ethanesulfonic acid (HEPES), Cell-Counting Kit-8 (CCK-8), and 4',6-diamidino-2-phenylindole (DAPI) were purchased from Biosharp Life Sciences (China) Co., Ltd. (Hefei, Anhui, China).

Synthesis of α-methoxy-ω-p-formylphenyl poly(ethylene glycol) (PEG-CHO)

The PEG-CHO was prepared as described elsewhere (Heathcote et al. 2007; Zhu et al. 2010). In brief, 20.0 g of PEG and 2.0 g of NaOH were added to 50 mL of tetrahydrofuran (THF), followed by stirring in an ice bath for 30 min. Afterward, 7.6 g of p-toluenesulfonyl chloride (TsCl) dissolved in 30 mL of THF was dropped into the PEG solution. After 4 h of reaction, 100 mL of ice water was added, and the mixture was extracted with 200 mL of CH₂Cl₂ three times. The organic phase was washed with deionized water and saturated NaCl solution. The washed organic layer was evaporated and concentrated, and 250 mL of absolute ether was poured to obtain a white precipitate. After filtration, the precipitate was vacuum dried at 50 °C for 1 d to obtain a white solid powder (PEG-OTs). Then, 15.0 g of the white powder, 19.3 g of anhydrous K₂CO₃, 2.6 g of p-hydroxybenzaldehyde, and 100 mL of anhydrous CH₃CN were added into a flask and refluxed for 48 h at 85 °C in an N₂ atmosphere. The reaction product was distilled under reduced pressure to remove the solvent, and 200 mL of deionized water was added to dissolve the crude product. The dissolved product was extracted with 50 mL CH₂Cl₂ three times and precipitated with 500 mL absolute ether to obtain a white solid PEG-CHO.

Synthesis of polyethylenimine-CO₂ adduct templated CaCO₃ nanoparticles

The CO₂ atmosphere was introduced into 80 mL of 0.1 g/mL PEI aqueous solution for 48 h to obtain the PEI-CO₂ adduct (Additional file 1: Figure S1), into which 800 mL of Ca(OH)₂ solution (saturated at 25 °C) was slowly dropped under vigorous stirring at 37 ℃ for 6 h to obtain a PEI@CaCO₃ white emulsion, which was equally divided into four parts. The first part was spiked with a solution of 4 g PEG-CHO in 20 mL deionized water. The second part was spiked with 0.4 g of FA, which had been dissolved in 20 mL deionized water and activated with 0.35 g of 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDC) for 40 min. After adding the functioning molecules, the two parts were stirred at 37 ℃ for 24 h in darkness. The third part was allowed to react with 4 g of PEG-CHO and then with 0.35 g of FA, the same as those described for the first and second parts. The last part was kept without any treatment. The obtained PEI@CaCO₃-PEG, PEI@CaCO₃-FA, PEI@CaCO₃-PEG-FA, and PEI@CaCO₃ suspensions were dialyzed against water through a regenerated cellulose membrane (Biosharp, Hefei, Anhui, China) with a molecular weight cutoff (MWCO) of 10,000 Da for 2 d, with refreshing water at every 8 h. The dialysis retentates were rotary evaporated at 45 ℃ for 3 h and freeze dried at 20–30 Pa for 1 d. The resultant powders were ground under a CO₂ purge gas and then saturated with CO₂ at 0.6 Mpa for 24 h in a steel container to obtain the final products, namely PEI-CO₂@CaCO₃-PEG, PEI-CO₂@CaCO₃-FA, PEI-CO₂@CaCO₃-PEG-FA, and PEI-CO₂@CaCO₃.

Synthesis of doxorubicin-loaded PEI-CO₂@CaCO₃ nanoparticles

Thirty mL of DOX aqueous solution (5 mg/mL) was added into 60 mL of PEI-CO₂ solution (containing 6 g PEI) under vigorous stirring. Into the mixture, 600 mL of Ca(OH)₂ solution (saturated at 25 °C) was slowly dropped in 50 min. When the solution had turned translucent, the reaction mixture was ultrasonicated for 1–2 min to guarantee homogenous mixing and, thereafter, the reaction went on until a purple emulsion was formed, which was equally divided into three parts. The first part was kept without treatment, the second part reacted with 0.4 g FA, and the third part was grafted with 4 g PEG-CHO and then with 0.4 g FA. The reaction process was the same as that described for the corresponding CaCO₃ particles without DOX (See the above section). Each of the three suspensions with DOX was concentrated to about 100 mL via rotary evaporation at 45 ℃ for 2 h. The concentrated specimens were placed into 50-mL ultrafiltration tubes (MWCO 30 k Da, Millipore, MA, USA) and centrifuged at 9500 rpm (i.e., 11,600 g) (TGL-20 M, Shanghai Lu Xiangyi Centrifuge Instrument Co., Ltd., Shanghai, China). The retentates were further treated with another three cycles of resuspension-centrifugation to remove water-soluble impurities. The obtained precipitates were freeze dried, ground, and saturated with CO₂, as described for the corresponding CaCO₃ particles without DOX (See the above section). The final products were designated by adding “DOX/” before the sample code of the corresponding blank CaCO₃ nanoparticles (e.g., DOX/PEI-CO₂@CaCO₃).

