Mono and Multiple Tumor-Targeting Ligand-Coated Ultrasmall Gadolinium Oxide Nanoparticles: Enhanced Tumor Imaging and Blood Circulation

Hydrophilic and biocompatible PAA-coated ultrasmall Gd2O3 nanoparticles (davg = 1.7 nm) were synthesized and conjugated with tumor-targeting ligands, i.e., cyclic arginylglycylaspartic acid (cRGD) and/or folic acid (FA). FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles were successfully applied in U87MG tumor-bearing mice for tumor imaging using T1 magnetic resonance imaging (MRI). cRGD/FA-PAA-Gd2O3 nanoparticles with multiple tumor-targeting ligands exhibited higher contrasts at the tumor site than FA-PAA-Gd2O3 nanoparticles with mono tumor-targeting ligands. In addition, the cRGD/FA-PAA-Gd2O3 nanoparticles exhibited higher contrasts in all organs, especially the aorta, compared with those of the FA-PAA-Gd2O3 nanoparticles, because of the blood cell hitchhiking effect of cRGD in the cRGD/FA-PAA-Gd2O3 nanoparticles, which prolonged their circulation in the blood.


Preparation of FA-PAA-Gd 2 O 3 Nanoparticles
FA-PAA was first prepared as described previously (Figure 1a) [40]. To obtain FA-NH 2 -Boc, 0.9 mmol FA was dissolved in DMSO (15 mL) in a 100-mL three-neck round-bottom flask at 60 • C under N 2 flow with magnetic stirring. After the solution attained room temperature, 1.0 mmol DCC and 0.1 mmol DMAP were dissolved in the solution by magnetic stirring for 1 h. Next, 1.8 mmol EDA-Boc was dissolved in the solution by magnetic stirring for another 12 h. The resulting solution was slowly poured into cold ethyl acetate, and finally FA-EDA-Boc (yellow precipitate) was washed several times with ethyl acetate. Synthesis of FA-NH 2 -TFA was carried out by dissolving the yellow precipitate in 2 mL TFA in a 100-mL three-neck round-bottom flask with magnetic stirring for 3 h at room temperature. Chloroform was slowly poured into the solution until a yellow precipitate was obtained. Next, the clear solution was removed and precipitate was washed three times with ethyl acetate. The obtained FA-NH 2 -TFA was dried to powdered form using a rotary evaporator. To obtain FA-PAA, FA-NH 2 -TFA was dissolved in 5 mL DMSO containing 40 µL TEA with magnetic stirring. Separately, 1.5 mmol PAA was dissolved in DMSO (20 mL) under N 2 flow at 60 • C in a 100-mL three-neck round-bottom flask with magnetic stirring. After the solution attained room temperature, 1.5 mmol DCC and 0.15 mmol DMAP were dissolved in the solution with continuous magnetic stirring for 1 h. Then, the above-prepared FA-NH 2 -TFA solution was slowly added to the PAA solution with continuous magnetic stirring for 12 h. The obtained solution was dialyzed against triple-distilled water for 24 h (MWCO = 1000 amu). The remaining solution inside the bag was filtered through Whatman filter paper (Sigma-Aldrich, USA) and evaporated using a rotary evaporator to collect FA-PAA (dark yellow solid).
FA-PAA-Gd 2 O 3 nanoparticles were obtained using a one-pot polyol method ( Figure 1b). Briefly, a mixture of 2.0 mmol GdCl 3 ·6H 2 O, 0.3 mmol of the above-synthesized FA-PAA, and 20 mL TEG was magnetically stirred in a three-neck round-bottom flask at 60 • C under atmospheric conditions to prepare a clear precursor solution. Next, NaOH (10 mmol) dissolved in 10 mL TEG was slowly poured into the precursor solution with magnetic stirring for 12 h at 120 • C until the pH reached~9.0. Subsequently, the solution was cooled to room temperature and 400 mL ethanol was poured with magnetic stirring for~30 min. FA-PAA-Gd 2 O 3 nanoparticles were obtained by centrifugation (4000 rpm) and removing the supernatant. The nanoparticles were finally dispersed in ethanol, followed by centrifugation, and this step was repeated five times to remove TEG, free ions (Gd 3+ , Na + , and Cl − ), and unreacted FA-PAA. Finally, the product solution was dialyzed against triple-distilled water (MWCO = 2000 amu) for two days to remove any remaining impurities from the FA-PAA-Gd 2 O 3 nanoparticles. temperature, 1.5 mmol DCC and 0.15 mmol DMAP were dissolved in the solution with continuous magnetic stirring for 1 h. Then, the above-prepared FA-NH2-TFA solution was slowly added to the PAA solution with continuous magnetic stirring for 12 h. The obtained solution was dialyzed against triple-distilled water for 24 h (MWCO = 1000 amu). The remaining solution inside the bag was filtered through Whatman filter paper (Sigma-Aldrich, USA) and evaporated using a rotary evaporator to collect FA-PAA (dark yellow solid). FA-PAA-Gd2O3 nanoparticles were obtained using a one-pot polyol method ( Figure  1b). Briefly, a mixture of 2.0 mmol GdCl3•6H2O, 0.3 mmol of the above-synthesized FA-PAA, and 20 mL TEG was magnetically stirred in a three-neck round-bottom flask at 60 °C under atmospheric conditions to prepare a clear precursor solution. Next, NaOH (10 mmol) dissolved in 10 mL TEG was slowly poured into the precursor solution with magnetic stirring for 12 h at 120 °C until the pH reached ~9.0. Subsequently, the solution was cooled to room temperature and 400 mL ethanol was poured with magnetic stirring for ~30 min. FA-PAA-Gd2O3 nanoparticles were obtained by centrifugation (4000 rpm) and removing the supernatant. The nanoparticles were finally dispersed in ethanol, followed by centrifugation, and this step was repeated five times to remove TEG, free ions (Gd 3+ , Na + , and Cl − ), and unreacted FA-PAA. Finally, the product solution was dialyzed against triple-distilled water (MWCO = 2000 amu) for two days to remove any remaining impurities from the FA-PAA-Gd2O3 nanoparticles.

