A facile aqueous production of bisphosphonated-polyelectrolyte functionalized magnetite nanoparticles for pH-specific targeting of acidic-bone cells

Bone malignancy treatment is being hindered due to the insufficient selectivity of therapeutic nanoparticles towards malignant bone sites. Polyelectrolyte functionalized magnetic nanoparticles having dually specific pH-sensing ability and bisphosphonate moieties, can be an effective solution for selective targeting of bone malignancies. First, polyelectrolyte was prepared via N-carboxycitraconyzation of chitosan (NCCS) followed by successive functionalization with alendronic acid (AL) and fluorescein isothiocyanate (FITC). Then, Fe3O4-NCCS-FITC-AL nanoparticles were synthesized by a facile one-step microwave-assisted aqueous method via in situ surface functionalization. The formation, crystal structure, and surface conjugation of Fe3O4 nanoparticles with polyelectrolytic stabilizer were confirmed by Fourier transform infrared spectroscopy, X-ray diffraction, and thermogravimetric analyses. Synthesized Fe3O4-NCCS-FITC-AL nanoparticles were superparamagnetic, colloidally stable and highly hemocompatible under physiological conditions. Moreover, at pH 5.0, Fe3O4-NCCS-FITC-AL nanoparticles formed a precipitate due to inversion of their surface charge. This pH-dependent charge-inversion drastically changed the interactions with erythrocytes and bones. Selective membranolysis of erythrocytes occurred at pH 5.0. The designed nanoparticles showed enough potential for selective targeting of pathological bone sites in early-stage magnetofluorescent imaging and as a therapeutics carrier to treat malignant bone diseases.


Introduction
Metastatic and metabolic bone malignancies occur in a wide range of ages including children, and the symptoms such as severe bone pain are distressing to various patients. 1 Current therapeutics like those based on selective estrogen receptor modulators and bisphosphonates are poorly bioavailable, require frequent and high dosing, and cannot precisely target disease sites. 2 Therefore, these drugs are oen ineffective due to their high systemic toxicity and off-targeted side effects including arterial pain, osteonecrosis of the jaw, musculoskeletal pain, and ulcers. 3,4 Various nanomaterials, such as polymeric micelles, 5,6 liposomes, 7 and gold, 8 and magnetic nanoparticles with targetable functional moieties are being used to target the pathological sites for diagnosis and therapy of bone malignancies. [6][7][8][9][10][11][12][13][14][15] Among them, magnetic iron oxide nanoparticles (MIONPs) have attracted signicant research attention. For instance, MIONPs have been used for detoxication of biouids, 10,11 drug targeting vehicles, [13][14][15] thermal ablation of tumor cells via apoptosis, 16,17 and contrast agents for magnetic resonance imaging. [13][14][15] MIONPs can be synthesized through nonaqueous routes like solvothermal, microemulsion, high-temperature decomposition of organo-iron precursors, and aqueous paths, such as hydrothermal and coprecipitation methods. However, non-functionalized MIONPs are highly prone to aggregate because of their strong magnetic attraction forces and high surface energy. The colloidal instability leads to opsonization and fast recognition by macrophages and mononuclear phagocytes. These unmodied MIONPs are rapidly removed through the reticuloendothelial system (RES). 13,15,18 So, the RES escaping is a vital challenge to any MIONPs used in diagnosis and therapy. In addition, the size, shape, surface functionality, nature of surface charge and colloidal stability of MIONPs also affect the evasion of RES. The RES engulfment can be minimized by prolonging circulation time of MIONPs via functionalization with mimicking ligands that masks phagocyte surfaces as well as coating their surface with hydrophilic polymers. 14 Therefore, appropriate surface functionalization of MIONPs is essential for bio-related applications because the functional moieties rst encounter with biological entities. 19 So far, various materials having different functional moieties like poly(ethylene glycol) (PEG), poloxamer, polyvinyl alcohol, polyaminoacid, carbohydrates, carboxylate and bisphosphonated polymers have been introduced on the surface of MIONPs. [20][21][22][23][24][25][26][27][28][29] As a result, surface functionalized particles having relatively wider size of 100-200 nm can easily be accumulated and retained in the targeted sites due to their enhanced permeability and retention effect.
Here, bisphosphonate moiety consisting of P-C-P bond is investigated as a bone targeting, pH-sensing, and dispersing milieu for MIONPs. Because, small organic molecules of bisphosphonates have been found useful to treat various bone malignancies like chondrosarcoma, osteosarcoma, Ewing's sarcoma, inammation of bones due to rheumatoid arthritis, periodontal disease, and so on. 3,15 They are also efficient to inhibit bone resorption due to their ability to bind bone mineral by bi-and tridentate ligation to Ca 2+ ions via osteoclastic differentiation and apoptosis. 3,15 Until now, a number of targeting nano-systems based on bisphosphonates have been developed, and some of them have been tested preclinically for bone metastasis. 3,9,10,[12][13][14][15]30,31 But bisphosphonates employed for bone-targeting are non-specic, and as a result, they target both healthy and malignant bone tissues without selectivity resulting in poor performances of therapeutics and adverse side effects. 3 Therefore, development of synthetic methods for MIONPs with targeting ability towards malignant bone cells via facile and biosafe procedures is still a great challenge. pH-responsive systems may introduce selectivity to therapeutics based on bisphosphonatefunctionalized MIONPs for metastatic bone diseases as well as primary bone malignancies. [32][33][34][35][36] Early-stage detection of bone malignancy enables more options for better treatments, resulting in higher survival rates and enhance quality of life. [37][38][39][40][41][42][43][44][45][46][47] Ferreira et al. developed bone-targetable pH-sensitive bisphosphonated liposomes and evaluated their cytotoxicity, cardiotoxicity, biodistribution, and used them for treating bone metastasis. 7 Chen et al. reported a bisphosphonated amphiphilic hyperbranched polymer and PEG-based micelles for targeting drugs to maligned bone cells. 6 Stewart et al. reported bisphosphonated acid-sensitive drug-loaded micelles for the treatment of osteosarcoma. 5 Au et al. designed pH-responsive folate conjugated nanoscale MOF bearing bisphosphonate moieties for treating a tumor. 34 The pH responsibility of 1,2dioleoyl-glycero-3-phosphatiethanolamine and cholesteryl hemisuccinate, ester/amide, hydrazone and phosphonate moieties are the key of these pH-sensitive systems.
To the best of our knowledge, there is no report on acidsensible charge-inversional bisphosphonated MIONPs for targeting and treatment of malignant bone cells. The present research focuses on a facial aqueous synthetic route to synthesize a novel magnetic nano-biosystem using a polyelectrolytic stabilizer for selective targeting of acidic bone cells. The polyelectrolyte stabilizer was rst prepared via N-carboxycitraconyzation of chitosan (NCCS) followed by successive functionalization. Then, in situ synthesis of Fe 3 O 4 and concomitant functionalization was carried out to obtain Fe 3 O 4 -NCCS-FITC-AL nanoparticles. The designed nanoparticles would enhance therapeutic efficacy and minimize systemic and off-targeted side effects. In addition, the conjugated polyelectrolytic-NCCS-FITC-AL layer is supposed to improve hemocompatibility, biocompatibility, acid-sensing chargestealth ability, and selective targeting ability of the MIONPs towards malignant bone sites. We selected AL molecule as a bisphosphonate for bone-targeting component, carboxylate moieties for the acid-sensing, and CS for the biocompatible and reactive scaffold of the polyelectrolytic coating. Fluorescein isothiocyanate (FITC) was employed as a uorescent probe for facile imaging and detection of materials. The formation, crystal structural, colloidal stability, pH-sensitive chargeinversion ability, hemocompatibility, magnetic property of the nanoparticles were conrmed and evaluated. Erythrocytes membranolytic efficiency, uorescent properties as well as selectivity of the Fe 3 O 4 -NCCS-FITC-AL nanoparticles towards bone mineral (HAp) and a native bone sample under physiological and acidic conditions were also studied.