The as-synthesized DOX-loaded CaCO₃ nanoparticles were vacuum dried at 100 °C for 1 h to release CO₂ from the PEI-CO₂ adduct in the particles, leading to specimens designated as DOX/PEI@CaCO₃, DOX/PEI@CaCO₃-FA, and DOX/PEI@CaCO₃-PEG-FA.

Characterizations of nanoparticles

The morphology of nanoparticles was characterized by scanning electron microscopy (SEM, Quanta 250, FEI, USA) and transmission electron microscopy (TEM, JEM-1200, JEOL, Japan). The polymorph of different nanoparticles was recorded by X-ray diffractometry (XRD, Ultima IV, Rigaku, Japan). Thermogravimetric analysis (TGA) was performed on a TG209F1 thermogravimetric analyzer (Netzsch, Germany). The structure of PEIs in the nanoparticles was analyzed by nuclear magnetic resonance (NMR, AVIII HD-400 MHz, Bruker, Germany) and X-ray photoelectron spectroscopy (XPS, Thermo Kalpha, Thermofisher, USA). Nanoparticles were dissolved in trifluoroacetic acid and then added to D₂O, ready for NMR analysis. The zeta potential and particle size of all nanoparticles were measured by particle size potentiometry (Zetasizer Nano ZSP, Marvin, UK).

pH-dependent decomposition of blank and drug-loaded PEI-CO₂@CaCO₃ nanoparticles

In vitro Ca²⁺ release experiments for various nanoparticles were performed in phosphate buffer solution (PBS) at pH 7.4, 6.5, or 5.5. The nanoparticle solution (10 mL at 1 mg/mL) was sealed in a dialysis bag (MWCO: 5000 Da) and shaken in 400 mL PBS with a predesigned pH value at 37 ℃. The PBS (4 mL) was withdrawn at predetermined time intervals and replaced with an equal volume of fresh medium. Into the withdrawn PBS, we added 2 mL Arsenazo III solution (0.2 mM, in HEPES-buffered saline at pH 7.4, where [HEPES] = 20 mM and [NaCl] = 150 mM). The absorbance of the solution at 656 nm was then measured, and the concentration of Ca²⁺ ions was calculated based on the standard curve.

To observe CO₂ bubble formation, about 2 mg DOX/PEI-CO₂@CaCO₃-PEG-FA and DOX/PEI@CaCO₃-PEG-FA powder were dispersed in 5 mL PBS (pH 7.4 or 6.5), respectively. The CO₂ release was observed under an optical microscope (Olympus BX 43, Japan) after 2 h.

In vitro DOX release profiles of drug-loaded nanoparticles were examined in PBS at pH 7.4, 6.5, or 5.5. Both the nanoparticle specimens with 600 µg DOX equivalent and free DOX of 600 µg were added to 10 mL PBS and transferred to a dialysis membrane bag (MWCO: 5000 Da). The latter served as a control. The release experiment was initiated by placing the dialysis bag in 100 mL of PBS at different pH values. The releasing medium was shaken at 37 ℃, and at predetermined time intervals, samples (5 mL) were withdrawn and replaced with an equal volume of fresh PBS. The concentration of released DOX in the samples was determined by the measurement of fluorescence emission intensity at 588 nm (excitation at 480 nm), based on the standard curve obtained using DOX. Note that the DOX loadings in CaCO₃ particles were determined in the same way after the dissolution of the CaCO₃ powders in HCl solution, and calculated as Drug loading (%) = Mass of encapsulated DOX/(Mass of carriers + Mass of encapsulated DOX) × 100%.

Cytotoxicity of blank and drug-loaded PEI-CO₂@CaCO₃ nanoparticles

For the cell viability measurements, HeLa cells in DMEM suspension were seeded in 96-well plates (5 × 10³ cells per well) and allowed to grow overnight. Subsequently, the medium was discarded, and all wells were washed twice with fresh PBS. Then, 200 μL of serum-free DMEM containing DOX or CaCO₃ nanoparticles at predesigned concentrations was added to each well. After 24 h and 48 h of co-incubation, cell viability was assessed with a CCK-8 assay. Briefly, the medium was replaced with an equal volume of fresh medium containing CCK-8 and incubated for 30 min at 37 °C. The viable cell number was proportional to the absorbance of each well at 450 nm, recorded by a microplate reader (Biorad Elizer, PA). The data were normalized to the cell viability of the control group that grew in the same DMEM without DOX or nanoparticles.

