Polymalic acid chlorotoxin nanoconjugate for near-infrared fluorescence guided resection of glioblastoma multiforme

Abstract

Maximal surgical resection of glioma remains the single most effective treatment. Tools to guide the resection while avoiding removal of normal brain tissues can aid surgeons in achieving optimal results. One strategy to achieve this goal is to rely upon interoperative fluorescence staining of tumor cells in vivo, that can be visualized by the surgeon during resection. Towards this goal we have designed a biodegradable fluorescent mini nano imaging agent (NIA) with high specificity for U87MG glioma cells and previously unmet high light emission. The NIA is the conjugate of polymalic acid (PMLA) with chlorotoxin for tumor targeting, indocyanine green (ICG) for NIR fluorescence and the tri-leucin peptide as fluorescence enhancer. PMLA as a multivalent platform carries several molecules of ICG and the other ligands. The NIA recognizes multiple sites on glioma cell surface, demonstrated by the effects of single and combined competitors. Systemic IV injection into xenogeneic mouse model carrying human U87MG glioblastoma indicated vivid tumor cell binding and internalization of NIA resulting in intensive and long-lasting tumor fluorescence. The NIA is shown to greatly improve tumor removal supporting its utility in clinical applications.

Graphical Abstract

1. Introduction

Maximal surgical resection remains a major treatment strategy for gliomas [1–3], with several studies demonstrating that extent of tumor removal was important for patient survival [4–8]. To precisely differentiate tumor margins from indispensable regions of normal brain [9], several strategies have been employed especially to limit the damage to surrounding brain such as intraoperative MRI and 5-amino levulenic acid (5-ALA) generated fluorescence of tumor cells [10, 11]. In the presence of 5-ALA, tumor specific mitochondrial protoporphyrin IX (pP IX) is spontaneously synthesized, which is detected by its fluorescence emission under ultraviolet light of 485 nm wavelength [12]. Despite of the many drawbacks of pP IX as a fluorescent marker such as absence of specificity, “bleeding” of fluorescence into normal tissue due to cell lysis, poor tissue penetration, poor signal to noise ratio, and side effects from 5-ALA administration [13, 14], this method has been demonstrated to improve the extent of resection and subsequent progression-free survival [12]. Recent interest has focused on the use of near infrared region (NIR) owing to greater spatial resolution than UV light, and a narrow emission spectrum notably of Indocyanine Green (ICG) permitting filter optimization for fluorescence detection [15, 16]. Additionally, there is little absorption by hemoglobin and minimal light scattering at these wavelengths, so that intervening normal tissue does not attenuate the signal by the extent seen for light of lower wavelengths [17]. Finally, tissue emits very low auto-fluorescence at the NIR emission wavelength that enables an excellent signal to noise ratio and a sharp definition of tumor boundaries.

To target ICG or other fluorophores into tumor, Chlorotoxin (CTX) has been introduced, which is a 36-amino acid scorpion toxin peptide. It consists of a small three-stranded antiparallel beta-sheet with eight cysteines forming four disulfide bonds packed against an alpha-helix [18]. At pH 7.0, CTX is highly positively charged with a tendency for cell penetrating [19]. It has affinity for selective surface binding to glioma cells and to cells of other malignant tissues [19–24], but with minimally affecting non-cancerous tissues [21]. Once bound to the cell surface, CTX is internalized. These unique properties, along with its small size and lack of immunogenicity, make CTX an attractive ligand for targeted cancer therapies. CTX conjugated to the fluorophore Cy5.5 [25], avidly bound gliomasin animal models and was, therefore, named “Tumor Paint”. CTX chemically attached to to ICG (BLZ-100) was approved by FDA for NIR imaging. The new agent was successfully introduced with a handy imaging device SIRIS (Synchronized near-InfraRed Imaging System) [15] to serve the interoperative tumor resection. However, BLZ-100 owing to the single conjugation of ICG and CTX provides less fluorescence intensity when the number of tumor receptors declines from high to low tumor grade [21, 26]. Since the edges of gliomas often represent lower grade tumors, this could be a limiting factor. A clear advantage of a nanoconjugate would be the simultaneous delivery of several ICG molecules per carrier, thereby increasing fluorescent signal and improving tumor detection [22, 23, 27]. Such carrier would permit less dosing and still achieve strong imaging, while reducing potential toxicity.

Our novel approach takes advantage of the previously demonstrated tumor-specific binding of CTX, adds highly improved NIR fluorescence for deep tissue imaging, and provides the efficiency of poly(β-L-malic acid) (PMLA) as a platform for simultaneously binding one or several ligands together under a single structural umbrella. We have previously engaged the PMLA platform in targeted drug delivery to treat breast and brain tumor [28] and provide a modality for imaging for “virtual biopsy” of primary and metastatic brain tumors [29]. Based on our experience, we now developed a novel imaging agent for high precision fluorescence-guided tumor resection. We achieved a NIR fluorescent multifunctional imaging agent for interoperative excellent visualization of glioblastoma removal and the possibility of chemotherapy at the same time. The NIA relies upon CTX for tumor-specific binding, and ICG in combination with tri-leucine peptide (LLL) for high intensity NIR fluorescence (Fig. 1). The novelty of engaging several functional groups was met by investigating the NIA binding to glioblastoma cells and its cellular distribution.

Figure 1. Chemical structure and functional composition of the imaging agent P/LLL/CTX/ICG.

(A) Chemical representation of nano imaging agent (NIA) showing functional groups. Per cent indicates fraction of platform malyl residues equivalent to loading 8 molecules of CTX, 10 molecules of ICG and 207 molecules of LLL on an average per chain. Glioblastoma-specific targeting ligand CTX is covalently attached via PEG linker to facilitate binding with cell surface receptors. Chemical structures of Chlorotoxin (CTX) and Indocyanine Green (ICG) are depicted in Fig. S1. (B) Cartoon depicting the assumed functional organization of the imaging agent, represented by separation of fluorophores with hydrophobic side chains of tri-leucine peptide. (C) Intra venous (IV) application of the imaging agent in preparation of fluorescence assisted resection of glioblastoma tumor on preclinical mice.

2. Materials and methods:

2.1. Materials

2.1.1. Reagents

Highly purified PMLA was obtained from the culture supernatant of Physarum polycephalum (>95% purity, Mw 60 kDa, polydispersity P = 1.3) [28]. Tri-leucine, H-Leu-Leu-Leu-OH (LLL), and CTX were purchased form Bachem Americas Inc. (Torrance, CA, USA). Maleimide-PEG20000-SCM (MAL-PEG2000-SCM) was obtained from Laysan Bio Inc. (Arab, Al, USA). PDP was synthesized as described [28]. ICG-MAL and ICG-NHS were obtained from Intrace Medical (Lausanne, Switzerland). Rhodamine Red® C2-Maleimide and fluorescein-5-maleimide was purchased from Thermo Fisher Scientific (Waltham, MA, USA). PD-10 columns were obtained from GE Healthcare (Uppsala, Sweden). Anti-Von Willebrand Factor antibody ab11713 was purchased from Abcam (Cambridge, MA, USA). Unless otherwise indicated, all chemicals and solvents of highest purity were purchased from Sigma-Aldrich (St. Louis, MO, USA).

2.1.2. Cell lines and culture conditions

Primary glioblastoma U87MG cell line was a gift from Drs. Webster Cavenee and Frank Furnari (UC San Diego), and cultured in minimum essential medium (MEM) supplemented with 10% fetal bovine serum (FBS), 1% MEM non-essential amino acids, 1 mM sodium pyruvate and 2 mM L-glutamine. MDA-MB-468 was cultured in Leibowitz’s L-15 medium with 10% FBS at 37 °C without CO₂.