Synthesis and characterization of polyelectrolytic NCCS-FITC-AL stabilizer
A simple three-step procedure was used to synthesize polyelectrolytic NCCS-FITC-AL (Scheme 1). In the rst step, NCCS was synthesized by partial amidation of the primary amine groups in CS with citraconic anhydride (CAn) for improving dispersity in aqueous media with pH ranging from acidic to weakly basic by the carboxy group. The carboxy group in NCCS is easily transformable to the mono-Na salt via neutralization with aqueous NaOH. Synthesized NCCS is an amorphous ne white powder (Fig. S1a †), whose dispersity was dramatically improved than CS. It was fully dispersible in neutral deionized distilled water (DDW) due to the addition of hydrophilic carboxylate moieties to CS. The uorescent group was conjugated in the second step through the conventional nucleophilic addition of the primary amine moieties of NCCS to the isothiocyanate groups of FITC. The resulting product is a yellow-colored ne powder (Fig. S1b †). The NCCS-FITC chain possesses the primary amine, hydroxy, carboxy, and unsaturated ester groups that can be easily conjugated with suitable reagents. Among them, the highly electron-decient unsaturated ester groups are very prone to nucleophilic addition of amines even under slightly acidic aqueous environments. In the last step, the bisphosphonate moieties were partially introduced via aqueous aza-Michael addition of the primary amine moieties of AL to the alkenyl groups of NCCS-FITC under mildly acidic conditions to form the desired polyelectrolyte, NCCS-FITC-AL. The formed polyelectrolyte was also a light-yellow-colored (Fig. S1c †) very ne powder that was also completely dispersible in DDW.
The successful synthesis of NCCS-FITC-AL was conrmed by energy-dispersive X-ray (EDX) (Fig. S2 †), Fourier transform infrared (FTIR) (Fig. S3 †), 1 H NMR (Fig. S4 †), and UV-vis ( Fig. S5 †) spectroscopic studies. The EDX spectroscopic analysis showed the respective elements present in the NCCS, NCCS-FITC, and NCCS-FITC-AL structures. Especially Na and P originated from the carboxylate and bisphosphonate moieties. The characteristic bands appeared in the FTIR spectra of NCCS, NCCS-FITC, and NCCS-FITC-AL indicated the introduction of the carboxy (n C]O at 1406 cm À1 ), uorescent (n C]C at 1450-1600 cm À1 ), and bisphosphonate (n P]O at 1108 cm À1 ) moieties on CS (Fig. S3 †). The 1 H NMR spectroscopic analysis of CS, NCCS, NCCS-FITC, and polyelectrolytic NCCS-FITC-AL further conrmed the modications of CS by the signals indicating the introduction of the alkenyl and methylene groups (Fig. S4 †). The introduction ratio of the N-citraconyl group was calculated to be 44% to the amine groups from the integral ratio of the peaks assigned to the acetyl and alkenyl protons of NCCS. The introduction ratio of the bisphosphonate moieties was calculated to be 53% with respect to the N-carboxycitraconyl group. The presence of the FITC residues in the NCCS-FITC-AL structure was also conrmed by an absorption band at 495 nm observed in the UV-vis spectrum (Fig. S5 †).
In situ aqueous synthesis of Fe 3 O 4 nanoparticles and their concomitant surface functionalization with polyelectrolytic NCCS-FITC-AL The charge-inversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles were prepared by a modied coprecipitation method (Scheme 2). This method involved with in situ production of Fe 3 O 4 nanoparticles and concomitant surface functionalization with polyelectrolytic stabilizer, NCCS-FITC-AL, via nucleation, seed generation, and particle growth. 49,50 When the solutions of the Fe(II) and Fe(III) precursors were added to the dispersion of NCCS-FITC-AL under N 2 , the iron ions were accumulated to the carboxy and bisphosphonate moieties. Rapid nucleation of Fe 3 O 4 particles was occurred by the addition of aqueous ammonia solution through hydroxylation and dehydration of the localized iron ions. Then, the dispersion was transferred to a microwave (MW) oven, where MW was irradiated to grow particles at 100 C for 30 min. In this method, NCCS-FITC-AL served as a scaffold to control the growth of the nanoparticles. NCCS-FITC-AL can also limit the active mass transfer of iron compounds to the surface of the  (Fig. S6d †). 13 This result demonstrates that Fe 3 O 4 nanoparticles were predominantly formed over Fe 2 O 3. The oxidation of the iron(II) ions in Fe 3 O 4 did not occur by the phosphonate groups. 13,51 The characteristic bands of the phosphonate group generally appeared in the range of 900-1200 cm À1 . Therefore, the precise locations of the bands of P]O and P-O bonds are quite difficult to assign as they are overlapping with other bands such as those of the C-O and C-O-C moieties in NCCS-FITC-AL. Hence, the conjugation of functional groups like carboxylate and phosphonate to the Fe 3 O 4 surface was conrmed by careful analysis on the comparison of the magnied spectra of Fe 3 O 4 -NCCS-FITC-AL and NCCS-FITC-AL (Fig. 1B, 700-1500 cm À1 ). The stretching vibration for P]O in the phosphonate moieties in NCCS-FITC-AL was observed at 1108 cm À1 . When the phosphonate groups interacted with iron atoms, the stretching vibration for P]O signicantly changed from 1108 to 1133 cm À1 . The stretching bands of P-OH were also signicantly shied from 1040 to 1033 cm À1 and 906 to 902 cm À1 . In addition, a new peak appeared at 1030 cm À1 assignable the stretching vibration mode of Fe-O-P indicating an effective conjugation of the phosphonate moieties with the surface iron atoms on the Fe 3 O 4 nanoparticles. 13,15,32 On the other hand, the stretching vibration bands of the carbonyl group in the carboxylate ions were also shied from 1410 to 1396 cm À1 and 1376 to 1372 cm À1 due to the bidentate chelation to iron atoms on the surface of the Fe 3 O 4 nanoparticles. In addition, the peak intensity of the absorption of the carboxylate ion was signicantly lowered than that of the phosphonate structure, which demonstrated dominant coordination of the carboxylates to the Fe 3 O 4 nanoparticles, and the bisphosphonate moieties mostly remained intact for targeting. The other peaks related to the structure of NCCS-FITC-AL also appeared in shied positions. For instance, some medium-strong bands were observed at 1645, 1625, and 1617 cm À1 , and these bands are assignable to the stretching bands of amide-I and II and C]C overlapped with the bending bands of the -OH, pNH, and -NH 2 moieties. Two distinct sharp bands were visible at 2922 and 2853 cm À1 assignable to the C-H stretching of the CH 2 groups in the amino sugar units and AL. A broad band appeared at 3000-3500 cm À1 is assignable to the stretching vibrations of O-H, pNH, -NH 2, and ]C-H bonds.