Preparation of cRGD/FA-PAA-Gd2O3 Nanoparticles
Three quarters of the synthesized FA-PAA-Gd2O3 nanoparticles, 1.0 mmol EDC•HCl, and 1.0 mmol NHS were added to 20 mL triple-distilled water at room temperature under atmospheric conditions (Figure 1c). The solution pH was maintained at 6.0 by adding 1.0 M HCl with magnetic stirring at room temperature for 1 h. The solution pH was then increased to 7.2 by adding 1.0 M NaOH, followed by adding 50 mg cRGD. The resulting

Preparation of cRGD/FA-PAA-Gd 2 O 3 Nanoparticles
Three quarters of the synthesized FA-PAA-Gd 2 O 3 nanoparticles, 1.0 mmol EDC·HCl, and 1.0 mmol NHS were added to 20 mL triple-distilled water at room temperature under atmospheric conditions (Figure 1c). The solution pH was maintained at 6.0 by adding 1.0 M HCl with magnetic stirring at room temperature for 1 h. The solution pH was then increased to 7.2 by adding 1.0 M NaOH, followed by adding 50 mg cRGD. The resulting solution was stirred magnetically for 12 h followed by dialysis against triple-distilled water (MWCO = 1000 amu) for one day to remove free cross-linking agents and unreacted cRGD. A portion of the COOH groups in PAA was conjugated with Gd 2 O 3 nanoparticles via hard acid (i.e., Gd 3+ ) and hard base (i.e., COO − ) bonding, and a portion of the remainder was conjugated with NH 2 groups of FA and cRGD via amide bonds.

Evaluation of Physicochemical Properties of the Nanoparticles
To measure the nanoparticle diameters, a high-resolution transmission electron microscope (HRTEM) (200 kV; FEI, Hillsboro, OR, USA; Titan G2 ChemiSTEM CS Probe) Pharmaceutics 2022, 14, 1458 5 of 17 was used. The colloidal nanoparticles dispersed in aqueous media were dropped using a micropipette (2-20 µL, Eppendorf, Hamburg, Germany) onto a carbon film supported by a 200-mesh copper grid (Ted Pella Inc., Redding, CA, USA; Pelco No. 160) and air-dried at room temperature. Subsequently, the elements (C, N, O, and Gd) present in the nanoparticles were identified by an energy-dispersive X-ray spectroscope (EDS) (Bruker, Berlin, Germany; Quantax Nano) installed inside the HRTEM. To measure the Gd concentration in nanoparticle suspension, inductively coupled plasma-atomic emission spectroscopy (ICP-AES) (Thermo Jarrell Ash Co., Waltham, MA, USA; IRIS/AP) was used. The hydrodynamic diameters (a) and zeta potentials (ζ) of the nanoparticle colloids (0.01 mM Gd) were characterized using a dynamic light scattering (DLS) particle size analyzer (Malvern, Malvern, UK; Zetasizer Nano ZS). The nanoparticle structures in the powdered samples were characterized using an X-ray diffraction (XRD) machine (Philips, The Netherlands; X'PERT PRO MRD) with unfiltered CuKa (λ = 0.154184 nm) radiation; a scan range of 15-100 • and a scanning step of 0.033 • in 2θ were used. FT-IR absorption spectra (Mattson Instrument Inc., Madison, WI, USA; Galaxy 7020A) were taken using the powdered sample pellets with KBr to investigate PAA conjugation with nanoparticles, cRGD, and FA within 400-4000 cm −1 . The surface-coating amount was quantified using a thermo-gravimetric analysis (TGA) instrument (TA Instrument, New Castle, DE, USA; SDT-Q600) between room temperature and 900 • C under air flow. The average amounts (in wt.%) of surfacecoating ligands (FA-PAA and cRGD/FA-PAA) were obtained from the mass drops in TGA curves after considering water and air desorption between room temperature and~105 • C. The amount of nanoparticles was obtained from the remaining mass followed by XRD analysis. Elemental analysis (EA) (ThermoFisher, Waltham, MA, USA; Flash 2000) was carried out to measure the composition (C/H/O/N) and amount of surface-coating ligands (in wt.%) using powdered samples.

