Robust Chemical Strategy for Stably Labeling Polyester-Based Nanoparticles with BODIPY Fluorophores

Abstract

Aliphatic polyesters are among materials most extensively used for producing biodegradable polymeric nanoparticles currently in development as delivery carriers and imaging agents for a range of biomedical applications. Their clinical translation requires robust particle labeling methodologies that allow reliably monitoring the fate of these formulations in complex biological environments. In the present study, a practical and versatile synthetic strategy providing conjugates of poly(D,L-lactide) representative of this class of polymers with BODIPY fluorophores varying in functional groups and excitation/emission maxima was investigated as a tool for making traceable nanoparticles. Polymer-probe conjugation was accomplished by carbodiimide-induced and 4-(dimethylamino)pyridinium 4-toluenesulfonate-catalyzed esterification of the polymer’s terminal hydroxyl group, either directly with a carboxy-functionalized fluorophore or with amine-protected amino acids (Boc-glycine or Boc-6-aminohexanoic acid). In the latter case, the amino acid-derivatized polymeric precursors were reacted with amine-reactive BODIPY dyes after the removal of the protective group. Unlike nanoparticles encapsulating a strongly hydrophobic BODIPY_505/515 (logP_o/w = 4.3), nanoparticles labeled covalently with its carboxy-functionalized analogue (BODIPY FL) demonstrated stable particle-tracer association under perfect sink conditions. Furthermore, in contrast to the encapsulated dye rapidly partitioning from particles onto cell membranes but not stably retained by cultured cells, the internalization of the covalently attached probe was an irreversible process requiring the presence of serum, consistent with active nanoparticle uptake by endocytosis. In conclusion, the conjugation of particle-forming polymers with BODIPY fluorophores offers an effective and accessible labeling strategy for making traceable polyester-based biodegradable nanoparticles and is expected to facilitate their development and optimization as therapeutic carriers and diagnostic agents.

Graphical Abstract

INTRODUCTION

Biodegradable nanoparticle (NP) formulations are being explored for a wide range of biomedical applications, including the delivery of different classes of therapeutic agents, cell therapy, and imaging.^1,2 The biodistribution and fate of NPs after administration are essential determinants of both safety and efficiency of their use as therapeutic carriers or imaging agents and require sensitive and reliable tools for their monitoring and optimization. The non-toxic and biocompatible polymers of the aliphatic polyester family, including polylactide (PLA), polyglycolide, polycaprolactone, and their co-polymers, have been used most extensively as matrix-forming materials for making biodegradable NPs.³ However, despite a formidable body of research conducted on biodegradable polyester-based particles over the last several decades, monitoring their fate in vitro or in vivo still remains a challenge due to the lack of practical and effective techniques for creating traceable formulations. The production of traceable NPs would require the development of new labeling approaches enabling sensitive and specific detection in different environments using probe compounds compatible with the existing imaging and analytical methods. In addition to the obvious requirements for detection specificity and sensitivity, an ideal approach should (1) provide a stable particle-probe association in order to avoid artifacts related to premature probe release and redistribution and (2) should have no (or minimal) effects on the physicochemical properties of the particles (e.g., colloidal stability, size distribution, degradability, etc.), as well as their biocompatibility.

Fluorophores of the BODIPY (boron dipyrromethene-derived) family exhibiting a range of emission maxima starting from ca. 500 nm and extending to the far-red region are particularly attractive probes for NP labeling applications due to their remarkable photochemical stability, fluorescence minimally affected by the surrounding medium, high absorption coefficients, and fluorescence quantum yields.^4–6 Select BODIPY derivatives are available commercially as non-functionalized small molecules amenable to encapsulation in polyester-based NPs due to their high organophilicity and a virtual lack of water solubility, or in chemically functionalized forms suitable for covalent attachment. Although the encapsulation of hydrophobic small-molecule fluorophores in NPs is technically straightforward and has been reported with a number of different fluorescent molecules,^7–10 the covalent labeling of aliphatic polyester-based particle formulations poses a formidable challenge because the labile ester bonds in the backbone of the particle-forming polymer are not compatible with harsh reaction conditions.¹¹ In addition, biodegradable polyesters employed for biomedical applications usually have only one functional group per molecule available for chemical modifications, that is, the terminal hydroxyl (terminal carboxylic groups are typically end-capped for preventing autocatalytic degradation and improving polymer stability¹²), which both limits the labeling capacity and further narrows the choice of chemical strategies suitable for probe conjugation.

The modification of hydroxyl-terminated polyesters with probes containing a free carboxylic group, such as BODIPY FL propionic acid, can potentially be achieved via direct carbodiimide-induced coupling. The feasibility of this approach is supported by the successful synthesis of BODIPY FL conjugates using mild reaction conditions reported previously.¹³ However, a number of BODIPY fluorophores are available commercially only as N-succinimidyl ester (SE) derivatives rather than carboxylic acids, and are, therefore, not suitable for direct coupling. Using the SE-functionalized BODIPY dyes for polyester labeling requires the preparation of an intermediate polymer derivative containing an amino group covalently bound to the macromolecule. Such an intermediate could be further reacted with a SE-functionalized fluorophore resulting in the covalent attachment of the probe to the particle-forming polymer. In the present study, we evaluated a strategy of preparing such amino-terminated PLA intermediates by linking N-protected amino acids to the terminal hydroxyl of PLA under mild conditions with subsequent deprotection of the covalently attached amino group. Next, we examined the feasibility of using these chemically activated constructs to prepare PLA conjugates with BODIPY fluorophores exhibiting emission maxima in the orange-red and red regions (BODIPY_558/568 and BODIPY_630/650-X, respectively), which, together with the green fluorescent PLA-BODIPY FL conjugate prepared using the direct coupling approach, can be used as NP labeling materials for a variety of imaging and quantitative analysis applications in vitro and in vivo.^14–19 In proof-of-concept experiments, we used the PLA-BODIPY FL conjugate to make covalently labeled fluorescent NPs and examined the stability of the NP-probe association in comparison with non-covalently labeled control NPs prepared with the incorporation of a highly lipophilic analogue of BODIPY FL lacking the propanoate substituent, BODIPY_505/515,²⁰ by measuring the release of the respective probes under perfect sink and non-sink conditions. These studies were extended to comparatively investigate the cell uptake, distribution, and retention patterns of the NP-associated BODIPY probes as a function of the tracer incorporation mode (covalently linked vs. entrapped), exposure duration, and serum presence.

