Acta Biomaterialia 6 (2010) 2212–2218

Contents lists available at ScienceDirect

Acta Biomaterialia

journal homepage: www.elsevier .com/locate /ac tabiomat

Polymeric micelle-templated synthesis of hydroxyapatite hollow nanoparticlesfor a drug delivery system

Feng Ye *, Haifeng Guo, Haijiao Zhang, Xiulan HeSchool of Materials Science and Engineering, P.O. Box 433, Harbin Institute of Technology, Harbin 150001, People’s Republic of China

a r t i c l e i n f o a b s t r a c t

Article history:Received 23 June 2009Received in revised form 30 November 2009Accepted 4 December 2009Available online 22 December 2009

Keywords:HydroxyapatiteHollow nanoparticlesPolymeric micellepH-controlled releaseDrug delivery system

1742-7061/$ - see front matter � 2009 Acta Materialdoi:10.1016/j.actbio.2009.12.014

* Corresponding author. Tel.: +86 451 86413921; faE-mail address: yf306@hit.edu.cn (F. Ye).

Hydroxyapatite (HA) hollow nanoparticles (HNPs) have great potential in nanoscaled delivery devicesdue to their small size, excellent biocompatibility and expected high capacity. However, the preparationof HA HNPs for their application in a drug delivery system has rarely been reported because HA has acomplicated crystal structure and it is difficult to obtain stable HA HNPs with hollows that are only nano-scaled in size. In the present study, HA HNPs were successfully produced through a novel polymericmicelle-templating method. The micelles were structured with completely insoluble Pluronic P123 mol-ecules at cloud point as the core and Tween-60 molecules as the shell by the hydrophobic interaction ofthe alkyl chains with the insoluble P123 core. The morphology of the HA HNPs could be transformed fromnanospheres to nanotubes by adding citric acid as a cosurfactant. The prepared HA HNPs had a muchhigher drug payload than traditional nanoparticles, using vancomycin as the model drug. Most impor-tantly, the HA nanotubes were coupled with a layer of citrate molecules on the HA surfaces, which couldfurther improve the drug load efficiency and could form an excellent pH-controlled open/closed gate fordrug release with the addition of cationic polyelectrolytes.

� 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Recently, the synthesis of inorganic hollow nanoparticles(HNPs) has attracted much attention in the chemistry and materi-als communities because of their low density, large specific areaand pore volume, small size, mechanical and thermal stabilities,and surface permeability [1]. Such HNPs have a wide variety of po-tential applications in cosmetics, catalysis, coatings, compositematerials, dyes, ink, artificial cells and fillers. Their hollow struc-ture can also be used as a microencapsulate for drugs in the phar-maceutical field [2]. Hydroxyapatite (HA, Ca10(PO4)6(OH)2), themain mineral component of bones and teeth, is native to the hu-man body. HA crystals have been widely used in delivery systemsfor genes [3], proteins [4] and various drugs [5,6] due to their non-toxicity and excellent biocompatibility, as have been experimen-tally proven by recent reports [7–9]. However, HA crystals have alimited drug load capacity. HA mesoporous nanoparticles or HNPshave greater potential in a nanoscaled delivery system due to theirpromising high capacity. Recently, Yang et al. [10] reported thatluminescent HA mesoporous nanoparticles possessed a high load-ing for ibuprofen (over 40 wt.%), though the carrier had uncon-trolled release kinetics which made it have a burst release ofibuprofen of over 50% within 1 h. However, to our knowledge,

ia Inc. Published by Elsevier Ltd. A

x: +86 451 86413922.

the preparation and application of HA HNPs in drug delivery sys-tems have been reported only rarely. The reason for this mightbe that HA has a complicated crystal structure (in comparison tocommon oxides, such as SiO2) and it is very difficult to obtain sta-ble HA HNPs whose hollow part is only tens of nanometers throughtraditional methods.