2.2. Methods

2.2.1. Overview of chemical synthesis and characterization

Glioma targeting NIAs P/CTX(1.5%)/ICG(2%), P/LLL(40%)/CTX(1.5%)/ICG(2%), and non-targeting controls P/ICG(2%), P/LLL(40%)/ICG(2%), schematically presented in Fig. S1, were synthesized using PMLA as a nano-platform. The pendant carboxylic groups of the polymer were chemically activated by N-hydroxysuccinimide (NHS) ester (Fig. 2A, D). Intermediates of functional ligands were prepared by covalently attaching CTX-NH₂ to the linker MAL-PEG2000-NHS forming CTX-PEG2000-MAL using borate buffer pH 8.0 in dimethyl formamide (DMF, 1:2 volume:volume), (Fig. S2c). Preconjugates (Fig. 2A, D) were prepared by attaching 2-mercapto-1-ethylamine (MEA) to chemically activated PMLA backbone in DMF and following the reaction by ninhydrin reaction and thin-layer chromatography. Thiol groups were reacted with maleimide groups of CTX-PEG2000-MAL and ICG-MAL to form stable covalent thioethers. CTX (Fig. 2C, F) was attached at 1.5% of total malyl residues corresponding to 7–8 molecules of CTX per polymer chain. ICG was attached at 2% corresponding to 10 molecules per polymer chain on an average Fig. 2B, C, E, F). Excess NHS-ester on preconjugate was hydrolyzed in 100 mM sodium phosphate (pH 5.5) and the remainder of −SH groups were masked by reaction with pyridyldithiopropionate (PDP). Formation of the synthesized agents was monitored by sec-HPLC and UV absorbance and/or fluorescence intensity. Highly water-soluble agents were pure after PD-10 column (60% yield, stored at −20 °C) and had the designed composition by chemical group analysis [28] and UV spectral analysis.

Figure 2. Scheme describing the synthesis of the imaging agent and its precursors.

PMLA pendant carboxylic groups are activated by forming succinimide (NHS) ester via the NHS-cyclohexyl carbodiimide method and followed either by amidation with mercapto-1-ethylamine (MEA) forming preconjugate-1 (A) or followed by amidation with MEA and tri-leucine (LLL) forming preconjugate-2. Preconjugate-1 is conjugated with ICG-maleimide (ICG-MAL) forming the imaging agent precursor P/ICG(2%) (B), and by conjugation with ICG-MAL and CTX-PEG-MAL the imaging agent precursor P/CTX(1.5%)/ICG(2%) (C). Precursor-2 (D) conjugates ICG-MAL forming the highly fluorescent precursor P/LLL(40%)/ICG(2%) (E). Preconjugate-2 (D) conjugating ICG-MAL and CTX-PEG-MAL forming the powerful nano-imaging agent (NIA) P/LLL(40%/CTX(1.5%)/ICG(2%) (F) for tumor imaging and resection.

2.2.2. Synthesis of Preconjugates

NHS (0.43 mmol) and N,N’-dicyclohexylcarbodiimide (DCC; 0.43 mmol) dissolved in 1 ml of DMF were added consecutively to the solution of 50 mg of PMLA (0.43 mmol with regard to malyl units) dissolved in 1 ml of anhydrous acetone under vigorous stirring at room temperature (RT). After stirring at RT for 2 h to complete the activation of carboxyl groups, a mixture of LLL (0.172 mmol in DMF; 40 Mol-% with regard to malyl units) and MEA (0.043 mmol 10 Mol-% with regard to malyl units) dissolved in 0.5 ml DMF and 13.25 μl of trifluoracetic acid was added to the reaction mixture. At this stage, triethylamine (TEA, 0.215 mmol in DMF 0.25 ml; 40 Mol-% with regard to malyl units) was added dropwise over a period of 30 min and reaction mixture was stirred at RT for further 1.5 h. A solution of phosphate buffer (100 mM sodium phosphate and 150 mM NaCl, pH 6.8) was added at a ratio of 1:3 (organic solvent: buffer) and the reaction mixture was stirred at RT for 1h. After centrifugation at 1,500 × g for 10 min the clear supernatant was passed over Sephadex PD-10 columns (GE Healthcare Waltham, MA, USA) pre-equilibrated with deionized (DI) water. The product P/LLL(40%)/MEA(10%) containing fractions were collected and lyophilized. Reaction yield was 78%. Preconjugate without the addition of LLL was prepared using a similar procedure to obtain P/MEA(10%). Reaction yield 82%.

2.2.3. Synthesis of CTX-PEG2000-MAL

A solution of CTX (1mg, 0.26 μmol) dissolved in 0.2 ml of sodium borate buffer (0.15 M, 0.1 mM EDTA, pH 8.0) was added to MAL-PEG2000-NHS (1.63 mg, 0.78 μmol) in 0.2 ml of DMF. Reaction mixture was stirred at ambient temperature for 1h and passed over Sephadex PD-10 columns pre-equilibrated with DI water. The product containing fractions were collected (confirmed by HPLC), mixed together and lyophilized. Reaction yield was 82%.

2.2.4. Synthesis of P/ICG(2%)

At all times, solutions containing ICG and its derivatives were protected against light. To a solution of preconjugate at 4 mg/ml dissolved in buffer (100 mM sodium phosphate, pH 5.5), was added a solution of ICG-MAL prepared as 2 mg/ml in DMF. Reaction mixture was stirred at RT for 1h. Leftover thiol groups were blocked by the reaction with large excess of PDP. The reaction mixture was purified over PD-10 column in PBS, passed through 0.2-micron pore filters, snap-frozen and stored at −20 °C.

2.2.5. Synthesis of P/CTX(1.5%)/ICG(2%)

A solution of CTX-PEG2000-MAL, 4 mg/ml dissolved in buffer (100 mM sodium phosphate, pH 6.3) was dropwise added to 4 mg/ml of preconjugate P/MEA(10%) at RT in the same buffer. The reaction was monitored by sec-HPLC. After reaction completion (30 min), the pH of the reaction mixture was adjusted to 5.5 with addition of 1 M citrate buffer, and 2 mg/ml of ICG-MAL in DMF was added. After reaction completion, remaining free-SH groups were blocked with PDP. The obtained imaging agent P/CTX(1.5%)/ICG(2%) was purified over PD-10 column in PBS, passed through 0.2-micron pore filters, snap-frozen and stored at −20 °C.

2.2.6. Synthesis of P/LLL(40%)/ICG(2%)

To a solution of preconjugate P/LLL(40%)/MEA(10%) at 4 mg/ml dissolved in buffer (100 mM sodium phosphate, pH 5.5) was added to a solution of ICG-MAL, 2 mg/ml in DMF. Reaction mixture was stirred at RT for 1h. Leftover thiol groups were blocked by the reaction with PDP. The reaction mixture was purified over PD-10 column in PBS, passed through 0.2-micron pore filters, snap-frozen and stored at −20°C.

2.2.7. Synthesis of P/LLL(40%)/CTX(1.5%)/ICG(2%)

A solution of CTX-PEG2000-MAL, 4 mg/ml dissolved in buffer (100 mM sodium phosphate, pH 6.3) was dropwise added to 4 mg/ml of preconjugate P/LLL(40%)/MEA(10%) at RT in the same buffer. The reaction was monitored by sec-HPLC. After reaction completion (30 min), the pH of the reaction mixture was adjusted to 5.5 with 1 M citrate buffer, and 2 mg/ml of ICG-MAL in DMF was added. Leftover thiol groups were blocked by the reaction with PDP. The obtained imaging agent P/CTX(1.5%)/ICG(2%) was purified over PD-10 column in PBS, passed through 0.2-micron pore filters, snap-frozen and stored at −20°C. Synthesis of P/LLL(40%)/CTX(1.5%)/Rh(2%) was performed using similar procedure by replacing ICG-MAL with Rhodamine-MAL.