RESULTS AND DISCUSSION

Synthesis and Characterization of PLA-BODIPY Conjugates.

In this study, we focused on establishing and comparatively evaluating a robust chemical approach for producing stably labeled NP-forming polymers with BODIPY derivatives covering a broad range of excitation/emission maxima as starting materials. The feasibility of a simple one-step process that involves direct coupling with carboxylated BODIPY fluorophores was first investigated using BODIPY FL acid as an example. Despite the relatively low molar concentrations of the carboxylic reactant and hydroxyl groups available for modification, the direct binding of BODIPY FL acid to PLA (Figure 1) was successfully carried out at room temperature with about two-thirds of the dye covalently attaching to the polymer within 24 h. The application of the nearly neutral N,N-dimethylaminopyridine tosylate (DPTS), a catalyst enabling direct carbodiimide-based coupling of alcohols and carboxylic acids at room temperature with the minimal generation of byproducts (corresponding acylureas), allowed using significantly milder reaction conditions compared to the basic 4-dimethylaminopyridine employed previously for labeling a Soraphen A derivative with BODIPY FL.¹³ This change fully prevented the degradation of the BODIPY fluorophore and preserved its structure in our study. Additionally, in a series of preliminary model experiments (with phenethyl alcohol as a non-polymeric model of PLA and with thin-layer chromatography monitoring) we found that 1-ethyl-3-(3’-dimethylaminopropyl)carbodiimide hydrochloride (EDC) is a no less efficient reaction promoter than 1,3-diisopropylcarbodiimide (DIPS) recommended in the literature,²¹ while at the same time facilitating the separation of the modified polymer from the EDC-derived reaction products, which are much more polar than those of DIPS.

Figure 1.

Scheme showing direct carbodiimide-induced binding of acids to PLA applicable either to make BODIPY labeled conjugates in one step where a BODIPY acid precursor is used (PLA-BODIPY FL in example 1) or to form Boc-amino acid-functionalized intermediates for subsequent deprotection and conversion to PLA-BODIPY conjugates.

The purification of the reaction product from non-polymeric impurities was easily achieved via several precipitations with methanol from dichloromethane or chloroform in the presence of potassium trifluoroacetate as an ionic compound added to accelerate phase separation. Potassium trifluoroacetate appears to be particularly well-suited for this purpose, as it is both neutral and readily soluble in methanol. ¹H NMR of the labeled conjugate 1 (Figure S1) failed to detect any traces of EDC or DPTS and their derivatives. The positions of the proton signals corresponding to the 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene core (δ 7.05, 6.85, 6.25, and 6.08 ppm), as well as those of the propanoyl bridge (δ 3.27 and 2.82 ppm) and of 5,7-CH₃ (δ 2.52 and 2.20 ppm) were in agreement with the literature data for BODIPY FL acid.²² Therefore, because 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene core of BODIPY dyes exhibit good stability in different reaction media except strongly acidic or strongly basic conditions,²³ the direct coupling of carboxy-functionalized BODIPY fluorophores offers a reliable one-step approach for the covalent fluorescent labeling of polyesters when a corresponding BODIPY acid is available.

To extend the covalent labeling strategy to a broader set of BODIPY dyes including those presently available only as SE derivatives, we explored modifying the polymer with reactive primary amino groups via the attachment of amino acids as linkers. Reactions of Boc-protected amino acids with PLA (Figure 1) proceeded smoothly under the conditions similar to those used for the direct coupling of BODIPY FL. Due to much higher molar concentrations of acids in this case, the reactions were complete already within 7 h. The uniform attachment of the acids to the polymer was evident from the disappearance of the ¹H NMR signal at δ 4.33 ppm (terminal HOCH of PLA) and from a new peak showing up on the shoulder of an intense PLA resonance (owing to CH₃ groups) and corresponding to t-butyl of Boc (δ 1.41 ppm, Figure S2). The intensity data indicated near-complete acylation of the PLA terminal OH with the corresponding Boc-protected amino acids: Boc-glycine and Boc-6-aminocaproic acid.