Morphology and core-structure of Fe 3 O 4 -NCCS-FITC-AL nanoparticles
The solid-and liquid-state morphologies and core-structure of the Fe 3 O 4 -NCCS-FITC-AL nanoparticles were analyzed by scanning electron microscopy (SEM), transmission electron microscopy (TEM), dynamic light scattering (DLS), and X-ray diffraction (XRD) (Fig. 2). The SEM image ( Fig. 2A) revealed that Fe 3 O 4 -NCCS-FITC-AL nanoparticles are singly dispersed and almost spherical. The average diameter of Fe 3 O 4 -NCCS-FITC-AL nanoparticles was calculated to be 27 nm, and the size-distribution histogram is shown in Fig. 2B. The welldispersed state demonstrates that polyelectrolytic NCCS-FITC-AL effectively covering the magnetic-Fe 3 O 4 core reduced the inter-particle attractions through steric and electrostatic repulsions. The morphology of the core of the Fe 3 O 4 -NCCS-FITC-AL nanoparticles was analyzed by TEM (Fig. 2D). The TEM image revealed almost similar shapes of the nanoparticles, while the obtained average core size is one-fourth of the size revealed by the SEM image showing the morphology of the particles covered by the polymeric coating in a dried state. Fig. 2C shows the DLS proles of the Fe 3 O 4 -NCCS-FITC-AL nanoparticles before and aer microwave treatment. The average hydrodynamic diameter (D h ) of the nanoparticles was estimated to be 160 nm (black) with a PDI of 0.30. The D h of the Fe 3 O 4 -NCCS-FITC-AL nanoparticles was increased from 160 to 190 nm (red) aer MW treatment and the PDI was also signicantly lowered from 0.30 to 0.18. The increment of the size is ascribable to the growth of the particles during the MW-irradiation. The difference between the sizes observed by DLS and SEM is ascribable to the thickness of the swollen layer of NCCS-FITC-AL 52 and the Brownian motion of the particles. XRD analysis was conducted to determine the crystal structure of the core component inside the