Endocytosis of blank PEI-CO₂@CaCO₃ nanoparticles

To monitor the cellular uptake of the blank PEI-CO₂@CaCO₃ nanoparticles, the nanoparticles were fluorescently labeled with FITC. In brief, 500 mg nanoparticle powder with 90 mg FITC dispersed in deionized water was stirred in the dark for 2 d and then freeze dried after dialysis (MWCO: 10 k Da) for another 2 d. The labeled nanopowder was redispersed in serum-free DMEM at a concentration of 50 μg/mL. Subsequently, HeLa or L929 cells suspended in 2 mL of DMEM supplemented with 10% FBS and 1% antibiotics were seeded onto round coverslips in individual wells of a 6-well plate (1 × 10⁶ cells/mL, 2 mL/well). After incubation to 80% confluence (5% CO₂, 75% humidity, 37 ℃), the medium was aspirated and replaced with 2 mL of the medium containing FITC-labeled nanoparticles. The cells were incubated for 4, 8, and 12 h, washed with PBS three times and fixed with 4% paraformaldehyde for 15 min. The fixed cells were stained with 10% DAPI for 5 min, washed with PBS again, sealed with antifading mounting medium (S2100, Solarbio, Beijing, China), and observed under a Leica TCS SP5II confocal laser scanning microscope (CLSM, Leica, Mannheim, Germany). In addition, the cells incubated with different nanoparticles for 12 h were digested with 0.25% trypsin, centrifuged and resuspended in PBS, and passed through a 60-mesh cell strainer to obtain a single-cell suspension; the uptake of nanoparticles was quantitatively analyzed by flow cytometry (FCM, BD LSRFortessa, USA).

To reveal the location of nanoparticles in cells, HeLa or L929 cells were seeded onto round coverslips in a six-well plate at a density of 1 × 10⁵ cells per well. After cell adherence, the original FBS-containing DMEM medium was replaced with an equal volume of suspension containing FTIC-labeled nanoparticles (10 μg/mL). After 4 h of incubation, cells were washed three times with PBS, followed by the addition of Lyso-Tracker Red (75 nM) medium to incubate for another 30 min. Then, the cells were washed with PBS again, fixed with paraformaldehyde, and finally stained with DAPI. Samples were subjected to CLSM observation. The location of Ca²⁺ ions in cells was observed in the same way, except that the nanoparticles were not labeled with FITC, and Ca²⁺ ions in the cells were labeled instead. In detail, the Ca²⁺ ions were labeled with Fluo-4 AM (5 µM) in the DMEM for 30 min after 4 h of incubation in the original medium. The cells were washed with PBS and further labeled with Lyso-Tracker Red to indicate lysosomes.

To observe the dynamic release of Ca²⁺ ions in cells, HeLa cells were incubated in the original medium for 3 h, and Ca²⁺ ions were labeled with Fluo-4 AM, as described previously. Live cell imaging was recorded every 5 min within 30 min under a CLSM.

Location of DOX/PEI-CO₂@CaCO₃-PEG-FA nanoparticles in cells by TEM

The HeLa cells were seeded into a 100-mm petri dish at 3 × 10⁷ cells/dish and incubated overnight in 15 mL of FBS-containing DMEM medium. The original medium was replaced with an FBS-free DMEM medium containing DOX/PEI-CO₂@CaCO₃-PEG-FA at 50 μg/mL. After 4 h of incubation, the cells were rinsed with PBS three times, scraped from the petri dish with a cell scraper, and placed in a centrifuge tube to obtain a mung bean-sized cell spheroid via centrifugation. The cell spheroid was fixed in glutaraldehyde solution at 4 ℃ overnight, embedded in paraffin, and sliced for observation under a biological transmission electron microscope (FEI-Tecnai SPIRIT, USA).

Intracellular drug release of various drug-loaded PEI-CO₂@CaCO₃ nanoparticles

The HeLa cells were seeded onto round coverslips in a six-well plate at 1 × 10⁵ cells per well and incubated overnight. The original medium was replaced with serum-free DMEM containing free DOX or DOX-loaded nanoparticles at 1 μg/mL DOX equivalent. After a predetermined incubation period, cells were washed several times with PBS, fixed with paraformaldehyde, and stained with DAPI, followed by CLSM observation.