2.2.8. Synthesis of CTX-ICG conjugate

A solution of CTX (1mg, 0.26 μmol) dissolved in 0.5 ml of sodium borate buffer (0.15 M, 0.1 mM EDTA, pH 8.0) was added to ICG-NHS (0.64 mg, 0.78 μmol) in 0.2 ml of DMF. Reaction mixture was stirred at ambient temperature in dark for 1h and passed over Sephadex PD-10 columns pre-equilibrated with PBS. The obtained imaging agent CTX/ICG was purified over PD-10 column in PBS, passed through 0.2-micron pore filters, snap-frozen and stored at −20 °C.

2.2.9. Analytical methods for synthesis of PMLA conjugates

The conjugation reaction of 2-MEA with PMLA was followed by thin layer chromatography (TLC) on pre-coated silica gel 60 F254 aluminum sheets (Sigma-Aldrich, St. Louis MO, USA) and visualization of spots by UV light and by ninhydrin staining. Size exclusion chromatography was performed on an Elite LaChrom analytical system with Diode Array Detector L 2455 (Hitachi) and MW was measured using PolySep-GFC-P 4000 (300 × 7.80 mm) columns (Phenomenex, Terence, CA. USA) with PBS as a mobile phase at a flow rate of 1 ml/min and polystyrene sulfonates of known molecular weight as standards. Thiol residues attached to PMLA were assayed by the method of Ellman. Content of CTX in nanoconjugates was determined by Pierce™ BCA Protein Assay Kit (Thermo Scientific, Canoga Park, CA, USA). Known amounts of Free CTX were used as standards. Quantification of malic acid in nanoconjugates was performed by the malate dehydrogenase assay after acid hydrolysis [30]. Percentage (%) of the nanoconjugate loading with CTX and ICG was calculated by using the formula % = 100 × (μmol ligand) / (μmol malic acid).

2.2.10. Hydrodynamic diameter and zeta potential

Synthesized conjugates were characterized with respect to their size and ζ potential using Zetasizer Nano ZS90 (Malvern Instruments, United Kingdom). For the particle size measurements at 25 °C, the solutions were prep ared in PBS at a concentration of 2 mg/ml. For the measurement of the ζ potential, the concentration of the sample dissolved in 10 mM NaCl solution was 2 mg/ml, and the voltage applied was 150 mV. Data represents the mean of three measurements ± standard deviation.

2.2.11. Stability of NIA in Human Plasma

To study stability of NIA in the human plasma, NIA was additionally labelled with fluorescein and was used as reporter to measure stability in the dark using sec-HPLC equipped with fluorescence detector. We only used fluorescence data as human plasma and its constituent has very large absorbance in UV range. We expected that the relatively low stoichiometry of fluorescein conjugation and the low concentrations in plasma had only minor effects on the measurement of P/CTX/LLL stability; nevertheless, since fluorescein and ICG are known to bind to serum proteins, the results are only considered for orientation. Fluorescence intensity at 512 nm wavelength (excitation wavelength 494 nm) was recorded after sec-HPLC. NIA concentration in human plasma was 5 μM with reference to fluorescein. Aliquots were incubated at 37 °C for 0, 0.5, 1, 2, 3, 4, 6, 8, 16, 24, 48 and 96 h and diluted 10-fold in PBS before sec-HPLC analysis. Recorded spectra were normalized to the peak with the highest fluorescence intensity. The area under the NIA fluorescence intensity curve (AUC) before the reaction (time zero) was set as reference for measurement of the degree of degradation during incubation. The % decay measured by the AUC was evaluated as a function of time (Fig. S4).

2.2.12. Spectral properties of NIAs

ICG absorbs mainly between 600 nm and 900 nm and emits fluorescence between 750 nm and 950 nm. We validated the absorbance and fluorescence properties of all PMLA-ICG conjugates and compared with free ICG. Spectroscopic photometric and fluorometric measurements were carried out by four conceptually different optical instruments in order to validate the results under different instrumental settings comparable with the device that will be used in surgical tumor resection: 1. Plate reader Spectra Max M2 from Molecular Devices (Sunnyvale, CA, USA). 2. PTI QuantaMaster spectrofluorometer, Model QM4 from HORIBA (Kyoto, Japan). 3. Odyssey ® CLx Imaging System (LI-COR Biosciences, Lincoln, NE, USA). 4. A fully functional prototype of a clinical imaging system developed for NIR fluorescence image guided surgery [15]. All optical systems yielded similar results in proof of the spectral data reported here for the tested free ICG and ICG conjugates and the selection of the lead NIA. Absorbance spectrum was recorded at 600–900 nm in PBS pH 7.4 at concentrations from 10 μm to 13.7 nM. Concentration is referred to ICG (either free or conjugated to NIAs). In order to avoid misunderstandings, graphical presentations of ICG absorbance and fluorescence use intentionally pseudo and not authentic colors.

2.3. Flow cytometry

Binding studies were conducted by flow cytometry below 4 °C [31] wi thout separation of reactants from the reaction mixture [32]. The binding activities of NIAs and CTX-ICG to human glioblastoma multiforme (GBM) U87MG cells and competition with ligands representing constituents of NIA were measured using a BD LSRFortessa™ (BD Biosciences, San Jose, CA) flow cytometry system. For ICG containing conjugates we used an APC-Cy7 fluorescence detector (bandpass filter 780/60 as recommended by detector setting for ICG) and for rhodamine containing conjugates the fluorescence detector PE. 10,000 events were recorded for each sample and analyzed using FlowJo (FlowJo LLC, Ashland OR). Human U87MG cells, grown in DMEM media with 10% FBS, were dissociated by trypsin. For each reaction, 2.5 × 10⁵ cells were diluted in 0.2 ml PBS on ice. For binding studies, cells were incubated with various concentrations of ligand, NIA or CTX-ICG containing 0.05 μM – 12 μM total CTX (eBioscience™ Flow Cytometery Staining Buffer, Thermo Fisher Scientific, Waltham MA). After 30 min, the reaction mixture was filtered through a 40 μm cell strainer directly into flow cytometry tubes (5 mL polystyrene round-bottom tube 12×75 mm style). In competition experiments, the reaction mixture contained at all times the cells and varied concentrations of NIA and competitor (single competition: 5–125 μM free CTX and 5–125 μM P/LLL(40%), mixed competition: each competitor 125 μM) or PBS buffer as no-ligand-reference) was incubated for 30 min before addition of the competed ligand (NIA or CTX-ICG) and additional 30 min incubation. Mean fluorescence intensities (MFI) were calculated from the flow histograms (peak area × count⁻¹). Competition with single ligands P/LLL(40%) or CTX or with the mixture was repeated under conditions of cell centrifugation to remove competitors and NIA. Cells were resuspended in PBS and, after 30 min on ice, measured by flow cytometry. In the case of titration with NIA or CTX-ICG, apparent dissociation constants were calculated for single site binding after establishing best fits (r value <99%) of MFI against ligand concentration using GraphPad Prism 7.04 software. The single site model was validated by linear plots of (MFI) against (MFI) × (ligand concentration)⁻¹ [33, 34]

2.4. Fluorescence microscopy and immunohistochemistry

Conventional hematoxylin and eosin (H&E) staining was carried out for tumor cell staining and Von Willebrand factor (vWF) staining for vasculature. The distribution of rhodamine instead of ICG substituted NIA fluorescence followed IV injection of NIA into the tail vein of experimental mice. After replacement, the composition of NIA was P/LLL(40%)/ICG(1.5%)/Rh(2%). The replacement with rhodamine greatly improved fluorescence microscopy enhancing intensity and contrast. At controlled times after administration, mice were euthanized, brains harvested, embedded in Tissue-Tek® O.C.T (Electron Microscopy Sciences, Hatfield, PA) and frozen in dry ice. Frozen tissue blocks were sectioned at 8 μm thickness using Leica CM 3050S cryostat. Before staining, tissue sections were air dried at RT, fixed with ice-cold acetone for 10 min, and then rinsed three times with PBS. Sections were incubated in a humidified chamber with blocking buffer (4% normal goat serum, 4% normal donkey serum, 1% BSA and 0.1% Triton X-100 in PBS) for 1 h at RT to block non-specific sites. The blocked sections were incubated overnight at 4 °C with prepared fluo rescein-labeled anti-vWF antibody diluted in staining buffer, and later washed with PBS. Finally, sections were mounted with Prolonged Gold Antifade (Thermo Scientific) mounting medium containing DAPI. Images were captured using a Leica DM6000B microscope (Germany).

