Nanostructure Empowers Active Tumor Targeting in Ligand‐Based Molecular Delivery

Cell‐selective targeting is expected to enhance effectiveness and minimize side effects of cytotoxic agents. Functionalization of drugs or drug nanoconjugates with specific cell ligands allows receptor‐mediated selective cell delivery. However, it is unclear whether the incorporation of an efficient ligand into a drug vehicle is sufficient to ensure proper biodistribution upon systemic administration, and also at which extent biophysical properties of the vehicle may contribute to the accumulation in target tissues during active targeting. To approach this issue, structural robustness of self‐assembling, protein‐only nanoparticles targeted to the tumoral marker CXCR4 is compromised by reducing the number of histidine residues (from six to five) in a histidine‐based architectonic tag. Thus, the structure of the resulting nanoparticles, but not of building blocks, is weakened. Upon intravenous injection in animal models of human CXCR4+ colorectal cancer, the administered material loses the ability to accumulate in tumor tissue, where it is only transiently found. It instead deposits in kidney and liver. Therefore, precise cell‐targeted delivery requires not only the incorporation of a proper ligand that promotes receptor‐mediated internalization, but also, unexpectedly, its maintenance of a stable multimeric nanostructure that ensures high ligand exposure and long residence time in tumor tissue.


Introduction
Targeting of drugs for precision medicine is a widespread popular challenge, since proper drug biodistribution is expected to enhance effectiveness and minimize undesired side effects. [1] This is especially desirable regarding cytotoxic drugs, as those used in cancer, whose administration is associated to severe toxicities. It is assumed that functionalizing drugs or drug complexes with selective cell ligands would confer active targeting and ensure their accumulation in target cells and organs where such receptor is overexpressed. However, the biodistribution analyses of antibody drug conjugates and other similarly targeted drug constructs have repeatedly revealed that the fraction of administered agent reaching the target organ is limited to around 1%. [2] On the other hand, physical properties of drug vehicles such as surface charge, geometry, and size, among others, appear as key factors influencing the tissue accumulation Cell-selective targeting is expected to enhance effectiveness and minimize side effects of cytotoxic agents. Functionalization of drugs or drug nanoconjugates with specific cell ligands allows receptor-mediated selective cell delivery. However, it is unclear whether the incorporation of an efficient ligand into a drug vehicle is sufficient to ensure proper biodistribution upon systemic administration, and also at which extent biophysical properties of the vehicle may contribute to the accumulation in target tissues during active targeting. To approach this issue, structural robustness of self-assembling, protein-only nanoparticles targeted to the tumoral marker CXCR4 is compromised by reducing the number of histidine residues (from six to five) in a histidine-based architectonic tag. Thus, the structure of the resulting nanoparticles, but not of building blocks, is weakened. Upon intravenous injection in animal models of human CXCR4 + colorectal cancer, the administered material loses the ability to accumulate in tumor tissue, where it is only transiently found. It instead deposits in kidney and liver. Therefore, precise cell-targeted delivery requires not only the incorporation of a proper ligand that promotes receptor-mediated internalization, but also, unexpectedly, its maintenance of a stable multimeric nanostructure that ensures high ligand exposure and long residence time in tumor tissue. www.advancedsciencenews.com www.particle-journal.com pattern upon systemic administration when the delivery platform is based on passive targeting, [3,4] for instance by exploiting the enhanced permeability and retention (EPR) effect. [5] However, the weight of nanoscale properties of the material itself in determining biodistribution in presence of selective cellligands, that is, during active targeting, remains unsolved, despite its critical value in the design of new drug delivery systems. Combining efficient homing peptides with carrier materials in their optimal configuration might largely enhance the local accumulation in target tissues above the ≈1% threshold and thus increase precision and effectiveness in the delivery process.