It is well known that poly(ethylene oxide)–poly(propyleneoxide)–poly(ethylene oxide) (PEO–PPO–PEO) block copolymerexhibits peculiar behaviors in aqueous solution [11–13]. Belowthe characteristic temperature – the critical micellar temperature(CMT), both the PEO and PPO blocks are soluble in water, and thecopolymer molecules remain in the form of unimers in solution.At the CMT the PPO blocks become insoluble and form sphericalmicelles, with PPO blocks as the hydrophobic core and the hy-drated PEO blocks as the shell. Meanwhile, the PEO blocks startto lose hydrated water with increasing temperature, and at thetemperature called the cloud point (CP), the PEO blocks becometoo insoluble and the copolymers get phase-separated from thewater. The behaviors of block copolymer in water between theCMT and the CP have been extensively used to template variousmesoporous materials [14–18], while its characteristics belowCMT and at CP have often been used in cloud point extractions[19,20]. Actually the behaviors below CMT and at CP of triblockcopolymer–Pluronic P123 (EO20PO70EO20), i.e. completely insolu-ble at CP and totally soluble below CMT, can be adjusted for thepreparation of HA HNPs by the use of Tween-60, whose CP is high-

ll rights reserved.

F. Ye et al. / Acta Biomaterialia 6 (2010) 2212–2218 2213

er than that of P123. The details are shown in Scheme 1: at lowtemperature (but above the CMT of P123) P123 and Tween-60molecules form micelles with a core–shell structure in the pres-ence of ethanol and a phosphorous source; at the CP of P123, theP123 molecules in these micelles become insoluble, but theTween-60 molecules remain soluble; therefore, the insolubleP123 molecules become the condensed cores of new micelles andTween-60 molecules form the shells by the hydrophobic interac-tion of their alkyl chains with the insoluble P123 core. WhenCa2+ is added, HA forms within the shells of the micelles, and thecores act as template for the hollow parts of HA HNPs; the tem-plates can then be easily removed by water and ethanol solutionat low temperature (<CMT) because P123 is totally soluble belowits CMT. In addition, polar citric acid was used as a cosurfactantto modify the shape of the micelles and ultimately to regulatethe morphology of HA HNPs as the polar additive can associatewith nonionic surfactant molecules, resulting in a change in thecritical packing parameters of the surfactants.

In this report, HA HNPs were successfully produced through theabove-designed polymeric micelle-templating method. With add-ing citric acid as the cosurfactant, the morphology of HA HNPstransformed from nanospheres to nanotubes. The obtained HAHNPs had a much higher drug payload than traditional nanoparti-cles, using vancomycin as the model drug. Most importantly, theHA nanotubes were coupled with a layer of citrate molecules uponthe HA surfaces, which could further improve the drug load effi-ciency and could form an excellent pH-controlled open/closed gatefor drug release with the addition of cationic polyelectrolytes.

2. Materials and methods

2.1. Investigation on the cloud points of surfactants

The cloud point of P123 aqueous solutions with different addi-tives (ethanol, Tween-60 and citric acid) were investigated by rais-ing the temperature at a rate of 2 �C min�1. The cloudy phenomenaof solutions were recorded by digital camera. The digital imagesare shown in Supplementary Fig. S1.

2.2. Preparation of HA HNPs

According to Scheme 1, calcium nitrate (Ca(NO3)2�4H2O) andphosphorous acid (H3PO4, 85%) were used as the calcium and phos-phorous sources for HA, respectively. A mixture of the nonionicsurfactants Pluronic EO20PO70EO20 (P123; EO = ethylene oxide,PO = propylene oxide) and polyoxyethylene (20) sorbitan monoste-arate (Tween-60) was utilized as the template agent, with citricacid as the cosurfactant. All chemicals were analytical grade andused as purchased without further purification. Based on the anal-ysis results of the CP for P123 in different additives (Supplemen-

Scheme 1. Fabrication of hollow HA nanoparticles from P123

tary Fig. S1), the temperatures at stages I and II in Scheme 1were valued as 20 and 90 �C, respectively. Two groups of sampleswere prepared (designated S1 and S2). For S1, 2.9 g of P123 wasdissolved in 50 ml of deionized water plus 10 ml of ethanol andthe solution was stirred to form micelles at 20 �C. The solutionwas supplemented with 1.31 g Tween-60 (in 20 ml of water) and0.1 mol PO3�

4 , then continuously stirred for 1 h. Next, the refluxingflask containing the mixed solution was put directly into a waterbath at 90 �C. Subsequently 0.167 mol Ca2+ (39.438 g of Ca(-NO3)2�4H2O dissolved in 30 ml of water, pH 9) was slowly addedto the solution, which was left to react for 48 h. The solution wascontinuously stirred and refluxed throughout the whole process.After the reaction was completed, the precipitate was eluted withethanol and water six times at 10 �C because the CMT of P123 wasaround 15 �C. The obtained products were dried at 100 �C for 12 hand calcined at 300 �C for 3 h. For S2, 10 g of citrate acid was addedto the P123 solution to change the micelle shape and ultimatelyregulate the morphology of the HA HNPs, using ammonia to makethe solution’s pH 9 in the first stage. The rest of the process was thesame as for S1.