In Vitro Cellular Cytotoxicity Assay
Normal mouse hepatocytes (NCTC1469) and human malignant glioma (U87MG) cell lines were cultured in DMEM and RPMI-1640 media, respectively. Cells (5 × 10 4 ) were seeded into 24-well plates (500 µL cells/well) and incubated for 24 h in 5% CO 2 at 37 • C. The concentrated nanoparticle suspension was diluted with sterile PBS solution to prepare five test concentrations. Subsequently, 2 µL aliquots were added to the cells to obtain 10, 50, 100, 200, and 500 µM Gd concentrations, followed by 48 h incubation. Next, 200 µL CellTiter-Glo reagent was added for cell lysis and the reaction was incubated on an orbital shaker for 30 min. The cellular cytotoxicity of the nanoparticle suspension samples was measured using a CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI, USA) according to the manufacturer's instructions. Intracellular adenosine triphosphate was quantified using a Victor 3 luminometer (Perkin Elmer, Waltham, MA, USA) in the 300-700 nm wavelength range. Cell viability was measured in triplicate to obtain average values, which were normalized to those of the untreated control cells (0.0 mM Gd).

Water Proton Spin Relaxivity and Map Image Measurements
The concentrated nanoparticle suspension was diluted with triple-distilled water to prepare various concentrations (1, 0.5, 0.25, 0.125, and 0.0625 mM Gd), which were subject to analysis of the longitudinal (T 1 ) and transverse (T 2 ) water proton spin relaxation times and longitudinal (R 1 ) and transverse (R 2 ) map images using a 3.0 T MRI scanner (Siemens, Munich, Germany; Magnetom Trio Tim). Next, inverse relaxation times (1/T 1 and 1/T 2 ) were plotted as a function of Gd concentration to estimate the r 1 and r 2 values from the corresponding slopes. An inversion recovery method was used to measure the T 1 relaxation times by recording MR images at 35 different inversion times (TI) in the range of 50-1750 ms. The T 1 values were estimated from nonlinear least-square fits to the mean signal intensities at various TI values. To measure T 2 relaxation times, the Carr-Purcell-Meiboom-Gill pulse sequence was used for multiple spin-echo measurements. The MR images were obtained at 16

Preparation of Murine Tumor Model
U87MG tumor cells were cultured in RPMI-1640 containing 10% (v/v) fetal bovine serum and 1% (v/v) penicillin streptomycin for 24 h in 5% CO 2 at 37 • . Six 5-week-old male BALB/c nude mice (~20 g) were injected subcutaneously with U87MG tumor cells (5 × 10 6 cells/100 µL of PBS) in the left rump tissue, and MRI experiments were carried out after three weeks.

In Vivo T 1 MR Image Measurements
Mice were anesthetized using 1.5% isoflurane in oxygen. Measurements were taken before and after injecting the two forms of aqueous nanoparticle suspensions (approximately 0.1 mmol Gd/kg) into the tail veins of mice (N = 3 each group). A warm water blanket was used to maintain the body temperature at 37 • C during measurements. The slight breathing movements of mice, even under anesthesia, were fixed using a small animal sleeve. In addition, the mice were wrapped with a band around their abdomens to minimize abdominal movements. After the measurements, the mice were revived from anesthesia and placed in cages with free access to food and water. Radio frequency-spoiled T 1 -weighted gradient-recalled echo (GRE) sequences were used for obtaining images. The experimental parameters were as follows: H = 3.0 T, T = 37 • C, TE = 7 ms, TR = 850 ms, pixel band width = 15.63 Hz, frequency = 256 Hz, phase = 256, NEX = 3, FOV = 60 mm, FOV phase = 1, slice thickness = 1.0 mm, number of slices = 24, and spacing gap = 1.1 mm. The signal-to noise ratio (SNR) was defined as the ratio of mean signal intensity of the anatomical region of interest (ROI) to that of the background noise. The T 1 -contrast ROI was defined as SNR (t)/SNR (0), with t the time after injection and 0 the time before injection.