To discriminate between the roles of the ligand and the architecture of the vehicle itself in the process of active tumor targeting, we have engineered the modular protein T22-GFP-H6 into related constructs and tracked selected resulting variants upon administration in animal models of human colorectal cancer. Such fusion protein is composed by T22, a potent ligand of the cell surface cytokine receptor CXCR4, [6] overexpressed in several metastatic human cancers, [7] a fully fluorescent GFP and a C-terminal polyhistidine tail. T22-GFP-H6 spontaneously self-assembles in physiological conditions as 12 nm nanoparticles, formed by around ten copies of the polypeptide, organized in a toroid architecture [8,9] and with some extent of structural flexibility. [9] The high selectivity of T22 for CXCR4 observed in cell culture [6] is fully kept in vivo. [10,11] When administered intravenously in orthotropic mice models of human CXCR4 + colorectal cancer the fluorescent nanoparticles accumulate in primary tumor and metastatic foci at unusually high levels, estimated to represent more than 85% of the whole-body detected fluorescence. [6] The accumulation of florescence in inhibited by SDF1-α, the natural ligand of CXCR4, [11] and it does not take place when T22 is absent in equivalent fusion proteins. [10] Used as a carrier of the cytotoxic drug floxuridine (FdU), the nanoconjugate T22-GFP-H6-FdU reduces the volume of primary tumor, prevents the development of metastasis, and precisely destroys already formed metastatic foci in absence of detectable systemic toxicity. [12] Similar antitumoral effectiveness has been observed when the nanoparticles deliver, accommodated in the building blocks by genetic fusion, proapoptotic factors, and other antitumoral peptides. [13] Interestingly, the self-assembling of T22-GFP-H6 and related materials is driven by the overhanging polyhistidine tails that coordinate divalent cations from the media to promote stable cross-molecular protein interactions. [14] If the structure of the nanoparticle beyond the ligand itself, is relevant for precise targeting, destabilizing the supramolecular complex by modifying the histidine tail sequence would result in a potentially altered biodistribution map of the material, even if this material still contains the active CXCR4 ligand T22. The comparison of the fluorescence maps of T22-GFP-H6 and one of its less stable variants T22-GFP-H5T, once intravenously (iv) injected in colorectal cancer models, revealed that the presence of the targeting peptide T22 in the protein, although necessary for CXCR4-mediated cell binding, [6] is not sufficient for a proper tumor targeting. On the contrary, the nanoarchitecture of the material as an oligomeric supramolecular complex has a critical and unexpected impact on the fate, dynamics, and final accumulation of the material at the different organs, allowing the desired biodistribution upon administration. Therefore, nanoscale organization is an unexpected key determinant of not only passive but also active targeting.

Results
Being the H6 tail critical for nanoparticle formation, [15] this end-terminal peptide was replaced in T22-GFP-H6 by alternative histidine-rich peptides of similar length, with lower content of histidine (His) residues (Table 1). Since His residues promote the cross-molecular protein-protein interactions that sustain the architecture of the oligomers, [14] the reduction in the number of His residues was expected to generate less stable nanoparticles. Then, T22-GFP-H3A, T22-GFP-H5T, and T22-GFP-H5E fusions were designed, constructed, and expressed in bacteria as soluble protein versions, for comparison with the parental T22-GFP-H6. The alternative His-rich segments were selected according to previous reports indicating that His residues, intersected with hydrophobic or negatively charged residues, could be still retained in Ni +2 -based chromatography purification that uses His residues as binders. [16] All proteins (the parental and the derived versions) were produced as proteolytically stable full-length forms of expected The nomenclatures 6, 3, and 5 refer to the total number of His residues in the C-terminal tag and A, T, and E refer to alanine, threonine, and glutamic amino acids, respectively; b) The sequence of T22 is MRRWCYRKCYKGYCYRKCR; c) Underlined segments correspond to the amino acids introduced in the study; d) The linker sequence is GGSSRSS; e) The concentration (× 10 −3 m) of imidazole needed to induce protein elution from immobilized metal ion affinity chromatography; f) The above values (e)  www.advancedsciencenews.com www.particle-journal.com molecular masses ( Figure 1A, Table 