2.5. In vivo treatments

2.5.1. Tumor xenografts and nanodrug treatment

All experiments with animals were performed in strict accordance with the protocols approved by the Cedars-Sinai Medical Center’s Institutional Animal Care and Use Committee (IACUC). Athymic NCr-nu/nu female mice were obtained from NCI-Frederick. We used glioma model of U87MG[35], which is one the highly studied rodent model to study GBM. Human U87MG GBM cells (5 × 10⁴ cells) were stereotactically implanted into the right basal ganglia. Animals were monitored regularly for any potential neurological symptoms. On day 20–25, at which point the tumor reached an average diameter of 4–5 mm, animals were injected via tail vein with NIAs and imaged by the SIRIS for global drug uptake in tumors and vital organs.

2.5.2. Pharmacokinetics

Nude mice bearing human U87MG xenografts were injected with 150 μl solution of NIA via the tail vein at a dose of 200 nmol/kg (ICG). Mice were bled via the tail at 0, 0.083, 0.5, 1, 2, 3, 4, 6, 8, 12, 16, 24 and 48 hours. Blood was processed for serum by centrifugation (6000 rpm for 5 min). Four animals were used for each timepoint. Fixed amount of serum sample from each mouse was transferred to a 96 well plate, diluted with PBS (3×) and fluorescence intensity was recorded at 800 nm by Odyssey CLx (LI-CO Biotechnology, Lincoln, NE). Data were analyzed using GraphPad PRISM Version 7.04.

2.5.3. Biodistribution

After IV administration of NIA at 200 nmol/kg (ICG dose), animals were euthanized by cervical dislocation at 0 (control), 2, 4, 8, 12, 24 and 48 hours post injection. Major organs (brain, lung, kidney, liver, heart, spleen and skin) along with tumors were collected, weighed, transferred into 1.5 mL Eppendorf centrifuge tubes, frozen and stored at −20 °C. For analysis of biodistribution s amples were thawed, mixed with PBS at a 1:10 weight/volume ratio, and subject to ultra-sonication using Q700 Ultra-Sonicator (QSonica, LLC Newtown, CT USA). Cell extraction was carried out in three-minute cycles (0.5-minute × 6) at 5-amp amplitudes with 30 second breaks. For sonication without loss, the lid of the Eppendorf tube was perforated to allow a tight sealing for the sonicator probe intruding into the contained sample. This configuration prevented sample loss due to spilling over during sonication. To avoid heating during sonication, samples were always kept on ice. When chunks of tissue were no longer visible, 50 μL portions of the sonicated samples were transferred to 96 well plates in triplicates. Fluorescence intensities, owing to NIA concentrations were independent of tissue/tumor mass when recorded at 800 nm by Odyssey CLx and plotted as function of time elapsed from NIA injection to animal dislocation. The reading at time 0 was used to correct for background. Standard curves were prepared separately for each organ after isolating organs from PBS injected mice by adding (“spiking”) known amounts of NIA. This curve was then used for quantification of NIA in each organ at a given endpoint. Recorded fluorescence intensities were analyzed with GraphPad Prism 7.04 software.

2.5.4. Tumor visualization in real time for fluorescence guided resection

The synthesized targeted imaging agent P/LLL(40%)/CTX(1.5%)/ICG(2%) and the non-targeted control P/LLL(40%)/ICG(2%) were tested in vivo using the SIRIS imaging system. Mice carrying tumor xenografts received IV injection through a tail of either control or ICG-containing NIA (200 nmol/kg) and were then sacrificed at various timepoints. No behavioral or physical abnormalities were observed. The tumor and organs were isolated and visualized using the custom-built SIRIS [15] imaging device. An imaging device SIRIS for intra-operative detection of NIR fluorescence and resection of tumor has been built for clinical use for the resection of brain tumor [15]. SIRIS simultaneously acquires and superimposes both white light (WL) and NIR images on a high definition (HD) video monitor. This system was used to simulate the process of fluorescence guided precision resection of xenograft tumor from mice brain, usually at a distance of 30 cm from the camera. Real time video rate images of white light, NIR, and superimposed NIR on White light tumor were displayed on a monitor during tumor removal surgeries.

2.5.5. NIR fluorescence and white light guided resection

After administration of NIA at a dose of 200 nmol/kg, animals were euthanized by cervical dislocation at certain (0, 2, 4, 8, 12, 24 and 48 hours) time points. The brain was harvested intact via craniectomy and placed onto the imaging platform to visualize the tumor fluorescence. An incision was made in the right premotor cortex (where previously intracranial inoculation of tumor cells was done) with the help of surgical scalpel No.15. The tumor was found immediately under the cortex as evident by a strong fluorescence signal under NIR mode of SIRIS imaging device. In most of the cases a demarcation line has been noticed in part of the perimeter of the tumor that was followed for resection of the tumor using micro forceps and micro scissors (Aesculap). The entire resection was visualized under SIRIS imaging system, providing magnification and illumination to aid the dissection, with simultaneously checking for fluorescence intensity as well as white light in the resection cavity. Tissue with fluorescence was removed until no further fluorescence signal was detected in the tumor cavity (Fig. 7). For comparison, resection was conducted on mice injected with PBS under similar conditions except using illumination of a ceiling positioned fluorescent light source (white light).

Figure 7. SIRIS assisted real time NIR-imaging and surgical resection of GBM on mouse model.

Lead NIA (P/LLL/CTX/ICG) was IV administered via tail vein. Animals were euthanized after 4 hours. Targeted NIA selectively accumulated only in tumor area as evident by clear demarcation of tumors (white arrows) without any noticeable signal from tumor free brain tissue. Tissue highlighted by strong fluorescence was resected and histologically evaluated by H&E staining to confirm the location of tumor. ICG excitation wavelength was 785 nm aided by use of a cleanup filter to allow for maximum excitation efficiency. In conjunction, a notch filter in front of the camera removes the excitation light from the image thus capturing only the fluorescence emission. This configuration allows imaging system to image fluorescence with maximum efficiency at high signal-to-noise ratio.

2.5.6. Evaluation of resection efficiency

After resection, brains were embedded in Tissue-Tek® O.C.T. Compound (Electron Microscopy Sciences, Hatfield, PA) and frozen in dry ice. Frozen tissue blocks were sectioned at 8 μm thickness using Leica CM 3050S cryostat in three different areas (top, middle and deep). Sections were stained for H&E and bright field images were recorded on Keyence BZ-X700 Series microscope (Keyence Corporation of America, Itasca IL). In each section, non-resected tumor portion was demarked and region of interest (ROI) were drawn as shown in Fig. 8 A (bottom panel) covering each residual tumor area. A separate ROI was drawn covering entire tumor area and set as 100%. Area were measured using ImageJ software. The efficiency of resection in in tumor sections referring to top, middle and deep tumor regions were calculated as (%)Resection = 100 × (Total tumor area − sum of non-resected area) × (Total tumor area)⁻¹. Similarly, resected tumor tissue was analyzed to detect the presence of any healthy tissue in the resected tissue (Fig. 8 B and Fig. S9).

