Novel strategies for enhanced fluorescence visualization of glioblastoma tumors based on HPMA copolymers conjugated with tumor targeting and/or cell‐penetrating peptides

Nano‐sized polymer systems are often used as carriers for drugs and contrast agents to increase circulation time and solubility and to reduce possible side effects. These nanomedicines usually accumulate in tumor tissue due to the enhanced permeability and retention (EPR) effect. However, a targeting group may be attached to the polymer carrier in addition to the active substance to further increase tumor accumulation and specificity. In this study, the oligopeptide sequence RGD was chosen to target αvβ3 integrins overexpressed in the tumor vasculature and on some tumor cells. A set of polymer conjugates bearing a fluorescent dye and RGD peptide of different structures (linear, cyclic, branched) was prepared for use in tumor diagnosis, with a potential future application in navigated surgery. The accumulation of the most promising candidate, a targeted fluorescent nanoprobe, increased by 35% in glioblastoma tumors compared to the non‐targeted control, which accumulated only due to the EPR effect. However, the administration of a polymer‐bound modified cilengitide as an antiangiogenic treatment did not show a beneficial effect in the suppression of angiogenesis.

formation of metastases. 2 Further investigation of RGD binding to integrins revealed that head-to-tail cyclization produces a more rigid structure that has up to 100-fold higher affinity for certain receptors compared to the linear RGD-based peptides. 3In the last decades, several cyclic RGD-based structures have been described, for example, cilengitide cyclo(RGDf(Me)V) also known as EMD 121974, and iRGD cyclo(CRGDKGPDC) also known as CEND-1.The former peptide entered phase I clinical trials in 2003 as an angiogenesis inhibitor (specifically α V β 3 and α V β 5 receptors) for the treatment of glioblastoma (GBM).Phase II clinical trials showed that cilengitide was well tolerated and exhibited moderate antitumor activity as monotherapy for GBM treatment. 4Unfortunately, phase III clinical trials did not show that cilengitide in combination with temozolomide and radiotherapy significantly prolonged patient survival, so further clinical research was terminated. 5The second iRGD structure was described in 2009 as "a tumor-penetrating peptide," which means that it targets α V β 3 -and/or α V β 5 -positive tumors, followed by neuropilin-1 receptor-mediated uptake. 6The iRGD sequence has been intensively studied over the last 10 years, resulting in phase I clinical trials in 2018 using iRGD in combination with Abraxane and Gemcitabine for the treatment of pancreatic cancer.][9][10] Advanced RGD-based therapies are based on the attachment to the drug-delivery system (DDS), resulting in a cumulative effect of targeting integrin-positive cells (endothelial and/or tumor) [11][12][13] and passive accumulation due to the so-called enhanced permeability and retention (EPR) effect, particularly in highly vascularized tumors.Nanomaterials such as nanoparticles, 14 micelles, 15 or polymer conjugates 16 bearing an anticancer drug and a targeting peptide with the RGD sequence have been evaluated for the treatment of many tumors.][19][20] Polymer conjugates based on N-(2hydroxypropyl)methacrylamide (HPMA) are an important example of the aforementioned DDS.These water-soluble copolymers are biocompatible and well tolerated in humans.Furthermore, when prepared by controlled polymerization techniques, their molecular weight distribution is very narrow and suitable for pre-clinical and clinical investigations since the low chain heterogeneity of the polymer precursor leads to uniform pharmacokinetic behavior.HPMA-based copolymers are not biodegradable but are safely eliminated from the body by glomerular filtration up to a certain molecular weight (up to range 50,000-70,000 g mol −1 ) and hydrodynamic radius (<10 nm). 21ll-penetrating peptides (CPPs) are moieties capable of increasing cellular uptake usually without energy consumption.CPPs are often attached to various cargoes that need to be transported to the extracellular area.Recently, we described the uptake ability of CPPs attached to the HPMA-based polymer chain, 22 which shows that the cellular uptake of fluorescently labeled polymer-peptide conjugates bearing the Tat peptide (GRKKRRQRRR) or the minimal sequence of Penetratin (RRMKWKK) was more efficient compared to the controls.Unfortunately, the lack of CPPs selectivity for tumor cells makes their application in tumor therapy challenging.
In this comparative study, fluorescently labeled HPMAbased copolymers were decorated with different RGDbased structures (linear, cyclic, and branched) to evaluate their integrin-targeting potential in vitro in presence or not of a CPP in order to evaluate the interest of a triple-functional system combining passive accumulation, active targeting, and increased cellular penetration.Finally, the selected systems were also evaluated in vivo in GBM-bearing mice intended for precise tumor boundaries visualization and antiangiogenic tumor treatment.

Synthesis of peptides
All the peptides (Table 1 and Scheme 1) were successfully synthesized using the standard Fmoc strategy achieving 83%-95.2%purity.The yields of the linear peptides ranged between 71.4% and 73.7%, which is a common yield due to the loss of some of the resin substitution during synthesis, especially during peptide isolation and purification.Cyclic peptides were synthesized from linear precursors by head-to-tail cyclization and the cyclization step was repeated until a negative bromophenol blue test was obtained ensuring full cyclization.The yields of the entire seven-step process were relatively low, approximately 30%-40%, which may be due to a reduction in substitution during numerous steps.In the case of Asp containing cyclic peptides cRGDx and C6cilengitide, significantly higher yields and purity were reached when preloaded Fmoc-Asp(Wang resin)-OAll resin was used instead of the TentaGel R PHB resin.Unfortunately, the TentaGel resin had to be first manually loaded with Fmoc-Asp-OAll and the first amino acid substitution was relatively low, causing a very low yield of the entire synthesis.N 3 -PEG x -spacers were introduced to the N-termini of all peptides to reach the peptides with a reactive azide group enabling copper-free click reaction with the polymer precursors and extending the distance in between the peptide and polymer chain.It was hypothesized that this could be a very important feature as the oligopeptide should be more exposed and thus, more effective.The prepared peptides contained the usual impurity, a peptide without an N 3 -PEG x -spacer, but this was not a problem since only the derivative containing the azide group can react with the polymer precursor containing DBCO groups, and the free peptide was removed during the final purification procedure.
A divergent approach was used to synthesize a branched peptide forkRGD12 containing four copies of the RGD motif, as shown in Scheme 3. The lysine-based fork-like structure containing four free amino groups suitable for subsequent formation of the peptide one amino acid at a time was prepared (the purity of the forked structure was 84%) and this approach ensured that the resulting branch contained four copies of the RGD peptide.Therefore, this process is preferable to the convergent approach where a fully synthesized RGD-based peptide is attached to a forklike structure.In that case, however, the RGD peptide must be fully protected.The solubility of the fully protected peptide is very poor and the final product may have branches that contain less than four RGD peptides.The yield of this process was relatively low, approximately 15% due to the resin substitution at the beginning of the process but when the yield was calculated from the resin substitution at the end of the process, it was around 30%.Nevertheless, the yield is reduced during the multi-step process but the purity of forkRGD12 was above 80%.A list of the prepared peptides is provided in Table 1.