It is noteworthy that though Fmoc-glycine was found to be similarly effective at forming the corresponding conjugate, the removal of Fmoc protection could not be achieved without harsh conditions requiring polymer exposure to diethylamine for several hours. Therefore, the use of Fmoc-protected amino acids for attaching amine-reactive BODIPY dyes to PLA was eventually not pursued. In contrast, we found that the Boc protective group can be easily removed from the conjugates in non-hydrolytic conditions (anhydrous trifluoroacetic acid in dichloromethane at room temperature for 30 min), minimizing the risk of polymer backbone degradation. The deprotected primary amino groups were then successfully reacted with the SE-functionalized BODIPY fluorophores: BODIPY_558/568 SE and BODIPY_630/650-X NHS ester (Figure 2). Importantly, the longer 6-aminocaproate bridge was found to be advantageous over the glycinate spacer: in equal conditions (the same molar amounts of the aminated PLA and the amine-reactive BODIPY_630/650-X), a larger fraction of the dye (85%) was attached to the 6-aminocaproate intermediate than to the glycinate derivative (63%). Because a part of amino groups remained unreacted due to BODIPY label amounts insufficient for the complete PLA modification, the residual amines were capped via acetylation with an excess of acetic anhydride in order to maximally recapitulate the properties of the unmodified polymer containing no ionogenic functions. The BODIPY-labeled polymers 3a-c were isolated and purified similarly as the PLA conjugates mentioned above. Although the signals of all the protons of the BODIPY_558/568 residue could be used for ¹H NMR analysis, the signals of the BODIPY_630/650-X fluorophore protons were more complex and only one of them (belonging to CH₂ of the bridge p-C₆ H₄ OCH₂ CO-) could be reliably used for the calculations. The spectrum of the conjugate 3c formed using the 6-aminocaproic acid linker was further complicated, most likely due to several conformations of the bridge containing two amide bonds, but still amenable to the quantitative NMR analysis.

Figure 2.

Scheme showing the covalent binding of amine-reactive BODIPY fluorophores (Su = N-succinimidyl) to amino acid-functionalized PLA intermediates.

Importantly, we observed that though the PLA-glycine intermediate was stable when the amino group was protonated with trifluoroacetic acid or when this construct was dissolved in dichloromethane prior to reacting it with SE-functionalized BODIPY fluorophores, the unprotected intermediate was susceptible to a spontaneous loss of the amino function when stored as a dry polymer: nearly a half of the unreacted amino groups were lost at 4 °C over 2 days, as was determined after the treatment with acetic anhydride. Additional experiments showed that at room temperature the loss of reactive amino groups was nearly complete within 3 days. A plausible explanation for their disappearance is that the free amino groups of the PLA-glycine intermediate attack the nearest carbonyl of the polymer backbone, resulting in scission of the ester bond and elimination of 3-methyl-1,4-morpholine-2,5-dione (Figure S3).

Between the two amino acid-mediated dye conjugation strategies, the chemical activation of the polymer with 6-aminocaproate appears to provide a more robust result, as the amino group of this intermediate shows greater availability toward amine-reactive species and may also be more stable upon storage compared to that of the glycine-derived conjugate. Furthermore, the ester bond formed using 6-aminocaproate as a linker is expected to be less prone to hydrolytic cleavage compared to the more labile ester linkage formed by the N-acylglycine carboxylic group, thereby improving the stability of the labeled polymer construct.

Nanoparticle Formulation and Comparative In Vitro Release Studies.

We next examined the properties of PLA-based NPs labeled with BODIPY FL-conjugated polymer. Because nanocarrier labeling by the physical entrapment of hydrophobic small-molecule dyes remains a popular approach due to the simplicity and lack of requirement for chemically modifying the particle-forming components, we also included in these studies NPs made of plain PLA with the incorporation of a non-functionalized, strongly hydrophobic BODIPY_505/515 dye (log P_O/W = 4.3).²⁰ BODIPY is analogous in its structure and fluorescence characteristics to BODIPY FL, whereas its hydrophobicity is comparable to that of another popular probe used for non-covalent NP labeling, Coumarin 6 (log P_O/W =5.0).²⁴ NPs with different sizes (280 and 140 nm, Figure 3A) and narrow, non-overlapping size distributions were prepared by the emulsification-solvent evaporation method using either a water-immiscible or a partially water-miscible volatile organic phase composition (chloroform or 1:1 chloroform/tetrahydrofuran, respectively) to allow particle size modulation without changing the amounts and ratios of the particle-forming components.¹⁸ For equivalent dye incorporation yields, the respective probe amounts were adjusted to 25 μg per 100 mg polymer in covalent- and entrapment-labeled NPs either by admixing the BODIPY FL-labeled polymer (containing 0.010 mmol dye per gram) to plain PLA or by using an appropriate dilution of free BODIPY_505/515 within the organic phase. The spectra and emission intensities of the resultant covalent- and entrapment-labeled NP suspensions were similar (Figure S4) with a sharp peak at 540 nm followed by a steep decline at higher wavelengths (9 and 4% of the maximum intensity determined at 590 and 640 nm, respectively).

Figure 3.

Size distribution and release kinetics of encapsulated and covalently labeled BODIPY probes from PLA-based NPs. Small and large NPs (mean diameters: 140 and 280 nm, respectively) with narrow size distributions were made using emulsion templates with or without the inclusion of a water-miscible solvent (tetrahydrofuran) in the organic phase as a way of adjusting the particle size without altering the amounts and ratios of the particle-forming materials (A). Solubility of a hydrophobic probe (BODIPY_505/515, logP_O/W =4.3) in aqueous media was determined as a function of bovine serum albumin concentration added as a solubilizer prior to release studies (B). Release of BODIPY_505/515 encapsulated in small and large NPs was examined under perfect sink conditions in comparison to a covalently attached analogue, BODIPY FL, using 5% albumin solution as an acceptor medium (C). Some error bars are too small to be seen. Unlike the rapid and NP size-dependent release of the encapsulated probe under perfect sink conditions, the dissociation of the non-covalently incorporated tracer over time was marginal from NPs of both sizes when measured in PBS under non-sink conditions (D).

