Guiding Brain‐Tumor Surgery via Blood–Brain‐Barrier ...€¦ · due to their infiltrated nature. Here, a pair of gold nanoprobes that enter a brain tumor by crossing the blood–brain

Co

mm

un

iCatio

n

© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (1 of 9) 1603917

Guiding Brain-Tumor Surgery via Blood–Brain-Barrier-Permeable Gold Nanoprobes with Acid-Triggered MRI/SERRS Signals

Xihui Gao, Qi Yue, Zining Liu, Mengjing Ke, Xingyu Zhou, Sihan Li, Jianping Zhang, Ren Zhang, Liang Chen, Ying Mao, and Cong Li*

DOI: 10.1002/adma.201603917

or indefinable in the operating room.[2] Incomplete excision usually leads to a poor prognosis due to recurrence induced by residual tumor foci. Conversely, aggres-sive resections may not benefit patients due to the resulting injuries to nonregen-erative nerves, vessels or “intricate” areas that control movement or language.[3] As a result, intraoperative definition of brain-tumor margins is crucial for improving surgical prognosis.

Stereotactic magnetic resonance imaging (MRI) is widely used for guiding brain-tumor resection in the clinic. The tumor margins delineated by preop-erative MRI, however, are often not fully aligned to the actual borders due to the inevitable brain shift during the crani-otomy.[4] Even though intraoperative MRI (iMRI)[5] can be applied to overcome this problem, its extremely high running cost and prolonged anesthetic period limit its applications to just a few major clinical research centers. Compared to iMRI, fluorescence-guided surgery (FGS) shows higher sensitivity, more rapid acquisi-tion rates, and more affordable running

costs. For example, the Food and Drug Administration (FDA)-approved 5-aminolevulinic acid (5-ALA) improves progression-free survival of glioma patients by marking tumor margins with its fluorescent metabolite.[6] 5-ALA-based FGS, however, suffers from less-than-desirable predictive values that could

Surgical resection is a mainstay in the treatment of malignant brain tumors. Surgeons, however, face great challenges in distinguishing tumor margins due to their infiltrated nature. Here, a pair of gold nanoprobes that enter a brain tumor by crossing the blood–brain barrier is developed. The acidic tumor environment triggers their assembly with the concomitant activation of both magnetic resonance (MR) and surface-enhanced resonance Raman spectroscopy (SERRS) signals. While the bulky aggregates continuously trap into the tumor interstitium, the intact nanoprobes in normal brain tissue can be transported back into the blood stream in a timely manner. Experimental results show that physiological acidity triggers nanoparticle assembly by forming 3D spherical nanoclusters with remarkable MR and SERRS signal enhancements. The nanoprobes not only preoperatively define orthotopic glioblastoma xenografts by magnetic resonance imaging (MRI) with high sen-sitivity and durability in vivo, but also intraoperatively guide tumor excision with the assistance of a handheld Raman scanner. Microscopy studies verify the precisely demarcated tumor margin marked by the assembled nano-probes. Taking advantage of the nanoprobes’ rapid excretion rate and the extracellular acidification as a hallmark of solid tumors, these nanoprobes are promising in improving brain-tumor surgical outcome with high specificity, safety, and universality.

X. Gao, Z. Liu, M. Ke, X. Zhou, S. Li, Prof. C. LiKey Laboratory of Smart Drug DeliveryMinistry of EducationSchool of PharmacyFudan University826 Zhangheng Road, Shanghai 201203, ChinaE-mail: [email protected]. Q. Yue, Dr. L. Chen, Prof. Y. MaoDepartment of NeurosurgeryHuashan HospitalFudan University12 Middle Wulumuqi Road, Shanghai 200040, China

J. ZhangDepartment of Nuclear MedicineShanghai Cancer CenterFudan University270 Dongan Road, Shanghai 200032, ChinaR. ZhangCenter of Analysis and MeasurementFudan University220 Handan Road, Shanghai 200433, ChinaProf. Y. MaoState Key Laboratory of Medical NeurobiologySchool of Basic Medical Sciences and Institutes of Brain ScienceFudan University138 Yixueyuan Road, Shanghai 200032, China

Surgical resection remains vital to the treatment of malignant brain tumors. The extent of cytoreduction directly correlates with the chances of survival of the patients.[1] The infiltrating nature of brain tumors, however, makes it difficult for surgeons to distinguish the tumor margins as they are usually invisible

Adv. Mater. 2017, 1603917

www.advancedsciencenews.com www.advmat.de

http://doi.wiley.com/10.1002/adma.201603917

Co

mm

un

iCati

on

© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimwileyonlinelibrary.com1603917 (2 of 9)

be the consequences of the variable metabolic potentials of the cancer cells and rapid fluorescence quenching after exci-tation by a violet-blue light.[7] As another important optical imaging technique, surface-enhanced resonance Raman scat-tering (SERRS) imaging shows great promise in guiding tumor resection because of its ultrahigh sensitivity, uncompromised stability, and fingerprint-like spectra.[8] Probes tracked by MR and SERRS imaging simultaneously could accurately delineate brain-tumor boundaries by overcoming the mismatched co-reg-istration between the preoperative and intraoperative images.[9]