1), and the specific fluorescence emission values were of the same order of magnitude than that shown by T22-GFP-H6 (Table 1). This fact indicated that native-like conformation was reached in individual GFPbased building blocks. The purification by His-tag-based affinity chromatography was efficient in all cases, but the concentration of imidazole required to elute the proteins was different in each case (Table 1). It was lower, as expected, at lower His residue content. The H5T-tagged polypeptide was eluted at an imidazole concentration that represented 86% of that required by H6-tagged materials, indicating that the strength of Hisdivalent cation interactions was weakened down to this relative level compared to the H6 tag. T22-GFP-H3A and T22-GFP-H5E required even less imidazole concentration for detachment from immobilized Ni +2 , representing 68% and 61% of that required for T22-GFP-H6, respectively (Table 1). This fact, and the resulting quantitative data about imidazole-mediated detachment, confirmed that the strength of His-based crossmolecular interactions can be regulated by the number of His residues in overhanging tags. In this context, since the self-assembling of His-tagged T22-carrying nanoparticles is based on the ability of His residues to interact with each other's through divalent cations from the media, the quantitative reduction in the interactivity with Ni +2 www.advancedsciencenews.com www.particle-journal.com of the engineered proteins should be translated into nanoparticles less stable than T22-GFP-H6, if they were actually formed. When checking the self-assembling of the materials in the standard carbonate buffer, all proteins spontaneously formed nanoparticles ( Figure 1B,C, Table 1), with hydrodynamic sizes and Z-potential values similar to those shown by the parental T22-GFP-H6 (Table 1). The microscopy scrutiny of all nanoparticles revealed a toroidal architecture ( Figure 1C), compatible with the previously obtained molecular model of T22-GFP-H6. [8] However, when challenging the assembled materials with ionic strength, T22-GFP-H3A and T22-GFP-H5E, those with less molecular interactivity (Table 1), immediately disassembled into smaller materials with sizes compatibles with the dimeric form of GFP (around 7 nm, Figure 1B). This was indicative of weak cross-molecular interactions between building blocks. Instead, T22-GFP-H5T tolerated well the presence of salt in the media. However, this construct showed high instability during freezing and thawing and it partially disassembled as structures smaller than 12 nm (Figure 2A), of size comparable to assembling intermediates described for T22-GFP-H6. [9] Some of these structures were also observed under transmission electron microscopy (TEM) ( Figure 1C). These small forms appeared together with a minor occurrence of larger protein clusters, indicative of supramolecular instability (Figure 2A), and conformational impact linked to freezing and thawing-induced damage. [17] To further assess the differential stability between H6-and H5T-based nanoparticles, they were incubated for 24 h at 37 °C in human sera, to better reproduce the conditions of in vivo administration. As observed (Figure 2A), T22-GFP-H5T (but not T22-GFP-H6) nanoparticles dissociated under these conditions, confirming again the lower stability of the H5T material. Such weaker structural robustness was not due to defects in the folding of H5T building blocks, as thermal stability analysis indicated that both modular polypeptides were  www.advancedsciencenews.com www.particle-journal.com equally stable (or even T22-GFP-H5T lightly more stable than T22-GFP-H6, Figure 2B). This was in agreement with the fluorescence data from Table 1.
In the light of these observations, we decided to comparatively determine the influence of nanoparticle stability on in vivo biodistribution by comparing T22-GFP-H6 and T22-GFP-H5T materials. Importantly, the modular polypeptides themselves, were both proteolytically resistant (Figure 1), structurally stable ( Figure 2C), targeted to the tumoral marker CXCR4 through T22, [6,8,18] and only differ in a few structural amino acids at their C-termini. Also, the electrophoretic motility of these proteins did not change in serum, as well as their specific fluorescence ( Figure 2C). All these data confirmed that despite the differences in the stability of the nanoparticles the building blocks were both structurally robust and competent. In addition, the interactivity between T22 and CXCR4 ( Figure 2D) and the ability of the peptide to mediate receptor specific endosomal internalization of nanoparticles ( Figure 2E) were fully confirmed in both constructs.