Prior to testing NP labeling stability, we first measured the solubility of BODIPY_505/515 in water and in serum albumin-containing aqueous solutions. Consistent with the strong hydrophobicity of the probe, its solubility was less than 0.1 mg/L in water (Figure 3B), but rapidly increased in the presence of the protein in accordance with its high solubilizing capacity,²⁵ reaching 1.2 mg/L at the physiologically relevant albumin concentration (5% w/v). As anticipated, when NPs were diluted in this medium to provide perfect sink conditions, 50% or more of the encapsulated BODIPY_505/515 dissociated rapidly within the first 2 h (Figure 3C). At the same time, the release rate showed a clear inverse dependence on the particle size as predicted theoretically or demonstrated experimentally in previous studies.^26–28 In contrast, covalently linked BODIPY FL exhibited a much more stable association with the carrier under the same conditions with only a small labile fraction (12 ± 1%) immediately redistributing into the acceptor medium and likely representing the dye linked to PLA oligomers initially present or newly formed in the polymer due to the effect of sonication applied during the emulsification step.^29,30 Remarkably, the remainder of the chemically linked probe stayed stably associated with small-sized NPs exhibiting no further release over 24 h. The suitability of non-covalent NP labeling for cell culture or in vivo studies is often justified by presenting the release data that show negligible redistribution of the probe from NPs diluted into an aqueous acceptor medium, most typically phosphate-buffered saline (PBS).^31,32 As anticipated, based on the extremely limited capacity of this medium toward the highly hydrophobic BODIPY derivative, the concentration of the probe released from NPs diluted in PBS to an equivalent of 1 mg/L of BODIPY_505/515 approached saturation already at the earliest examined time point (30 min), with no further release observed over the next 24 h regardless of the NP size (Figure 3D).

Nanoparticle-Cell Interactions Correlated with the Behavior of Covalently Linked or Encapsulated BODIPY Dyes.

We next examined the cell uptake patterns of BODIPY probes incorporated in small labeled NPs either as a free compound or as a PLA-linked conjugate as a function of serum presence and exposure times. Both microscopic studies and fluorometric analysis revealed markedly dissimilar patterns of the cell interactions by the two formulations. Although in both cases the accumulation of the probe within the cells increased with exposure duration (p < 0.001), the NP-entrapped probe exhibited no dependence on the serum conditions (full vs. depleted, Figures 4 and 5A), suggesting that the dye was distributed into the cells primarily through the interaction with NPs coming directly in contact with cell membranes rather than through its release into the medium.^33–35 This mechanism is also supported by the highly diffuse pattern of the cell staining (Figures 4 and 6), suggesting that the probe is distributed uniformly throughout the volume of the plasma membrane and the lipid bilayer of endocytic vesicles stained with LysoTracker Red, both acting as a sink accommodating the released lipophilic tracer,³⁶ in agreement with previous reports showing the localization of BODIPY_505/515 in intracellular lipid bodies.^37,38 Notably, the uptake of the probe slowed down substantially after the first 4 h with a rate about seven times lower compared to initial (0.22 ± 0.08 vs. 1.47 ± 0.28 RFU/h), consistent with a saturable uptake mechanism driven by passive probe partitioning between the cellular compartment and the particles. In contrast, the uptake of the chemically linked BODIPY FL, while generally exhibiting slower kinetics, was strongly dependent on the serum conditions with only marginal accumulation observed in the presence of a serum-depleted medium over 24 h (p = 0.006, Figures 4 and 5B). The intracellular localization of the dye exhibited a punctate pattern that only partially corresponded to the subcellular distribution of LysoTracker Red, suggesting the localization of a substantial label fraction in early, non-acidified endosomes. Furthermore, though most cells internalized the chemically linked probe (Figure S5), there was a clear distinction between cell groups with higher and lower avidities (Figure 6), consistent with previously reported variations in NP endocytosis rates within a cycling cell population.³⁹ Similar trends in the uptake pattern and strong dependence on serum presence were observed with analogously made small NPs labeled by covalently attached orange-red BODIPY_558/568 (Figure S6): probe internalization measured at 24 h increased near-linearly with serum concentration within the 0–10% range and showed a direct, also near-linear dependence on exposure duration (13 ± 1 and 108 ± 5 RFU at 2 and 24 h, respectively, in the presence of 10% serum). Remarkably, in accordance with the higher extinction coefficient and the red-shifted fluorescence spectrum of this probe in comparison to BODIPY FL, the use of NPs labeled covalently with BODIPY_558/568 resulted in a stronger cell-associated signal and negligible interference due to autofluorescence, potentially enabling undistorted observations and accurate measurements carried out longitudinally in a serum-containing environment, which can offer an important advantage for in situ studies of cell-nanocarrier interactions.¹⁹

Figure 4.

Comparative study of cell uptake and the probe distribution of analogously made small NPs formulated either with encapsulated BODIPY_505/515 (entrapment-labeled) or with PLA-BODIPY FL conjugate 1 (covalently labeled) as a function of exposure time and serum presence. Fluorescence images were captured at an original magnification ×100 using the fluorescein isothiocyanate channel setting (λ_ex /λ_em = 488 nm/492–519 nm). The scale bar represents 100 μm.

Figure 5.