Gold nanospheres (AuNS) are commonly used in the devel-opment of SERRS probes because of their good biocompatibility and ease of synthesis.[10] Gambhir and co-workers reported an AuNS-based MRI/Raman/photoacoustic triple-modality nano-probe that visualized glioma xenografts with ultrahigh sensi-tivity (10−12 M).[11] Huang et al. similarly showed the capability of AuNS nanoprobes to detect multifocal malignant residues in a genetically engineered glioma model.[12] Their common fea-tures included: i) a large diameter (>120 nm), which is required to resonantly absorb the incident NIR light upon excitation of their surface plasmon oscillations, and ii) intratumoral delivery via the enhanced permeability and retention (EPR) effect; how-ever, this effect may also hinder their clinical translation. The bulky AuNSs are easily captured by the reticuloendothelial system, resulting in their nonspecific localization in normal tis-sues. Additionally, the EPR effect of brain tumors is reported to be much weaker than that of periphery tumors.[13] Therefore, improving the pharmacokinetics and vasculature permeability of the AuNS nanoprobes is essential to demarcate brain-tumor margins with high clarity and safety.

The assembly of AuNSs remarkably increases the SERRS signal by forming numerous interparticle “hotspots”[14] and enhances longitudinal MR signals by lengthening the rotational correlation time (τR) of the conjugated Gd3+ chelators.[15] Acidi-fication of the tumor extracellular fluid (TEF) is a hallmark of solid tumors and plays an important role in the invasion of cancer cells by activating proteases, the remodeling of the extra-cellular matrix, and the suppression of immune responses.[16] We hypothesized that the assembly of small volume AuNSs in the acidic TEF would visualize tumors with concomitant MR/SERRS signal activation and reduce potential side effects due to their rapid excretion rate. Recently, Ellingon and co-workers obtained a pH map of the brains of glioma patients using chem-ical exchange saturation transfer MRI technology,[17] which not only verified the existence of an acidic gradient at the margins of brain tumors but also histologically confirmed the correla-tion between the acidic signature and viability of glioma cells. Therefore, the acid-triggered MR and SERRS signal enhance-ment in the TEF would help to achieve maximal excision of the neoplastic tissues while minimizing injuries to the nearby functional areas of the brain.

The blood–brain barrier (BBB) precisely regulates brain homeostasis and restricts the brain’s uptake of most exogenous compounds.[18] Clinically used MRI-contrast agents such as gadolinium-diethylenetriamine penta-acetic acid (Gd3+ –DTPA) image brain tumors by diffusing into the tumor interstitium where the BBB is disrupted.[19] However, tumor-associated BBB breakdown in high-grade brain tumors is heterogeneous, with the tumor core being the most permeable in comparison to

the impermeable proliferating tumor periphery.[20] Addition-ally, BBB is kept intact in most low-grade brain tumors that cannot be visualized by Gd3+–DTPA.[21] Therefore, it is neces-sary to develop probes that visualize brain-tumor-infiltrated margins by circumventing the BBB. Receptor-mediated trans-cytosis (RMT) is a natural process through which endogenous molecules are transported within endocytic vesicles from the luminal to abluminal side of the brain capillaries after binding to their corresponding endothelial receptors. Low-density lipoprotein-receptor-related protein-1 (LRP1) is an endothe-lial receptor that not only carries multiple substrates from the blood into the brain but also actively transports brain-derived metabolites from the brain into the blood.[22] By taking advan-tage of the bidirectional BBB traversing capability of LRP1, it is possible to enhance intratumoral uptake of AuNS nanoprobes while rapidly eliminating nonspecific delivery to the normal brain tissues.

Here, we developed a pair of AuNS-based nanoprobes to guide brain-tumor surgery by simultaneously activating MR and SERRS signals via their specific assembly in the acidic TEF. These nanoprobes, Au–AZ and Au–AK, existed as mono-disperse nanoparticles in a neutral environment (Figure 1a). After intravenous (i.v.) injection of the mixture of the two, both of the nanoprobes entered brain tumors by crossing the BBB via LRP1-mediated RMT. The removal of the shielding layer in an acidic tumor environment exposed the azide and alkyne functionalities on the gold core surface. The click cycloaddi-tion between the decoated Au–AZ and Au–AK triggered their aggregation with the concomitant activation of both MR and SERRS signals. While the bulky AuNS aggregates were retained in tumor extracellular interstitium, the intact nanoprobes in the normal brain tissue were transported back into the blood stream (Figure 1b). Considering the overexpression of LRP1 in the glioma neovasculature,[23] which enables specific MR and SERRS signal activation in tumor sites and rapid excretion of the intact nanoprobes, these acid-responsive nanoprobes are promising for guiding brain-tumor resection with high sensi-tivity and safety and broad applicability.