Figure 8. U87MG GBM resection guided by NIA imaging.

Representative examples are shown for resection under NIR light and white light. (A) Panel shows H&E stained sections of brain after resection. The tumor had been removed under NIR light at 4h following injection of lead NIA (200 nmol/kg of ICG equivalent), P/LLL(40%)/CTX(1.5%)/ICG(2%). Resection was smooth revealing clean separation of the tumor from the brain. Resection was incomplete indicated by H & E staining. The brains resected under NIR (n=3) were compared with examples under white light (in the absence of NIA) (n=5) by calculating the areas of interest for non-resected tumor relative to the total tumor area before the resection. (B) H&E staining of resected tumors revealed healthy brain tissue, unintentionally resected together with tumor under the condition of NIR or white light. Black arrows indicate the healthy tissues. (C) Per cent of residual over total tumor are compared for NIR and white light resections. Measurements included sections in the top, the middle and the deep regions of the tumors (Fig. S8). No significance was noticed in resection efficiency under NIR or white light at the top. Analysis of middle and deep sections of resection cavity showed significant difference with p=0.027 and 0.011 respectively. (D) For comparison, we also measured the per cent fractions of co-resected healthy tissue in resected tumor. Significantly less healthy tissue was co-resected in case of NIR compared with white light (p=0.04). Student’s t-test was used for comparisons, with significance level set to *p < 0.05. The results are summarized for the middle region of tumors. In several cases, the location of the co-resected healthy brain was embedded within the tumor (Fig. S9). Notably, since error bars in panel C are similar in size and more or less independent of tumor position under NIR light, the precision of resection refers mainly to an optimal recognition of tumor fragments.

3. Results

3.1. Syntheses and characterization

Syntheses were highly reproducible with yields for preconjugate 70–82% (reference to PMLA), and final conjugates >85% (reference to preconjugate). Chemical analysis of attached ligands [28] met an accuracy of ± 10% in agreement with design. All imaging agents were pure by sec-HPLC (Fig. S3) and dynamic light scattering. The hydrodynamic diameter and zeta potential by dynamic light scattering (Table 1, SD ± 10%) meet the “Mini-Nano Drug” classification [36].

Table 1.

Summary of nanoconjugates, their abbreviation and physicochemical characterization

Nano-imaging agents and intermediates	Hydrodynamic diametera (nm)	Zeta potentialb (eV)
PMLA	6.1 (± 0.4)	−22.8 (± 1.3)
PMLA/LLL(40%)	6.5 (± 0.6)	−29 (± 0.3)
PMLA/ICG(2%)c	8.5 (± 0.8)	−33.1 (± 1.2)
PMLA/CTX(1.5%)/ICG(2%)	9.8 (± 1.1)	−21.2 (± 0.7)
PMLA/LLL(40%)/ICG(2%)	8.2 (±1.4)	−24.8 (± 1.2)
PMLA/LLL(40%)/CTX(1.5%)/ICG(2%)	11.82 ((±1.6)	−20.47 (±1.8)

^aHydrodynamic diameter by number distribution at 25°C measured in PBS at a concentration of 2 mg/ml, calculated from DLS data by the Malvern Zetasizer software (Malvern Instruments, Malvern, UK), which assumes spherical shapes of particles.

^bzeta potential at 25 °C in aqueous solution of 10 mM NaCl at 150 mV.

^ccomposition of nanoconjugates; percentage refers to total number (100%) of pendant carboxyl groups in unsubstituted PMLA.

3.2. Stability of NIAs

During storage in the dark (−20 °C, PBS, pH. 7.4) the nanoconjugates did not show any loss of ICG fluorescence intensity in agreement with previous observations for other nano constructs [37]. To study the stability in human plasma, NIAs were additionally labelled with 2% fluorescein. In human plasma, the fluorescein labelled NIA transformed from single peak in PBS buffer (pH 7.4) to double peak indicated by sec-HPLC with subsequent decay by degradation over time (Fig. S4a). The decay followed approximately first order kinetics with t_1/2 ≈ 10 h while resulting in various degradation products at 37 °C (Fig. S4b).

3.3. NIR absorbance of free and conjugated ICG

Absorbance spectra in PBS pH 7.4, 22 °C (Fig. S5) p eaked for free ICG at 695 nm wavelength (maximum for 100 μM/minimum for 3 μM concentration) and 780 nm (minimum for 100 μM/maximum for 3 μM). A similar dependence has been reported recently [38]. After conjugation with PMLA, maxima located at 725 nm & 790 nm wavelength (P/CTX/ICG), and at 730 nm and 795 nM (P/LLL/CTX/ICG) for both 3 μM and 100 μM ICG.

3.4. Selection of lead NIA

The degree of ligand loading was considered on the following arguments: (1) Tri-leucine (40%) was optimal with regard to membrane destabilization [39, 40], without noticeable effect on nanoconjugate solubility. (2) CTX (1.5%) was in an optimal range favoring multivalence, not exhibiting unspecific binding and only modest degree of inhibition of ICG fluorescence emission. (3) ICG (2%) provided excellent fluorescence intensity without self-aggregation, precipitation and unspecific absorbance to membranes and lipid protein binding in the concentrations used for imaging. The finding that the absorbance of P/LLL(40%)/CTX(1.5%)/ICG(2%), is shifted towards the NIR-region, promoting optical transmission and high fluorescence intensity for in vivo imaging, prompted us to choose the nanoconjugate as the lead NIA in guided tumor resection. The biodegradability and non-toxicity of PMLA, and the multiplicity of pendant carboxylic groups for drug attachment is optimal for combining imaging guided resection with CTX targeted pharmacological treatment. Our investigation revealed that ICG fluorescence intensity is significantly influenced by conjugation with PMLA. In conjugates free ICG, P/ICG(2%) and P/CTX(1.5%)/ICG(2%) fluorescent intensities are lower by over 7-fold compared with intensities for conjugates P/LLL(40%)/ICG(2%) and P/LLL(40%)/CTX(1.5%)/ICG(2%) (Fig. 3A). Reduction in self quenching [41] and simultaneous introduction of a hydrophobic milieu by placing LLL between ICG are considered the main mechanisms increasing the intensity [42].

Figure 3. Fluorescence intensity of free and precursors of the imaging agent.

(A) Fluorescence intensities of free ICG and imaging agent precursors were measured at 22 °C at pH 7.4 using Odyssey® CLx Imaging System, wavelength 800 nm. The fluorescence intensities of LLL containing conjugates P/LLL(40%)/ICG(2%) (◊) and P/LLL(40%)/CTX(1.5%)/ICG(2%) (O) are manifold enhanced in comparison with intensities of free ICG (*). Whereas, fluorescence of ICG conjugates P/ICG(2%) (□) and P/CTX(1.5%)/ICG(2%) (Δ) without LLL shows decreased intensities. (B) Fluorescence intensity of lead NIA, P/LLL(40%)/CTX(1.5%)/ICG(2%), (+LLL), exceeds by far that of conjugate P/CTX(1.5%)/ICG(2%), (-LLL). (C) Cartoon suggesting fluorescence quenching by proximal conjugation of ICG residues (left) and (D) absence of quenching by interspaced conjugation of hydrophobic LLL (right).