Fluorometric analysis of the probe uptake kinetics by cultured A10 cells as a function of the tracer incorporation mode (entrapment-labeled, A vs. covalently labeled, B), exposure time, and serum presence. Cell-associated signal was measured after replacing the medium at predetermined time points with PBS (λ_ex/λ_em = 485 nm/535 nm). The results were analyzed for the effects of the experimental variables by two-factor ANOVA and multiple linear regression. The increase in fluorescent signal was strongly dependent on the exposure duration for NPs prepared with the tracer incorporation mode (p < 0.001), but only covalently labeled NPs exhibited significant dependence on serum concentration (p = 0.006 vs. p = 0.28 for entrapment-labeled NPs).

Figure 6.

Characterization of the intracellular probe distribution following A10 cell exposure to small NPs formulated either with encapsulated BODIPY_505/515 (entrapment-labeled) or with PLA-BODIPY_505/515 FL conjugate 1 (covalently labeled). One hour prior to acquiring microscopic images, cells were incubated with LysoTracker Red DND-99 (50 nM) to stain the lysosomal compartment. Fluorescence images were captured at an original magnification ×200 using the fluorescein isothiocyanate (λ_ex/λ_em = 488 nm/492–519 nm) and rhodamine (μ_ex/λ_em = 543 nm/551–622 nm) channel settings to visualize the fluorescent probes and lysosomes, respectively. The scale bar represents 100 μm.

Besides their dissimilar uptake kinetics and serum dependence patterns prompting distinct internalization mechanisms (passive tracer redistribution vs. stimulation-dependent carrier endocytosis), the physically entrapped and covalently attached BODIPY probes also showed marked differences in their cell residence profiles. Cells exposed to dye-loaded NPs for 24 h in the presence of 10% serum lost 75 ± 5% of their fluorescence regardless of the time after the medium replacement (Figure 7), again consistent with passive probe partitioning that rapidly establishes a new equilibrium between the cells and the surrounding protein-containing medium. In contrast, cells pre-exposed to covalently labeled particles retained both the microscopic pattern of the probe distribution and most (92 ± 8%) of the cell-associated fluorescent signal 8 h after the medium replacement, in accordance with negligible export rates characteristic of the endocytosed NPs.³⁹

Figure 7.

Cell-associated fluorescent signal retention after A10 cell exposure to small NPs formulated either with encapsulated BODIPY_505/515 (entrapment-labeled) or with PLA-BODIPY FL conjugate 1 (covalently labeled). Following a 24 h exposure, unincorporated NPs were removed. The medium was replaced with PBS at predetermined time points, and fluorescence images were captured at an original magnification ×100 using the fluorescein isothiocyanate channel setting (λ_ex/λ_em = 488 nm/492–519 nm). The scale bar represents 100 μm.

Optimal Labeling Strategy and Potential Applications of BODIPY-Labeled Polyester NPs.

These results show that a reliable NP labeling strategy based on stable association of the probe with the carrier requires its incorporation in a molecular form that essentially has no diffusivity through the particle matrix, thus preventing its redistribution on a time scale comparable to that of the particle disintegration (or, for any practical purposes, greater than the duration of experiments involving the labeled formulation). Although this condition can theoretically be reasonably satisfied with small-molecule probes possessing extremely high hydrophobicities, the suitability of this labeling approach has to be reexamined for each specific nanocarrier formulation and experimental settings.²⁴ In a recently reported example, a small-molecule fluorophore with a log P_O/W greater than 10 was shown to rapidly partition from polyester-based NPs coming in contact with lipid membranes.²⁴ The same probe escaped from a polymeric carrier within minutes after intravenous injection in another study.⁴⁰ It is noteworthy that the small-molecule probes that are typically used for NP labeling are orders of magnitude less hydrophobic^7,41 and consequently are expected to dissociate even more rapidly from the carrier under perfect sink conditions. Besides the stability of its association with the carrier, an ideal labeling construct should be chemically similar to the particle-forming material in order not to interfere with its colloidal properties and assembly/disassembly rates, cargo encapsulation capacity, interaction with its surroundings and cells, and so forth. In this regard, chemically attaching the probe to the particle-forming polyester itself, whereas not being as technically straightforward as incorporating it in the carrier physically, generates a construct whose hydrophilic-lipophilic balance and molecular size accurately match those of the parent polymer, in turn minimizing its diffusivity and precluding premature tracer release. Indeed, the dissociation of the covalently linked probe is governed by the same chemical mechanism (i.e., the hydrolytic cleavage of the ester bond) that drives NP disassembly and therefore is expected to occur at a rate comparable to that of the nanocarrier disintegration. The goal of this study was to establish and characterize a reliable methodology for NP labeling, whose effect on the physical and chemical properties of the formulation can be regarded as “vanishingly small”, so that such labeled NPs can serve as adequate models of the parent, unlabeled formulations made of preformed polyester polymers best suited for a particular purpose. A labeling strategy using preformed polymers as the starting material (as opposed to de novo synthesized polymers, whose properties may differ substantially from those of the original particle-forming polymer) appears, therefore, to be the optimal choice. The ability to use fluorophores in their ready-made, commercially available forms is another advantage compared to the de novo synthesis of labeled polymers typically requiring specially designed, chemically modified dyes and necessitating exposure to harsh conditions usually employed for polymerization.^42,43 By combining adequate carrier-probe association stability with physical and chemical properties consistent with those of the original particle-forming polymer and the parent (unlabeled) NP formulation, the “post-polymerization” covalent labeling strategy is appropriate for applications requiring reliable measurements that faithfully reflect the carrier fate and interactions with minimal or no distortions due to the label presence.