Au–AZ and Au–AK possessed a similar structure, including an inner AuNS core, an intermediate functional layer, and an outer shielding layer (Figure 1a). AuNSs with a diameter of 20 nm were chosen as the metallic core because of its appro-priate circulation lifetime and inherently low SERRS signal, which enabled the minimization of background signals. The intermediate layer consisted of Raman reporters, paramagnetic chelators and either azide or alkyne functionalities on the mod-ified metallic core surface. Heptamethine cyanine derivative IR783B was chosen as the Raman reporter because its absorp-tion maxima could resonate with near-infrared (NIR) excitation light, which significantly increases the sensitivity to a magni-tude of 10−12–10−14 M.[8,12] Additionally, the use of NIR light could remarkably reduce the phototoxicity and autofluorescence background due to the inherently low tissue absorption in the NIR wavelength range (650–900 nm). The strain-promoted copper-free cycloaddition between the azide and dibenzocy-clooctyne groups labeled on to Au–AZ and Au–AK, respectively, facilitated the rapid assembly of the AuNSs in vivo. The outer shielding layer was comprised of a flexible polyethylene glycol (PEG) chain (MW: 2 kDa). While the inward ends of the PEG

Adv. Mater. 2017, 1603917

www.advancedsciencenews.comwww.advmat.de

Co

mm

un

iCatio

n


chains coated the metallic surface through acid-labile hydra-zone bonds, their outward chain ends were partially modified with BBB-permeable angiopep2 peptides. The acid-labile PEG coating not only improves the circulation half-life and biocom-patibility of the nanoprobes but also prevents their nonspecific assembly.

The synthetic procedure of Au–AZ and Au–AK is outlined in Figure S1 in the Supporting Information. For comparison, con-trol nanoprobes AuC–AZ and AuC–AK without pH-sensitive groups were also prepared. The synthesis of Au–AZ and Au–AK was monitored by infrared (IR) spectroscopy (Figure S2 and S3, Supporting Information). Both of the nanoprobes shared sim-ilar spectra, including a strong peak at 1720 cm−1 attributed to the CO stretching in DTPA, scissor-like peaks at 1044 and 1193 cm−1 assigned to the SO stretching in IR783B, a broad peak at 2800–2960 cm−1 attributed to CH stretching in the PEG chain and peaks at 1650 and 1550 cm−1 assigned to the typical amide bending in the angiopep-2 peptide. Au–AZ also showed a characteristic peak at 2100 cm−1 corresponding to NNN stretching in the azide moiety. Transmission electron

microscopy (TEM) images showed that the nanoprobes were distributed in a dispersed manner as spherical core–shell struc-tures (Figure S4a, Supporting Information). The metallic core with an average diameter of 20 nm was coated by a semitrans-parent PEG corona with a thickness of 3 nm. Both Au–AZ and Au–AK displayed a surface plasmon absorption at 521 nm (Figure S4b, Supporting Information), and they shared similar SERRS spectra with strong twin peaks with Raman shifts of 509 and 541 cm−1 (Figure S4c, Supporting Information) that were assigned to CS stretching and CCC bending vibra-tion, respectively.[24] The average hydrodynamic diameters and zeta potentials of the nanoprobes were determined to be 25 nm and −16 mV, respectively, by dynamic light scattering (Figure S4d, Supporting Information). The physical parameters of these probes are listed in Table S1 in the Supporting Infor-mation. Approximately 10568 Gd3+ ions and 1029 IR783B mol-ecules were conjugated to each nanoprobe. The average molar ratio of angiopep-2/PEG/azide or alkyne/AuNP was determined to be 54/5047/11312 or 11361/1, respectively. Both of the nano-probes were well dispersed in 10 × 10−3 m phosphate buffered

Adv. Mater. 2017, 1603917


Figure 1. Guiding brain-tumor surgery by acid-responsive gold nanoprobes. a) The cleavage of the PEG coating in physiological acidity triggers aggrega-tion between Au–AZ and Au–AK via “click” cycloaddition reactions. b) Due to the LPR1-mediated, bidirectional, BBB-traversing strategy, while the bulky AuNS aggregates are continuously trapped in tumor acidic interstitium, the intact nanoprobes in the normal brain tissue can be transported back into the blood stream, which increases the sensitivity for the brain-tumor margins.

Co

mm

un

iCati

on


saline (PBS) buffer (pH 7.4) containing 10% bovine serum albumin (BSA) without any noticeable aggregation and precipi-tation over 48 h.

The acid-triggered assembly between Au–AK and Au–AZ was tested at pH 7.4, 6.5, and 5.5, which mimic the physi-ological environments in normal tissue, tumor extracellular fluid, and lysosomal lumen, respectively. Figure 2a shows the hydrodynamic diameters of the Au–AZ/Au–AK mixture (0.75 × 10−9 m for each nanoprobe) as functions of pH and incubation time. At pH 7.4, the mixture showed an initial

diameter of 26 nm and a polydispersity index (PDI) of 0.251. The average diameter increased slightly to 31 and 43 nm at 2 and 8 h postincubation. The assembly rate of the nanoprobes accelerated substantially in the more acidic environments. The nanoparticle diameters were determined to be 49, 92, and 238 nm at 0.5, 2, and 8 h postincubation, respectively, at pH 6.5. The highest assembly rate was observed at pH 5.5, and the corresponding diameters were 107, 231, and 249 nm at the same time points. Notably, the PDI value increased to 0.438 and 0.471 at pH 6.5 and 5.5, respectively, after 2 h incubation,

Adv. Mater. 2017, 1603917


Figure 2. Physiological acidity triggered assembly of the Au–AZ/Au–AK mixture. a) Hydrodynamic diameters of the mixture (0.75 × 10−9 m for each nanoprobe) as functions of pH and incubation time. b) TEM images of the mixture (0.75 × 10−9 m for each) at 2 h postincubation at selected pH values. Insets: enlarged AuNSs or AuNS aggregates. The arrows denote the semitransparent PEG coating. c) Absorption spectra of the mixture (0.75 × 10−9 m for each nanoprobe) as functions of pH and incubation time.