Magnetism of Fe 3 O 4 -NCCS-FITC-AL nanoparticles
The magnetic response of the dispersed Fe 3 O 4 -NCCS-FITC-AL nanoparticles in Dulbecco's phosphate-buffered saline (DPBS) was tested using a permanent magnet. The Fe 3 O 4 -NCCS-FITC-AL nanoparticles are fully dispersible in the absence of external magnetic elds even at the concentration as high as 70 mg mL À1 (Fig. 3A). By contrast, in the presence of a strong  28 and a cationic polyelectrolyte (46 emu g À1 ) 48 prepared without using MW treatment, while it is still lower than that of the typical bulk Fe 3 O 4 (98 emu g À1 ). The higher M s value of the composite particles is ascribable to the presence of the iron phosphonate structures on the surface of the particles, which were secured more stabilized spin contributions than that of the magnetically inactive dead surface layer. 13,14 The M s augmentation has probably originated from the high crystallinity of the Fe 3 O 4 core. The obtained result validates that the charge-inversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles have enough magnetic efficacy for magnetically-induced manipulation for in vivo applications. This MW-assisted synthesis achieves both the excellent magnetic property and the high physiological stability discussed later, which are prerequisites for bio-related applications.

Optical properties of Fe 3 O 4 -NCCS-FITC-AL nanoparticles
The optical behaviors of the charge-inversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles were analyzed by UV-vis absorption, uorescence excitation and emission spectroscopies (Fig. 4). The polyelectrolytic stabilizer, NCCS-FITC-AL, shows a strong absorption band at 495 nm in the UV-vis spectrum. This clear band originates from the FITC moieties of the stabilizer (Fig. S5 †). The UV-vis spectrum of Fe 3 O 4 -NCCS-FITC-AL also showed a shoulder peak around 494 nm for the absorption of Fe 3 O 4 core, while the intensity of the absorption by the FITC moieties was lower than that of the polyelectrolytic polymer dispersion due to the lower relative content of the FITC moieties in the composite. The emission spectrum of Fe 3 O 4 -NCCS-FITC-AL showed an emission peak with the maximum at 516 nm by excitation at 500 nm. Both the emission and excitation peaks accord with the typical uorescent behavior of the FITC moieties even in the presence of the carboxy and bisphosphonate moieties and the Fe 3 O 4 core. 54,55 The observable uorescence is sufficient for imaging as described later. This nding agrees well with previous reports. 7,34,54 pH-dependent aggregation-dispersion, charge-inversion, and physiological stability of Fe 3