In vivo biodistribution assay of blank PEI-CO₂@CaCO₃ nanoparticles

The BLBC/c-Nude mice were used for animal experiments. All animal experiments were performed in compliance with the guidelines for the care and use of laboratory animals. The tumor model was established subcutaneously by inoculating 1 × 10⁸ HeLa cells into the right back of a mouse. To avoid fluorescence from DOX, blank nanoparticles labeled with FITC were used to analyze the in vivo biodistribution. The labeled nanoparticles (5 μg in 100 μL PBS) were administrated via tail vein injection into tumor-bearing nude mice. At 24 h post-injection, the fluorescent images were recorded using an in vivo imaging system (IVIS Spectrum, Perkin Elmer, USA). After live imaging, mice were killed, and the main organs and tumor tissues were excised for ex vivo imaging using the same imaging system.

In vivo antitumor effect

When the xenograft tumor had reached around 50 mm³, mice were randomly divided into five groups (n = 7) and intravenously injected with 200 μL saline containing free DOX or DOX-loaded CaCO₃ particles at 1 mg/kg DOX equivalent. The group injected with 200 μL saline served as a control. The day of administration was defined as Day 0. The administration was performed every 2 d over a 14-d therapeutic period. Tumor volume and body weights were measured. At Day 14, three mice in each group were randomly selected and euthanized. The tumors were collected, weighed, washed with saline thrice, and fixed in 10% formalin. The fixed tumors were embedded in paraffin to prepare sections for hematoxylin–eosin (HE) staining. Tissue images were recorded at 200 × magnification under an optical microscope (Pannoramic 250, 3DHISTECH Ltd., Budapest, Hungary). The survival rate was calculated for up to 30 d.

Statistical analysis

Quantitative data were presented as mean ± standard deviation (S.D.) from triplicate experiments (n = 3) performed in a parallel manner unless otherwise stated. Statistical significance was tested by two-tailed Student’s t-test or one-way ANOVA, where P < 0.05 (*) and P < 0.01 (**) were considered statistically significant.

Results

Chemical structures and composition of as-synthesized CaCO₃ particles

The presence of aldehyde C=O stretching at 1690 cm^–1 and benzene ring breathing stretching at 1600 cm^–1 indicated the successful synthesis of PEG-CHO (Additional file 1: Figure S2). Accordingly, the signals of aldehyde proton at 9.98 ppm and benzene ring protons at 7.77 and 7.33 ppm appeared in the ¹H NMR spectrum of PEG-CHO (Additional file 1: Figure S2).

The organic chemicals in the as-synthesized CaCO₃ particles were characterized by ¹H-NMR in D₂O after the dissolution of the CaCO₃ particles with trifluoroacetic acid. As shown in Fig. 2, the hydrolysis of the Schiff base group in an acidic environment would restore the aldehyde group and, therefore, the proton signal from aldehyde could be found at 9.13 ppm in the PEI@CaCO₃-PEG specimen, with benzene ring protons at 6.43 and 7.22 ppm. The PEG protons (–CH₂CH₂O–, peak d) at 3.01 ppm partially overlapped with PEI ethylene signals, ranging from 2.25 to 3.36 ppm. After grafting with folic acid (FA), additional signals associated with FA appeared in the spectra of the PEI@CaCO₃-PEG-FA specimen, such as pyrimidine ring proton at 8.23 ppm (peak j). Reasonably, the PEI@CaCO₃-FA specimen displayed proton signals from both PEI and FA.

Fig. 2

¹H NMR spectra of three CaCO₃ particles dissolved in trifluoroacetic acid and D₂O. A Chemical structure of modified PEIs, where the PEG side chains had restored the aldehyde terminal groups (see red groups). B Assignment of the NMR spectra. The presence of Ca²⁺ had altered the chemical shift. For instance, the PEG proton signal usually appears at 3.5 ppm, but here, it appears at 3.0 ppm

All FTIR spectra (Fig. 3A and B) showed absorptions at 1485 and 875 cm^–1 due to the unique symmetrical stretching and bending vibration of CO₃²⁻ in CaCO₃. In addition, the weak N–H and C–H vibration peaks due to the presence of PEI backbones could be seen at 3300 and 2887 cm^–1, respectively. The specimens containing PEG chains displayed absorption of C–O–C stretching at 1100 cm^–1. After being loaded with DOX, the specimens showed a much stronger absorption at 3300 cm^–1, resulting from hydroxyl and amine groups in DOX, as well as an additional absorption at 1630 cm^–1 from the carbonyl groups in DOX. Moreover, the survey scan of the XPS analysis (Additional file 1: Figure S3) showed the presence of C, N, O, and Ca in PEI-CO₂@CaCO₃-PEG-FA. The high resolution of the C_1S scan displayed a major C = N(C–O) peak at 286.5 eV (from FA), being companied by a C–C peak at 284.5 eV (from PEI and PEG) and a C = O peak at 288 eV (from FA). The high resolution of the N_1S scan displayed a major C–N peak at 399 eV (from PEI and FA) and a minor C = N peak at 401 eV (from both FA and the Schiff base linkage between PEI backbone and PEG side chains).