Co

mm

un

iCatio

n


which indicates the various assembly degrees of the nano-probes under acidic conditions.

TEM images showed that while the Au–AK/Au–AZ mix-ture was monodisperse (20 nm) in a neutral environment, the diminishing of the PEG coating with concomitant interpar-ticle assembly was observed under physiological acidic condi-tions (Figure 2b). Interestingly, all these aggregates showed a 3D structure with well-organized spherical geometry. After 2 h incubation at pH 6.5, the diameters of aggregates were distrib-uted in a range of 20–115 nm (mean: 72 nm). Larger aggregates with diameters ranging from 60 to 200 nm (mean: 163 nm) were observed at pH 5.5. As expected, no noticeable aggregates were detected after incubation of pH-inert AuC–AZ/AuC–AK mixture in either acidic or neutral environments. Since AuNS assembly could significantly change the solution color or the Raman or photoacoustic signal,[25] extensive efforts have been made to trigger AuNS aggregation via ionic strength,[25] elec-trostatic affinity,[26] and hydrophobic interaction.[27] Most of the above strategies, however, are not suitable for clinical transla-tion due to the production of agglomerates with amorphous morphology, poor water solubility, and low physiological sta-bility. Crosslinker-mediated assembly is a promising approach to controlling the size as well as the geometrical morphology of the aggregates because the concentration and coupling activity can be conveniently adjusted. Brust and co-workers showed that the spherical aggregates only occurred when the number of crosslinkers on the AuNS building blocks (≈5 nm) was in a range of 60–14 000 (≈0.76–178 crosslinkers nm−2).[28] In this work, ≈10 000 click agents were conjugated to the nanoprobes (≈7.9 clickable agents nm−2), which satisfies the prerequisite to form the spherical aggregates. Interestingly, the globular

aggregates under acidic conditions ceased their growth after reaching a diameter of ≈200 nm. Berlin and co-workers con-firmed that the diameters of the aggregates depended linearly on the concentration of the AuNS building blocks and logarith-mically on the concentration of the crosslinkers.[29] Therefore, it is feasible to adjust the morphology and size of the AuNS aggregates to amplify their signal intensities.

At neutral pH, the Au–AK/Au–AZ mixture showed a single absorption centered at 523 nm, which was assigned to the transverse plasmon band (TPB) of monodisperse AuNPs (Figure 2c). While a bathochromic shift from 523 to 536 nm was observed after incubation at pH 6.5 for 2 h, the TPB moved to 558 nm with an appearance of a new broad absorption at 685 nm, which was assigned to the longitudinal plasmon band (LPB) of the AuNS aggregates. Zhang and co-workers claimed that the changes in the TPB and LPB were results of the inter-action between the AuNS building blocks.[30] While weak inter-actions with large interparticle distances led to a redshift of the TPB, strong interactions with short interparticle distances generated the LPB. The appearance of the LPB indicated the initiation of compact stacking between the AuNSs in the aggre-gates, which was beneficial to amplifying the SERRS signal by generating “hotspot” matrices. Additionally, the LPB could reso-nate with the optical absorption of the Raman reporter and the wavelength of excitation source, which was a prerequisite for the occurrence of SERRS under irradiation with NIR light.[31]

The longitudinal relaxivity (r1p) of the Au–AK/Au–AZ mix-ture (0.75 × 10−9 m for each nanoprobe) as functions of pH and AuNS concentration was investigated at 7.0 T (Figure 3a). At 2 h postincubation, while the r1p value of the mixture increased by 14.8% to 5.6 mM−1 s−1 at pH 5.5, r1p enhancements of 2.3%

Adv. Mater. 2017, 1603917


Figure 3. Au–AZ/Au–AK mixture responding to physiological acidity with simultaneous enhancement of both MR and SERS signals. a) T1 relaxivity of the nanoprobe mixture or Gd3+–DTPA as function of the incubation time at pH 7.4, 6.5, and 5.5. The relaxivity was measured in a 7.0 T magnet. b) Time-dependent Raman spectra of the mixture (0.75 × 10−9 m for each nanoprobe) measured at pH 7.4, 6.5, and 5.5.

Co

mm

un

iCati

on


and 9.8% were observed at pH 7.4 and 6.5, respectively. After 8 h incubation, the r1p value reached 5.6 mM−1 s−1 at both pH 5.5 and 6.5, which were significantly higher than the value of 5.1 mM−1 s−1 observed under pH 7.4. In contrast, the r1p of Gd–DTPA (3.7 mM−1 s−1) remained constant regardless of the incubation time or environmental pH. The r1p of Gd3+ chela-tors depends on a complex interplay between the structural, dynamic, and electronic properties of the Gd3+ center.[32] The most convenient way to increase r1p is to conjugate the Gd3+ chelators with slowly moving macromolecules, such as BSA, by lengthening the molecular rotational correlation time (τR).[33] The increased r1p value after acid-triggered assembly could be interpreted as the prolongation of the τR value of the Gd3+ che-lators labeled on the AuNS aggregates.