O 4 -NCCS-FITC-AL nanoparticles
The aggregation-dispersion and charge-inversional behaviors of the Fe 3 O 4 -NCCS-FITC-AL nanoparticles were examined at a controlled pH range of 3.5-9.0 (Fig. 5). The Fe 3 O 4 -NCCS-FITC-AL nanoparticles were initially well-dispersed in DBPS at the examined pH range (Fig. 5A), while selectively aggregated only at pH 5.0 aer 2.5 h (Fig. 5B). This pH responsibility was further studied by DLS. The initial D h of Fe 3 O 4 -NCCS-FITC-AL was almost identical (190-200 nm) regardless of the pH values, and the D h at pH below 4.5 and above 6.5 was stable. By contrast, the D h at the pH range of 5.0-6.0 was gradually increased during the  incubation. The increment of D h at pH 5.0 was most signicant, and the D h aer 2.5 h reached 690 nm. Aer that, the nanoparticles were fully settled down and the size exceeded the measurable limit of DLS. This pH-dependent variation of D h can be explained by the z-potential (Fig. 5C). At the pH range of 3.5-4.5, the Fe 3 O 4 -NCCS-FITC-AL nanoparticles have positive zpotentials stabilizing the dispersion by electrostatic repulsions of the ammonium and carboxy groups formed by the protonation of the amine and carboxylate groups. By contrast, at pH 5.0 where the Fe 3 O 4 -NCCS-FITC-AL nanoparticles were selectively precipitated and pH 5.5 and 6.0 where the D h was increased, the z-potentials became less positive close to zero charge. The precipitation and aggregation of the nanoparticles occurred by the inter-particle attractions among the zwitterionic surface structure. The z-potentials were decreased as the increase of the pH values due to the partial deprotonation of the ammonium structure transforming into the amine groups. The weakly positive z-potentials in this precipitating region, not isoelectric and higher than those of our previously reported material 56 implies that a part of the cationic groups does not contribute to the precipitation behavior. A plausible reason is the lower concentration of the free bisphosphoric and carboxyl groups around the surface contributing to the ionic crosslinkage due to the ligation to the Fe 3 O 4 core as suggested by the FTIR spectroscopic study. Above pH 6.0, the Fe 3 O 4 -NCCS-FITC-AL nanocomposite particles showed negative z-potential due to the deprotonation of carboxy and phosphonic groups to form carboxylate and phosphonate groups, respectively, resulting in stable dispersion by the ionic repulsions between the nanoparticles. This pH-dependent charge-conversion property of the nanoparticles having targetable bisphosphonate moieties would be benecial for selective targeting of acidic bone cells by adhering to the negatively charged cell surface and endocytosis. The physiological stability of the charge-inversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles dispersed in saline was conrmed by studying their D h and PDI at physiological conditions (pH 7.4, 37 C) for 3 weeks (Fig. 5E). Both the D h ($200 nm) and PDI (below 0.20) of the particles were not signicantly changed throughout the observation period because of the electrostatic repulsions among the negatively charged carboxylate and phosphonate moieties of the polyelectrolytes. This result demonstrates that this magnetic and uorescent nanosystem has capabilities of prolonged circulation and selective deposition in locally acidic environments.

Hemocompatibility of Fe 3 O 4 -NCCS-FITC-AL nanoparticles
Hemocompatibility is one of the important prerequisites of biomaterials for intravenous applications. The concentrationdependent hemolytic behaviour of Fe 3 O 4 -NCCS-FITC-AL was examined using sheep erythrocytes in saline at pH 7.4 and 37 C for 5 h. Sheep erythrocytes dispersed in a 1% aqueous solution of Triton X 100 and saline were employed as the positive and negative controls, respectively (Fig. 6A). The treatment with Triton X-100 completely hemolyzed the erythrocytes through membrane disruption, and the color was turned into dark red, while the red color of the erythrocytes was retained in the negative control sample. The colors of the erythrocyte suspensions containing Fe 3 O 4 -NCCS-FITC-AL gradually became slightly darker by trace degrees of hemolysis enhanced as the increase of the concentration of Fe 3 O 4 -NCCS-FITC-AL (Fig. 6A). At pH 7.4, Fe 3 O 4 -NCCS-FITC-AL nanoparticles showed only 0.21% hemolysis at 100 mg mL À1 , while they showed hemolysis approximately three times higher at 400 mg mL À1 (Fig. 6C). The hemolysis degrees of Fe 3 O 4 -NCCS-FITC-AL are signicantly lower than that of previously reported magnetic nanocomposite particles coated with cationic polyelectrolyte, 48 carboxy-CS, 28 and functionalized carboxy-CS. 56 The morphological change of the sheep erythrocytes in the presence and absence of Fe 3 O 4 -NCCS-FITC-AL was conrmed by optical microscopy (Fig. 6B). Most of the erythrocytes incubated with Fe 3 O 4 -NCCS-FITC-AL retained their native morphologies in the same manner with the negative control (Fig. S8 †). The improved hemocompatibility of Fe 3 O 4 -NCCS-FITC-AL plausibly originates from the weakly negative surface charge, which is smaller than natural erythrocytes, resulting in negligible interactions with the negative membrane surface of erythrocytes under these conditions. The hemocompatibility of Fe 3 O 4 -NCCS-FITC-AL meets the acceptable limit of blood-contacting biomaterials (5%) under the physiological conditions, 57,58 indicating the potential of Fe 3 O 4 -NCCS-FITC-AL for intravenous applications.