Polymer precursors and polymer-peptide conjugates
Linear copolymers were prepared by the controlled reversible addition-fragmentation chain-transfer (RAFT) polymerization yielding polymers with a low dispersity (Ð below 1.2).
HPMA copolymers with the Ma-β-Ala-TT, Prec1, and Prec2 were prepared using the monomer:chain-tranfer agent-:initiator ratios of 400:2:1 and 550:2:1, respectively, to achieve a molar mass and hydrodynamic diameter below the limit for renal elimination (M w approximately 50,000 g mol −1 and D H ≈10 nm). 21The obtained molar mass of both precursors was according to the desired calculated theoretical molar mass: 30,000 g mol −1 for Prec1 and 40,000 g mol −1 for Prec2 (Table 2).Removal of the potentially cytotoxic dithiobenzoate (DTB) groups did not lead to significant changes in molar mass or dispersity.Importantly, the requirements for renal filtration of the polymer carriers were achieved with a sufficient amount of reactive 1,3-thiazolidine-2-thione (TT) groups for the following conjugations.
The prepared polymer precursors were used to synthesize the polymer-peptide conjugates.First, TT groups were partially replaced by DBCO groups, with the amount of DBCO groups being 1.2 equiv.to the molar mass of the peptide that was subsequently attached and ranged between 2 and 9 mol%.High-performance liquid chromatography (HPLC) monitoring showed a shift in the retention time of the polymer peak and a change in its UV-vis spectrum (DBCO absorbs at 291 and 308 nm).Before the reaction with the unprotected azide-containing peptide, the remaining reactive TT groups along the polymer precursor were removed by the addition of 1-aminopropan-2-ol (AMP) to prevent crosslinking of the polymer chains via the amino groups of lysines or guanidine groups of arginines of the RGD peptide.Polymer-peptide conjugates with RGD or/and CPP were prepared by an uncatalyzed strain-promoted click reaction (86%-94% yield).The reaction rate and selectivity of this reaction allowed the unprotected peptide binding specifically to the polymer without any by-products.All potential impurities were removed in the final purification by precipitation and column chromatography.A representative example of conjugate preparation is provided in Scheme 2 and the conjugate characteristics are summarized in Table 3.Neither M w nor Ð could be accurately determined because the polymer conjugates containing the fluorescent dye Dy633 absorb in the near-infrared region that collides with the  wavelength of the light scattering detector laser (658 nm).Nonetheless, size exclusion chromatography (SEC) analysis revealed that attachment of the peptide did not significantly change the molar mass of the polymer-peptide conjugate or broaden the distribution compared to the corresponding precursor.A representative example of refractive index (RI) chromatograms of labeled polymer conjugates compared to Prec1 is shown in Figure S1.The preparation of fluorescently labeled polymer conjugates with indocyanine green azido derivative (ICG) was similar to the synthesis described above with the only difference being the fluorophore attachment method.The azido derivative of fluorescent dye ICG was bound to the polymer precursor by the same click reaction as the peptide.The resulting polymer conjugates (Table 3) showed similar characteristics as the polymer precursors and the polymer conjugates (in terms of SEC measurement shown in Figure S2).
The polymer conjugate P12-CIL intended for antiangiogenic treatment and the control polymer Ref5 were not fluorescently labeled.The hydrodynamic radii of fluorophore-free conjugates were 5.5 ± 0.1 nm for Ref5 and 9.2 ± 0.5 nm for P12-CIL, which correspond to the SEC characterization.Similarly, the molar mass and hydrodynamic radius were slightly increased by the attachment of DBCO groups and C6cilengitide, with the increase in M w being in line with the mass of the attached peptide.

Expression of α v β 3 integrin receptor on glioblastoma cell lines
A positive cell line that overexpresses and a negative cell line that minimally expresses the integrin receptor were selected to compare potential differences in the binding efficacy of the tested polymer-peptide conjugates.Thus, we examined α v β 3 integrin expression on GBM cells, U87-MG and LN-18.Western blot analysis confirmed strong expression of α v and β 3 subunits of the integrin receptor in U87-MG in contrast to LN-18, in which α v and β 3 subunits expression were very low (Figure 1A).However, the β 3 subunit expression in U87-MG cells (Figure S3A) decreased with increasing cell passage (>10) to the expression level in LN-18 cells.The α v subunit expression in U87-MG cells also decreased after the 10th passage but less substantially than β 3 subunit.In agreement with the literature, expression of both integrin receptor subunits was low in the LN-18 cell line; thus, these cells were considered α v β 3 integrin negative. 23Unmodified raw data of Western blot are shown in Figure S3C.
Since the polymer-peptide conjugates are intended to bind to the receptor on the cell surface, we also investigated the expression of α v β 3 integrin receptor on the U87-MG and LN-18 cell surface using an anti-α v β 3 integrin antibody at 4 • C to avoid energy-dependent endocytosis (Figure 1B).The flow cytometric analysis confirmed strong expression of α v β 3 integrin receptor in U87-MG and only modest (similar to control) expression in LN-18 (Figure 1B).Likewise, the Western blot results confirmed that α v β 3 integrin receptor expression decreased with increased passages (>5) of the U87-MG cells (Figure S3B).Based on these results, U87-MG until the fifth passage was used as the α v β 3 integrin receptor-positive cell line and LN-18 was employed as a negative cell line in the following experiments.