Fig. 3

Structural analyses of bare (top) and drug-loaded (bottom) CaCO₃ nanoparticles

The XRD patterns (Fig. 3C, D) revealed that all CaCO₃ nanoparticles had a vaterite crystal form, with minor calcite crystals. All specimens showed a typical broad peak at 2θ of 25° due to the presence of amorphous PEI. Loading with DOX did not change the crystal form of CaCO₃.

The TG curves before and after drug loading, as shown in Fig. 3E, F, displayed three weight loss processes, centered at about 60 °C, 390 °C, and 740 °C, which can be assigned to the CO₂ release from PEI-CO₂ adduct, thermal decomposition of the organic substances, and the decomposition of CaCO₃, respectively. The detailed composition based on TG analysis can be found in Table 1. Grafting with PEG and FA could reduce the CaCO₃ content in the nanoparticles, and the loading of DOX (5.79–6.55%) could further lower it down to about 58%.

Table 1 Composition of as-synthesized CaCO₃ nanoparticles

Morphology and particle size of as-synthesized CaCO₃ particles

All specimens displayed spherical morphology, without obvious changes before and after being loaded with DOX (Fig. 4 and Additional file 1: Figure S4). A thin, gray layer was found around individual nanoparticles under TEM, which was due to the presence of PEI polymers (Fig. 4C). The average particle size was slightly enhanced after loading with DOX (Fig. 4E), whereas the zeta potential was decreased (Fig. 4F) a little. All specimens showed positive zeta potentials, consistent with the cationic PEI nature. Specimens with PEG chains demonstrated lower zeta potentials than their counterparts without PEG chains, suggesting a charge shielding effect due to the neutral PEG chains (Additional file 1: Figure S4).

Fig. 4

Morphological observation of PEI-CO₂@CaCO₃-PEG-FA nanoparticles A–D and the dynamic light scattering results E, F

In vitro particle decomposition and drug release of as-synthesized CaCO₃ particles

The Ca²⁺ release behavior of drug-loaded particles at different pH values is shown in Fig. 5A. The particles were stable at pH 7.4, and Ca²⁺ release was faster at lower pH values. The CO₂ saturation of PEI could accelerate Ca²⁺ release. A reduction in Ca²⁺ release rate was observed when the PEI-CO₂ adduct in CaCO₃ was heated to restore the PEI polyamine structure (Fig. 5B). The same trends of DOX release could be observed for DOX-loaded specimens (Fig. 5C, D), suggesting that DOX release was due to the acidic dissolution of CaCO₃ nanoparticles. As expected, free DOX diffused very quickly through the dialysis membrane (indicated by an arrowed line, Fig. 5D). In addition, the Ca²⁺ release behavior of blank CaCO₃ particles was almost the same as that of drug-loaded counterparts (Additional file 1: Figure S5), indicating that drug loading could not affect the dissolution of CaCO₃ particles. Besides the release of Ca²⁺, the formation of CO₂ bubbles could be observed as another indicator of CaCO₃ dissolution (Additional file 1: Figure S6).

Fig. 5

In vitro Ca²⁺ and DOX release behavior of drug-loaded CaCO₃ nanoparticles. The Ca²⁺ and DOX released more swiftly in the particles with PEI-CO₂ A, C than in those with PEI B, D. For free DOX, over 80% of DOX was released through the dialysis membrane in 2 h D

Cytotoxicity, endocytosis, and drug release into cells

The blank CaCO₃ nanoparticles showed toxicity to both HeLa and L929 cells, albeit only at high concentrations (e.g., 100 μg/mL) (Fig. 6A, B). At the same dosage, HeLa cells were less viable that L929 cells, possibly due to the strong metabolism of HeLa cells and the strong acidity of the cell environment, which caused a fast decomposition of the CaCO₃ nanoparticles. Among the blank CaCO₃ specimens, PEI-CO₂@CaCO₃-PEG demonstrated the lowest toxicity, due to the presence of a biocompatible PEG surface layer. After being loaded with DOX, obvious cytotoxicity was observed at very low dosages (Fig. 6C, D) for all specimens.