Figure 3b shows the SERRS signal enhancement of the Au–AK/Au–AZ mixture as functions of pH and incubation time. While the SERRS signals remained at the background level regardless of incubation time at pH 7.4, a fingerprint SERRS signal was observed clearly at 30 min postincubation at pH 5.5 or 2 h postincubation at pH 6.5. Quantification of the char-acteristic twin peaks with Raman shifts of 509 and 541 cm−1 showed that the SERRS signal increased 17.3- and 11.0-fold after 2 h incubation at pH 5.5 and 6.5, respectively. Maximal 19.7- and 19.2-fold SERRS signal enhancements were recorded after 8 h incubation at pH 6.5 and 5.5. “Hotspots” were defined as the nanogaps between the two or more adjacent nano-particles, which remarkably increased the SERRS signal by

significantly enhancing the local electromagnetic field. Using the Monte Carlo method,[34] maxima of 461 AuNSs and 590 hotspots were calculated in the 3D spherical aggregates with a diameter of 200 nm when the average interparticle distance was set to 2 nm (Figure S5, Supporting Information). As the SERRS effect was proportional to the number of interparticle hotspots, the generation of spherical aggregates with well-organized 3D geometry was beneficial to maximize the “hotspot” number and SERRS effect.

In vivo MRI studies were conducted on nude mice bearing orthotopic U87 glioblastoma xenografts before and after i.v. injection of the Au–AK/Au–AZ mixture (1:1 molar ratio, totaling 0.02 mmol kg−1 [Gd3+]) in PBS solution. T2-weighted MRI located the glioma position by highlighting the area with high water concentration (Figure 4a). T1-weighted (T1W) MRI offered nanoprobe-induced MR signal enhancement in the brain at selected time points postinjection (PI) of the Au–AZ/Au–AK mixture. While the contrast between the tumor and surrounding normal brain tissues was negligible before the probe injection, the tumor margin was clearly visualized with remarkable MR signal enhancement at 30 min PI of the nanoprobes. In contrast to the complete diminishment of the intratumoral T1W–MR signal in the first 1 h postadministra-tion of Gd3+–DTPA (Figure S6, Supporting Information), the tumor boundary was presented clearly even at 24 h PI of the nanoprobes. Importantly, the tumor margin delineated by the Au–AZ/Au–AK mixture correlated well with that defined

Adv. Mater. 2017, 1603917


Figure 4. Au–AZ/Au–AK mixture preoperatively delineating orthotopic glioma margin. a) Representative MR images of mouse brains bearing glioma xenograft before the injection and at selected times PI of the pH-responsive or inert nanoprobe pair (1:1 molar ratio, totaling 0.02 mmol kg−1 [Gd3+]) via i.v. injection. The arrows denote the tumors. b) Histological H&E image verified the tumor margin delineated by MRI. Scale bar: 2.0 mm. c) Representa-tive color-coded ΔT1 maps indicated T1 value changes in the brain pixel by pixel at 24 h PI of the nanoprobe pair. d,e) Average T1 values of the tumor and contralateral normal brain tissue at 24 h PI of the pH inert (d) or pH-responsive (e) nanoprobe pair. Data are the mean ± SD. PI, post injection.

Co

mm

un

iCatio

n

© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (7 of 9) 1603917Adv. Mater. 2017, 1603917


by the histological H&E image of an identical mouse brain (Figure 4b). Even though T1W–MR signal enhancement was also observed in tumor region after i.v. injection of the pH-inert AuC–AZ/AuC–AK mixture, its T/B ratio was significantly lower than that of the Au–AZ/Au–AK mixture. Figure 4c shows the ΔT1 map images that quantified the variation of the T1 values pixel by pixel before and at 24 h PI of the pH-responsive or con-trol nanoprobes. While the value in tumor region decreased by 28.2% to 2300 ms at 24 h PI of the Au–AZ/Au–AK mixture, the average T1 value was reduced by 14.7% to 2910 ms after admin-istration of the control AuC–AZ/AuC–AK mixture (Figure 4d). Weak MR signal intensities were observed in the contralateral hemisphere after injection of either the pH responsive (5.1%) or inert (5.5%) nanoprobe mixtures, which supported their rapid clearance from normal brain tissues. Notably, the average T1 value in the tumor was only reduced by 13.3% to 2937 ms at 24 h PI of the Au–AZ/Au–AK mixture when LRP1 was pre-saturated by free angiopep2 peptides (100 mg kg−1, Figure S7, Supporting Information) as a receptor binding competitor. The above study verified that the nanoprobes crossed the BBB via LRP1-mediated transcytosis. The capability of Au–AZ/Au–AK mixture to persistently visualize brain-tumor margins with high T/B ratio could be interpreted by: i) the enhanced r1p value of the conjugated Gd3+ chelators after nanoprobe assembly, ii) the increased local Gd3+ chelator concentration due to the retention

of the AuNS aggregates in the tumor interstitium, and iii) the rapid clearance of the intact nanoprobes from the normal brain tissues via the LRP-mediated bidirectional BBB traverse.