3.5. Interaction of the lead imaging agent, P/LLL(40%)/CTX(1.5%)/ICG(2%), with glioma cells

Glioma grading is independent of size of tumor; in the applied model, tumor is classified as grade IV GBM. Animals formed tumors of different sizes (Fig. 8 and Fig. S9), and tumors of small and large size adopted equally intensive staining by NIA. The degree of glioblastoma imaging by fluorescence depends in the first place on the binding of the lead NIA P/LLL/CTX/ICG to the surface of glioma cells. Several kinds of interactions could be possible based on the diverse multiple groups, CTX, LLL, and ICG, and could empower selective binding to several tumor surface molecules and invoke more than one internalization pathways into glioma cells. To shed light on possibly diverse modes of interactions, we focused our studies on binding to the cell surface of glioblastoma cells using flow cytometry, focusing on contributions of nondrug vehicle P/LLL(40%), CTX and ICG.

3.5.1. Ligand binding to glioma cell surface.

Based on our strong in vivo results for tumor specific accumulation of NIA with very high contrast ratios (Fig. 6B) and fluorescent microscopy data (Fig. 5C, D) provided a convincing evidence that healthy tissue does not react with NIA. Hence, we focused our efforts to study specific ligand binding ability of NIA to malignant U87MG glioma cells. The reaction of glioma cells with our NIA and its molecular constituents was followed by flow cytometry at 4 °C [31]. Cells and reagents were incubated at this temperature before analysis. The degree of fluorescence labeling increased as a function of the concentration of varied ligands seen in the histograms of Fig. 4A–D. The mean fluorescence intensity was analyzed as function of total ligand concentrations in the insets of Fig. 4A, B and Fig. S6. To have a common denominator, we used the total chlorotoxin content of NIA (P/LLL(40%)/CTX(1.5%)/ICG(2%) equal to 7.75 molecules CTX per NIA molecule) as the variable (Fig. 4A inset). The calculated binding curves showed the best fit for an apparent dissociation constant Kd(CTX_tot) = 4.79 μM. A similar analysis carried out for binding of CTX-ICG (1 molecule of CTX per molecule of CTX-ICG) indicated at 2-fold higher value Kd(CTX_tot) = 8.52 μM (Fig. 4B). The difference is indicative of higher binding affinity for each CTX residue in the multi component NIA in comparison with the two-component CTX-ICG. A calculation also indicates that the NIA molecule benefits from a higher affinity indicated by the global dissociation constant Kd-global(NIA) = 4.79 μM / 7.75 = 0.62 μM.

Figure 6. Pharmacokinetics and biodistribution of lead NIA, P/LLL(40%)/CTX(1.5%)/ICG, in glioblastoma mouse model.

(A) Pharmacokinetics in human serum (blue), in tumor (red) and in tumor free brain (yellow). (B) NIA accumulation in tumor and tumor free brain at various times after IV injection into tail. Significantly high contrast was noticed at all time points with 4 h being the highest and recommended for surgical resection. (C) Biodistribution of lead NIA in brain, tumor, kidneys, spleen, lung, liver and heart 4 h after IV injection into tail vain. lower panel shows picture of resected organs by SIRIS imaging system [15]. Spleen, liver and kidneys visually appear to have same intensities under SIRIS imaging system as the signal reaches the saturation in those organs under instrument setting.

Figure 5. Ex vivo fluorescence microscopy showing extravasation of NIA across BBB and accessing tumor cells.

Tumor and brain sections 16 h after IV injection into mouse tails: (A) After IV injection of P/Rh(2%); (B) P/LLL(40%)/Rh(2%); and (C) Lead NIA(Rh), P/LLL(40%)/CTX(1.5%)/Rh(2%). In this experiment, rhodamine substitutes ICG, since NIR(ICG)-fluorescence cannot not be recorded in the conventional microscopic set up. The rhodamine and ICG versions are similar by their values of dissociation constants Kd(CXT_tot) for binding glioma cells, Fig. S6 and Fig. 4A. Red color distribution indicates fluorescence in vesicular and diffusely distributed within the tumor interstitial. Vessels are stained with antibody to vWF (green) and nuclei by DAPI (blue). Merge mode (yellow) shows superposition of capillary and lead NIA staining. Note that low fluorescence in red is still visible in the vasculature after injection of P/LLL(40%)/Rh(2%) at a time when most of the agent has been released from the mouse body. (D) Confocal microscopic analysis of lead NIA(Rh). Intense distribution of NIA stained vesicles outside blood capillary and next to nuclei is seen only after injection of P/LLL(40%)/CTX(1.5%)/Rh(2%) in Panel C and in Panel D (“Rhodamine and Merge”), demonstrating that NIA containing CTX (in contrast to agent not containing CTX, P/LLL(40%)/Rh(2%), in panel B) is required for permeation across tumor BBB. Staining is not observed outside the tumor area (the dotted line marks the tumor border).

Figure 4. Binding of NIA, P/LLL(40%)/CTX(1.5%)/ICG(2%), and CTX/ICG to glioma cells measured by flow cytometry.

(A) Flow cytometry histogram for binding of NIA at 4 °C [31] as a function of total concentration of C TX, (CTX_tot). Inset: median fluorescence intensity (MFI) as a function of concentration following saturation dependence with an apparent dissociation constant Kd(CTX_tot) = 4.79 μM (“operational dissociation constant”). Concentrations CTX_tot are (1) 0 μM; (2) 0.3 μM; (3) 0.75 μM; (4) 1.5 μM; (5) 2.25 μM; (6) 3.75 μM; (7) 5.62 μM; (8) 7.5 μM; (9) 11.25 μM.(B) Flow cytometry histogram for binding of CTX-ICG as function of CTX_tot. Inset: MFI as a function of concentration following saturation dependence with an apparent dissociation constant Kd(CTX_tot) = 8.52 μM, the operational dissociation constant calculated for the dissociation of the cell CTX-ICG complex. Concentrations CTX_tot are (1) 0 μM; (2) 0.5 μM; (3) 1.0 μM; (4) 2 μM; (5) 3 μM; (6) 5 μM; (7) 7.5 μM; (8) 10 μM; (9) 15 μM. (C) Flow cytometry histogram showing the fluorescence intensity of 5 μM NIA at varied concentrations of competing CTX. Inset: The percent decrease in MFI, %ΔMFI, is shown as function of reciprocal concentration, [CTX]⁻¹. At [CTX]⁻¹ = 0, the extrapolated value for %ΔMFI = 70% is interpreted as the decrease under CTX saturation. NIA is 5 μM at all concentrations of added CTX. Concentration of CTX_tot are: (1) PBS, no ligand; (2) 0 μM; (3) 5 μM; (4) 50 μM; (5) 75 μM; (6) 125 μM. (D) Flow cytometry histogram for competition of P/LLL(40%) against binding of NIA at a fixed concentration of competitor CTX = 5 μM. Concentrations are: (1) PBS, no ligand; (2) 5 μM NIA, no competing ligand; (3)-(5) 5 μM NIA and one or two other competing ligands; (3) competitor 125 μM P/LLL(40%); (4) competitor125 μM CTX; (5) competitors (3) and (4). MFI was calculated for (2), (3), (4), (5). %MFI was 100% for NIA alone (2); %ΔMFI was calculated for (2)-(3), (2)-(4), (2)-(5).

Above calculations suggest that the NIA binding benefits from multivalences by combination binding to two or more CTX binding sites and/or by interaction by one or more structural constituents other than CTX. We tested for this possibility by questioning, whether competition of CTX against NIA could predict full NIA displacement in Fig. 4C. However, the maximum displacement of NIA under the condition of extrapolated saturating concentration of CTX (Fig. 4C, inset) was 70% instead of 100% expected for competitive binding of NIA and CTX. This suggested that some other structural component(s) of NIA contributed to binding at sites which did not overlap the binding site of CTX. Such a candidate could be P/LLL(40%). We have shown that this component contributes to membrane permeation and endosomal release of PMLA nano conjugates through hydrophobic interactions [39, 40]. Since P/LLL(40%) is not fluorescent, competitive binding of P/LLL(40%) could be recognized by MFI reduction during NIA binding. A decrease in mean fluorescence intensity was indeed observed (printout histogram Fig. 4D). Competition of 5 μM NIA (100% fluorescence intensity) with 125 μM P/LLL(40%) was indicated by a 41–46% intensity decrease, and, in the mixture of 125 μM P/LLL(40%) plus 125 μM CTX, by an overall decrease of 56–58%. The results indicated that under combined competition of P/LLL(40%) and CTX, NIA could not be fully displaced from binding to glioma cells. The incomplete degree is best explained by mixed competitive/noncompetitive inhibition. Moreover, ICG could be another structural component contributing to binding. These subsites could be independent, allosterically coupled or partially overlapping NIA bindings sites. Whatever the details, the results indicate that NIA binding involved three or more subsites.