pH-triggered membrane disruption of erythrocytes by Fe 3 O 4 -NCCS-FITC-AL nanoparticles
The pH-and concentration-dependent membrane disruption abilities of the charge-inversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles were tested by hemolysis assay using sheep erythrocytes as a model cell (Fig. 7). The optical images (Fig. 7A)   the negative surface charge of the nanoparticles that lead to enhanced adherence to the erythrocytes membrane surface. The hemolysis degree became signicantly higher (18.7%) at the pH range of 5.5 to 5.0, where the negative surface charge of the Fe 3 O 4 -NCCS-FITC-AL nanocomposite particles changes to positive (Fig. 7B). The positive charge originates from the protonation of the acid-sensing secondary amine, carboxy, and phosphonate groups, enhancing electrostatic interactions with the negatively charged lipid membrane of the erythrocytes under the acidic environments. Another plausible reason for the increase of the hemolysis degree is the hydrophobic interactions between the pendant hydrophobic structures in the polyelectrolyte coating with the lipid membrane of the erythrocytes. At pH 5.0 in the presence of the charge-inversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles, the erythrocyte membranes were signicantly damaged and the erythrocytes were obviously agglutinated (Fig. S9 †). This pH-sensible membrane disruption suggests the acidity-responsible membranolytic ability of Fe 3 O 4 -NCCS-FITC-AL nanoparticles at the pH range from 5.0 to 5.5 identical to the endosomal pH. Similar behaviour was also reported for MIONPs for selective disruption of cancer cells only under the acidic pH. 58,59 Fe 3 O 4 -NCCS-FITC-AL meets a requirement for intravenous applications that nanoparticles spontaneously become membranolytic under the acidic conditions with high hemocompatibility under the physiological conditions.
The selective interactions and adherence of the chargeinversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles to the erythrocyte membranes were further conrmed by uorescence microscopy (Fig. 8). At pH 7.4, the erythrocytes show their regular biconcave circular shape in the light microscopic image (Fig. 8A), and uorescent parts are invisible in the uorescent microscopic image of the erythrocytes (Fig. 8C). These images indicate that Fe 3 O 4 -NCCS-FITC-AL charged negatively under physiological conditions without interaction with the erythrocytes. By contrast, the erythrocyte membranes were signicantly broken at pH 5.0 due to the strong adherence of positively charged Fe 3 O 4 -NCCS-FITC-AL on the erythrocyte membranes (Fig. 8B). The accumulation of Fe 3 O 4 -NCCS-FITC-AL on the erythrocyte membranes was clearly conrmed by the uorescent part in the uorescence image matching with the morphology observed in the optical image (Fig. 8D). This observation implies the practical application of Fe 3 O 4 -NCCS-FITC-AL for uorescence imaging visualizing cellular interactions.

pH-dependent bone mineral affinity of Fe 3 O 4 -NCCS-FITC-AL
The charge-inversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles were designed for bone-seeking diagnosis and therapeutic applications. Therefore, we investigated the pH-dependent affinity of Fe 3 O 4 -NCCS-FITC-AL towards bone minerals. HAp with a monoclinic structure having 2.5 mm of the average length was used as a model bone mineral. HAp has the chemical formula of 3Ca 3 (PO 4 ) 2 $Ca(OH) 2 , and its surface composes of Ca 2+ , OH À , and PO 4 3À ions with the protonated phosphate ions depending on pH and coexisting ions. 61 HAp and Fe 3 O 4 -NCCS-FITC-AL were incubated at 37 C in saline and the dispersion pH was adjusted to acidic (5.0) and physiological (7.4) conditions. The adhesion ability of the Fe 3 O 4 -NCCS-FITC-AL nanoparticles to HAp surface was studied by SEM and EDX analyses (Fig. 9). Blank experiments without Fe 3 O 4 -NCCS-FITC-AL were also performed at both pHs of 5 and 7.4 to observe the effects of the acidic environment (H + rich) and saline (Na + and Cl À ions) on the surface of HAp. Any changes caused by these ions were not observable for both of the control samples ( Fig. 9A and C). In addition, adhered particles were not found in the SEM image This stronger adherence under the acidic conditions, approximately 140% higher affinity than that under the physiological conditions (Fig. S10 †), is plausibly ascribed to both the physisorption and chemisorption of the nanoparticles on the HAp surface (Fig. 10). The surface of Fe 3 O 4 -NCCS-FITC-AL is covered by the anionic bisphosphonate and carboxylate ions and cationic quaternary ammonium ions that electrostatically interact with Ca 2+ and H 2 PO 4 À on the surface of HAp, respectively, even though the zeta potential is not so high due the buffering ability of the polyampholytic skeleton. In addition, the dissolution-deposition equilibrium of the phosphonate ions of HAp results in the incorporation of the bisphosphonate groups of Fe 3 O 4 -NCCS-FITC-AL into the lattice structure of HAp (Fig. 10, right part). These interactions of phosphonates with HAp were reported in previous NMR spectroscopic ( 31 P, 15 N, 13 C, 2 H, and 1 H) and chromatographic studies. 61,63,66,67 This nding revealed that the designed nanosystem has adequate potentials for selective targeting of acidic bone sites.

Affinity of Fe 3 O 4 -NCCS-FITC-AL nanoparticles towards native bone sample
The affinity and adherence of the charge-inversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles towards native bone samples were studied by SEM analysis. Bovine leg bone samples were treated with Fe 3 O 4 -NCCS-FITC-AL at pH 7.4 and 5.0 (Fig. 11). The optical and SEM images of the bone sample are shown in   (Fig. 11C). The poor affinity is due to the electrostatic repulsions between the negatively charged Fe 3 O 4 -NCCS-FITC-AL nanoparticles and the bone surface under the physiological conditions. By contrast, at pH 5.0, Fe 3 O 4 -NCCS-FITC-AL nanoparticles were adhered on the bone surface, which was almost covered by these nanoparticles (Fig. 11D). This pH dependence, namely high affinity only under acidic conditions, is identical to HAp employed as the model system, while the enhancement in the affinity to the bone under the acidic conditions is more signicant than that to HAp. Bone mainly consists of HAp and collagen, whose solubility is higher at pH 5.0 by the positively charged nature. 42,68 The increase in the active surface of HAp by  the dissolution of charged collagen is a plausible factor for higher affinity. In metastasized bone sites, degradation of collagen is enzymatically promoted, and this acidity-sensitive adsorption is presumably more enhanced. Osteoclasts become acitivated by cancers and overexpress cathepsin-K degrading collagen bers networks. 5 This synergy of the pH-triggered charge-conversion and targetable bisphosphonate moieties in the Fe 3 O 4 -NCCS-FITC-AL nanoparticles enabled selective accumulation on HAp unvailed under acidic conditions.