2.3. Drug load and in vitro release

The resultant HA HNPs were used as carriers for vancomycin.Typically, 0.4 g of HA HNPs was added to 20 ml of phosphate-buf-fered solution (PBS, 10 mM) contained 0.4 g vancomycin, whichwas then stirred at 37 �C for 12 h. For HA nanotubes coupled witha layer of citrate, 0.04 g of 40 wt.% cationic polyelectrolyte(poly(dimethyldiallyl ammonium) chloride, PDDA) was added tothe PBS solution before the final completion of drug loading, toassociate with the citrate carboxyl of the HA nanotubes and forman open/close gate. After being centrifuged (5000 rpm) and washedthree times to remove the drug molecules adsorbed on the outersurface of the HA HNPs, the obtained products were freeze-driedand then vancomycin-loaded HA HNPs carriers were prepared. Inorder to determine the drug payload of these HA carriers, 0.02 g(MC) of vancomycin-loaded HA HNPs was dissolved in 10 mMPBS solution (pH 1) and the concentration of vancomycin in theresultant solution was measured by ultraviolet–visible (UV–vis)spectroscopy at a wavelength of 280 nm. The PBS solutions withdifferent concentrations of vancomycin (0.1, 0.2, 0.25, 0.4, 0.6, 0.8and 1 g l�1) were also measured by UV–vis spectroscopy to evalu-ate the Lambert–Beer mass absorption coefficient (f) of the vanco-mycin in the PBS solution, as shown in Supplementary Fig. S2. Theproduct of the concentration and volume of the resultant solutionis equal to the mass (MV) of vancomycin-loaded in the HA HNPs.The drug payload ratio (L) for the HA HNP carrier was calculatedas: L = MV/MC � 100%, and the loaded efficiency of the drug wasE = L/(1 � L) � 100% due to the mass of vancomycin and HA inthe loading process was the same. The actual value of L was the

and Tween-60 core–shell structured micelles templates.

2214 F. Ye et al. / Acta Biomaterialia 6 (2010) 2212–2218

average drug payload ratio for three groups of experiments whichwere carried out at the same time and under the same conditions.Thermogravimetric (TG) analysis of drug-loaded HA and blank HAwas also carried out to calculate the drug payload ratio for the HAHNPs. In addition, for comparison, HA nanoparticles (S0) synthe-sized by a traditional sol–gel method [21] were also used in thedrug-loading experiment. The vancomycin-loaded as-dried S1(nanospheres) and as-dried S2 (nanotubes coupled with citrateon the surfaces) and the calcined S2 and S0 (traditional HA nano-particles) samples were designated S1-V, S2-V, cS2-V and S0-V,respectively. The as-dried S2 with PDDA added was designatedS2-PDDA-V.

For the in vitro release study, the vancomycin-loaded HA HNPs(0.04 g) were added to semipermeable bags (3000 Da porosity),then immersed in 50 ml of PBS solutions of different pH (7.4, 5.4and 2) and stirred at a rate of 150 rpm at 37 �C. The concentrationof the released vancomycin at different set times was monitored byUV–vis spectroscopy, and the cumulative release mass at differentset times were calculated. The cumulative release ratio equals thecumulative release mass divided by the total mass of the carrier(0.04 g). The actual value of release was the average of three groupsof experiments which were carried out at the same time and underthe same conditions. The corresponding standard deviations wereconsidered as the statistical variations for all experiments.

10 20 30 40 50 600

2000

4000

6000

8000

10000

d

c

bRel

ativ

e In

tens

ity (

a.u.

)

2 theta (deg.)

All peaks from HAP

a

Fig. 1. XRD patterns of hollow HA samples: (a) as-dried S1, (b) calcined S1, (c) as-dried S2 and (d) calcined S2.