Physicochemical Properties of FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 Nanoparticles
The FA-PAA-Gd 2 O 3 ( Figure 2(a-i,a-ii)) and cRGD/FA-PAA-Gd 2 O 3 nanoparticles ( Figure 2(b-i,b-ii)) were nearly monodispersed and ultrasmall, with diameters ranging from 1.5-3.0 nm. The average particle diameters (d avg ) of FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 were 1.7 nm, as estimated from log-normal function fits to the observed particle diameter distributions (Figure 2c). The EDS spectra confirmed the presence of Gd, C, N, and O in the nanoparticles (Figure 2d,e). The observed values are listed in Table 1.  The hydrodynamic diameters (aavg) of FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles dispersed in aqueous media and physiological solution (0.9 NaCl wt.% in water) were measured to be 11.4 and 13.8 nm, respectively, by their DLS patterns (Figure 3ai,a-ii) using log-normal function fits to the observed hydrodynamic diameter distributions (Table 1). DLS patterns were measured three times. Similar hydrodynamic diameters were observed for both samples at all times, indicating the presence of stable colloids in aqueous and physiological solutions. PAA contains a large number of hydrophilic COOH groups; therefore, the FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles can attract a large number of water molecules, which explains the observed large aavg values and excellent colloidal stability. Moreover, the cRGD/FA-PAA-Gd2O3 nanoparticles had a higher aavg value than the FA-PAA-Gd2O3 nanoparticles due to the additional cRGDs in their surface-coating layers. Additionally, the lesser number of free COO − groups in the cRGD/FA-PAA-Gd2O3 nanoparticles resulted in their lower zeta potential (ζ; −16.6 mV) than that (−33.9 mV) of the FA-PAA-Gd2O3 nanoparticles ( Figure 3b and Table 1). As shown in Figure 3c, the aqueous nanoparticle suspensions exhibited excellent colloidal stability (i.e., no precipitation after synthesis for >1 year). The dispersion of nanoparticle colloids in aqueous media was confirmed by the Tyndall effect ( Figure 3d); laser light scattering was observed only in nanoparticle suspension samples (two cuvettes on the right), unlike in triple-distilled water (left cuvette). The hydrodynamic diameters (a avg ) of FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles dispersed in aqueous media and physiological solution (0.9 NaCl wt.% in water) were measured to be 11.4 and 13.8 nm, respectively, by their DLS patterns (Figure 3(a-i,a-ii)) using log-normal function fits to the observed hydrodynamic diameter distributions (Table 1). DLS patterns were measured three times. Similar hydrodynamic diameters were observed for both samples at all times, indicating the presence of stable colloids in aqueous and physiological solutions. PAA contains a large number of hydrophilic COOH groups; therefore, the FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles can attract a large number of water molecules, which explains the observed large a avg values and excellent colloidal stability. Moreover, the cRGD/FA-PAA-Gd 2 O 3 nanoparticles had a higher a avg value than the FA-PAA-Gd 2 O 3 nanoparticles due to the additional cRGDs in their surface-coating layers. Additionally, the lesser number of free COO − groups in the cRGD/FA-PAA-Gd 2 O 3 nanoparticles resulted in their lower zeta potential (ζ; −16.6 mV) than that (−33.9 mV) of the FA-PAA-Gd 2 O 3 nanoparticles (Figure 3b and Table 1). As shown in Figure 3c, the aqueous nanoparticle suspensions exhibited excellent colloidal stability (i.e., no precipitation after synthesis for >1 year). The dispersion of nanoparticle colloids in aqueous media was confirmed by the Tyndall effect ( Figure 3d); laser light scattering was observed only in nanoparticle suspension samples (two cuvettes on the right), unlike in triple-distilled water (left cuvette).
cRGD/FA-PAA-Gd2O3 nanoparticles resulted in their lower zeta potential (ζ; −16.6 mV) than that (−33.9 mV) of the FA-PAA-Gd2O3 nanoparticles (Figure 3b and Table 1). As shown in Figure 3c, the aqueous nanoparticle suspensions exhibited excellent colloidal stability (i.e., no precipitation after synthesis for >1 year). The dispersion of nanoparticle colloids in aqueous media was confirmed by the Tyndall effect ( Figure 3d); laser light scattering was observed only in nanoparticle suspension samples (two cuvettes on the right), unlike in triple-distilled water (left cuvette).

Crystal Structures of the Nanoparticles
The XRD patterns of FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles were recorded before and after TGA ( Figure 4). Prior to TGA, the nanoparticles were not fully crystallized owing to their ultrasmall particle size, resulting in broad amorphous XRD patterns [41]. However, crystal growth during TGA up to 900 °C led to sharp peaks of bodycentered cubic (bcc) Gd2O3 [42]. Moreover, the powdered samples subjected to TGA showed a lattice constant of 10.814 Å, which is consistent with the reported value (10.813 Å) [42].