Figure 5a presents the diagrams describing the procedure of SERRS-guided glioma resection using a handheld Raman detector after intravenous administration of the Au–AZ/Au–AK mixture. Due to its high portability and maneuverability, the handheld Raman scanner can monitor the tumor resec-tion in real time by overcoming the enveloping tissues that would obstruct the visual field.[35] The 785 nm laser not only minimizes autofluorescence and phototoxicity but also helps to detect tumor foci hidden below the normal brain tissue. As illustrated in Figure 5b, SERRS-guided surgery was conducted on an isolated mouse brain implanted with a U87MG glio-blastoma xenograft at 24 h PI of the Au–AZ/Au–AK mixture. To establish a standard procedure of the tumor resection, the glioma location delineated by preoperative MRI was partitioned by a 5 × 5 square grid (3 mm × 3 mm) that covered the whole tumor bed. The average SERRS intensity in each square was determined by quantifying the characteristic twin peaks. While the square located in the tumor core offered the highest SERRS intensity of approximately 600 (AU), the values in the tumor margin area were measured to be in the range of 70–600 (AU). With the progress of excision guided by the handheld Raman detector, the SERRS intensities gradually reduced, and the brain

Figure 5. Au–AZ/Au–AK mixture intraoperatively guiding glioma resection using a handheld Raman detector. a) Cartoons illustrating the procedure of Raman-spectroscopy-guided glioma resection. b) Photographs of the excised mouse brain bearing U87MG glioma xenografts and Raman spectra at the tumor site during the image-guided tumor resection. Surgery was conducted at 24 h PI of the Au–AZ/Au–AK mixture (1:1 molar ratio, totaling 600 pmol kg−1). The intensities of the characteristic Raman twin peak were quantified in a 5 × 5 grid covering the tumor cutting bed. c) Histological H&E and Raman spectroscopic images in a representative glioma margin area at 24 h PI of the Au–AZ/Au–AK mixture. The dashed line denotes the tumor margin. Scale bar: 10 µm. d) TEM images of the glioma invasive margin and contralateral normal brain at 24 h PI of Au–AZ/Au–AK. Scale bar: 0.5 µm. The yellow arrows denote the aggregated AuNSs with diameters above 60 nm. The red arrows denote the monodisperse AuNSs. e) Diameter distribution of the nanoparticles in glioma and normal brain tissue at 24 h PI. f) Average density of individual AuNSs at the tumor and normal brain tissue area. Data are the mean ± SD. *p < 0.05 (Mann–Whitney U-test). PI, post injection.

Co

mm

un

iCati

on

© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimwileyonlinelibrary.com1603917 (8 of 9) Adv. Mater. 2017, 1603917


tumor was believed to be excised completely when the charac-teristic SERRS peaks in all the squares could not be detected. The full excision of the brain tumor was verified by H&E histo-logical staining, and no residual tumor foci were found in the operating bed (Figure S8, Supporting Information). Meanwhile, all the excised tissues were verified to be malignant, which indi-cated the minimized injury to normal brain tissues during the surgery.

Figure 5c shows the Raman spectroscopic and H&E histo-logical images of a representative brain-tumor margin area at 24 h PI of the Au–AZ/Au–AK mixture. Impressively, strong but heterogeneous SERRS signals were observed in the tumor site but were barely detected in adjacent normal brain tissues. The tumor boundary delineated by the Raman microscopy image correlated well with that defined by H&E staining. In contrast, SERRS signals in tumor site were reduced remarkably at 24 h PI of the pH-inert AuC–AZ/AuC–AK mixture (Figure S9, Sup-porting Information). TEM images of the brain-tumor section displayed both AuNS aggregates with diameters in the range of 40–90 nm and monodisperse AuNSs with diameters of 20 nm (Figure 5d). The average diameter of the AuNSs in the brain tumor was measured as 67 nm, which was significantly larger than that (24 nm) of AuNSs located in normal brain tissues (Figure 5e), thus verifying the specific nanoprobe assembly in tumor sites. Notably, the average density of individual AuNSs in the tumor was measured as 166 pieces µm−2, which was 520 times higher than that of the AuNSs in normal brain tissue (Figure 5f). Zlokovic and co-workers showed that the expression of LRP1 in the abluminal side of the human brain capillaries and that this receptor actively regulated Aβ transcytosis from the brain into the blood site.[36] Additionally, the upregulation of LRP1 in transgenic Alzheimer’s disease mouse models remark-ably decreased Aβ deposition in the aging stage.[37] In contrast to the LRP1-mediated bidirectional BBB transport of the nano-probes in normal brain tissues, the acid-triggered decoating and self-aggregation in tumors prevented the delivery of the nano-probes from the brain back to the blood, which may explain the significantly higher AuNS concentration in the tumor than that in the normal brain tissue. Immunohistochemistry images of the brain section demonstrated the remarkably higher LRP1 expression level in the brain tumor compared to that of sur-rounding normal tissue (Figure S10, Supporting Information), which indicated its active role in mediating the intracerebral delivery of the nanoprobes.