Fig. 2. SEM micrograph of (a) as-dried S1 and (b) as-dried S2, and T

2.4. Characterization

The phases of the samples were analyzed by X-ray diffraction(XRD) on a Philips X’pert diffraction instrument with Cu Ka inci-dent radiation (40 kV, 35 mA). The morphology of the productswas observed by scanning electron microscopy (SEM; Hitachi4700,Japan) and transmission electron microscopy (TEM; Philips EM20).The voltages used for the TEM observation were 120 and 200 kV.The obtained HA powders were separately dispersed in ethanoland the resultant HA suspension was dropped onto transmissionelectron microscope copper grids to prepare samples for TEM. Fou-rier transform infrared (FTIR) spectra were collected on a NicoletNexus spectrometer. Nitrogen sorption was operated at 77 K withan Autosorb-1 Gas Sorption instrument (Quantachrome Corpora-tion, USA) to detect the pores of the samples, and the pore size dis-tribution was plotted based on the Barrett–Joyner–Halenda (BJH)method. The pore volume was calculated according to the BJHcurve from 2 to 100 nm. The particle size distribution was analyzedusing a Zeta Plus/PALS instrument (Brookhaven, USA). TG measure-ment (Netzsch Thermoanalyzer STA 449C, Germany) was used todetermine the amount of vancomycin-loaded in the materials.UV–vis measurement was carried out with a UV–vis absorptionspectroscope (Varian Cary 3000, USA).

3. Results

3.1. Morphology, microstructure and surface characteristic of HA HNPs

Fig. 1 shows the XRD patterns of the as-dried and calcined HAsamples. All the diffraction peaks corresponded to the standardcharacteristic peaks of hexagonal HA, indicating that the phase ofthe samples was highly pure HA. Fig. 2 illustrates the SEM andTEM micrographs of S1 and S2. For S1, HA hollow nanosphereswith an average diameter of 60 nm and a hollow diameter ofaround 3 6 nm were obtained (the voltage used for the TEM obser-vation was 200 kV). Very interesting stripe-like patterns spaced�4 nm apart were found in the hollow part of the HA nanospheres.With the addition of citrate (for S2), the morphology was trans-formed into nanotubes. These HA nanotubes were open-ended,with a diameter of about 35 nm, a length range of 50–250 nmand a hollow diameter of 13 nm (Fig. 2d–f). These sizes were fur-ther confirmed by the results of the particle size distribution anal-ysis (Fig. 3) and gas sorption (Fig. 4). Fig. 3 shows the averageparticle diameter was 60 and 36 nm for S1 and S2, respectively.All three nitrogen adsorption/desorption isotherms of the HA sam-ples can be categorized as type IV, with a distinct hysteresis loop.

EM micrograph of (c) as-dried S1 and (d, e, and f) as-dried S2.

Fig. 3. The particle size distribution of the as-dried S1 and S2.

F. Ye et al. / Acta Biomaterialia 6 (2010) 2212–2218 2215

The BET surface area is calculated to be 66.11 m2 g�1 for S1,97.76 m2 g�1 for the as-dried S2 and 116.8 m2 g�1 for the calcinedS2 (at 300 �C). The typical H1 hysteresis loop of the as-dried andcalcined S2 (Fig. 4b and c) indicated that there were a great amountof cylinder-like open-ended pores inside the HA nanoparticles,

Fig. 4. (a) Nitrogen adsorption/desorption isotherm of (a) as-dried S1, (b) as-driedS2 and (c) calcined S2. (b) BJH pore size distribution for (a) as-dried S1, (b) as-driedS2 and (c) calcined S2.

which supported the TEM result well (Fig. 2d–f). The BJH pore sizedistribution of S1 was centered at 36 nm. The as-dried S2 had anarrow pore size distribution centered at 13.3 nm and the centerfor the calcined S2 was at 15.6 nm, suggesting that the calcinationincreased the hollow diameter of the HA nanotubes (2.3 nm). Thepore volume was evaluated as 0.471 cm3 g�1 for S1,0.34104 cm3 g�1 for the as-dried S2 and 0.35809 cm3 g�1 for thecalcined S2. The above results prove that small HA HNPs with ahollow diameter of only tens nanometers could be successfullyachieved according to Scheme 1.