Crystal Structures of the Nanoparticles
The XRD patterns of FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles were recorded before and after TGA (Figure 4). Prior to TGA, the nanoparticles were not fully crystallized owing to their ultrasmall particle size, resulting in broad amorphous XRD patterns [41]. However, crystal growth during TGA up to 900 • C led to sharp peaks of body-centered cubic (bcc) Gd 2 O 3 [42]. Moreover, the powdered samples subjected to TGA showed a lattice constant of 10.814 Å, which is consistent with the reported value (10.813 Å) [42].

Crystal Structures of the Nanoparticles
The XRD patterns of FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles were recorded before and after TGA (Figure 4). Prior to TGA, the nanoparticles were not fully crystallized owing to their ultrasmall particle size, resulting in broad amorphous XRD patterns [41]. However, crystal growth during TGA up to 900 °C led to sharp peaks of bodycentered cubic (bcc) Gd2O3 [42]. Moreover, the powdered samples subjected to TGA showed a lattice constant of 10.814 Å, which is consistent with the reported value (10.813 Å) [42].

Surface Coatings
The surface coating of ultrasmall Gd2O3 nanoparticles with FA-PAA and cRGD/FA-PAA was supported by FT-IR absorption spectra (Figure 5a). The C=O stretching vibration

Surface Coatings
The surface coating of ultrasmall Gd 2 O 3 nanoparticles with FA-PAA and cRGD/FA-PAA was supported by FT-IR absorption spectra (Figure 5a). The C=O stretching vibration of the COOH groups of PAA at 1695 cm −1 exhibited red-shift and split into COO − antisymmetric and symmetric stretching vibrations at 1540 and 1400 cm −1 , respectively [43], confirming the successful coating of PAA on the ultrasmall Gd 2 O 3 nanoparticle surface. The red-shift and split resulted from the hard acid-base bonding between the COO − (hard base) of PAA and Gd 3+ (hard acid) of the Gd 2 O 3 nanoparticles [44]. Additionally, the C-H stretching vibrations of PAA, FA, and cRGDs at~2953 cm −1 were observed in the FT-IR absorption spectra of FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles, supporting the presence of these ligands in the nanoparticles. Amide-I C=O stretching vibration of FA and cRGD (at 1642 cm −1 ) [45,46] was observed as well, confirming the successful conjugation of NH 2 groups of FA and cRGD with the COOH groups of PAA.  The surface-coating amount (P; in wt.%) was obtained by TGA. As shown in Figure  5b, the p values were 47.5 and 51.3% for FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles (Table 1), respectively, as determined by the mass loss after taking into account water and air desorption between room temperature and ~105 °C. The remaining mass was ascribed to Gd2O3 nanoparticles ( Figure 5b and Table 1). The cRGD/FA-PAA-Gd2O3 nanoparticles had a higher p than that of the FA-PAA-Gd2O3 nanoparticles due to additional cRGDs in their structure. Based on the EA, p values were 52.4 and 56.8% for FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles, respectively, as determined by adding the C/H/O/N atomic wt.%, i.e., 22.57/3.48/25.01/1.32 and 24.44/3.75/25.86/2.76, respectively. These values were slightly higher than those estimated by TGA because the water and air content in the samples could not be eliminated from the measured wt.% in EA. The estimated wt.% difference (i.e., 3.8% by TGA and 4.4% by EA, for an average of 4.1%) between cRGD/FA-PAA-Gd2O3 and FA-PAA-Gd2O3 nanoparticles represented the wt.% of cRGD. Assuming that the PAA/FA molar ratio of 1.5/0.9 used in FA-PAA synthesis was maintained in the nanoparticle samples, the wt.% of cRGD/FA/PAA was estimated as 4.1/6.8/45.9. Based on the bulk density of Gd2O3 (7.407 g/cm 3 ) [47], p values estimated from TGA and EA, average mass of FA-PAA (2064 g) and cRGD/FA-PAA (2225 g) obtained using the above-estimated ligand wt.% ratio, and davg value determined by HRTEM imaging, the grafting density (σ, i.e., the average number of FA-PAA and cRGD/FA-PAA coating a Gd2O3 nanoparticle unit surface area) [48] was found to be 0.6-0.7 nm −2 . By multiplying σ with the Gd2O3 nanoparticle surface area (πdavg 2 ), the average number (NNP) of FA-PAA and cRGD/FA-PAA coating each Gd2O3 nanoparticle was found to be 6-7. The surface-coating results are listed in Table 1.  (Table 1), respectively, as determined by the mass loss after taking into account water and air desorption between room temperature and~105 • C. The remaining mass was ascribed to Gd 2 O 3 nanoparticles (Figure 5b and Table 1 [47], p values estimated from TGA and EA, average mass of FA-PAA (2064 g) and cRGD/FA-PAA (2225 g) obtained using the above-estimated ligand wt.% ratio, and d avg value determined by HRTEM imaging, the grafting density (σ, i.e., the average number of FA-PAA and cRGD/FA-PAA coating a Gd 2 O 3 nanoparticle unit surface area) [48] was found to be 0.6-0.7 nm −2 . By multiplying σ with the Gd 2 O 3 nanoparticle surface area (πd avg 2 ), the average number (N NP ) of FA-PAA and cRGD/FA-PAA coating each Gd 2 O 3 nanoparticle was found to be 6-7. The surface-coating results are listed in Table 1.