When both T22-GFP-H6 and T22-GFP-H5T were administered intravenously in mice bearing subcutaneous SP5 CXCR4 + colorectal tumors, the accumulation pattern of both proteins in tumor was clearly divergent. While T22-GFP-H6 was progressively found in tumor (Figure 3A), with a plateau of fluorescence reached at 24 h, T22-GFP-H5T was only transiently found in tumoral tissues at 5 h postadministration, followed by a fast decline ( Figure 3B). This might be indicative of lake or poor cell uptake in the tissue, through which the material appears to transiently pass by. Moreover, a background (offtarget) fluorescence emission of T22-GFP-H5T was observed in liver and kidney, having an increase during the 24-48 h period postinjection, whereas T22-GFP-H6 emission during  www.advancedsciencenews.com www.particle-journal.com this period was declining in these organs. The much more extensive and sustained T22-GFP-H6 tumor accumulation was clearly evidenced by the quantitative ex vivo analyses of relevant organs ( Figure 3B). Thus, T22-GFP-H6 reached a tumor exposure (AUC = 5.04 × 10 8 emitted fluorescence intensity-FLI-units × hour) 2.7 fold higher than T22-GFP-H5T (AUC = 1.90 × 10 8 ) (Figure 4A,B, Table 2). Mostly, background signal was observed in other nontarget organs, except for T22-GFP-5HT in the 24-48 h period, which registered increases of 64% in the kidney and 14% in the liver ( Figures 3C and 4B, Table 2). Consequently, T22-GFP-H6 had an AUC ratio tumor/ (kidney+liver) of 2.2, while in T22-GFP-H5T this ratio was 0.8 ( Figure 4C). Since the divergence in the biodistribution maps of the two tested related proteins is irrespective of the common N-terminal ligand (T22, binding CXCR4) but dependent on the amino acid sequence of the C-terminal architectonic peptide, we conclude that a multimeric organization of the modular proteins offers an appropriate nanoscale presentation of the ligand, with a geometry supporting its targeting function in the body.

Discussion
Two N-terminal homologous GFP modular proteins, namely T22-GFP-H6 and T22-GFP-H5T (Table 1), targeted to CXCR4 tumors, showed a very dissimilar biodistribution upon iv administration in mice models of human, CXCR4 + colorectal cancer ( Figure 3). Both protein versions are proteolytically stable upon bacterial production ( Figure 1) and upon incubation in human serum ( Figure 2C), showing no loss, in any case, of relevant protein fragments that might abort the cell binding process. Both polypeptides are also highly fluorescent (Table 1), show robust structural stability ( Figure 2B) and spontaneously assemble as regular nanoparticles of comparable size and physicochemical properties ( Figure 1C, Table 1) that equally penetrate CXCR4 + cells in culture ( Figure 2D). However, the minor sequence differences at the His-rich C-terminal peptide (Table 1), responsible for cross-molecular interactions and divalent cation-mediated nanoparticle formation [14] resulted weakened in T22-GFP-H5T relative to the parental T22-GFP-H6, to around 86% (Table 1). This is because of the reduction in the number of His residues in such architectonic peptide, from six to five, which minimizes the binding of the protein to divalent cations, including the Ni +2 of the purification columns (Table 1). Other two constructs with five and three His residues in the C-terminus, respectively, are not able to form nanoparticles in high salt buffer ( Figure 1B)   Measures of ex vivo fluorescence emission by subcutaneous CXCR4+ SP5 patient-derived tumors and normal mouse organs, as measured by FLI (Protein-buffer Radiant Efficiency/10 5 ) at the indicated time after iv injection of the material, using the IVIS Spectrum equipment. www.advancedsciencenews.com www.particle-journal.com at 37 °C indicated structural instability of T22-GFP-H5T nanoparticles ( Figure 2A) that was not apparent by the mere hydrodynamic size analysis upon biological fabrication ( Figure 1B). Such less stable T22-GFP-H5T nanoparticles reached the target tumor tissue at 5 h post iv administration ( Figure 3). However, they failed to accumulate in the tumor (being undetectable at 24 h), while displaying a much lower tumor exposure than the parental H6-tagged protein (Figure 4). Moreover, the amounts of this protein were progressively fading in tumor tissue, while an increased in its fluorescence signal was observed, at later times and at important levels, in nontumor organs such as liver and kidney. Therefore, T22-GFP-H5T had a lower accumulation in tumor than in nontumor tissues (AUC ratio = 0.8). This was in sharp contrast with T22-GFP-H6, for which most of the injected dose accumulated in tumor rather than in nontumor tissues (AUC ratio = 2.2) (Figure 4, Table 2). Thus, despite T22-GFP-H6 started their tumor uptake at later times, it reached a total tumor exposure 2.7 fold higher than this achieved for T22-GFP-H5T, and also maintained a high fluorescence exposure in tumor tissue beyond 48 h. In addition to