Although most of the experiments focusing on the evaluation of the covalent labeling approach in our study were carried out with BODIPY FL as a model probe, this approach can be readily applied to a wide range of BODIPY dyes, including those with more extensive π-conjugated systems and red-shifted fluorescence. The use of these compounds can provide increased sensitivity and specificity by strengthening the signal and by reducing interference due to autofluorescence in protein-containing environments. Remarkably, this labeling strategy is compatible with different NP formulation methodologies^14,18 and does not adversely affect the capacity of the nanocarriers to accommodate therapeutic payloads^14,17 or incorporate nanocrystalline iron oxide in the particle matrix for magnetically targeted delivery.^15,44 BODIPY-labeled NPs can in turn be used to label cells without compromising their viability or proliferation rates for cell therapeutic applications.^15,45 The broad applicability of this covalent labeling methodology to a variety of BODIPY probes also allows co-labeling particulate formulations with spectrally complementary fluorophores, offering a way to monitor nanocarrier disassembly with high precision in different media of interest or inside living cells using Förster resonance energy transfer.¹⁹ Besides providing a valuable tool for fluorescence microscopy, optical imaging, or quantitative fluorometric analysis,^14,46 BODIPY-labeled biodegradable NPs can potentially be used diagnostically for photoacoustic and magnetic resonance imaging or positron emission tomography studies,^47–49 or therapeutically as boron carriers for boron neutron capture therapy or the more recently introduced proton boron capture therapy,⁵⁰ where stable incorporation of these compounds in the carrier is critical for their effective use.

CONCLUSIONS

Studying the behavior of nanoparticulate formulations in biological systems often relies on monitoring the localization and signal of molecular reporters associated with the particle. Stability of the particle-probe association on a time scale comparable to the lifetime of the particle is, therefore, a prerequisite for performing these measurements without introducing significant distortions. Here, we report a chemical strategy for covalently labeling particle-forming polyester polymers with BODIPY probes varying in fluorophore structure and available reactive groups, demonstrating the suitability of this labeling methodology for producing traceable NP formulations, and confirming the utility of the BODIPY-labeled NPs for studying the interactions with living cells. The results of these studies are consistent with NP uptake being an active and unidirectional process with comparatively slow kinetics markedly affected by the presence of serum. This pattern is different from the serum-independent and rapidly reversible, partitioning-driven accumulation of the hydrophobic BODIPY probe incorporated in NPs, in accordance with the distinct (biological vs. physicochemical) mechanisms governing the internalization, intracellular distribution, and fate of submicron-sized carriers and their payloads. Based on these results, physically incorporated BODIPY dyes may be useful as model compounds for studying the behavior of similarly hydrophobic small-molecule cargos. However, as shown previously with several other small-molecule probes, the encapsulation of free tracers does not offer a valid approach for stably labeling NP formulations, paving the way to significant inconsistencies in the interpretation of experimental data if intended for that purpose.⁵¹

We anticipate that the efficient and broadly applicable chemical strategy providing polyester derivatives stably labeled with BODIPY fluorophores and constructed using the original polymer as the starting material will address the need for a robust and practical way of making traceable NP formulations well-suited for in vitro and in vivo investigations. The implementation of this labeling approach to reliably monitor the behavior of biodegradable polymer-based NPs in complex biological environments or to study their interactions with living cells can facilitate the development and optimization of new therapeutic nanocarriers and diagnostic agents.

METHODS

Poly(d,l-lactide) polymers (PLA) were purchased from Sigma-Aldrich (St. Louis, MO) and Lakeshore Biomaterials (Birmingham, AL). From the signal intensities in ¹H NMR spectra (terminal CH₃ of C₁₂ H₂₅ to CH of the polymer backbone), the M_n were estimated to be ca. 30 and 50 kDa for PLA from Sigma-Aldrich and Lakeshore Biomaterials, respectively. Both polymers were capped with a dodecyl group at the carboxylic end (Figure 1). BODIPY fluorophores were purchased from Thermo Fisher Scientific (Waltham, MA). DPTS catalyst was prepared as described previously.²¹ All other reagents were purchased from Sigma-Aldrich (St. Louis, MO). Dichloromethane was dried over molecular sieves (4 Å, 1.6 mm pellets) and trifluoroacetic acid was distilled under argon prior to use. NMR spectra were recorded with deuterated chloroform as a solvent in all cases using a Bruker AVANCE DMX 400 spectrometer (400 MHz for ¹H).

Polymer-dye conjugates evaluated in this study were synthesized and characterized as follows.

One-Step Synthesis via Direct Esterification.

Poly(d,l-lactide) (Sigma-Aldrich, 0.742 g, ca. 25 μmol) was dissolved in dichloromethane (4 mL), and BODIPY FL acid (2.8 mg, 9.6 μmol) and DPTS (0.120 g, 0.41 mmol) were added. The mixture was stirred at room temperature to homogenization, and EDC (80 mg, 0.42 mmol) was introduced. The mixture was further stirred at room temperature for 24 h, and the polymer was precipitated with methanol (150 ml). After stirring at 4 °C for several hours (to a complete coagulation), the supernatant was decanted off, the polymer was rinsed with methanol, dissolved in chloroform (6 ml), and re-precipitated with methanol (150 mL). To accelerate the coagulation of the colloid-like polymer particles, potassium trifluoroacetate (80 mg) was added. The mixture was stirred at 4 °C to a complete transparency of the supernatant, the polymer was separated, dissolved in chloroform, and re-precipitated as above two more times. Finally, the polymer was dried at 10 Torr and dissolved in dichloromethane. The solution was filtered, dried, and the residue was evacuated at 20–50 mTorr and room temperature for 77 h, yielding 0.736 g of BODIPY FL-labeled PLA (1). Fluorescence measurements of combined dried supernatants (aliquot amounts of each) and of 1 dissolved in acetonitrile detected 67% of the total fluorescence in the polymer fraction. ¹H NMR of 1 (Figure S1) showed BODIPY FL bound to 28% of PLA macromolecules.