Conclusions
In this report, we demonstrated a magnetic nanomaterial,  Fe 3 O 4 -NCCS-FITC-AL particle is a promising candidate for earlystage diagnosis and therapy of bone malignancies.

Experimental section
Material and methods Materials. CS (M w ¼ 5.6 Â 10 5 g mol À1 ), CAn, and AL were purchased from Tokyo Chemical Industry (Tokyo, Japan). FeSO 4 $7H 2 O and FITC were purchased from Sigma Aldrich (St. Louis, MO, USA). NH 4 OH (28%) and FeCl 3 $6H 2 O were purchased from Kanto Chemical (Tokyo, Japan). HAp was purchased from Wako Pure Chemical (Osaka, Japan). DPBS was purchased from Gibco Life Technologies (Paisley, UK). Preserved sheep whole blood was purchased from Cosmo Bio (Tokyo, Japan). Bovine leg bone was purchased from Baticrom Halal Shop (Tokyo, Japan). All other reagents used in this study were lab grade and used without further purication. Water was puried using a Nomura Micro Science (Kanagawa, Japan) MINIPURE TW-300RU water purication system. Deionized distilled water (DDW) was used throughout the study.
Measurements. FTIR spectra were documented on a JASCO (Tokyo, Japan) FTIR-460 plus spectrometer using pressed KBr pellets with a scan rate of 4 cm À1 sec À1 . 1 H NMR spectra were recorded on a JEOL (Tokyo, Japan) ECX-400 instrument using tetramethylsilane as an internal standard (400 MHz) at room temperature. Hydrodynamic diameters were measured by DLS on a Malvern Instrument (Malvern, UK) Zetasizer Nano ZS. The size of dried nanoparticles was measured using SEM images taken with a Hitachi (Tokyo Japan) SU8000 microscope operated at an accelerating voltage of 30 kV. The powder X-ray diffraction (XRD) patterns were measured on a Rigaku (Tokyo, Japan) Ultima IV RINT D/max-kA diffractometer with Cu Ka radiation (l ¼ 1.54178Å). TEM images were taken using a JEOL (Tokyo, Japan) TEM-2100F eld emission scanning electron microscope. Thermogravimetric analysis was carried out on a Seiko Instrument (Tokyo, Japan) TG/DTA 6200 (EXSTER6000) at a heating rate of 10 C min À1 under N 2 . The magnetic properties of the nanoparticles were analyzed on a Riken Denshi (Tokyo, Japan) BHV-30 series vibrating sample magnetometer at ambient conditions. Energy-dispersive X-ray spectra were recorded on a JEOL (Tokyo, Japan) JSM-6510A analytical scanning electron microscope. The optical absorbance was taken using an AS ONE ASV11D UV-visible spectrophotometer. The erythrocytes specimens were observed with an Olympus (Tokyo, Japan) CKX53 microscope. The color pictures of the specimens were captured using a Visualix Pro2 (Kobe, Japan) camera. All images were taken at the same magnication (20Â).