Fig. 5 gives the FTIR spectra of HA HNPs samples and pure citricacid. Phosphate absorption bands occurred at 1092, 1034, 962, 602,567 and 472 cm�1, and the hydroxyl absorption bands at 3572 and634 cm�1 were characteristic of a typical HA FTIR spectrum. Forpure citric acid, the bands at 1755 and 1703 cm�1 were ascribedto the stretch vibration of two carboxyl groups which were associ-ated with hydrogen bonds (Fig. 5e). For the as-dried S2 (Fig. 5c), thedisappearance of the 1755 and 1703 cm�1 citric acid bands and theappearance of a band at 3149 cm�1, which was ascribed to the flex-ible vibration of the hydroxyl groups of citric acid, indicated thatcitric acid molecules were associated with the hydrophilic partsof the surfactant molecules by hydrogen bonds. The bands at1591 and 1390 cm�1 were ascribed to the symmetry and anti-sym-metry stretch vibration of fully deionized carboxyl acid groupswhich were chemically bound to Ca2+ on the HA surface [22], indi-cating that the as-dried S2 was coupled with citrate on the sur-faces. The FTIR results (Fig. 5a and c) also demonstrated that nonoticeable bands of residual P123 or Tween-60 were present inthe as-dried hollow HA samples, indicating the template removalstep in Scheme 1 was feasible. In addition, it is worth noting thatthe calcinations removed the layer of citrate from the hollow HAnanorod surfaces, and caused an increase in the hollow diameterfrom 13.3 to 15.6 nm. This suggests that the thickness of the citratelayer on the inner surface of the HA nanotubes was 1.15 nm. Thisvalue is close to the citrate molecule length when two –COO�

groups graft with the Ca2+ of HA in a neutral environment and, inci-dentally, is half the size of a vancomycin molecule.

3.2. Drug load and in vitro release properties

Fig. 6 shows the FTIR spectra of drug-loaded HA HNPs and purevancomycin. The characteristic bands of HA were also present inthe drug-loaded HA samples, indicating that the inclusion of van-comycin in HA HNPs did not cause any detectable change in the

4000 3500 3000 2500 2000 1500 1000 500

50

100

150

200

250

300

472

13901591

1385

17031755

32963496

831

9623572

567602

634

3149

10341092

e

d

c

b

Tra

nsm

ittan

ce (

a.u.

)

Wave number (cm-1)

a

Fig. 5. FTIR spectra of (a) as-dried S1, (b) calcined S1, (c) as-dried S2, (d) calcined S2and (e) pure citric acid.

4000 3500 3000 2500 2000 1500 1000 500

40

80

120

160

200

1700~1200

1497

e

dc

bTra

nsm

ittan

ce (

a.u.

)

Wave number (cm-1)

a

Fig. 6. FTIR spectra of vancomycin-loaded (a) as-dried S1, (b) calcined S2, (c) as-dried S2, (d) as-dried S2 with the addition of PDDA and (e) pure vancomycin.

2216 F. Ye et al. / Acta Biomaterialia 6 (2010) 2212–2218

structure of HA. Additional well-pronounced bands in the rangefrom 1700 to 1200 cm�1 are due to the presence of vancomycin.Compared to pure vancomycin, the drug-loaded HA samplesshowed no noticeable shift in the vancomycin bands. For instance,a strong intensity band, characteristic of a C–C mode of vibration inaromatic rings, was located in all spectra at 1497 cm�1. Based onthis analysis, we suggest that all HA HNPs were successfully loadedwith vancomycin and the drug-loading process did not affect thechemical structure of the vancomycin-loaded in the HA HNPs.

Fig. 7 shows a TEM micrograph of S2-PADA-V. The HA nano-tubes showed little change in morphology after drug loading, andinside the nanotubes were many nanoparticles with a diameterof 8–10 nm, which were probably the condensed aggregates ofvancomycin molecules (the voltage used here was 120 kV). Itshould be mentioned that the drug molecules could also decoratethe outer surfaces of the HA.

The UV–vis results gave the drug payload ratio (L) and theloaded efficiency of drug (E) for the HA delivery systems, as shownin Table 1. The drug payload ratios obtained from TG analysis arealso presented in Table 1 (the TG/differential thermal analysis re-sults of the as-dried S2 and S2-PDDA-V are presented in Supple-mentary Fig. S5). Both results were close to each other, indicating

Fig. 7. TEM micrograph of S2-PADA-V delivery system (as-dried S2 being loadedwith drug and with PDDA added).

that the two methods were effective in determining the drug pay-load ratio.

Table 1 demonstrates that the high specific surface areas andlarge pore volumes gave all HA HNPs much higher values of Land E than traditional nanoparticles (S0, with a size range of 40–60 nm). The 1.15 nm citrate layer upon the HA nanotube surfacesfurther increased the drug payload ratio from 16.93% to 24.14%.This increase was possibly the result of the association of the cit-rate layer with vancomycin molecules by hydrogen bonding. Forthe S2-PDDA-V system, the cationic polyelectrolyte PDDA, with apositive charge, could form a complex with the negatively chargedcarboxyl ions of citrate upon the hollow HA nanorods surfaces,which could increase the payload of vancomycin and could capturethose vancomycin molecules that would otherwise have beenrinsed from the HA nanotubes in the washing process. Therefore,S2-PDDA-V had the highest payload ratio of vancomycin, at about35.8%.