this, the injected equal dose and highly similar fluorescence emission of the two compared proteins lead to much higher tumor exposure for T22-GFP-H6 than in nontumor tissues, while the opposite happened for T22-GFP-H5T, suggesting a more intense and faster clearance from the body of T22-GFP-H5T, since its total (tumor + nontumor) fluorescence emission showed a 43% reduction as compared to total T22-GFP-H6 FLI emission (Figure 4). In this regard, the higher accumulation in kidney and liver at longer times (48 h) for T22-GFP-H5T, together with its lower nanostructure stability as determined in vitro (Figure 2A), strongly suggests the possible occurrence of a much higher renal excretion and/or hepatic metabolism than T22-GFP-H6. The tumor accumulation pattern followed by T22-GFP-H6 was in agreement with previous experiments in related mice models. [8] This was indicative of the robustness of the material regarding biodistribution to tumor, leading to high exposure in that tissue by achieving a high uptake peak and a long residence time, while displaying low uptake in offtarget tissues. [6,8,10,19] In fact, the present data also suggested a lack of intracellular penetration of T22-GFP-H5T in tumor. When stable nanoparticles that effectively internalize in target CXCR4 + tumors cells are administered, [6] a residence time of around 48 h in tumor is consistently observed. During this time period, the nanocarrier is probably degraded within uptaking cells. [20] A 48 h residence time or longer, occurs also in therapeutic protein-only nanoparticles targeting CXCR4+ cancer cells. [21] The shorter tumor residence time of T22-GFP-H5T suggests that this protein carrier, despite interacting with the CXCR4 receptor through its T22 ligand, is not effectively internalized in target cells. Consistently, an early and short residence time of GFP-H6 (lacking the T22 ligand) in tumor has been also observed. [11] In summary, once rule out proteolytic degradation of both compared proteins for at least a 48 h period, equal to the biodistribution study time, it seems likely that having or lacking a nanostructure is a main driver for their biodistribution. Thus, T22-GFP-H5T lack of nanostructure and smaller size (6 nm) allows for early and rapid entrance in tumor extracellular space, despite precluding its internalization in target cancer cells, which is followed by an easy return to blood and renal clearance. In contrast, the higher size of the nanostructured T22-GFP-H6 particles slow their entrance in the tumor, but increases their recirculation in blood, because of lack of renal flirtation, and their internalization in CXCR4 + target cancer cells along time, significantly enhancing their whole exposure to the tumor.
These data were compatible with a robust structure of T22-GFP-H6 nanoparticles compared to a progressively disassembling T22-GFP-H5T materials, provided the nanostructure is assumed as a critical component of the active targeting process. While the role of nanostructure as an element influencing passive targeting has been largely discussed and recognized, [4,22] its potential impact on active targeting (that mediated by a cellsurface ligand) has been a rather neglected issue. Nanoscale organization of a targeted material might enhance its interaction with target cells by the multimeric binding of nanoparticles to cell surface receptor molecules. [23,24] Multivalent ligands generally show lower dissociation rates than individual versions ligands in the interaction with the receptor, [25] apart from a cooperative cell binding that promotes a more efficient early interaction and endosomal internalization. [26] Such cooperativity in both signaling and internalization of artificial constructs has been already described in different therapeutic platforms, [24,27] what could be specially efficient in the case of symmetrically ordered materials. [28] In the case of recombinant proteins, multivalent presentation of ligands in supramolecular constructs might be more efficient than monovalent versions, [29] what has been already discussed in the context of virus-like presentations of cell interactors and the consequent enhanced endosomal cell uptake. [26] In this regard, the results presented here support again the convenience of multivalent presentation, that also enhances the specificity in cell-receptor recognition. In this context, hybrid nanoparticles in which peptides R9 (an unspecific cell-penetrating peptide) and T22 (a specific CXCR4 ligand) are combined show lower CXCR4specificity than T22 only-based nanoparticles. [15] Besides, the size increase derived from oligomerization, in the case of the modular proteins described here from ≈4 nm (the hydrodynamic size of a GFP monomer) to ≈12 nm, above the renal cut-off or ≈6-8 nm, [10] might also increase circulation time and in consequence opportunities for a tight interaction with target tissues, promoting the desired tumor accumulation of tumorhoming materials.