Polymer-peptide conjugates efficiently bind and visualize the α v β 3 integrin receptor-positive glioblastoma cells
Importantly, the proper selection of the fluorescent dye is a key parameter for visualization of the interaction of polymer-peptide conjugates with cells.Initially, two fluorescent dyes with similar spectral characteristics but differing in structure and charge were selected.The initial flow cytometry experiments showed strong nonspecific cellular binding of Cy5.5-labeled polymer conjugates after incubation at 4 • C, at which the unwanted energy-dependent endocytosis of conjugates should be blocked.This was studied in detail with both cell lines (U87-MG and LN-18) in a wide range of dye concentrations (Figure 2A,B) and two identical polymers with different Cy5.5 dye content, specifically 0.4 wt% (Ref1-Cy-L) and 1.0 wt% (Ref2-Cy-H), were tested.The third polymer contained the Dy633 dye (conjugate Ref3-Dy with 1.4 wt% of dye). Figure 2A,B shows that polymer-bound Cy5.5 binds nonspecifically to the cellular membranes to a much greater extent than Dy633.In addition, a strong dependence on the increased amount of Cy5.5 content was observed on the nonspecific cellular binding.In contrast, the nonspecific cellular binding of polymer-bound Dy633 was more than threefold lower in both cell lines even though the content of Dy633 was 1.4 times higher than for Cy5.5.It was hypothesized that the aforementioned nonspecific interactions of Cy5.5 with cell membranes are caused by the positive charge in the fluorophore, which is attracted to negatively charged cell membrane components such as the phospholipid bilayer and heparan sulfate.The difference in the strength of this interaction between the two cell lines could be due to the different representations of these negatively charged components in their membrane structures.In contrast, Dy633 did not exhibit this nonspecific interaction because its overall charge is negative; therefore, Dy633 conjugates were used to further investigate RGD-targeted polymerconjugate binding efficacy to exclude the nonspecific binding of the conjugates.Nevertheless, the use of Cy5.5 conjugates is possible in comparative experiments where all the conjugates contain the same label.Indeed, the influence of the dye used should always be investigated to minimize the influence of this labeling on the results.
The cell binding activity of various cyclic RGD-based peptides attached to the polymer backbone on the α v β 3 integrin receptor was evaluated by flow cytometry after 1 h incubation with U87-MG cells at 4 • C (Figure 2C).First, the influence of small structural changes on the receptor binding between P1-cRGD with cyclic RGD peptide (RGDfK), P2-cRGN with Asp replaced with Asn (RGNfK), and P3-CIL with a peptide derived from antiangiogenic drug cilengitide by enrichment with one additional amino acid Lys was evaluated.According to the literature, the affinity of the peptide derived from cilengitide should not be affected by this structural change. 24The P1-cRGDcontaining cycloRGD12 peptide (see Figure 2C) demonstrated the highest binding affinity to integrin receptors and was efficient at a concentration of 83 µM.P1-cRGD showed stronger binding to integrins (1.6×) as compared to P2-cRGN and P3-CIL.At a concentration of 170 µM, there was no further increase in P1-cRGD and P2-cRGN MFI, suggesting that surface receptor saturation probably occurred at 83 µM and a plateau effect was observed at a higher concentration.In contrast, there was a further increase in MFI for P3-CIL suggesting a lower affinity of P3-CIL than P1-cRGD.In summary, replacing aspartic acid with asparagine reduced the binding efficiency of P2-cRGN by approximately 35%, as did the insertion of one extra amino acid to the peptide structure in the P3-CIL conjugate.
Furthermore, we investigated the influence of the linear versus the cyclic forms of RGD, various lengths of the PEG spacer, or fork-like presentation of linear RGDs on cellular binding.After incubation with U87-MG cells (see Figure 2D), the linear version of the RGD peptide did not improve cell binding as compared to the control Ref3-Dy, whereas the P7-fRGD conjugate with the branched RGD peptide containing four copies of a linear RGD presented a 2−2.5-fold increased binding at all tested concentrations.However, the best results were achieved with P5-cRGD4 and P6-cRGD conjugates containing a cyclic RGD, which increased the cellular binding five-to sixfold over Ref3-Dy at 50 and 100 µM (example of histogram is shown in Figure S4A).As expected, the rigid cyclic structure binds to the receptor significantly more than the linear analogs, as observed for the peptides. 25owever, the differences were much smaller when compared to the low-molecular-weight peptides.In contrast, our hypothesis that a branched peptide formed by the assembly of linear copies could induce the formation of integrin clusters, leading to a stronger specific binding, was not as pronounced as expected. 10,26,27This suggests that the branched peptide is actually containing linear copies that do not bind significantly to the receptor.
We hypothesized that longer ethylene glycol linkers would increase the distance between the targeting peptide and the polymer carrier and thus that the targeting peptide should be more "uncovered" and accessible in the polymer coil.This effect has been already observed in our previous work for CPP Tat and minipenetratin. 22Figure 2D shows that the effect of linker length, four unit in P5-cRGD4 and a longer 12 unit in P6-cRGD, on specific binding to the α v β 3 integrin receptor is negligible at all concentrations tested.A slight improvement (about 20%) in the binding of the P6-cRGD conjugate with the longer linker over P5-cRGD4 was observed at the highest concentration of 100 µм but the difference was not statistically significant.
The integrin specificity of the P5-cRGD4 and P6-cRGD conjugates was verified by incubation with α v β 3 integrinnegative LN-18 cells.Figure 2E reveals that the binding of the P5-cRGD4 and P6-cRGD conjugates to LN-18 was significantly lower compared to the positive cell line U87-MG (Figures 2D and S4B).The absolute MFI values increased with concentration, indicating a low level of specific binding to the receptor.
Adhesion inhibition analysis, also called competition analysis, is used to evaluate the binding efficiency of substances with the RGD motif 16,28 using integrin-positive cells (here hβ3 cells) as described in Duret et al. 8 The method assumes that cells added to a vitronectin-coated surface will attach via integrins but if an integrin antagonist is added to the cells, binding to the vitronectin surface does not occur.Thus, a potential integrin antagonist was tested over a wide range of concentrations (0.1-10 µм of cRGD), showing the statistically significant concentration dependence of adhesion inhibition of polymer conjugate P11-cRGD, with an IC 50 of 7.9 µм (Figure 2F), which is in line with published data, the IC 50 of polymer-bound cyclic RGD or cilengitide is reported to be around 10 µм. 28The negative control polymer Ref4-ICG, containing the same concentration of ICG did not show a significant inhibitory effect over the entire range tested.