Gold nanoparticles with low toxicity, tunable size, and well-established surface chemistry have been explored as drug delivery systems.[38] The intracellular uptake is thought to be a predominant reason leading to the cytotoxicity of AuNP. Jiang and co-workers found that AuNP aggregates larger than the receptor-mediated endocytosis threshold (50 nm) preferred to nonspecifically attach to the cancer cell surface without obvious cytotoxicity.[39] Albanese and Chan showed that the viabilities of different cell lines were barely affected in the presence of AuNS aggregates with a diameter of 98 nm.[40] de Jong and co-workers reported no obvious inflammatory response in rat lungs after treatment with AuNS agglomerates.[41] Considering the attenu-ated cytotoxicity of the bulky AuNS aggregates and the rapid excretion of the intact nanoprobes, the potential side effects induced by the nanoprobes should be tolerable.

Here, we have developed two pH-responsive nanoprobes to guide brain-tumor surgery via the MR/SERRS signals activated upon their self-assembly in acidic tumor extracellular fluid. The novel features of the pH-responsive nanoprobes include: i) simultaneous activation of MR and SERRS signals for pre-operative and intraoperative imaging, respectively, which helps overcome the image distortion caused by brain shifts during surgery, ii) demarcation of the tumor invasive margin with high sensitivity and durability through the retention of the nanoprobe aggregates in the tumor site and the clearance of the intact nanoprobes in normal brain tissue, iii) brain-tumor visualization regardless of their genotypes or phenotypes since extracellular acidification is a hallmark of all solid tumors, and iv) low systemic toxicity due to their minimized uptake in normal brain tissues and rapid excretion rate. As far as we are aware, this pair of nanoprobes is the first example of a system that guides brain-tumor resection by sensing acidic tumor microenvironments. They are promising candidates to improve the outcome of brain-tumor surgery and accelerate the clinical translation of AuNS-based imaging probes.

Supporting InformationSupporting Information is available from the Wiley Online Library or from the author.

AcknowledgementsX.G. and Q.Y. contributed equally to this work. This work was supported by the National Basic Research Program of China (973 Program, 2013CB932500, and 2015CB755500), the National Natural Science Foundation of China (Nos. 81371624, 81571741, 81572483, and 81611130223), the Key Basic Research Program of Shanghai Science and Technology Committee (16JC1420100), and the Shanghai Foundation for Development of Science and Technology (No. 15140901300). C.L. thanks the helpful discussion with Prof. Ruimin Huang. All animal experiments were carried out in accordance with guidelines approved by the ethics committee of Fudan University (Shanghai, China).

Received: July 24, 2016Revised: February 3, 2017

Published online:

[1] W. Stummer, T. Meinel, C. Ewelt, P. Martus, O. Jakobs, J. Felsberg, G. Reifenberger, J. Neurooncol. 2012, 108, 89.

[2] S. A. Toms, W.-C. Lin, R. J. Weil, M. D. Johnson, E. D. Jansen, A. Mahadevan-Jansen, Neurosurgery 2005, 57, 382.

[3] N. Sanai, M. S. Berger, Neurosurgery 2008, 62, 753.[4] S. Dimou, R. Battisti, D. Hermens, J. Lagopoulos, Neurosurg. Rev.

2013, 36, 205.[5] N. Foroglou, A. Zamani, P. Black, Br. J. Neurosurg. 2009, 23, 14.[6] J.-C. Tonn, W. Stummer, Clin. Neurosurg. 2008, 55, 20.[7] Q. T. Nguyen, R. Y. Tsien, Nat. Rev. Cancer 2013, 13, 653.[8] S. Harmsen, R. Huang, M. A. Wall, H. Karabeber, J. M. Samii,

M. Spaliviero, J. R. White, S. Monette, R. O’Connor, K. L. Pitter, Sci. Transl. Med. 2015, 7, 271ra7.

[9] a) K. Y. Ju, S. Lee, J. Pyo, J. Choo, J. K. Lee, Small 2015, 11, 84; b) M. V. Yigit, L. Zhu, M. A. Ifediba, Y. Zhang, K. Carr, A. Moore, Z. Medarova, ACS Nano 2011, 5, 1056.

Co

mm

un

iCatio

n

© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (9 of 9) 1603917Adv. Mater. 2017, 1603917


[10] K. Saha, S. S. Agasti, C. Kim, X. Li, V. M. Rotello, Chem. Rev. 2012, 112, 2739.

[11] M. F. Kircher, A. De La Zerda, J. V. Jokerst, C. L. Zavaleta, P. J. Kempen, E. Mittra, K. Pitter, R. Huang, C. Campos, F. Habte, R. Sinclair, C. W. Brennan, I. K. Mellinghoff, E. C. Holland, S. S. Gambhir, Nat. Med. 2012, 18, 829.

[12] R. Huang, S. Harmsen, J. M. Samii, H. Karabeber, K. L. Pitter, E. C. Holland, M. F. Kircher, Theranostics 2016, 6, 1075.

[13] a) X. Gao, C. Li, Small 2014, 10, 426; b) X. Gao, J. Qian, S. Zheng, Y. Changyi, J. Zhang, S. Ju, J. Zhu, C. Li, ACS Nano 2014, 8, 3678.

[14] G. Braun, I. Pavel, A. R. Morrill, D. S. Seferos, G. C. Bazan, N. O. Reich, M. Moskovits, J. Am. Chem. Soc. 2007, 129, 7760.