Conclusion
The occurrence of an effective ligand of a tumor cell marker is necessary but not sufficient to ensure a proper tumor biodistribution of functional proteins upon systemic administration, as proved here by using a model self-assembling protein.
Contrarily, a supramolecular architecture of such targeted polypeptide, in form of multimeric nanoscale materials, enables the tumor homing peptide, here modeled by the CXCR4 ligand T22, to drive the accumulation of the material in the target tumor tissue. Several factors, including the multimeric regular presentation of the ligand and the nanoscale size of the complex are probably involved in the complex process of active targeting. In active targeting, the administered material www.advancedsciencenews.com www.particle-journal.com needs to overcome several biological barriers, including renal and hepatic clearance, to achieve higher exposure and residence time in tumor. The concept presented here might represent a convincing explanation of the poor biodistribution so far reached by tumor-targeted medicines, including antibody-drug conjugates. In addition to this, it is offering a potential developmental roadmap for the improvement of these drugs, of high intrinsic therapeutic potential, to reach satisfactory efficiencies in the clinical context.

Experimental Section
Genetic Design, Protein Production, and Purification: The genetic design of newly His-derived modular proteins was based on the parental T22-GFP-H6 construction. The C-terminal H6 poly-His tail was exchanged for alternative His-rich human peptides under specific criteria (explained during the work). The already displayed abbreviations -H6, -H3A, -H5T, and -H5E correspond to HHHHHH, HAAHAH, HTHTHTHTH, and HEHEHEHEH amino acid sequences respectively. Nomenclature has been established from N to C terminal according to their modular organization. All protein sequences were designed in house as codon-optimized genes, synthetized and inserted into pET22b plasmids using NdeI and HindIII restriction enzymes and provided by Geneart (ThermoFisher).
Protein Purity, Integrity, and Concentration: Protein purity was determined by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and western blot (WB) immunoassay with an anti-GFP monoclonal antibody (Santa Cruz Biotechnology). Protein integrity was also analyzed by matrix-assisted laser desorption ionization time-offlight (MALDI-TOF) mass spectrometry and concentration determined by Bradford's assay.
Volume Size Distribution, Z-Potential, and Fluorescence Emission: Volume size distribution (VSD) and protein surface charge (Z p ) of all proteins were determined by dynamic light scattering (DLS) and Z-potential measurements respectively at 633 nm and 25 °C in a Zetasizer Nano ZS (Malvern Instruments Limited) using ZEN2112 3 mm quartz batch cuvettes and DTS10170 capillary cells respectively. Measurements were performed in triplicate for error estimation and VSD peak values referred to the average mode of the populations with a rendered standard error lower than 0.01. Fluorescence emission of each GFP variant was determined at 513 nm using an excitation wavelength of 488 nm with a Varian Cary Eclipse Fluorescence Spectrophotometer (Agilent Technologies). For that, all the proteins were equally diluted in the corresponding sodium carbonate buffer w/o salt until 1 mg mL −1 in a final volume of 100 µL.
Ultrastructural Morphometry: The nanoscale morphometry (size and shape) of self-assembled nanoparticles was determined at nearly native state, both by deposition on silicon wafers with field emission scanning electron microscopy (FESEM) and by negative staining with TEM. Drops of 3 µL of T22-GFP-H6, T22-GFP-H3A, T22-GFP-H5T, and T22-GFP-H5E samples diluted at 0.4 mg mL −1 in sodium carbonate buffer were directly deposited on silicon wafers (Ted Pella Inc., Reading) for 30 s, excess of liquid was blotted with Whatman filter paper number 1 (GE Healthcare), air dried for few min, and immediately observed without coating with a FESEM Zeiss Merlin (Zeiss) operating at 0.8 kV and equipped with a high resolution in-lens secondary electron detector. Drops of 3 µL of the same four samples were directly deposited on 200 mesh carbon-coated copper grids (Electron Microscopy Sciences, Hatfield) for 30 s, excess blotted with Whatman filter paper, contrasted with 3 µL of 1% uranyl acetate (Polysciences Inc.) for 1 min, blotted again and observed in a TEM Jeol 1400 (Jeol Ltd.) operating at 80 kV and equipped with a Gatan Orius SC200 CCD camera (Gatan Inc.). For each sample and technique, representative images of a general field and a nanoparticle detail were captured at high magnifications (from 100 000× to 600 000×).