Synthesis of Amino Acid-Functionalized Polymeric Precursors.

To obtain poly(d,l-lactide) Boc-glycinate (2a), poly(d,llactide) (Lakeshore Biomaterials, 1.145 g, ca. 23 μmol) was first dissolved in dichloromethane (6.6 mL), then Boc-glycine (0.177 g, 1.0 mmol) and DPTS (0.285 g, 0.97 mmol) were added, followed by EDC (0.209 g, 1.09 mmol). The mixture was stirred at room temperature for 7 h, and the polymer was precipitated with methanol (200 mL). After stirring at 4 °C for several hours, the clear supernatant was decanted off, the polymer was rinsed with methanol, dissolved in chloroform (6 mL), and re-precipitated with methanol (200 mL) and potassium trifluoroacetate (100 mg). The mixture was stirred at 4 °C overnight to a complete transparency of the supernatant. The polymer was separated, dissolved in chloroform, and re-precipitated as above one more time. After drying at 10 Torr, the polymer was dissolved in dichloromethane. The solution was filtered, dried, and the residue was evacuated at 20–50 mTorr and room temperature for 21 h, yielding 1.147 g of PLA Boc-glycinate (2a). ¹H NMR of 2a is in agreement with the structure (Figure S2). To make poly(D,L-lactide) Boc-6-aminocaproate (2b), poly(D,L-lactide) (Lakeshore Biomaterials, 1.149 g, ca. 23 μmol) was dissolved in dichloromethane (7 ml), Boc-6-aminocaproic acid (0.234 g, 1.0 mmol) and DPTS (0.285 g, 0.97 mmol) were added, followed by EDC (0.209 g, 1.09 mmol). The mixture was stirred at room temperature for 7 h, and the polymer was precipitated and purified as described above for the glycine conjugate 2a, except that the amount of potassium trifluoroacetate in the precipitations with methanol had to be increased to 200 mg for an efficient polymer coagulation. Yield of 2b: 1.154 g. ¹H NMR of 2b corresponds to the structure (Figure S2).

BODIPY_558/568-Labeled Poly(D,L-Lactide) with Glycinate Spacer (3a).

Poly(d,l-lactide) Boc-glycinate 2a (1.147 g, ca. 23 μmol) was dissolved in dichloromethane (6.7 ml), and trifluoroacetic acid (1.75 ml, 22.7 mmol) was added. The mixture was reacted at room temperature for 0.5 h, and the volatiles were removed in vacuo (10 Torr) at room temperature. The residue was dissolved in chloroform (30 ml) and dried as above. Finally, the residue was dissolved in toluene (25 ml), dried at 10 Torr and 35 °C, then evacuated at 20–50 mTorr and room temperature for 2 h, yielding 1.236 g of Boc-deprotected polymer. The polymer was dissolved in dichloromethane (5 mL), triethylamine excess (0.05 mL, 0.36 mmol) was added, and BODIPY_558/568 SE (5 mg, 11 μmol) in dichloromethane (1.5 mL) was introduced dropwise. The mixture was reacted at room temperature for 23 h, the volatiles were removed in vacuo (10 Torr, room temperature), and the residue was dissolved in chloroform (6 mL). Acetic anhydride (0.06 mL, 0.63 mmol) was added, the mixture was reacted at room temperature for 20 min, and the polymer was isolated and purified as described above for 2a. Yield of 3a: 1.138 g. ¹H NMR of 3a detected the signals of the fluorophore (with the propanoyl bridge): δ, ppm 2.68 (t, 7 Hz, 2H), 3.34 (t, 7 Hz, 2H), 6.37 (d, 4 Hz, 1H), 6.77 (d, 5 Hz, 1H), 6.96 (d, 4 Hz, 1H), 7.00 (d, 5 Hz, 1H), 7.10 (s, 1H), 7.16 (t, 5 Hz, 1H), 7.46 (d, 5 Hz, 1H), 8.11 (d, 4 Hz, 1H), and the N-acetyl group at δ 2.00 ppm. Intensity data showed that 30% of the aminated PLA macromolecules were modified with the BODIPY_558/568 fluorophore and 60% were bearing the acetyl group.

BODIPY_630/650-Labeled Poly(d,l-Lactide)with Glycinate Spacer (3b).

Poly(d,l-lactide) Boc-glycinate 2a (1.157 g, ca. 23 μmol) was deprotected and reacted with BODIPY_630/650 NHS ester(5 mg,7.6 μmol) followed by acetic anhydride as described above. Prior to the treatment with acetic anhydride, the dried reaction mixture was stored for 2 days at 4 °C. After acetylation, the polymer was precipitated and purified as above. Yield of 3b: 1.038 g. ¹H NMR of 3b showed that 18% of the PLA macromolecules modified with the BODIPY_630/650-X fluorophore and only 30% acetylated. To quantify the amount of bound BODIPY_630/650-X, the most prominent signal of ArOCH₂CO (δ 4.50 ppm, s) was used.

BODIPY_630-650-X-labeled Poly(d,l-,L-Lactide) with 6-Aminocaproate Spacer (3c).