Cooperative effect of RGD-based targeting and Tat-mediated cell penetration
Polymer nanomedicines must also be efficiently internalized in cells, which could be achieved by the attachment of a CPP to a polymer chain.Recently, 70-fold improved cellular internalization was described using the Tat peptide 22 ; however, this was counterbalanced by a lack of selectivity toward tumor cells.][31] The cellular labeling with bifunctional conjugates containing both targeting and CPP peptides on the same polymer backbone was analyzed by flow cytometry after incubation (1 h at 4 • C) with α v β 3 -positive U87-MG and α v β 3 -negative LN-18 cells, respectively.The effect of Tat on cell binding and/or internalization (Figures 3A and  S5) is very important and observed as well on the two cell lines.In comparison, the contribution of α v β 3 integrin binding is sevenfold lower.The impact of the Tat peptide is observed at the three tested concentrations (10, 50, and 100 µм).However, at 10 µм, the difference between the RGD-Tat (P9-cRGD-Tat and P10-fRGD-Tat) and Tat only (P8-Tat) was significantly more elevated (p < .001)and probably due to the concentration dependence of the combined effects of RGD-specific binding plus the nonspecific Tat-binding effect since the internalization step should be reduced or abrogated at 4 • C, the condition used in this experiment.This is confirmed when using the LN-18 integrin-negative cell line since no differences are observed between the P8-Tat and P9-cRGD-Tat conjugates at any of the concentrations tested (Figure 3B).
In summary, the bifunctional conjugate containing the Tat and RGD-based peptides exhibits a partial cooperative effect on cell binding.Importantly, this effect is observed only at the low concentrations and thus could be interesting for in vivo targeting of the tumor tissue.It is of note that the in vitro experiments were not performed under flow-through conditions, but in a closed system that favors the effect of cell penetration over integrin targeting.We hypothesize that following in vivo application, the targeting effect of RGD would be more important to ensure the localization of the nanomedicine in the proximity of the cell membrane where the following CPP effect could occur.

In vivo tumor accumulation and biodistribution
Since the polymer conjugate bearing a cyclic form of the RGD peptide exhibited the most promising targeting results in vitro, the conjugate P11-cRGD was evaluated in vivo in subcutaneous (s.c.) U87-MG tumors compared to a control conjugate Ref4-ICG.Conjugates containing the fluorophore ICG were administered intravenous (i.v.) into the tail vein at a concentration of 10 or 100 µM of the ICG dye (corresponding to 11.9 and 119 µM peptide, respectively).The fluorescent signals from the tumor and the skin corresponding to the same area on the other side of the mice were acquired at different time intervals for 48 h using a Pearl Trilogy small animal imaging system at a wavelength of 800 nm, followed by fluorescence quantification.Significant differences in the tumor-to-skin (T/S) ratio between the actively targeted P11-cRGD and the non-targeted control conjugate Ref4-ICG over time were observed at 10 µM ICG (see Figure 4A).From 1 h after administration until the end of the experiment at 48 h, the difference between the conjugates was significant (p < .001).To better distinguish the differences between the conjugates, the data were plotted as the percentage of Ref4-ICG contributing to the EPR effect (Figure 4B,C, green section), and the additional contribution of RGDmediated active targeting to integrins (Figure 4B,C, blue section).
The data imply that it is advantageous to administer a lower dose of the conjugate (10 µм related to ICG) because the targeting effect was more pronounced.At a higher dose of 100 µм ICG, the EPR effect was more prominent and the effect of the targeting peptide was masked.The best tumor visualization effect mediated by active targeting was achieved in units of hours from administration, the maximum was reached at 5 h when the targeting effect increased tumor accumulation by 35% compared to the non-targeted control.After prolonged circulation in the bloodstream, the influence of passive accumulation caused by the EPR effect gradually increased, so there was a gradual increase in the accumulation of the untargeted polymer conjugate.This effect was more pronounced a long time after administration, as evidenced by no observable difference between the targeted and the control polymer conjugate at 72 h after administration (data not shown).The importance of active targeting is, therefore, more advantageous in the short time after administration.It should be noted that the accumulation and especially the retention of polymer-bound ICG (even without a targeting moiety) exceeds the accumulation of low-molecular-weight ICG, which is the standard clinically approved visualization probe for fluorescence-guided surgical procedures.The half-life of the low-molecularweight ICG after i.v.administration in rats was 80 min. 29igure 4E shows the contrast between healthy tissue and tumor in fluorescence imaging (ICG fluorescence shown in green over the image of a grayscale mouse) 48 h after administration of the targeted P11-cRGD conjugate and control Ref4-ICG at 10 µм, reaching a T/S ratio of 3.5 and 2.6, respectively, at 10 µм at 48 h with the highest T/S of 3.7 and 2.8, respectively, reached 24 h after administration.
The experiment was terminated after 48 h, the animals were euthanized and the fluorophore content in individual organs was determined ex vivo (Figure S6).At that time, the tumors had a volume of 0.41 ± 0.22 cm 3 .The strong fluorescence was observed in tumor, skin, spleen, kidneys, and liver.Overall, increased accumulation of fluorescently labeled polymer conjugates in the kidney and liver is a commonly observed phenomenon as their excretion occurs primarily via the renal and secondarily via the hepatic route. 30he ex vivo tumor-to-tissue ratios shown in Figure 5A,B showed important results in the tumor-to-fat (T/F) and tumor-to-muscle (T/M) ratios, which are most important parameters for the design and development of probes for optically guided surgery.During surgery, the autofluorescent skin is removed and the tissue surrounding the tumor is usually fat and muscle.The T/F and T/M ratios remained very high even after 48 h after administration (up to 18.2 and 12.4, respectively, for 10 µM, and 13.1 and 10.1, respectively, for 100 µM).Moreover, these ratios were higher for the targeted probe P11-cRGD compared to the control Ref4-ICG demonstrating the benefit of the RGD-based targeting for a potential future use in optically guided surgery in both administered concentrations.On contrary, the T/S ratios were generally lower (not exceeding 2 for both administered concentrations) and were not significantly different for targeted and non-targeted probes, since the effect is faded due to the autofluorescence of the skin (as visible in Figure S6).The difference between both administered concentrations in vivo for targeted and non-targeted polymer, was no longer observed ex vivo, probably due to the removal of the highly fluorescent skin, since the tumor fluorescent signal is no longer burdened by the signal coming from the skin.Therefore, such targeted polymer probe does not have real benefit for conventional fluorescence imaging in vivo, but by our opinion would be more suitable as probe for navigated surgery where skin is removed and does not further affect the visualization efficiency.
The s.c.U87-MG tumor model does not correspond fully to human GBM tumors.Indeed, U87-MG presents a very strong EPR effect due to its rapid growth as well s.c.localization than after orthotopic brain implantation. 31Because of this strong EPR effect, the RGD-based active targeting, which comes in addition to the passive ERP-based effect could be partly masked.