Methods
Synthesis of bisphosphonated polyelectrolytic CS derivative: N-carboxycitraconylation of CS (NCCS). CS was dissolved in 200 mL of aqueous acetate buffer (pH 4.6) by magnetic stirring overnight. The solution was adjusted at pH 5.6 using dilute aqueous NaOH solution. CAn was diluted in methanol (10 mL) and dropwise added to CS solution for 1 h. The reaction was continued for 12 h at room temperature. Aer the reaction, 10 M NaOH aqueous solution was added to the reaction mixture and stirred for an additional 10 min. The reaction mixture was added dropwise to an excess amount of acetone to precipitate NCCS. The precipitate was separated using a Buchner funnel. The resulting product was puried via dispersing in DDW and reprecipitated in acetone. The puried NCCS was dried in a vacuum desiccator and stored in a refrigerator.
FITC labeling of NCCS. NCCS (800 mg) was dissolved in DDW (300 mL). FITC (4.00 mg) was dissolved in 10 mL methanol and the solution was added dropwise to the mixture for 1 h. The reaction was continued in dark conditions for 4 h at room temperature. The reaction mixture was added dropwise to an excess amount of methanol to precipitate the NCCS-FITC. The resulting product was separated by centrifugation at 4000 rpm and washed repeatedly until the decant became completely colorless. The obtained NCCS-FITC was dried in a desiccator in dark and stored in a freezer at À4 C.
Bisphosphonylation of NCCS-FITC. In dark conditions, NCCS-FITC was dissolved in DDW (200 mL) by magnetic stirring at room temperature. AL (100 mg) was dissolved in 10 mL DDW and the solution was added dropwise to the dispersion of NCCS-FITC. The reaction was performed at room temperature in dark conditions for 24 h. Aer the reaction period, the product was precipitated by pouring the mixture into an excess amount of acetone and separated by ltration. The resulting product was puried by dispersing in DDW and reprecipitating in methanol. Bisphosphonated polyelectrolytic NCCS-FITC-AL was obtained by drying in a vacuum desiccator (yield ¼ 85%) and kept at À15 C.
MW-assisted in situ synthesis and surface functionalization of Fe 3 O 4 nanoparticles. Bisphosphonated polyelectrolytic NCCS-FITC-AL was dissolved in deoxygenated DDW (175 mL) by magnetic stirring. FeCl 3 $6H 2 O (270 mg, 1.00 mmol) and FeSO 4 $7H 2 O (139 mg, 0.50 mmol) were dissolved in 15 mL DDW with purging N 2 . These solutions were mixed in a three-necked round bottom ask (300 mL) under N 2 . The ask was placed in a preheated thermostat oil bath maintained at 80 C. Aqueous ammonia solution (5 mL, 28%) was injected at a rate of 1 mL min À1 . The mixture was magnetically stirred for 10 min. The reaction mixture was, then, quickly transferred into a 500 mL conical ask. The ask was placed in a MW oven to be heated at 100 C for 30 min. The resultant black precipitate was magnetically washed repeatedly with DDW until the supernatant became neutral pH. The solid precipitate was separated at the bottom of the ask and dried in a vacuum desiccator for 24 h. Bare Fe 3 O 4 was synthesized following the aforementioned procedure without adding the stabilizer, NCCS-FITC-AL.
Hemocompatibility of Fe 3 O 4 -NCCS-FITC-AL. Hemolysis activities of the Fe 3 O 4 -NCCS-FITC-AL nanoparticles were assayed by using the following protocol. Preserved sheep whole blood (40 mL) was centrifuged at 2500 rpm for 5 min. The sheep erythrocytes were separated by centrifugation and plasma removal and washed three times with saline (150 mM NaCl aq.). The washed erythrocytes were redispersed in saline (40 mL). The nanoparticles were also dispersed in saline and the dispersion was mixed with erythrocytes suspension. The pH of the mixtures was adjusted to 7.4 using 0.01 M HCl aq. The nal concentrations of the Fe 3 O 4 -NCCS-FITC-AL nanoparticles were adjusted to 100, 200, and 400 mg mL À1 . Triton X-100 (1% aqueous solution) and saline were added to the erythrocytes suspension to prepare the positive (100% lysis) and negative (0% lysis) control samples, respectively. All the samples and controls were incubated for 5 h at 37 C. The samples were gently swirled once per 20 min for resuspending the erythrocytes. Aer incubation, the suspension was centrifuged at 10 000 rpm for 5 min to separate the erythrocytes and the supernatants were incubated for 30 min at room temperature to oxidize the released hemoglobin. The optical absorbance of oxyhemoglobin was assessed by absorption at 540 nm. The hemolysis percentage of erythrocytes was calculated using the following equation: % hemolysis ¼ {(Abs sample À Abs negative control )/(Abs positive control À Abs negative control )} Â 100 Membrane morphology study of erythrocytes. The morphology of erythrocytes was studied using a light microscope. The erythrocytes were added to saline containing the charge-conversional Fe 3 O 4 -NCCS-FITC-AL nanoparticles at 400 mg mL À1 , and the pH values were set to 7.4 and 5.0 using 0.01 M HCl aqueous solution. The resulting dispersions were incubated at 37 C for 5 h, and mildly shaken once every 20 min. Aer the incubation period, the dispersions were treated with a magnet to remove free nanoparticles. Then, the erythrocytes were cleaned by centrifugal washing and resuspended in saline. A control sample was also made without using nanoparticles. The washed erythrocytes suspension (10 mL) was placed on a glass slide and covered with a coverslip glass. The resulting specimens were observed microscopically.
pH-selective adhesion of Fe 3 O 4 -NCCS-FITC-AL to HAp and native bone sample. Fe 3 O 4 -NCCS-FITC-AL nanoparticles (500 mL) were dispersed in saline (10 mL) and the pH values of the dispersions were adjusted to 7.4 and 5.0 using 0.01 M HCl aqueous solution. HAp (20 mg) was added to each dispersion and dispersed by mild shaking. The dispersions were incubated at 37 C for 3 h and shaken once every 20 min for redispersion. Aer incubation, the HAp particles were separated by centrifugation at 2000 rpm and washed 7 times with saline. Finally, the washed HAp was analyzed by SEM. For the native bone sample, a small piece of clean bone was added to the dispersion instead of HAp, incubated under the aforementioned conditions, and washed 7 times with saline and water to remove the free and loosely bound nanoparticles. The control bone samples were made without using Fe 3 O 4 -NCCS-FITC-AL nanoparticles. All the bone samples were also analyzed by SEM and EDS.
Author contributions M. A. Rahman conceptualized the idea and did formal experiments. He collected and analysed the data, and also explained the results. This research was supervised by B. Ochiai. The manuscript was draed by M. A. Rahman and revised by B. Ochiai. The nal version of the manuscript has approved by both the authors.

Conflicts of interest
The authors declare no competing nancial interests that would have been arised to inuence the work reported in this paper.