r 1 , r 2 Values and R 1 , R 2 Map Images
To investigate the potential of the synthesized FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles as T 1 MRI contrast agents, T 1 and T 2 relaxation times were measured at various Gd concentrations at 3.0 T MR field and 22 • C. For 0.25, 0.5, and 1.0 mM Gd, nonlinear curve fits to obtain the T 1 and T 2 relaxation times are provided in Figure 6(a-i,a-ii), respectively. Subsequently, 1/T 1 and 1/T 2 inverse relaxation times were plotted as a function of Gd concentration to obtain r 1 and r 2 values from the corresponding slopes (Figure 6b and Table 2). As shown in Table 2, the estimated r 1 values were approximately four times higher than those of commercial Gd-chelates [49]. In addition, the synthesized nanoparticles exhibited dose-dependent contrast changes in R 1 and R 2 map images (Figure 6c). Considering that the r 2 /r 1 ratios were close to 1, these results indicate that the synthesized nanoparticles could act as high-performance T 1 MRI contrast agents.  Table 2). As shown in Table 2, the estimated r1 values were approximately four times higher than those of commercial Gd-chelates [49]. In addition, the synthesized nanoparticles exhibited dose-dependent contrast changes in R1 and R2 map images ( Figure  6c). Considering that the r2/r1 ratios were close to 1, these results indicate that the synthesized nanoparticles could act as high-performance T1 MRI contrast agents. The r2/r1 ratio is greater than 1 because longitudinal relaxation accompanies transverse relaxation, whereas the reverse is not feasible. Therefore, r2/r1 ratios close to 1 and as large as possible are ideal for T1 and T2 MRI contrast agents, respectively. Therefore, Gd-chelates and iron oxide nanoparticles are suitable for use as T1 and T2 MRI contrast agents, respectively. Similarly, FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles are potential T1 MRI contrast agents, as their r2/r1 ratios are close to 1.

Cellular Toxicity of the Nanoparticles
The toxicity of FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles was investigated by measuring the viability of NCTC1469 normal and U87MG tumor cells. As shown  The r 2 /r 1 ratio is greater than 1 because longitudinal relaxation accompanies transverse relaxation, whereas the reverse is not feasible. Therefore, r 2 /r 1 ratios close to 1 and as large as possible are ideal for T 1 and T 2 MRI contrast agents, respectively. Therefore, Gd-chelates and iron oxide nanoparticles are suitable for use as T 1 and T 2 MRI contrast agents, respectively. Similarly, FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles are potential T 1 MRI contrast agents, as their r 2 /r 1 ratios are close to 1.

Cellular Toxicity of the Nanoparticles
The toxicity of FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles was investigated by measuring the viability of NCTC1469 normal and U87MG tumor cells. As shown in Figure 7a, NCTC1469 cells exhibited good viability when treated with up to 500 µM Gd in both nanoparticle samples. However, the viability of U87MG cells decreased with increasing Gd concentration (Figure 7b). The toxicity observed in U87MG cells was attributed to the overexpressed receptors and integrins in tumor cells compared with those in normal cells and the resultant targeting effect of nanoparticles [31][32][33][34][35][36][37][38][39]. In addition, at high Gd concentrations, increased cellular toxicity of the cRGD/FA-PAA-Gd 2 O 3 nanoparticles compared to that of the FA-PAA-Gd 2 O 3 nanoparticles was attributed to multiple targeting by cRGD and FA in the cRGD/FA-PAA-Gd 2 O 3 nanoparticles. in Figure 7a, NCTC1469 cells exhibited good viability when treated with up to 500 μM Gd in both nanoparticle samples. However, the viability of U87MG cells decreased with increasing Gd concentration (Figure 7b). The toxicity observed in U87MG cells was attributed to the overexpressed receptors and integrins in tumor cells compared with those in normal cells and the resultant targeting effect of nanoparticles [31][32][33][34][35][36][37][38][39]. In addition, at high Gd concentrations, increased cellular toxicity of the cRGD/FA-PAA-Gd2O3 nanoparticles compared to that of the FA-PAA-Gd2O3 nanoparticles was attributed to multiple targeting by cRGD and FA in the cRGD/FA-PAA-Gd2O3 nanoparticles.
Recently, enhanced cytosolic concentration of reactive oxygen species (ROS) and autophagic vesicles has been reported as a result of internalized gadolinium oxide nanoparticles in human umbilical vein endothelial and breast cancer cells (MCF-7) [50,51]. Consequently, potential mitochondrial membrane collapse, cell viability reduction, and cell death via necrosis and apoptosis were observed. In addition, growing evidence supports nanoparticle-induced ROS and subsequent ROS-mediated cellular apoptosis and necrosis for various nanoparticle systems [52][53][54]. Similar cytotoxic effects probably decreased U87MG cell viability with increasing Gd concentration in the present study. However, detailed studies are needed to unfold the mechanisms underlying FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticle-mediated cytotoxicity in U87MG tumor cells.