Antiangiogenic therapy using a polymer-bound Cilengitide derivative
The possible antiangiogenic activity of RGD-based peptides was investigated in mice bearing s.c.U87-MG tumors.Indeed, it has been shown that RGD targeting of α v β 3 receptors can induce apoptosis of endothelial cells. 25More recently, the antiangiogenic drug Cilengitide has been clinically evaluated as a potential angiogenic inhibitor in the treatment of GBMs and other solid tumors.Herein, Cilengitide was enriched with lysine and an N 3 -PEG 12 linker for the attachment to a polymer carrier to form the peptide derivative C6cilengitide (P12-CIL) and a pHPMA homopolymer (Ref5) was prepared as a control.
Mice were treated according to the dosing schedule (shown by the purple arrows in Figure 6) with 0.19 mg of the conjugate P12-CIL administered per dose (total of 10 doses), corresponding to a concentration of 24.2 µм C6cilengitide.Polymer conjugates (in PBS or plain PBS as negative control) were administered into the tail vein of mice from the fourth day after inoculation of U87-MG tumors and the size of tumors was recorded every second day (Figure 6A), showing that the administration of antiangiogenic treatment did not affect tumor growth in the treated group (green curve) compared to the controls.The tumors were subjected to photoacoustic measurements (raw data shown in Figure S7) on day 26 to determine the tumor hemoglobin level and oxygen saturation (Figure 6B,C).It was hypothesized that the values of these two parameters would be reduced in response to the antiangiogenic activity of the P12-CIL treatment.As can be seen, no particular effect of P12-CIL treatment could be detected.
We hypothesize that the main reason for ineffective treatment could be escribed to the low effectivity of cilengitide itself, which was found in the unsuccessful international randomized phase II clinical trial of cilengitide as an antiangiogenic agent despite promising phase I and II trials of cilengitide in combination with temozolomide and radiotherapy for the treatment of GBMs. 5

CONCLUSIONS
Actively targeted RGD-based conjugates targeting the α v β 3 integrin receptor were successfully prepared and biologically evaluated.Receptor binding was dependent on the RGD peptide structure with the cyclic form being the most effective.The cRGD-bearing conjugate specifically blocked integrins in the competitive assay to a high extent.The cRGD-polymer conjugates could potentially serve to visualize tumor vasculature and α v β 3 integrin-positive tumors and be used as a nanotheranostic if the fluorescent dye was replaced by a suitable photosensitizer.Unfortunately, the administration of an antiangiogenic polymer therapeutic containing a cilengitide-derived peptide did not slow tumor progression possibly due to the overall low antiangiogenic efficacy of cilengitide, which led to the termination of its clinical trials.3-Amino-1-(2-azatricyclo[10.4.0.0 4,9 ]hexadeca-1( 16),4,6,8,12,14-hexaen-10-yn-2-yl)propan-1-one (DBCO-NH 2 ) was purchased from Click Chemistry tools.Methanol, acetonitrile, dichloromethane (DCM) and all other organic solvents were purchased from VWR International s. r. o.All chemicals and solvents were analytical grade.The solvents were dried and purified by conventional procedures.

Physicochemical characterization
Molecular weight averages and dispersity of all prepared polymers and conjugates were determined by SEC using an HPLC system (Shimadzu) equipped with a TSK3000SWXL column (Tosoh Bioscience) and multiangle light scattering DAWN 8 EOS (LS, Wyatt Technology Corp.), RI, and UV-vis (Shimadzu) detectors.Measurements were performed using a mobile phase composed of 80 vol% of methanol and 20 vol% of acetate buffer (0.3 M; pH 6.5) with a flow rate of 0.5 mL min −1 and RI increment dn/dc = 0.167 mL g −1 .
Peptide purity measurements and monitoring of the reaction course were determined by HPLC (Shimadzu) equipped with a Chromolith Performance RP-18e column (100 × 4.6 mm; Merck).The mobile phase with a linear gradient was composed of a water−acetonitrile mixture (5−95 vol% of acetonitrile) with 0.1 vol% TFA at a flow rate of 5 mL min −1 in 5 min.Detection was performed using a fluorescent RF-10AXL and a UV−vis photodiode array detector (Shimadzu).
The content of peptides attached along the polymer chain was determined by amino acid analysis.Samples were first hydrolyzed (6 M HCl, 16 h, 115 • C in sealed ampoule) and then derivatized with OPA and SH-acid in borate buffer (0.65 M), followed by HPLC analysis of the fluorescent derivative excited at 229 nm and detected at 450 nm using a fluorescent detector.A linear gradient of 10%−100% of buffer B in 35 min and a flow rate of 1 mL min −1 was used, where buffer A was 97.5 vol% sodium acetate buffer (0.05 M) (pH 6.0) and 2.5 vol% acetonitrile; buffer B was 70 vol% MeOH and 30 vol% sodium acetate buffer (0.17 M) (pH 6.0).
Molecular weights of the prepared peptides and peptide derivatives were verified by Matrix Assisted Laser Desorption-Ionization with Time-of-flight mass spectroscopy (MALDI-TOF MS) on a Bruker Brifex III mass spectrometer.
Hydrodynamic diameter (D h ) measurements of polymer precursors and conjugates without fluorescent dye were performed using a Nano-ZS instrument (ZEN3600, Malvern).Samples were dissolved in PBS buffer (concentration 5 mg mL −1 ) and filtered through a 0.45 µm polyvinylidene fluoride (PVDF) filter, and then the intensity of the scattered light was detected at an angle θ = 173 • with a laser wavelength of 632.8 nm.Values were determined as a mean ± standard deviation (SD) of at least five independent measurements.