Fig. 6

Test of cytotoxicity of as-synthesized CaCO₃ nanoparticles. Cell viability is normalized to that of the control group in the DMEM medium

Various CaCO₃ nanoparticles labeled with FITC were incubated with HeLa and L929 cells and observed under a CLSM. As shown in (Additional file 1: Figure S7), both cells could phagocytose the CaCO₃ nanoparticles in 4 h, with HeLa cells containing more particles. The corresponding flow cytometry data (Fig. 7A) further revealed that HaLa cells phagocytosed more nanoparticles grafted with FA, whereas L929 cells responded equally to all types of CaCO₃ specimens. This suggests that the expression of FA receptors on HeLa cells might facilitate particle endocytosis.

Fig. 7

Endocytosis of various CaCO₃ nanoparticles. A Flow cytometry showed HeLa cells endocytosed more FA-grafted particles. B Fluorescence staining revealed both particles and the released Ca²⁺ ions located in lysosomes. Scale bars: 25 μm.  C Dynamic change of local Ca²⁺ fluorescent intensity after incubation with PEI-CO₂@CaCO₃-PEG-FA particles. Scale bar: 10 μm 

As shown in Fig. 7B, the fluorescence staining technique revealed that both the endocytosed CaCO₃ nanoparticles and the released Ca²⁺ ions were located in the lysosomes of HeLa cells. The same location could be found for L929 cells (Additional file 1: Figure S8). The internalized CaCO₃ nanoparticles entered lysosomes where the acidic environment would decompose CaCO₃ nanoparticles, releasing both Ca²⁺ and CO₂ in situ. To deeply explore the release behavior of CaCO₃ nanoparticles in cells, we observed the dynamic changes in local fluorescence intensity of Ca²⁺ ions from endocytosed PEI-CO₂@CaCO₃-PEG-FA particles. Within a period of 30 min (Fig. 7C), we observed a gradual increase (red arrows) or decrease (white arrows) of local Ca²⁺ fluorescence intensity, as well as the maintenance of the intensity (yellow arrows), corresponding to the dissolution of CaCO₃, Ca²⁺ ion escape from lysosomes, and the balance between the dissolution and the escape, respectively.

The TEM could directly reveal the morphological changes in lysosomes due to encapsulation of the as-synthesized CaCO₃ nanoparticles (Fig. 8). Besides vesicle-like normal lysosomes (L, Fig. 8), the black spheres (C, Fig. 8) in the view field indicated those lysosomes encapsulated with the CaCO₃ nanoparticles. A bursting lysosome was attached to a small bubble surrounded by a black thick wall (B1, left, Fig. 8) or composed of several bubbles with thick walls (B1, right, Fig. 8). Some burst lysosomes (B2, left and right, Fig. 8) displayed irregular main bodies with gray debris scattered far away. Within some debris, empty bubbles surrounded by thick walls could be seen as well. The black or gray regions were due to the presence of a CaCO₃ phase, and the bubble-like structure was the result of the formation of CO₂ bubbles that pushed away the surrounding CaCO₃ phase. The TEM images clearly proved the burst of lysosomes due to the dissolution of the encapsulated CaCO₃, which released CO₂ bubbles to blow and break the lysosomes.

Fig. 8

PEI-CO₂@CaCO₃-PEG-FA particles were entrapped in lysosomes imaged by TEM. L: normal lysosomes; C: particles in lysosomes; B1: bursting lysosomes; B2: burst lysosomes

The delivery of DOX into cells was associated with the endocytosis and dissolution of the CaCO₃ carriers (Fig. 9). Free DOX entered the nuclei in only 2 h, almost without DOX in the cytoplasm. This is because that DOX can easily bind to nuclear DNA (Agudelo et al. 2016), and free DOX cannot stay in the cytoplasm in large amounts (left, Fig. 9). However, the DOX-loaded CaCO₃ specimens delivered DOX (via lysosomes) into the cytoplasm and then into the nuclei (via lysosomal escape). This could be clearly seen in the “PEI@” group (middle, Fig. 9), where the PEI-CO₂ adducts in the CaCO₃ particles had been transformed into the corresponding PEIs. For this group, DOX began to appear in the nuclei at 4 h and further accumulated there at 6 h. The particle type affected the DOX delivery considerably. Those specimens containing FA demonstrated more DOX in the nuclei at the same timepoint. For the “PEI-CO₂@” group (right, Fig. 9), where the PEI-CO₂ remained in the particles, DOX spread in the cytoplasm and reached the nuclei in 2 h. Thereafter, DOX accumulated in the nuclei as well. No obvious difference could be observed across the specimens in this group. The relatively fast decomposition of the PEI-CO₂@CaCO₃ particles in the acidic environment (Fig. 5) was responsible for the fast DOX delivery into nuclei, masking the influence of the particle type.