In Vivo T1 MRI
T1 MR images of the tumor and organs including the liver, kidneys, and aorta were measured before and after intravenous injection (up to 3 h) of the aqueous nanoparticle suspension samples into mice tails (Figure 8). Positive contrasts were observed in the tumor and all organs after injection, confirming that the nanoparticle samples acted as T1 MRI contrast agents. To study the contrast changes with time, the T1-contrast of the ROI were plotted as a function of time (Figure 9a-d), and they increased to reach maxima within an hour after injection, followed by a decrease thereafter. Notably, the T1-contrast ROIs were the highest in the aorta, followed by the kidneys for both FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles, indicating their prolonged blood circulation and delayed renal excretion. In addition, the T1-contrast ROIs of the cRGD/FA-PAA-Gd2O3 nanoparticles were higher than those of the FA-PAA-Gd2O3 nanoparticles for tumors and all organs, confirming that cRGD enhanced tumor imaging and prolonged the blood circulation duration. Recently, enhanced cytosolic concentration of reactive oxygen species (ROS) and autophagic vesicles has been reported as a result of internalized gadolinium oxide nanoparticles in human umbilical vein endothelial and breast cancer cells (MCF-7) [50,51]. Consequently, potential mitochondrial membrane collapse, cell viability reduction, and cell death via necrosis and apoptosis were observed. In addition, growing evidence supports nanoparticle-induced ROS and subsequent ROS-mediated cellular apoptosis and necrosis for various nanoparticle systems [52][53][54]. Similar cytotoxic effects probably decreased U87MG cell viability with increasing Gd concentration in the present study. However, detailed studies are needed to unfold the mechanisms underlying FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticle-mediated cytotoxicity in U87MG tumor cells.

In Vivo T 1 MRI
T 1 MR images of the tumor and organs including the liver, kidneys, and aorta were measured before and after intravenous injection (up to 3 h) of the aqueous nanoparticle suspension samples into mice tails (Figure 8). Positive contrasts were observed in the tumor and all organs after injection, confirming that the nanoparticle samples acted as T 1 MRI contrast agents. To study the contrast changes with time, the T 1 -contrast of the ROI were plotted as a function of time (Figure 9a-d), and they increased to reach maxima within an hour after injection, followed by a decrease thereafter. Notably, the T 1 -contrast ROIs were the highest in the aorta, followed by the kidneys for both FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles, indicating their prolonged blood circulation and delayed renal excretion. In addition, the T 1 -contrast ROIs of the cRGD/FA-PAA-Gd 2 O 3 nanoparticles were higher than those of the FA-PAA-Gd 2 O 3 nanoparticles for tumors and all organs, confirming that cRGD enhanced tumor imaging and prolonged the blood circulation duration.

Discussion
In the present study, mono (i.e., FA) and multiple (i.e., cRGD and FA) tumor-targeting ligand-coated ultrasmall Gd 2 O 3 nanoparticles were synthesized. FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles were nearly monodispersed with an average particle diameter of 1.7 nm. The hydrodynamic diameters were 11.4 and 13.8 nm and zeta potentials were −33.9 and −16.6 mV for FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles, respectively. Their colloidal stability was excellent, as the nanoparticles did not precipitate for more than one year after synthesis. Both nanoparticle samples exhibited approximately four times higher r 1 values compared with those of the commercial molecular chelates [49], confirming their potential as high-performance T 1 MRI contrast agents.
Both forms of the nanoparticles did not show any toxicity in NCTC1469 cells up to 500 µM Gd concentration. However, increased toxicity was observed in U87MG cells with increasing Gd concentration (Figure 7b). This was attributed to the tumor-targeting effect of the nanoparticles. In addition, the toxicity of the cRGD/FA-PAA-Gd 2 O 3 nanoparticles was slightly higher than that of the FA-PAA-Gd 2 O 3 nanoparticles because of multiple tumor targeting by cRGD and FA in the cRGD/FA-PAA-Gd 2 O 3 nanoparticles.
Additionally, the T 1 -contrast ROIs of cRGD/FA-PAA-Gd 2 O 3 nanoparticles in the tumor were higher than those of the FA-PAA-Gd 2 O 3 nanoparticles (Figure 9a). This demonstrates the superiority of the multiple-targeting over the mono-targeting approach for tumor imaging. As shown in Figure 10

