Labelling Primary Immune Cells Using Bright Blue Fluorescent Nanoparticles

Abstract

Tracking cell movements is an important aspect of many biological studies. Reagents for cell tracking must not alter the biological state of the cell and must be bright enough to be visualized above background autofluorescence, a particular concern when imaging in tissue. Currently there are few reagents compatible with standard UV excitation filter sets (e.g. DAPI) that fulfill those requirements, despite the development of many dyes optimized for violet excitation (405 nm). A family of boron-based fluorescent dyes, difluoroboron β-diketonates, has previously served as bio-imaging reagents with UV excitation, offering high quantum yields and wide excitation peaks. In this study, we investigated the use of one such dye as a potential cell tracking reagent. A library of difluoroboron dibenzoylmethane (BF₂dbm) conjugates were synthesized with biocompatible polymers including: poly(L-lactic acid) (PLLA), poly(ε-caprolactone) (PCL), and block copolymers with poly(ethylene glycol) (PEG). Dye-polymer conjugates were fabricated into nanoparticles, which were stable for a week at 37 °C in water and cell culture media, but quickly aggregated in saline. Nanoparticles were used to label primary splenocytes; phagocytic cell types were more effectively labelled. Labelling with nanoparticles did not affect cellular viability, nor basic immune responses. Labelled cells were more easily distinguished when imaged on a live tissue background than those labelled with a commercially available UV-excitable cytoplasmic labelling reagent. The high efficiency in terms of both fluorescence and cellular labelling may allow these nanoparticles to act as a short-term cell labelling strategy while wide excitation peaks offer utility across imaging and analysis platforms.

Graphical Abstract

Using nanoparticles with a bright boron-based fluorescent dye in the core allows for cell tracking across multiple short wavelength excitation sources.

Introduction

Tracking fluorescently labelled cells is a common strategy to assess cell behavior in vitro and in vivo, particularly for highly motile cells such as lymphocytes.^1–3 Many fluorescent dyes provide intracellular labelling by binding in the cytoplasm or to DNA, but labelling reagents continue to be an area of active research, particularly to take advantage of the further ends of the UV-visible spectrum.⁴ Cytoplasmic labelling reagents are preferred for live cell imaging over time, as DNA-binding dyes can interrupt transcription and cell replication.^4,5 Common cytoplasmic dyes align well with standard green (Ex 470/40, Em 525/50) and red (Ex 550/25, Em 605/70) filter sets for widefield microscopy, including carboxyfluorescein succinimidyl ester (CSFE), calcein-AM, tetramethylrhodamine isothiocyanate (TRITC), and similar newer commercial fluorophores. However, the UV-excitable blue filter (Ex 365/50, Em 445/50) set is often underutilized for cell tracking.

Many traditional blue fluorophores, such as DAPI and Hoechst, bind directly to DNA, while others suffer from low intensities, such as derivatives of pyrene or coumarin.⁶ Dim fluorescence signal under the classic UV-excited blue filter makes it difficult to distinguish labelled cells from tissue autofluorescence. Newer, brighter blue and violet fluorophores are available, but are primarily designed for multiplexed flow cytometry and laser-based microscopy, with a maximum excitation at either 405 or 395 nm and narrow excitation peaks. This makes translating the same dyes from one excitation source to another difficult; concentration, time of labelling, and detector sensitivity must be optimized for each read-out. A bright blue reagent that can be used across multiple platforms such as flow cytometry, confocal microscopy, and widefield fluorescent microscopy would enable a single optimized labelling procedure to be used with flexible choice of read-out.

Boron-based organic dyes serve as effective fluorescence reporters.^7–10 Previously, we reported luminescent difluoroboron β-diketonate (BF₂bdk) complexes as well suited for biomolecular imaging^11–14 and oxygen sensing.^15–18 Here we explore them as reagents for tracking cells in the blue channel. In particular, fluorescent BF₂bdk complexes offer outstanding optical properties such as large extinction coefficients and quantum yields,¹⁹ two-photon absorption,²⁰ solvatochromism,²¹ and photostability.²² One of the brightest of these dyes, methoxy substituted difluoroboron dibenzoymethane (BF₂dbmOMe), emits intense blue fluorescence with a quantum yield approaching unity (Φ_F = 0.99).²³ This is significantly larger than most commercially available blue fluorophores (e.g., DAPI: Φ_F = 0.58;²⁴ pyrene: Φ_F = 0.75;²⁵ 7-hydroxy-4-methylcoumarin: Φ_F = 0.63 ²⁶). Therefore, we sought to harness this bright blue dye emission for in vitro and ex vivo imaging agents.