Fig. 9

Particles were incubated with HeLa cells, and the encapsulated DOX was released into the cells. The particles with PEI-CO₂ released DOX more rapidly (right group) than did the particles with PEI (middle group). Red: DOX; Blue: nuclei. Scale bars: 25 μm

In vivo biodistribution and antitumor effects

The results of the in vivo distribution study (Fig. 10) showed that CaCO₃ nanoparticles were mainly concentrated in the liver and tumor in the HeLa-xenografted mice. The particle size was in the range of 200–300 nm, and the liver was the main metabolic organ; thereby, CaCO₃ nanoparticles accumulated in the liver. Besides the liver, nanoparticles could accumulate in the tumor via the passive enhanced permeability and retention (EPR) effect (Kalyane et al. 2019) or via FA-mediated targeting of tumor cells. Both could contribute to tumor accumulation in PEI-CO₂@CaCO₃-FA and PEI-CO₂@CaCO₃-PEG-FA specimens, whereas only the EPR effect played a role in the PEI-CO₂@CaCO₃ specimen, leading to lower tumor accumulation in the latter (Fig. 10C). The specimen containing PEG accumulated less in the liver than did the other two CaCO₃ specimens; PEG grafting reduced toxicity to the liver.

Fig. 10

Blank particles (without DOX) biodistribution in vivo

The DOX-loaded nanoparticles showed an antitumor effect (Fig. 11) in vivo. Among the saline- and nanoparticle-treated groups, tumor growth was significantly inhibited in the DOX/PEI-CO₂@CaCO₃-FA and DOX/PEI-CO₂@CaCO₃-PEG-FA groups, with a final tumor volume of 138 ± 82 mm³ and 124 ± 67 mm³, respectively, corresponding to a tumor inhibition rate of 79.6% and 81.7% with respect to the tumor volume (678 ± 107 mm³) of the saline group. The inhibition rate was 47.2% and 26.8% for the DOX/PEI-CO₂@CaCO₃ and the free DOX group, respectively (Fig. 11A). The highest survival rate (Fig. 11B) and the largest gain of body weight (Fig. 11C) both proved that DOX/PEI-CO₂@CaCO₃-PEG-FA possessed the best antitumor effect. Direct observation of the tumor size clearly displayed the different antitumor effects across the investigated groups (Fig. 11D).

Fig. 11

Antitumor effects of various DOX-loaded CaCO₃ particles. The saline and free DOX served as controls

The tumor was sectioned and observed histologically (Additional file 1: Figure S9). The specimens treated with DOX/PEI-CO₂@CaCO₃-FA and DOX/PEI-CO₂@CaCO₃-PEG-FA had the largest necrosis areas, with a significantly reduced density of cell nuclei. At the final stage of the treatment, the main organs from each experimental group were taken for histological observation (Additional file 1: Figure S10). Only the free DOX group had a certain organic toxicity, whereas the other groups showed no abnormalities, indicating that the as-synthesized CaCO₃ nanoparticles effectively reduced the toxicity of free DOX.

Discussion

The unique porous structure of the vaterite polymorph of calcium carbonate makes it a good carrier to deliver drugs (Zhao et al. 2015; Min et al. 2015; Dong et al. 2016; Dizaj et al. 2019) and even vaccines (Wang et al. 2018) for cancer therapy. Previous investigations adopted various polymers (Som et al. 2016; Min et al. 2015) and biomolecules (e.g., ovalbumin (Wang et al. 2018), hesperidin (Yang et al. 2016), and capsaicin (Xu et al. 2022)) to stabilize the metastable vaterite particles. These stabilizers usually possess carboxylate anions or chelating groups that bind to Ca²⁺ cations during CaCO₃ formation. The obtained particles often maintain the vaterite polymorph for several days (Volodkin et al. 2004; Trushina et al. 2014; Boyjoo et al. 2014), beyond which it transforms into a thermodynamically stable and nonporous calcite phase. We used PEI-CO₂ adduct to template CaCO₃ formation. The PEI-CO₂ as the CO₂ source restricted the nucleation and growth of CaCO₃ crystals within the branched PEI backbones. The calcium source was a saturated Ca(OH)₂ solution. In this context, no salt byproduct was formed, simplifying the purification process. Moreover, PEI-CO₂ released CO₂ to form CaCO₃, leaving the PEI chains being tightly entangled with the primary CaCO₃ crystals. The entangled PEI chains could stabilize the obtained vaterite phase as colloidal particles in water for over 8 months (Additional file 1: Figure S11). Such long-term stability has not been observed in previously synthesized vaterite crystals. For example, high molecular PEIs have been observed to induce vaterite CaCO₃ formation at air–water interface (Park et al. 2004), using Ca(HCO₃)₂ as the raw material. The vaterite phase remained for 3 d in water.