Conclusions
Hydrophilic and biocompatible PAA-coated ultrasmall Gd2O3 nanoparticles (davg = 1.7 nm) were successfully conjugated with the tumor-targeting ligands FA and/or cRGD. The FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles exhibited excellent colloidal stability (no precipitation for >1 year after synthesis). They were successfully applied for tumor imaging in U87MG tumor-bearing mice via T1 MRI. The salient outcomes of our study can be summarized as follows: (1) Both nanoparticles displayed r1 values approximately four times higher (12.0 and 11.2 s −1 mM −1 for FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles, respectively) than those of commercially available Gd-chelates. (2) The cRGD/FA-PAA-Gd2O3 nanoparticles exhibited higher contrasts at the tumor site than the FA-PAA-Gd2O3 nanoparticles owing to their multiple tumor-targeting effects. (3) Both nanoparticles exhibited the highest contrast in the aorta among the various organs analyzed, because of prolonged blood circulation. This is due to their ideal hydrodynamic diameters (11.4 and 13.8 nm for FA-PAA-Gd2O3 and cRGD/FA-PAA-Gd2O3 nanoparticles, respectively), which are small enough to minimize opsoniza- Both FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles exhibited the highest positive contrasts in the aorta among the organs analyzed, which included the liver, kidneys, and tumors (Figure 9a-d), showing their prolonged circulation in the blood. Nanoparticles that can circulate in the blood for prolonged durations should have hydrodynamic diameters small enough to minimize opsonization [14,15] and evade RES uptake, and large enough (>10 nm) to delay renal excretion [14,16]. The hydrodynamic diameters of the nanoparticles synthesized in this study ranged from 11 to 14 nm, thereby satisfying these conditions. As shown in Figure 9b-d, cRGD/FA-PAA-Gd 2 O 3 nanoparticles exhibited higher positive contrasts in all organs, especially in the aorta, than the FA-PAA-Gd 2 O 3 nanoparticles. This was likely due to the blood circulation-enhancing effect of the cRGD present in the cRGD/FA-PAA-Gd 2 O 3 nanoparticles. As cRGD binds to integrins expressed on blood cells (termed cell hitchhiking) [14,55], the nanoparticles can circulate for a longer duration in the blood and provide a better contrast.
It is known that nanoparticles with ultrasmall particle and hydrodynamic diameters (d < 3 nm and a < 5 nm) are excretable via the renal system [56,57]. The synthesized nanoparticles (d avg = 1.7 nm) in the present study exhibited a avg = 11.4 nm for FA-PAA-Gd 2 O 3 nanoparticles and 13.8 nm for cRGD/FA-PAA-Gd 2 O 3 nanoparticles. Therefore, a portion of the nanoparticles could be slowly excreted through the renal system, as can be noticed from the gradual decrease in SNR with time in the kidneys (Figure 9c). However, detailed studies are needed to clarify the excretion pathway of the nanoparticles. Ultrasmall nanoparticles exhibited no or negligible contrast enhancements in healthy normal brain MRI [58], supporting that they cannot pass the blood-brain barrier (BBB) for the normal brain; however, they can pass the BBB for brain tumors, possibly through damage to the BBB, as observed in brain tumor MRI with D-glucuronic acid-coated ultrasmall Gd 2 O 3 nanoparticles [59]. For other organ tumors, tumor-targeting ligand-conjugated Gd 2 O 3 nanoparticles have been successfully applied to tumor imaging via various imaging modalities [5]. The toxicity of Gd 2 O 3 nanoparticles is of great concern owing to the release of Gd 3+ ions [60][61][62]. For commercial molecular Gd 3+ -chelates, it is known that if free Gd 3+ ions are liberated in the body, this can promote nephrogenic systemic fibrosis, which is a rare disease that can lead to hardening or thickening of the skin and deposits [63]; therefore, Gd 2 O 3 nanoparticles should be completely excreted through the renal system after injection.

Conclusions
Hydrophilic and biocompatible PAA-coated ultrasmall Gd 2 O 3 nanoparticles (d avg = 1.7 nm) were successfully conjugated with the tumor-targeting ligands FA and/or cRGD. The FA-PAA-Gd 2 O 3 and cRGD/FA-PAA-Gd 2 O 3 nanoparticles exhibited excellent colloidal stability (no precipitation for >1 year after synthesis). They were successfully applied for tumor imaging in U87MG tumor-bearing mice via T 1 MRI. The salient outcomes of our study can be summarized as follows: (1) Both nanoparticles displayed r 1 values approximately four times higher (12. Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding authors.