The two systems with the highest drug payload ratios, S2-V andS2-PDDA-V, were selected separately for in vitro release in PBSsolution at different pHs (7.4, 5.4 and 2). The cumulative drug re-lease profiles as a function of release time in PBS solution at differ-ent pHs for S2-V and S2-PDDA-V are shown in Figs. 8 and 9,respectively. As demonstrated in Fig. 8, S2-V had a high release ratewhen the pH of PBS was 7.4 during the first 2 h, but then reachedits release equilibrium with little further release. In the acid envi-ronment (pH 5.4 and 2), the release rate was low for the first 3 h,but then increased sharply in the subsequent 2 h. This was possiblydue to the acid environment partially protonizing the –COO�

groups. The association between –COOH and vancomycin byhydrogen bonding resulted in a low drug release rate, but withtime the combination between citrate and the Ca2+ of HA becamegreatly reduced or had vanished completely, causing a burst re-lease of drug.

In contrast, for the profile of S2-PDDA-V shown in Fig. 9, the ini-tial release rate was low and increased very little with time whenthe PH of the PBS was 7.4. However, when the pH was 5.4 (i.e. aweak acid environment) the system had a high initial release thatincreased almost linearly with time, the amount of vancomycin re-leased increasing incrementally to 89.28% in 5 h. Furthermore, theprofile of S2-V (pH 5.4) in Fig. 8 shows that the release ratio from0.5 to 3 h remained almost constant. If the release ratio at 0.5 h isconsidered to be the rate of release of the drug loaded on the outersurfaces of the HA (the real value should actually be less that this,as 0.5 h was enough time for drug coating the outer surfaces to becompletely released), the rate of release of the drug loaded insidethe HA tubes for the S2-V system was over 93%. For the S2-PDDA-V system (see S2-PDDA-V (pH 7.4) in Fig. 9), the rate wasover 90%. Generally, these results indicate that the S2-PDDA-Vdelivery system is a smart drug carrier with a pH-controlled releas-ing property.

4. Discussion

Unlike the HA HNPs prepared at the CP of P123 (90 �C, as de-scribed in Scheme 1), the HA samples synthesized at 20 �C pos-sessed an irregular aggregated morphology with a pore size ofabout 6 nm when P123 was 2.9 g and around 8 nm when P123was 11.6 g (see Supplementary Fig. S3). These irregular HA aggre-gates were probably the result of aggregation of unstable sphericalmicelles of P123 and Tween-60 at low temperature (20 �C). At60 �C, the product obtained was HA nanoparticles without anyobvious pores. This was a result of the formation of reverse mi-celles, because P123 was found to form a reverse phase in solutionat 60 �C in our previous study [23]. Therefore, the temperature atwhich the micelles in the solution were stable and the core of

Table 1The parameters, drug payload ratio (L) and loading efficiency of drug (E) for different delivery systems determined by UV–vis and TG analysis.

Delivery system S0-V S1-V S2-V S2-PDAD-V cS2-V

Surface area (m2 g�1)* 11.26 66.11 97.76 97.76 116.8Pore volume (cm3 g�1)* – 0.471 0.34104 0.34104 0.35809Hollow diameter (nm)* – 36.013 13.3 13.3 15.6

L (Er) (%) 2.20 (±0.27) 16.01 (±0.37) 24.14 (±0.49) 35.83 (±0.58) 16.93 (±0.50)UV–vis

E (%) 2.25 19.06 31.82 55.84 20.38TG (%) 1.8 15.9 24.5 38.0 16.5

Note: The parameters of HAP samples were measured before drug loading. Er was the corresponding standard deviation.* The parameters of HAP samples that were measured before drug loading.

Fig. 8. The in vitro vancomycin release profile of the S2-V system in PBS solution atdifferent pHs (7.4, 5.4 and 2).

Fig. 9. The in vitro vancomycin release profile of d S2-PDDA-V system in PBSsolution at different pHs (7.4 and 5.4).

pH =5.4 pH =7.4

H+

OH-

= Positively charged PDAD = Vancomycin

=ionized carboxyl of citrate =protonated carboxyl

Close Open

Release Enclose

Fig. 10. Schematic representation of a pH-responsive drug carrier based on theinteraction between negatively charged carboxyl groups of citrate upon hollow HAnanorods surfaces and positively charged PDDA molecules. The interaction formedthe gates to open or close to store or release vancomycin from the hollow nanorods,as controlled by environment pH.