[15] R. J. Holbrook, N. Rammohan, M. W. Rotz, K. W. MacRenaris, A. T. Preslar, T. J. Meade, Nano Lett. 2016, 16, 3202.

[16] B. A. Webb, M. Chimenti, M. P. Jacobson, D. L. Barber, Nat. Rev. Cancer 2011, 11, 671.

[17] R. J. Harris, T. F. Cloughesy, L. M. Liau, R. M. Prins, J. P. Antonios, D. Li, W. H. Yong, W. B. Pope, A. Lai, P. L. Nghiemphu, B. M. Ellingson, Neuro-Oncology 2015, 17, 1514.

[18] W. M. Pardridge, Drug Discovery Today 2007, 12, 54.[19] S. Cha, AJNR Am. J. Neuroradiol. 2006, 27, 475.[20] H.-L. Liu, M.-Y. Hua, P.-Y. Chen, P.-C. Chu, C.-H. Pan, H.-W. Yang,

C.-Y. Huang, J.-J. Wang, T.-C. Yen, K.-C. Wei, Radiology 2010, 255, 415.[21] C. Ewelt, F. W. Floeth, J. Felsberg, H. J. Steiger, M. Sabel,

K.-J. Langen, G. Stoffels, W. Stummer, Clin. Neurol. Neurosurg. 2011, 113, 541.

[22] K. Kingwell, Nat. Rev. Drug Discovery 2016, 15, 84.[23] L. Maletínská, E. A. Blakely, K. A. Bjornstad, D. F. Deen, L. J. Knoff,

T. M. Forte, Cancer Res. 2000, 60, 2300.[24] a) T. B. Neves, G. F. Andrade, J. Spectrosc. 2015, 2015, 1; b) M. Jiang,

Z. Qian, X. Zhou, X. Xin, J. Wu, C. Chen, G. Zhang, G. Xu, Y. Cheng, Phys. Chem. Chem. Phys. 2015, 17, 21158.

[25] R. A. Sperling, P. R. Gil, F. Zhang, M. Zanella, W. J. Parak, Chem. Soc. Rev. 2008, 37, 1896.

[26] S. Jung, J. Nam, S. Hwang, J. Park, J. Hur, K. Im, N. Park, S. Kim, Anal. Chem. 2013, 85, 7674.

[27] A. Sánchez-Iglesias, M. Grzelczak, T. Altantzis, B. Goris, J. Perez-Juste, S. Bals, G. Van Tendeloo, S. H. Donaldson Jr, B. F. Chmelka, J. N. Israelachvili, ACS Nano 2012, 6, 11059.

[28] I. Hussain, Z. Wang, A. I. Cooper, M. Brust, Langmuir 2006, 22, 2938.

[29] D. Van Haute, J. M. Longmate, J. M. Berlin, Adv. Mater. 2015, 27, 5158.

[30] C. D. Grant, A. M. Schwartzberg, T. J. Norman, J. Z. Zhang, J. Am. Chem. Soc. 2003, 125, 549.

[31] R. Weissleder, Nat. Biotechnol. 2001, 19, 316.[32] P. Verwilst, S. Park, B. Yoon, J. S. Kim, Chem. Soc. Rev. 2015, 44,

1791.[33] P. Caravan, Chem. Soc. Rev. 2006, 35, 512.[34] H. Liu, Z. Yang, L. Meng, Y. Sun, J. Wang, L. Yang, J. Liu, Z. Tian,

J. Am. Chem. Soc. 2014, 136, 5332.[35] M. Jermyn, K. Mok, J. Mercier, J. Desroches, J. Pichette,

K. Saint-Arnaud, L. Bernstein, M.-C. Guiot, K. Petrecca, F. Leblond, Sci. Transl. Med. 2015, 7, 274ra19.

[36] Z. Zhao, A. P. Sagare, Q. Ma, M. R. Halliday, P. Kong, K. Kisler, E. A. Winkler, A. Ramanathan, T. Kanekiyo, G. Bu, Nat. Neurosci. 2015, 18, 978.

[37] J. M. Castellano, R. Deane, A. J. Gottesdiener, P. B. Verghese, F. R. Stewart, T. West, A. C. Paoletti, T. R. Kasper, R. B. DeMattos, B. V. Zlokovic, Proc. Natl. Acad. Sci. USA 2012, 109, 15502.

[38] Y. Cheng, J. D. Meyers, R. S. Agnes, T. L. Doane, M. E. Kenney, A. M. Broome, C. Burda, J. P. Basilion, Small 2011, 7, 2301.

[39] W. Cui, J. Li, Y. Zhang, H. Rong, W. Lu, L. Jiang, Nanomedicine 2012, 8, 46.

[40] A. Albanese, W. C. Chan, ACS Nano 2011, 5, 5478.[41] I. Gosens, J. A. Post, L. J. de la Fonteyne, E. H. Jansen, J. W. Geus,

F. R. Cassee, W. H. de Jong, Part. Fibre Toxicol. 2010, 7, 1.

Documents

Guiding Brain‐Tumor Surgery via Blood–Brain‐Barrier ...€¦ · due to their infiltrated nature. Here, a pair of gold nanoprobes that enter a brain tumor by crossing the blood–brain