As most nanomedicines for cancer therapy are administered via vein injection, colloidal stability in an aqueous environment is a prerequisite for blood circulation prior to accumulation in the tumor. For CaCO₃ drug carriers, the vaterite phase should be kept until delivery of the payload into tumor cells. Therefore, both colloid and phase stability of a period of at least several days (e.g., 2 d) should be required for a vaterite drug delivery system. However, for a real nanomedicine used clinically, such stability should be months to years, considering the time for manufacture, transport, and storage. Our observation of stability for 8 months is highly valuable as long-term stability data for vaterite drug carriers are lacking.

The reactivity of the amino groups from the PEI chains on and in the CaCO₃ nanoparticles facilitates the further functionalization of the nanoparticles. As shown in this study, PEG could be grafted to extend blood circulation (Cao et al. 2021) and to reduce the toxicity of encapsulated DOX to remote organs (e.g., liver) (Fig. 10C). Once reaching the tumor tissues, the PEI chains would be detached from the nanoparticles due to the acidic breakdown of the Schiff base linkages. The cationic PEI chains on the nanoparticles would bind to the negatively charged cell membrane, facilitating endocytosis. In addition, FA could be also grafted to generate tumor targeting. It should be noted that PEG grafting and FA targeting have been widely used to enhance antitumor effects for decades (Greenwald 2001; Wang and Low 1998). The preferential biodistribution in mice tumor tissues (Fig. 10C) could account for the strong tumor inhibition of the drug-loaded vaterite nanoparticles that were grafted with FA (Fig. 11A), despite the fact that they possessed similar in vitro cytotoxicity as free DOX (Fig. 6C, D).

Previous investigations have shown several advantages of CaCO₃ in vaterite form as an antitumor drug carrier. First, it can dissolute in the acidic environment of tumor tissues (pH 6.8–7.2) to release CO₂, allowing for ultrasound imaging of cancers during chemotherapy via drug delivery (Min et al. 2015). Second, it can further dissolute in lysosomes (pH 4.0–4.5) after internalization to generate CO₂ bubbles to disrupt lysosomes, releasing the payload (Wang et al. 2018) to target subcellular counterparts. Compared with polymeric nanocarriers, CaCO₃ can easily be synthesized and is more economical but lacks in reactivity for further functionalization. Our formulation of CaCO₃ drug carriers keeps the intrinsic advantages of the vaterite polymorph and further provides reactivity due to the polyamine structure of the PEI stabilizer. Moreover, the PEI chains on and in the nanoparticles can absorb CO₂, forming the zwitterionic PEI-CO₂ adduct in situ. This adduct can release CO₂ in acidic environments, which can react with CaCO₃ to form water-soluble calcium bicarbonate (Ca(HCO₃)₂) (Park et al. 2004), accelerating the dissolution of CaCO₃. Therefore, PEI-CO₂-containing CaCO₃ particles released Ca²⁺ and DOX more swiftly than did the PEI-containing analogs (Fig. 5). Accordingly, faster delivery of DOX into the nuclei in the former was observed (Fig. 9). The additional CO₂ from the PEI-CO₂ templates would enhance the pressure of lysosomes, making them more easily disrupted and further accelerating DOX escape from lysosomes (Fig. 8). The PEI itself could act as a proton sponge, with the protonation in lysosomes causing osmotic swelling, which should also have contributed to the lysosomal rupture (Kircheis et al. 2001). The rapid delivery of DOX into nuclei must have contributed to the efficient therapy of the HeLa tumor in mice (Fig. 11).

The simple synthesis, long-term stability of the colloidal vaterite phase, ease of functionalization, and high CO₂ content of PEI-CO₂-templated CaCO₃ nanoparticles could provide a new platform for the delivery of therapeutic agents into cells, beyond the antitumor drugs shown in this study.

Conclusions

Here, we report a simple preparation of CaCO₃ nanoparticles using PEI-CO₂ and Ca(OH)₂ as the CO₂ and calcium sources, respectively. The PEI chains could template the formation of the vaterite polymorph of CaCO₃, stabilize the hydrocolloidal vaterite particles, and provide reactive amino groups for further potential functionalization. Both PEG and FA were grafted onto the CaCO₃ nanoparticles via acid-cleavable Schiff base linkage and relatively stable amide linkage, respectively. The PEI chains in the nanoparticles could further absorb CO₂ to restore the PEI-CO₂ adduct. The functionalized porous vaterite particles could deliver DOX into HeLa cells via lysosomes. During the endocytosis process, the PEI-CO₂ adduct could release extra CO₂ to accelerate the CaCO₃ dissolution in acidic lysosomes, facilitate the disruption of lysosomes, and rapidly deliver DOX into the nuclei, demonstrating an in vivo antitumor effect against HeLa xenograft in mice.