Synthesis of polymer precursor poly(HPMA-co-Ma-β-Ala-TT)
A biocompatible polymer precursor with a narrow distribution of molecular weight was prepared by controlled radical polymerization using the RAFT technique previously described. 33Briefly, monomers HPMA (838 mg; 5.85 mmol) and Ma-β-Ala-TT (168 mg; 0.65 mmol) were dissolved in a mixture of t-BuOH (8.36 mL; 85 vol%) and DMA (15 vol%) with chain transfer agent DTB-AIBN (5.23 mg; 23.6 µmol) and initiator V-70 (3.64 mg; 11.8 µmol).The ampule with the polymerization mixture was bubbled with argon, sealed, and reacted in a water bath for 16 h at 40 • C. The polymer was isolated by repeated precipitation to the mixture of acetone and diethylether (1:1, v/v) and dried under a vacuum.The polymer was then dissolved in DMA (5 mL) with AIBN (153 mg; 0.93 mmol) and heated in a sealed ampoule under an argon atmosphere at 80 • C. 34 The final polymer precursor Prec1 (660 mg; 65.6%) was precipitated and characterized by SEC.The content of reactive TT groups was determined spectrophotometrically.The polymer precursor Prec2 was prepared analogously and a summary of the results is provided in Table 2.

4.5
Synthesis of peptides

Linear peptides
All linear peptides and linear precursors for cyclization were prepared by microwave-assisted solid-phase peptide synthesis (SPPS) using a microwave peptide synthesizer Liberty Blue (CEM) starting from the C-terminus using standard Fmoc procedures with the consecutive addition of the N-Fmoc-protected amino acid derivative (Fmoc-AA; 2.5 equiv.),DIC (2.5 equiv.)as an activator, and Oxyma (2.5 equiv.)as an activator base in fresh DMF.TentaGel R RAM and TentaGel R PHB were used as solid support for SPPS.Preloaded Fmoc-Asp-(Wang resin)-OAll was used to prepare cRGDx and C6cilengitide cyclic peptides.After the last Fmoc group removal, the final attachment of N 3 -PEG 4 -COOH or N 3 -PEG 12 -COOH (3 equiv.) to form the azide-peptide derivatives was performed using PyBOP (3 equiv.),HOBt (3 equiv.), and DIPEA (6 equiv.)for 3 h.The final peptide derivative was cleaved from the resin using the mixture of TFA/TIPS/H 2 O (95/2.5/2.5, v/v/v) for 3 h and obtained after solvent evaporation, precipitation to a diethylether, and drying.The prepared peptide derivatives are listed in Table 1.

Cyclic peptides
All cyclic peptides were prepared from linear precursors by the same procedure as described for linear peptides using TentaGel R RAM and TentaGel R PHB, including Fmoc-Lys(Mtt)-OH for a side reaction after the cyclization.The head-to-tail cyclization was performed as described earlier. 35Subsequently, the protecting Mtt group was selectively cleaved from Lys by the reaction with TIPS/TFE/HFIPA/DCM (0.5/1/2/6.5, v/v/v/v) mixture for 1 h (repeated twice).Finally, spacer N 3 -PEG 4 -COOH or N 3 -PEG 12 -COOH was attached (as described above) and the peptide derivative was cleaved from the resin by the same procedure as described for linear peptides.Prepared cyclic peptides are listed in Table 1 and their structures are shown in Scheme 1.
The success of the condensation was qualitatively verified by the bromophenol blue test 36 and the purity of the intermediates was monitored by HPLC (sample obtained by cleaving a small amount of peptide from the resin).

Synthesis of polymer-peptide conjugates via the "click" reaction
Polymer conjugates bearing fluorescent dye and targeting RGD-based peptide azido derivative intended for in vitro evaluation were prepared by a three-step reaction similar previously described. 37In the first step, the polymer precursor Prec1 (80 mg; 55.9 µmol TT) was dissolved in DMA to form a 10% solution.DBCO-NH 2 (11.34 mg; 41 µmol) was dissolved in DMA and mixed with the polymer solution with 7 µL of DIPEA (41 µmol).The course of the reaction was monitored by HPLC.After the complete attachment of DBCO-NH 2 , the fluorescent dye Cy5.5-NH 2 (0.8 mg; 1.06 µmol) and Dy633-NH 2 (1.6 mg; 2.07 µmol) dissolved in 20 and 40 µL DMA with 0.2 and 0.36 µL DIPEA, respectively, were added.Finally, AMP (4.3 µL; 55.9 µmol) was added to remove unreacted TT groups and the resulting polymer precursor was precipitated into diethyl ether and dried.In the next step, the strain-promoted azide-alkyne cycloaddition reaction of peptide azido derivatives to the polymer precursor took place, either in DMA or in water depending on the solubility of the corresponding peptides.The resulting theoretically bound amount of pure peptide (without linker) was set to 15 wt% and the course of peptide attachment was monitored using HPLC.The resulting fluorescently labeled polymer-peptide conjugates precipitated using a mixture of acetone:diethyl ether (1:1, v/v) and dried.The conjugates were then dissolved in water and purified by column chromatography using Sephadex G-25 (PD 10 columns, GE Health care) before freeze drying.A similar synthesis was performed to prepare all conjugates listed in Table 3.
Polymer conjugates containing ICG were prepared in the same way, except that the peptide derivative and azido-dye were both attached via the strain-promoted azide-alkyne cycloaddition ("click") reaction to a precursor containing DBCO groups.The polymer conjugate P11-cRGD was also prepared similarly, only excluding the fluorescent dye addition step.