Boron-based fluorophores are often incorporated into a polymer matrix to improve their physical and chemical properties.²⁷ Many BF₂bdk complexes have large dipole moments (e.g., μ= 6.7 Debye for BF₂dbm), making their emission sensitive to the polarity of media and the local concentration. Therefore, emission spectra are tunable by varying the chemistry and/or length of the polymer chains, fabricated either in film or nanoparticle format.^28,29 Suitably chosen polymers also can increase solubility in aqueous solution and offer protection to the dye (e.g. against hydrolysis), improving their shelf life.³⁰ Poly(_D,L-lactic acid) (PLA), a biocompatible and biodegradable material, has been utilized extensively for nanoparticle formation.³¹ The first-generation boron-based polymer BF₂dbmPLA (Figure 1a-(1)), has found application in fluorescence imaging of cells and tissues, including intracellular uptake and oxygen sensing.^11,32 Poly (_L-lactic acid) (PLLA), a stereoisomer of PLA, offers a higher degree of crystallinity, which may slow the rate of hydrolysis in aqueous solution.^33,34 Poly (ε-caprolactone) (PCL) is a more hydrophobic polyester³⁵ that is even slower to degrade than PLLA, making it useful for extended retention of loaded cargo.^36,37 Blends of other polymers with PCL provide tunable properties useful for drug delivery and tissue engineering applications.³⁸ Finally, although pure hydrophobic polymers can assemble as nanoparticles in water, hydrophilic segments, such as poly (ethylene glycol) (PEG), are often incorporated to enhance the water solubility for biological imaging and drug delivery.¹²

Figure 1:

Nanoparticle Design and Compositions. (a) Chemical structures of boron-based dye-polymer conjugates. (b) Chemical structures of each polymer tested. (c) Schematic of polymeric nanoparticles, which adopt a micellar structure with the boron-based dye in the core and the polymer in the outer shell.

Well-controlled labelling of immune cells is of particular interest, as these highly motile cells are frequently tracked in vitro and in vivo, or labelled for flow cytometric analysis. Fabricating dye-polymer conjugates into nanoparticles offers the opportunity to tune its uptake by immune cells.³⁹ Lymphocytes (T cells and B cells), dendritic cells, and macrophages each have been targeted for delivery of drugs or probes by using nanoparticles.^39–42 Particles similar in size to viruses (20 – 200 nm) are readily internalized via endocytosis, particularly by phagocytic cells such as B cells or professional antigen presentation cells such as dendritic cells after adsorption of serum proteins.⁴³ Altering the hydrophobicity of the nanoparticle can significantly affect internalization,^40,44 and we hypothesized that PLA, PCL, and block co-polymers of PLA-PCL would exhibit differential uptake by endocytic cells.⁴⁵ Furthermore, PEGylation hinders particle uptake in other systems, and we hypothesized that it would similarly prevent uptake of these boron-based nanoparticles.^46–49

In this paper, we tested the utility of labelling and tracking primary immune cells (murine splenocytes) with polymeric nanoparticles containing a blue boron-based fluorophore. After confirming the stability of the dyes and nanoparticles in solution, we assessed internalization and suitability for cellular staining as a function of polymer chemistry. We tested whether cells labelled with these reagents were detectable by fluorescent microscopy under standard blue fluorescence filter sets and compared to a commercial reagent, and utilized these particles to enable four-color fluorescent labelling and cell tracking in live tissue. This is the first demonstration of a materials-based blue fluorescent labelling reagent for cell tracking in the context of tissue autofluorescence.

Materials and Methods

Polymer synthesis and characterization.