3.5.2. NIA permeation of blood-brain barrier (BBB) and subcellular distribution in GBM

The different sites involved in the binding of NIA on glioma cells are considered potential internalization sites. To prove internalization, the cellular distribution of the imaging agent in xenogeneic GBM and especially in the tumor cells was studied by ex vivo fluorescence microscopy of glioblastoma and healthy brain sections using lead NIA which had 2% ICG substituted by 2% rhodamine. According to evidence for similar binding of NIA(Rh) by flow cytometry (Fig. S6 in comparison with Fig. 4A), we assumed that the substitution was a valid manipulation. The substitution improved the staining intensity and contrast. The staining is compared for sections of tumor-free brain (non-tumor) and tumor in Fig. 5. Sections probed with P/Rh(2%) are devoid of fluorescence staining (Fig. 5A). Tumor-free brain remained unstained in all cases after injection of NIA or CTX-free precursor (Fig. 5A). NIA, P/LLL(40%)/CTX(1.5%)/Rh(2%), exhibited fluorescence in tumor, inside and outside blood vessels (Fig. 5C). In the absence of CTX, precursor, P/LLL(40%)/Rh(2%), showed some fluorescence which merged with the anti-vWF-stained vessels (Fig. 5B). In summary, nanoagent not containing CTX are consistently free of vascular staining in healthy brain (non-tumor) sections after 16 h (Fig. 5 A–C). At this time, the agents would have been washed out according to the PK half-life of 72 min measured in the case of NIA and probably holds for the other nanoconjugates. In contrast, tumor vessels retained moderate red staining (Fig. 5B) after injection of P/LLL(40%)/Rh(2%), which did not contain CTX, and could have expected to be washed out. Because vessel endothelia of tumor including glioblastoma are known to be compromised, it is possible that the tumor not only changed the surface of glioma cells, but also the membranes in the endothelia of tumor vessels. The staining could be explained by exposure of the CTX-free agent to P/LLL(40%)-binding sites which were similar with the ones indicated by the competition experiment at the surface of tumor cells (Fig. 4D, section 3.5.1). Moreover, the finding of red fluorescence in tumor vesicles indicates that the postulated sites for P/LLL(40%) alone are not sufficient for internalization. In contrast, and in agreement with the role of CTX, when NIA (P/LLL/CTX/ICG) had been injected (Fig. 5D), the red color was found to have permeated into the tumor cells, recognized as diffused or particle-like deposits. Overall, the results are consistent with a glioma specific imaging that involves multiple surface binding sites on tumor and endothelial cells and only CTX dependent internalization. The function of several different active sites in the NIA nanoconjugate could explain the intense fluorescence staining and the long residence time (t_1/2 ~ 10 h) in the tumor analyzed in Fig. 6A. The manifestation of CTX dependent staining within glioma cells demonstrate the ability of NIA to cross BBB and to stain the tumor cells.

3.6. Pharmacokinetics and Biodistribution of lead NIA

The optimal time of resection was selected by comparison of lead NIA clearance from blood and tumor (Fig. 6A). The blood/plasma clearance followed the single exponential time dependence of a one compartment kinetics with t_1/2 = 72 min. This compares with t_1/2 = 2–4 min of free ICG known to be rapidly bio-eliminated through the kidneys [37]. The life time is in sharp contrast with a slow first phase of tumor clearance following t_1/2 = 564 min and a second phase exceeding the time of observation (50 h). The fluorescence intensity was converted into concentration of NIA on the basis of calibration measurements by spiking samples with known amounts of NIA for each organ. On the bases of the concentration ratio of tumor over normal brain in Fig. 6B, the best times for image guided resection are ≥4h. On the same basis and in comparison, using images obtained from SIRIS camera during vivisection, the organ distribution is shown in Fig. 6C. The highest fluorescence in spleen, liver and kidneys correspond with their functions in clearance. Fluorescence in the brain is low or below detection limit and agrees with absence of toxicity.

3.7. Method of fluorescence staining of glioblastoma qualifying for image guided resection

The fluorescence staining performance of the lead imaging agent during tumor resection, P/LLL(40%)/CTX(1.5%)/ICG(2%), was tested in vivo using the SIRIS imaging device (Fig. 7). The test followed IV injection into the tail of nude mouse and surgery using SIRIS at optimal settings of the camera [15] 4 h after injection. Tumor was located under NIR by its blue (pseudo color) fluorescence (white arrows, Fig. 7). After incision and removal of tissue under NIR light, the tumor appeared with sharp boundaries and could be resected in a few steps. Very little or no fluorescence was seen with the control P/ICG(2%) (data not shown). Targeted agent (P/CTX(1.5%)/ICG(2%) without LLL showed selective accumulation in tumor but with lower intensities (data not shown). After resection, other vital organs showed maximum to no fluorescence depending on their involvement in blood clearing (Fig. 6C and Fig. 7).

3.8. Tumor resection and resection precision

A cohort of mice (n=3–5) carrying human GBM at comparable tumor size, were injected with lead NIA, P/LLL(40%)/CTX(1.5%)/ICG(2%), or PBS (control group), euthanized and brains harvested. For optimization of resection methods, tumors were resected from isolated brains in absence and presence of lead NIA fluorescence 4 h after injection. After resection, brains were fixed, sectioned and stained with H&E for tumor tissue identification under the microscope. Sections in the top, middle, and deeper part of the resection cavity were inspected (Fig. 8 A and further examples in Fig. S8). For each section, 2-dimensional regions of interest (ROI) were measured and related to the area for tumor occupancy before resection using the equation % Resection = 100 × (Total tumor area − Sum of ROI)/Total tumor area). Fig. 8 A and Fig. S8 show examples for the resection of a NIA injected mouse under NIR in comparison with resection of buffer-injected mice under white light. Fig. 8 C shows the analytic results of the precision to remove tumor tissue. Fig. 8 B and Fig. S9 show examples of resected tumor containing healthy brain tissue. Because prevention of healthy tissue from resection is highly desired, the evaluation under Fig. 8 A, C shows the precision of maintaining healthy brain. Clearly, the resection obtained after injection of NIA and resection under NIR light is much superior over the resection under white light and absence of NIA injection. Except for the resection in the tumor top region, the degree under NIR light, 98.4 ± 3.1%, was similar to the one for white light, 93.87 ± 7.7% (no significant difference, p = 0.11). However, resection under NIR showed benefit when the middle and deeper tumor cavity was analyzed (Fig. 8). Thus, resection efficiency was significantly higher at 98.19 ± 2.9% under NIA light than under white light, 74.68 ± 22.98% (p=0.027), in the middle region and for tumor in the deep region, under NIR light, 98.11 ± 3.64% (p= 0.011), compared with white light, 64.67 ± 25.54%, (Fig 8 A, C). Thus, under white light fragments were increasingly less resected when tumor was deep within the brain. Since the resection was equally well under NIR and white light in the top region, it can be concluded that the visibility/differentiation of tumor from healthy brain to the experimenter under white light, but not under NIR light decreases towards the tumor bottom and was the cause of the incomplete resection. In addition, adherent parts of tumor interdigitated into tumor free areas and were less efficiently identified as tumor fragments (Fig. S7, S8).