F. Ye et al. / Acta Biomaterialia 6 (2010) 2212–2218 2217

the micelles was condensed was the key to preparing dispersed HAHNPs. At the CP, the P123 molecule condensed to become the coreof the micelles, making the micelles stable and resulting in the for-mation of dispersed HA HNPs. In addition, the interesting stripe-patterns found in HA hollow nanospheres (Fig. 2b) are possibly aresult of the surfactant chains in the micellar shell region being ar-ranged in a orderly manner [12].

The complex between the positively charged PDDA and the neg-atively charged carboxyl of the HA nanotubes endowed S2-PDDA-Vsystem with an open/close gate for enclosing or releasing vanco-mycin molecules from HA nanotubes depending upon the environ-ment pH, because citric acid has three carboxyl groups with threedifferent dissociation constants (at pH 3.14, 4.77 and 6.39) [24].The open/close process for drug release is shown in Fig. 10. In com-parison with the mesoporous HA delivery system of Yang et al.[10], our HA HNPs systems possess larger surface areas and greaterpore volumes, and most importantly the citrate-coupled HA nano-tube system has an excellent pH-controlled releasing property fordrugs. These HA HNPs systems also have great potential for thedelivery of genes, proteins and other materials.

5. Conclusion

In summary, HA hollow nanospheres and nanotubes were suc-cessfully prepared using polymeric micelles of P123 and Tween-60as the template. The morphology of the HA HNPs could be trans-formed from nanospheres to nanotubes by adding citric acid as acosurfactant. All the synthesized HA hollow nanoparticles had anarrow size distribution. The HA nanospheres had an averagediameter of 60 nm and a hollow diameter of 3 6 nm, with a specificsurface area of 66.11 m2 g�1 and a pore volume of 0.47 cm3 g�1.The as-dried HA nanotubes were modified with citrate with athickness of 1.15 nm, and had an average diameter of about36 nm. The calcinations increased the hollow size and the specificsurface area of the hollow HA nanorods, from 13.3 to 15.6 nm andfrom 97.76 to 116.8 m2 g�1, respectively. The hollow property gaveall the hollow HA nanoparticles a much higher drug payload thantraditional HA nanoparticles. This smart drug system based onHA nanotubes functionalized with citrate carboxyl–cationic poly-electrolyte gates has an excellent pH-responsivity and a high drug

2218 F. Ye et al. / Acta Biomaterialia 6 (2010) 2212–2218

payload ratio of 35.83 wt.%. The loaded drug was stably stored in-side the hollow HA nanorods at neutral pH (7.4) and was releasedover time when the pH was at weak acidity.

Acknowledgement

The present study was supported by the Program for New Cen-tury Excellent Talents in University China (Grant No. NCET-04-0336).

Appendix A. Figures with essential colour discrimination

Certain figures in this article, particularly Figures 3, 8–10, aredifficult to interpret in black and white. The full colour imagescan be found in the on-line version, at doi:10.1016/j.actbio.2009.12.014.

Appendix B. Figures with essential colour discrimination

Certain figures in this article, particularly Figures 3, 8, 9, 10 andS1, are difficult to interpret in black and white. The full colourimages can be found in the on-line version, at doi: 10.1016/j.actbio.2009.12.014.

References

[1] Caruso F. Nanoengineering of particle surfaces. Adv Mater 2001;13:11–22.[2] Son SJ, Bai X, Lee SB. Inorganic hollow nanoparticles and nanotubes in

nanomedicine. Part 1. Drug/gene delivery applications. Drug Discov Today2007;12:650–6.

[3] Sibilla P, Sereni A, Aguiari G, Banzi M, Manzati E, Mischiati C. Effects of ahydroxyapatite-based biomaterial on gene expression in osteoblast-like cells. JDent Res 2006;85:354–8.

[4] Dasgupta S, Bandyopadhyay A, Bose S. Reverse micelle-mediated synthesis ofcalcium phosphate nanocarriers for controlled release of bovine serumalbumin. Acta Biomater 2009. doi:10.1016/j.actbio.2009.04.031.