Poly(D,L-lactide) Boc-6-aminocaproate 2b (1.154 g, ca. 23 μmol) was deprotected, reacted with BODIPY_630/650-X NHS ester (5 mg, 7.6 μmol), and acetylated as described above for 3a. The polymer was precipitated and purified as described for 2b. Yield of 3c: 1.127 g. ¹H NMR of 3c showed 26% of the PLA macromolecules modified with the BODIPY_630/650-X fluorophore and 68% acetylated.

Formulation and Characterization of BODIPY-Labeled NPs.

Larger and smaller sized PLA-based NPs were formulated using modifications of the emulsification-solvent evaporation method with water-immiscible (chloroform) and partially water-miscible (chloroform/tetrahydrofuran, 1:1) volatile solvent compositions in the organic phase, respectively. To produce fluorescent NPs with either a covalently linked or encapsulated BODIPY probe, the organic phase (volume adjusted to 6 mL) was used to dissolve either (i) 10 mg of BODIPY FL-labeled PLA (1) admixed to 90 mg of plain PLA or (ii) 25 μg BODIPY_505/515 added to 100 mg plain PLA, respectively. The indicated compositions contained equivalent molar amounts of the BODIPY tracers. The organic phase was emulsified by sonication on ice (2 min) in 15 ml of an aqueous solution containing 1% w/v of bovine serum albumin (BSA) as a colloidal stabilizer.¹⁵ Solvents were removed using a rotary evaporator, and the obtained NPs were passed through a sterile 5.0 μm polyvinylidene difluoride membrane (EMD Millipore; Billerica, MA). Orange-red fluorescent, small-sized NPs were prepared as above using PLA covalently labeled with (3a) instead of 1.

The amounts of the NP-associated BODIPY probes were determined spectrophotometrically (λ = 500 nm for BODIPY FL and BODIPY_505/515, and 560 nm for BODIPY_558/568, respectively) against suitable calibration curves after separating the particles by centrifugation (14 000 rpm for 15 min) and dissolving the residue in acetonitrile (NP-probe association yields: 88.8 ± 0.1% for chemically labeled small NPs; 88.4 ± 4.4% and 93.8 ± 1.6 for entrapment-labeled small and large NPs, respectively). The mass balance was confirmed by analyzing the NP-unbound probe fractions after diluting the supernatants 1 to 5 with acetonitrile. The NP size was determined using dynamic light scattering (90Plus particle size analyzer, Brookhaven Instruments; Holtsville, NY).

Prior to release studies, the solubility of BODIPY_505/515 was determined in PBS and in a series of BSA solutions in water at room temperature after allowing the probe to reach equilibrium overnight. Samples were passed through an aluminum oxide membrane with a 0.02 μm pore size (Anotop, Whatman Inc., NJ, USA) impermeable to intact NPs. The filtrates were extracted in two steps into chloroform using sodium chloride (5 M) as a salting-out agent. The extracts were combined, evaporated, dissolved in acetonitrile, and analyzed spectrophotometrically. Based on the solubility results, 5% w/v albumin concentration was chosen as the acceptor medium for release studies using perfect sink conditions. NPs diluted in this medium to a concentration equivalent to ca. 0.1 mg of the BODIPY probe per liter (i.e., ≤ 10% of its solubility) were loaded into syringes capped with aluminum oxide Anotop filters impermeable to the particles and mounted on a tube revolver shaker. Samples were collected and analyzed fluorometrically for the dye content at predetermined time points (λ_ex/λ_em = 505 nm/535 nm; cutoff: 530 nm). The results were verified by UV/vis spectrophotometry against a suitable calibration curve after diluting samples 1:4 in acetonitrile. Background-corrected absorbances of the samples and calibration standards were determined according to the formula: OD_{505 nm}-(OD_455nm+OD_555nm)/2. Dye release in PBS was determined under non-sink conditions using a similar setup with NPs diluted to a concentration equivalent to 1 mg of the probe per liter.

Cell Culture Studies.

Cell interactions were studied as a function of exposure duration and serum conditions using cultured rat aorta smooth muscle cells (A10), a vascular cell line exhibiting the defining characteristics of neointimal smooth muscle cells.⁵² The cells were seeded at a density of 10⁴/well on 96-well plates using Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum as a culture medium and allowed to spread and reach a near-confluent state overnight. A10 cells were then incubated at 37 °C with NPs diluted to a concentration corresponding to 0.1 μg BODIPY probe per ml in Dulbecco’s modified Eagle’s medium containing indicated fetal bovine serum concentrations. In some experiments, LysoTracker Red DND-99 (Thermo Fisher Scientific, Waltham, MA) was added to the medium to a final concentration of 50 nM 1 h prior to imaging for lysosome visualization. At predetermined time points, the cells were washed with 2% BSA solution in PBS to remove the non-internalized NPs. The medium was then replaced with PBS, cells were imaged under a fluorescence microscope using the green or orange-red fluorescence filters where appropriate (λ_ex/λ_em= 488 nm/492–519 nm or 543 nm/551–622 nm), and the cell-associated signal was determined fluorometrically using a microplate reader λ_ex/λ_em= 485 nm/535 nm or 540 nm/575 nm).

Cells pre-exposed to NPs for 24 h using high serum conditions were used to determine the probe dissociation patterns as a function of the particle-dye binding mode, serum presence, and incubation time. The cells were washed at predetermined time points (2, 4, and 8 h) as above, imaged, and analyzed fluorometrically for signal intensity.

Statistical Analysis.

Experimental data are expressed as mean ± standard deviation. Probe uptake and dissociation kinetics were analyzed by two-factor analysis of variance (ANOVA) and multiple linear regression for the effects of exposure duration and serum concentration. Differences were termed significant at p < 0.05.