4. Discussion

4.1. The importance of image guided tumor resection

Despite decades of efforts in developing more effective treatments for gliomas, surgery remains the single most successful strategy [2, 3]. The success of surgery is highly dependent on the extent of resection obtained, with increased survival demonstrated when >95% of the enhancing tumor volume is resected [9, 43]. More recent data have demonstrated even better survival when the surrounding tissue indicated by FLAIR signal from MRI is also resected [44].

Today’s interest is focused on the use of ICG fluorescence in the NIR range and of CTX for targeting to the site of interest. Fluorescence in the NIR spectral range has several advantages over fluorophores in the visible light range. First, penetration depth in the NIR range is high allowing deeper tissue imaging [45]. Second, lack of endogenous auto fluorescence such as hemoglobin in the NIR range enhances the signal to noise ratio, improving the image quality. For imaging glioblastoma, the uptake of CTX into tumor includes permeation across BBB, which could include transcytosis as the most effective delivery mechanism for brain treatment and imaging [46, 47]. Some reports also relate delivery across tumor BBB to the Enhanced Permeability and Retention (EPR) effect [48, 49]. To achieve a more powerful imaging than with CTX-ICG, our NIA has been engineered with high ICG-loading, specific tumor cell affinity and the availability to conjugate further nano-medicines and nano-imaging agents.

4.2. Additional factors contributing to powerful tumor imaging

Besides improving the fluorescence intensity delivering several ICG molecules per carrier platform, we have in addition substantially improved the fluorescence intensity of ICG bound to PMLA by disruption of fluorescence quenching (Fig. 3A–D). This phenomenon is referred to the relief from self-quenching of proximal ICG fluorophores [38]. In P/LLL(40%)/CTX(1.5%)/ICG(2%), the disruption is achieved systematically by incorporation of LLL in sites between ICG molecules on PMLA backbone. This strategy is highly effective as it not only keeps ICG molecules away from each other but also brings them in close proximity to hydrophobicity which contributes to fluorescence enhancement. Proximity of ICG to hydrophobicity has been reported in the case of ICG binding to lipids and lipoproteins resulting in fluorescence enhancement [42].

In order to understand the tumor cell multi targeting and prospective internalization mechanisms, we considered the multi functionality of the NIA construct, P/LLL(40%)/CTX (1.5%)/ICG(2%). By structural and chemical analysis, the polymalic acid (PMLA) platform and thus the nanoconjugates have linear chemical structures and do not assemble into higher order structures when 50–60% of the pendant carboxylates are free and ionized. Because the polymer backbone contains only single bonds (C-C-C and C-O-C), the unsubstituted PMLA and its derivatives exist in numerous interchangeable more or less structured coil conformations [28, 29, 36, 47, 50, 51]. According to this flexibility, ligands can be largely accessible for biological reactions. Although the polymer/copolymer is highly negatively charged and contributes in variable degrees to conjugate’s negative zeta potential (Table 1), the hydrophobic ligands LLL can bind to cell membranes [39, 40, 52], and potentially modulate the interactions of hydrophobic co-ligands ICG and positively charged CTX with biological receptors. In particular, LLL residues could promote the permeation of NIA through membranes similar as demonstrated for endosomes permeation into cytoplasm [39]. The amphiphilic nature of ICG has been demonstrated for interactions with lipoproteins [53], and cell types such as vascular endothelial or adipose tissue [54, 55], and furthermore for cancer cells, e.g. of colon cancer [52]. The preferential uptake of ICG by colon cancer cells is an example for its cell membrane-binding and permeation (endocytic) ability [52]. Similarly as in the case of P/LLL(40%) permeation through endosomal membranes, the ICG activity could be the result of cooperative hydrophobic / polar interactions with membranes. The contribution by CTX in glioma has been described as the selectively blocking of chloride channels functioning in tumor cell migration [56–58] and blocking annexin-A2 activity in membrane repair [59, 60]. CTX has been assumed to inhibit MMP-2 protease activity involvement in glioma invasiveness [57, 58, 61] in advanced tumor stages [18, 27, 62]. CTX is also known for cell penetrating activity [19]. Thus, the combination of CTX, ICG, LLL assembled on PMLA is prone to unfold novel powerful interactions at the tumor cell level and hence render NIA a more powerful agent than CTX-ICG biconjugate alone. The complexity of NIA reactions visualized by flow cytometry and microscopy could be causative for the longevity of fluorescence in glioma cells that exceeds by far the clearance from the blood stream (570 min vs. 72 min). The benefit is a long-lasting period of strong tumor fluorescence and thus for improved tumor resection.

The distribution of fluorescence of IV injected NIA in the recipient indicates high levels in spleen, liver, and kidneys; but fluorescence was absent in brain, heart and at low levels in lung. The high fluorescence in liver and kidneys parallels system clearance through the blood, and the high level in spleen likely reflects clearance through macrophages. The low background in tumor free brain underlines the high specificity of NIA for glioblastoma and absence of penetration across BBB of healthy brain.

4.3. Facilitated tumor resection

The infiltrative nature of glioma makes it difficult for surgeons to accurately identify the boundary between tumor and non-tumor tissue. A clear advantage of our polymalic acid-based NIA is the delivery of a much higher number of ICG fluorophores per CTX targeting site resulting in superior fluorescent imaging. It permeates the tumor BBB and internalizes with high efficacy into single tumor cells thereby improving tumor detection, even in the lower tumor grade periphery [22, 23, 27]. This has been achieved by multisite binding of polymalic acid as the molecular platform conjugated with LLL, CTX and ICG. The NIA is a mini-nano agent carrying ligands for multi-site attachment to glioma cells. A further benefit is the availability of multiple attachment sites for a plethora of drugs for glioma treatment [28, 29, 47, 63–65].

The increase of the PK half-life from 2–4 min of free ICG to 72 min of our NIA provides an increased time period for NIA uptake [54, 66] and superior tumor visibility. Furthermore, the NIR fluorescence merits its strong intensity to the imbedding of multiple ICG molecules between platform conjugated LLL residues. This effect is consistent with the observed binding of ICG to lipoproteins and phospholipids that influences fluorescence spectrum and intensity [67]. The binding to serum proteins including a variety of enzymes explains the instability in Fig. S3 due to degradation with a half-life of ≈ 10 h, which, however, is too long in order to explain the above PK half-life of 72 min.

4.4. Achievements

Respectable routine achievements of high precision better than 98% tumor removal were obtained under NIR light that was independent of the sampled tumor location top, middle and deep. The novelty and power of our multivalent imaging agent is its multipronged functions: (i) high degree of glioblastoma staining including border adjacent parenchyma infiltration, (ii) fluorometric demarcation of high intensity and contrast, (iii) deep tissue penetration favored by the small NIA size, (iv) intense cellular uptake and tumor persistence, (v) convenient time window for conducting postresection chemotherapy, (v) multiple functional attachment sites enable multimodal chemotherapy, (vi) referenced examples of chemotherapy, (vii) biodegradability [50], absence of toxicity for PMLA platform [50], low, if any immunogenicity [50], (viii) controlled stability through covalent (macromolecular) structure.

4.5. Outlook

We have demonstrated a method for precise fluorescence guided resection of glioblastoma through development of glioma specific NIA. The high resolution, and the degree of resection could be further developed including appropriate optical and surgical techniques to remove tumor infiltration. Remaining elimination of deep in brain located infiltration could be addressed by adjuvant chemotherapeutic drugs delivered by the same imaging platform used in resection. If successful in the glioblastoma mouse model, efficacy should be tested in other tumor models to demonstrate general applicability, and then treatment expanded to include big animals seeking approval for human treatment.