Cell lines
The human GBM U87-MG and LN-18 cell lines were purchased from American Type Culture Collection through an authorized distributor LGC Standards Sp. z.o.o.The cells were cultured in Eagle's minimal essential medium (Sigma-Aldrich/Merck) or Dulbecco's modified Eagle medium (DMEM, Thermo Fisher Scientific) supplemented with 1% penicillin-streptomycin and 10% fetal bovine serum (FBS).The human embryonic kidney HEK293T cell line and hβ3 variant transfected to strongly express the human β 3 integrin subunit were used as described previously. 38,39The modified cell line (hβ3) was cultured in DMEM supplemented with 1% glutamine, 10% FBS, and 700 µg mL −1 Geneticin (G418 sulfate; Gibco).All cells were cultivated in a humidified incubator at 37 • C with 5% CO 2 .

Quantification of α V β 3 integrin expression
Harvested U87-MG and LN-18 cells were incubated with mouse monoclonal anti-α V β 3 primary antibody (LM-609; Abcam) for 1 h at 4 • C, then washed three times with ice-cold PBS with 3% bovine serum albumin (BSA) before incubation with AlexaFluor488-conjugated goat anti-mouse secondary antibody (Thermo Fisher Scientific) for 1 h at 4 • C in the dark.After three washes with ice-cold PBS containing 3% BSA, cells were stained with SYTOX Blue Dead Cell Stain (Thermo Fisher Scientific) for 5 min to distinguish between live and dead cells.

4.8.2
Quantification of conjugate binding to integrin U87-MG and LN-18 cells were incubated with polymer-peptide conjugates for 1 h at 4 • C in the dark, washed three times with ice-cold PBS with 0.5% BSA, then stained with SYTOX Blue Dead Cell Stain.The median fluorescence intensity (MFI) of the live cells was determined by relating all targeted polymer-peptide conjugate values to the appropriate control conjugate Ref#.In the case of cooperation effect experiments, the amount of P8-Tat, a Tat peptide-bearing conjugate was calculated to match the amount of Tat peptide in the bifunctional conjugates at the given concentrations.The corresponding molar concentrations of the Tat peptide tested were 8, 40, and 80 µM.All samples were measured using a BD FACSVerse flow cytometer (Becton Dickinson Czechia, s.r.o.) and analyzed using FlowJo software version 10 (Tree Star Inc.).
The results were plotted as mean ± SD of at least three experiments performed in duplicate.

Adhesion inhibition assay
The ability of the RGD-based polymer-peptide conjugate P11-cRGD to inhibit the adhesion of hβ3 cells expressing increased α v β 3 integrin to a surface coated with a layer of vitronectin was determined using an adhesion assay as described previously. 9,40Briefly, a 96-well plate was coated with vitronectin (5 µg mL −1 in 50 µL/well) for 1 h at room temperature, then blocked with 3% BSA in PBS (50 µL/well) at room temperature.A dilution series of P11-cRGD (0.1−10 µм) and Ref4-ICG (7.75−775 µg mL −1 corresponding to the concentration of ICG in the P11-cRGD conjugate) was added to the wells (20 µL) and PBS was added to the control wells.Trypsin-detached hβ3 cells were seeded at a concentration of 25,000 cells/well in 180 µL of medium and incubated for 30 min at 37 • C in a humidified incubator with 5% CO 2 .Unattached cells were gently washed away with PBS (3×) and the attached cells were fixed with ethanol (200 µL/well) for 15 min before staining with methylene blue (50 µL/well).After drying, the absorption was measured at 620 nm using a spectrophotometer CLARIOstar Plus (BMG LABTECH) equipped with a 96-well plate reader.Cells incubated with PBS buffer were used as a control, with the average absorbance taken as 100% hβ3 cell attachment.Two independent experiments were performed in triplicate.

Mouse model
The animal experiments and procedures complied with ethical rules for in vivo testing in mouse models and were approved by the relevant authority following French law to minimize animal suffering and the number of animals used.All animal experiments were approved by the Animal Ethics Committee of the French Ministry, under the agreement number APAFIS#8868-2015093015035547-V8, and were performed according to the Institutional Animal Care and Use Committee of Grenoble Alpes University.The experiments were performed within OPTIMAL Grenoble, the Small Animal Imaging Platform of the Institute for Advanced Biosciences.All animal manipulations were performed with sterile techniques.U87-MG cells were grown as described and trypsin-detached (0.05%) cells were washed and resuspended in PBS at a concentration of 5 × 10 6 per 100 µL.Six-week-old immunodeficient female nude NMRI mice (Janvier, Le Genest) were anesthetized (air/isoflurane 4% for induction and 1.5% thereafter) and the cell suspension was s.c.injected into the right thigh.

In vivo imaging and biodistribution
Twenty-one days after inoculation, when the tumor size reached an average volume of 400 mm 3 , mice were randomly divided into four groups of three mice.P11-cRGD and Ref4-ICG samples were dissolved in PBS at a total concentration of ICG 10 and 100 µM, respectively (corresponding to a concentration of 60 or 600 µM cRGD12 peptide).Polymer conjugates were administered i.v.into the tail vein in a volume of 200 µL per mouse.Near-infrared fluorescence of anesthetized mice was measured using a Pearl Trilogy Small Animal Imaging System, LI-COR Biosciences channel 800 nm (excitation 785 nm, emission 820 nm) at different time points.After 48 h, the mice were euthanized and images of selected organs and tumors were taken for ex vivo fluorescence quantification using Image Studio Lite.