Determination of GFP Chromophore Fluorescence: The GFP chromophore fluorescence dependence on the temperature of each protein was also evaluated. Fluorescence spectra were recorded in a Varian Cary Eclipse spectrofluorimeter (Agilent Technologies). A quartz cell with 10 mm path length and a thermostated holder was used. The excitation slit was set at 2.5 nm and emission slit at 5 nm. λ ex was 488 nm. Protein concentration was 0.2 mg mL −1 in the corresponding buffer.
Structural Stability of Protein Constructs upon Human Serum Incubation: T22-GFP-H6 and T22-GFP-H5T protein nanoparticles were incubated at 37 °C with agitation (250 rpm) at proportion 1:1 in relation to human serum (Sigma-Aldrich) for 24 and 48 h. Protein VSD was determined at 24 h by a Zetasizer Nano ZS (Malvern Instruments Limited) and protein fluorescence and motility by a Varian Cary Eclipse spectrofluorometer (Agilent Technologies) and WB immunoassay respectively.
Protein Internalization: HeLa CXCR4 + cells (ATCC CCL-2) were cultured in 24-well plates (60 000 cells per well during 24 h for different time/concentration assays, in MEM Alpha 1× GlutaMAX medium (Gibco) supplemented with fetal bovine serum (FBS) at 37 °C in a 5% CO 2 humidified atmosphere, until reaching a confluence of 70%. Protein internalization was monitored at different concentrations (50 and 1000 × 10 −9 m) and times (1 and 24 h). After protein exposure, cells were detached and external hooked protein removed by adding Trypsin-EDTA (Gibco) at 1 mg mL −1 for 15 min and 37 °C. Intracellular protein fluorescence was determined by flow cytometry using a fluorescence assisted cell sorting (FACS)-Canto system (Becton Dickinson) at 15 mW with an air-cooled argon ion laser exciting at 488 nm. Measurements were performed in duplicate. Additionally, the specific protein CXCR4mediated internalization was proved by the addition of the receptor antagonist AMD3100 [30] that inhibits the interaction between T22 and CXCR4. This chemical compound was added at a final concentration of 500 × 10 −9 m (ten times protein concentration) for 1 h prior to protein incubation at 50 × 10 −9 m.
In Vivo Biodistribution Assays: All in vivo experiments were approved by the institutional animal Ethics Committee of Hospital Sant Pau. Five-week-old female Swiss nu/nu mice weighing between 18 and 20 g (Charles River, L-Abreslle) and maintained in specific-pathogen-free (SPF) conditions, were used for the in vivo biodistribution studies. A subcutaneous colorectal cancer mouse model, derived from the CXCR4 + patient sample SP5, was used. To generate this model, 10 mg of SP5 tumor tissue obtained from donor animals was implanted in the mouse subcutis. When tumors reached a volume of ≈500 mm 3 biodistribution assays of T22-GFP-H6 and T22-GFP-H5T nanoparticles were performed at three different times after nanoparticle injection, namely 5, 24, and 48 h. Mice received 100 µg single iv bolus of T22-GFP-H6 (n = 2) or 100 µg single iv bolus of T22-GFP-H5T (n = 2) in sodium carbonate buffer with salt. Control animals (n = 2) were iv administered with 150 µL of the same buffer.
At 5, 24, and 48 h after the iv injection, mice were euthanized and subcutaneous tumors and normal organs, including lung and heart, www.advancedsciencenews.com www.particle-journal.com Part. Part. Syst. Charact. 2019, 36, 1900304 kidney, liver, and bone marrow were collected. Biodistribution of GFP fluorescent nanoparticles was determined measuring ex vivo the fluorescence emitted by tumors and normal organs using the IVIS Spectrum equipment (PerkinElmer Inc, Waltham). The fluorescent signal (FLI) was first digitalized, displayed as a pseudocolor overlay, and expressed as radiant efficiency. FLI values were calculated subtracting the FLI signal from the protein-treated mice by the FLI auto-fluorescent signal of control mice.