[5] Liu TY, Chen SY, Liu DM, Liou SC. On the study of BSA-loaded calcium-deficienthydroxyapatite nano-carriers for controlled drug delivery. J Control Release2005;107:112–21.

[6] Palazzo B, Iafisco M, Laforgia M, Margiotta N, Natile G, Bianchi CL. Biomimetichydroxyapatite-drug nanocrystals as potential bone substitutes withantitumor drug delivery properties. Adv Funct Mater 2007;17:2180–8.

[7] Zhu X, Eibl O, Scheideler L, Geis-Gerstorfer J. Characterization of nanohydroxyapatite/collagen surfaces and cellular behaviors. J Biomed Mater ResA 2006;79A:114–27.

[8] Zhu X, Eibl O, Berthold C, Scheideler L, Geis-Gerstorfer J. Structuralcharacterization of nanocrystalline hydroxyapatite and adhesion of pre-osteoblast cells. Nanotechnology 2006;17:2711–21.

[9] Cai YR, Liu YK, Yan WQ, Hu QH, Tao JH, Zhang M, et al. Role of hydroxyapatitenanoparticle size in bone cell proliferation. J Mater Chem 2007;17:3780–7.

[10] Yang P, Quan Z, Li C, Kang X, Lian H, Lin J. Bioactive, luminescent andmesoporous europium-doped hydroxyapatite as a drug carrier. Biomaterials2008;29:4341–7.

[11] Wang Y, Wang J, Liu HZ. FTIR spectroscopic study on effects of temperatureand polymer composition on the structural properties of PEO–PPO–PEO blockcopolymer micelles. Langmuir 2002;18:5370–4.

[12] Kumbhakar M, Goel T, Nath S, Mukherjee T, Pal H. Microenvironment in thecorona region of triblock copolymer micelles: temperature dependentsolvation and rotational relaxation dynamics of coumarin dyes. J Phys ChemB 2006;110:25646–55.

[13] Ganguly R, Choudhury N, Aswal VK, Hassan PA. Pluronic L64 micelles nearcloud point: investigating the role of micellar growth and interaction in criticalconcentration fluctuation and percolation. J Phys Chem B 2009;113:668–75.

[14] Zhao DY, Huo QS, Feng JL, Chmelka BF, Stucky GD. Nonionic triblock and stardiblock copolymer and oligomeric surfactant syntheses of highly ordered,hydrothermally stable, mesoporous silica structures. J Am Chem Soc1998;120:6024–36.

[15] Kim JM, Stucky GD. Synthesis of highly ordered mesoporous silica materialsusing sodium silicate and amphiphilic block copolymers. Chem Commun2000:1159–60.

[16] Burleigh MC, Markowitz MA, Wong EM, Lin JS, Gaber BP. Synthesis of periodicmesoporous organosilicas with block copolymer templates. Chem Mater2001;13:4411–2.

[17] Zhao YF, Ma J. Triblock co-polymer templating synthesis of mesostructuredhydroxyapatite. Microporous Mesoporous Mater 2005;87:110–7.

[18] Sierra L, Valange S, Barrault J, Guth JL. Templating behavior of a triblockcopolymer surfactant with very long hydrophilic PEO chains(PEO140PPO39PEO140) for the synthesis of cubic mesoporous silica with largecage-like cavities. Microporous Mesoporous Mater 2008;113:352–61.

[19] Huddleston JG, Willauer HD, Griffin ST, Rogers RD. Aqueous polymericsolutions as environmentally benign liquid/liquid extraction media. Ind EngChem Res 1999;38:2523–39.

[20] Tani H, Suzuki Y, Matsuda A, Kamidate T. Enhancement of the excluded-volume effect in protein extraction using triblock copolymer-based aqueousmicellar two-phase systems. Anal Chim Acta 2001;429:301–9.

[21] Kim Il-S, Kumta PN. Sol–gel synthesis and characterization of nanostructuredhydroxyapatite powder. Mater Sci Eng B 2004;111:232–6.

[22] Cifuentes I, González-Díaz PF, Cifuentes-Delatte L. Is there a citrate apatite inbiological calcified systems? Calcif Tissue Int 1980;31:147–51.

[23] Ye F, Guo HF, Zhang HJ. Biomimetic synthesis of oriented hydroxyapatitemediated by nonionic surfactants. Nanotechnology 2008;19:245605.

[24] Gritti F, Guiochon G. Elution of propranolol as an ion–pair complex by buffersolutions on C18-silica. Anal Chem 2004;76:7310–22.