Antiangiogenic therapy of U87-MG tumor-bearing mice
Five days after inoculation, the mice were randomly divided into three groups of six mice and administrated a different treatment, specifically, 200 µL of solution (in total 10 doses) i.v. to the tail vein three times per week.The first group was administered a solution of the polymer conjugate P12-CIL (active component C6cilengitide) at a concentration of 0.95 mg mL −1 , which corresponded to a concentration of 24.2 µM peptide in one dose.The second group was administered the control polymer Ref5 and the third group was given a placebo, that is, 200 µL of phosphate-buffered saline (PBS).Mouse tumor sizes were measured using a caliper throughout the experiment.Moreover, at the end of the experiment (day 25), mice were subjected to photoacoustic measurement to determine the tumor O 2 saturation and hemoglobin concentration using a Vevo LAZR device Fujifilm, Visualsonics Inc.Total hemoglobin content and oxygen saturation were calculated from oxy-hemo data (Vevo Lab software) as previously described. 41

Statistical analysis
The data are plotted as mean ± SD and analyzed using two-way analysis of variance followed by Bonferroni's test (***p < .001,**p < 0.01, and *p < 0.05).
The half maximal inhibitory concentration (IC 50 ; the concentration of the RGD peptide reducing the cell adhesion ability to a half) was obtained from calculations corresponding to the fitting of the dose-response curve by nonlinear regression.All statistical analyses were performed using GraphPad Prism 5.03 (GraphPad Software Inc.).

A C K N O W L E D G M E N T S
This work was supported by the Czech Science Foundation (project 22-12483S), the Czech Health Research Council (project NU21-08-00280), and the project National Institute for Cancer Research (program EXCELES, ID project no.LX22NPO5102) funded by the European Union, Next Generation EU.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare no conflict of interest.

TA B L E 2 a
List of prepared polymer precursors.Molecular weight (M w ) and distribution (Ð) were determined by size-exclusion chromatography using refractive index (RI) and LS detection.b Hydrodynamic diameter (D H ) measurement was performed using a Nano-ZS instrument in PBS buffer at 25 • C.c  The content of 1,3-thiazolidine-2-thione (TT) groups was determined by UV-vis spectrophotometry in MeOH (ε 305 = 10,300 L mol −1 cm −1 ).

F I G U R E 1
Quantification of α v β 3 integrin expression in U87-MG and LN-18 glioblastoma cells.(A) Representative Western blots of the α v and β 3 subunits of the integrin receptor in low passage (PSG ≤ 10) of U87-MG and LN-18 cells.β-Tubulin was used as a loading control.(B) Representative histograms of α v β 3 integrin expression in LN-18 (blue) and U87-MG (orange) cells.Control-the cells non-subjected to α v β 3 integrin antibody (red).

F I G U R E 2
In vitro evaluation of RGD-based conjugates.Comparison of reference polymer conjugates bearing either Cyanine 5.5 (Ref1-Cy-L, Ref2-Cy-H) or Dyomics 633 (Ref3-Dy) fluorescent dyes in a concentration range of 0.1-10 µg mL −1 by flow cytometry (represented by median fluorescence intensity [MFI]) after incubation with (A) U87-MG and (B) LN-18 cells at 4 • C. The binding of polymer-peptide conjugates was measured by flow cytometry after incubation with (C and D) U87-MG cells and (E) LN-18 cells for 1 h at 4 • C. (F) hβ3 adhesion assay showing polymer conjugates P11-cRGD (pink squares) and the control Ref4-ICG (green squares) in a wide range of concentrations (here shown as the logarithm of concentration) incubated for 30 min at 37 • C. ***p < .001,**p < .01,and *p < .05.ICG, indocyanine green azido derivative.

F I G U R E 3
In vitro evaluation of the potential cooperation effect of targeting RGD-based and cell-penetrating peptides attached to one polymer backbone assessed by flow cytometry after incubation with (A) U87-MG and (B) LN-18 cells for 1 h at 4 • C. The rate of internalization is expressed on a relative scale to the control Ref3-Dy conjugate (gray bar) and ***p < .001.

F
I G U R E 4 (A) In vivo fluorescence of the tumor-to-skin ratio over time after injection of the fluorescently labeled polymer conjugates P11-cRGD (blue columns, n = 3) and Ref4-ICG (cyan columns, n = 3) at a concentration of 10 µм related to dye indocyanine green azido derivative (ICG).Percentage ratios of RGD-mediated active targeting (blue) and enhanced permeability and retention (EPR) influence (green) at (B) 10 µм and (C) 100 µм ICG.The number shown represents the % proportion assigned to active targeting and ***p < .001.(D) Representative example of the fluorescence (shown in pseudo color) distribution in tumor-bearing mice over time after i.v.administration of P11-cRGD and Ref4-ICG at 10 µм.(E) Tumor visualization (green fluorescent image was overlaid on grayscale mouse image) 48 h after i.v.administration of P11-cRGD and Ref4-ICG at 10 µм.

F
I G U R E 6 (A) The size of murine subcutaneous U87-MG tumors (average of six per group) and the treatment dosing schedule (purple arrows) with the P12-CIL (green curve), the control group the Ref5 (pink curve), and the untreated NT group received PBS (gray curve).The polymer conjugates were administered at a concentration of 0.19 mg per dose.U87-MG tumor photoacoustic measurements of (B) oxygen saturation and (C) hemoglobin concentration on day 26 of the treatment with P12-CIL (green), Ref5 (pink), and NT control (gray).

S C H E M E 2
Synthesis of polymer precursors Prec1 and Prec2, and polymer conjugates bearing RGD-based and/or cell-penetrating peptides.
List of prepared RGD-based and cell-penetrating peptide derivatives.
TA B L E 1

TA B L E 3 List and characterization of prepared polymer conjugates. Sample Name of the peptide M w a (g mol −1 ) Ð a Fluorescent dye Content of dye (wt%) Content of peptide e (wt%)
e Peptide content determined by an amino-acid analysis.