The boron-based dye was prepared as either a primary alcohol (BF₂dbmOCH₂CH₂OH)³¹ or a phenol (BF₂dbmOH),⁵⁰ to act as an initiator or a coupler respectively (Figure S1). These were used to synthesize polymer conjugates using methods similar to those previously described (Scheme S1, SI Methods).^{31 1}H NMR spectra were recorded on a Varian VMRS/600 (600 MHz) instrument in CDCl₃ unless otherwise indicated. ¹H NMR resonance was referenced to the residual protiochloroform signal at 7.260 ppm. Coupling constants are given in hertz. Polymer molecular weights (MW) and polydispersity indices (Đ) were determined by gel permeation chromatography (GPC) (THF, 25 °C, 1.0 mL / min) using multi-angle laser light scattering (MALLS) (λ = 658 nm, 25 °C) and refractive index (RI) (λ = 658 nm, 25 °C) detection. Polymer Laboratories 5 µm mixed-C columns (guard column plus two columns) along with Wyatt Technology (Optilab T-rEX interferometric refractometer, miniDAWN TREOS multi-angle static light scattering (MALS) detector, ASTRA 6.0 software) and Agilent Technologies instrumentation (series 1260 HPLC with diode array (DAD) detector, ChemStation) were used in GPC analysis. The incremental refractive index (dn/dc) was calculated by a single-injection method assuming 100% mass recovery from the columns. UV-vis spectra were recorded on a Hewlett-Packard 8452A diode-array spectrophotometer.

Luminescence measurements.

Steady-state fluorescence spectra for the boron dye initiator, polymer and nanoparticle suspensions were recorded on a Horiba Fluorolog-3 Model FL3–22 spectrofluorometer (double-grating excitation and double-grating emission monochromator) after excitation. Optically dilute aqueous solutions of the nanoparticles, with absorbance <0.1 au, were prepared in 1 cm path length quartz cuvettes. Fluorescence spectra were obtained under ambient conditions (i.e., air, ∼21% oxygen in volume).

Nanoparticle fabrication and characterization.

Nanoparticles were fabricated as previously reported.⁵¹ Nanoparticles were fabricated as a suspension in water and were stored as such at 4 °C. The polymer (~3.0 mg) was dissolved in DMF (3 mL), then the dye solution was added dropwise to rapidly stirred DI water (27 mL). The homogeneous mixture was stirred for 30 min, then the nanoparticle suspension was transferred into dialysis tubing (Specra/Pro, 12–14 kDa MWCO, Fisher Scientific) followed by dialysis against water for 24 hours. Nanoparticle size and polydispersity were analyzed by dynamic light scattering (DLS, Wyatt, DynaPro). Zeta potentials were determined by Zetasizer Nano Z (Malvern instruments, UK) and data were analyzed using DTS Nano software. UV-Vis absorbance was recorded by diluting ~1 mg/mL stock nanoparticle suspensions to 50 μg/mL in DI water. The extinction coefficient was estimated based on the Beer-Lambert law.

Nanoparticle stability.

The stock suspensions of nanoparticles (1 mg/mL) were serially diluted (200, 100, 50, 20, 10 μg/mL) with DI water, PBS, water/glucose/serum, and supplemented RPMI. Each sample (100 μL) was injected into a 96-well microtiter plate. Mineral oil was added on the top of each well via syringe to form a thin layer to prevent evaporation. The plate was put into the DLS instrument, protected from light, set to 37 °C and the sizes and polydispersities of the nanoparticles were recorded every 12 h for one week. Nanoparticle aggregates larger than 2000 nm exceed the DLS detection limit. Separately, samples were incubated at 37 °C for five days to obtain daily photographs, as well as the emission spectra, to capture evidence of aggregation and fluorescence changes. GPC was used to monitor the polymer molecular weights (i.e. polymer stability) before and after incubation for specified times. To prepare samples for GPC analysis, nanoparticle aliquots in water were freeze-dried then dissolved in THF for injection into the GPC instrument. The degradation of boron dyes and hydrolysis of polyester were also analyzed by ¹H NMR spectroscopy in CDCl₃.

Cell culture.

All animal work was approved by the Animal Care and Use Committee of the University of Virginia under protocol #4042, and was conducted in compliance with guidelines from the University of Virginia Animal Care and Use Committee and the Office of Laboratory Animal Welfare at the National Institutes of Health (United States). Mice were housed in a vivarium and given food and water ab libitium. Spleens, and lymph nodes where appropriate, were collected from male and female C57Bl/6 mice aged 6–10 weeks (Jackson Laboratories, USA) after isoflurane anesthesia and cervical dislocation. To isolate splenocytes, the spleen was processed through a 70-µm pore size nylon filter (Fisher Scientific, USA) and rinsed with sterile 1x phosphate buffer saline (PBS) supplemented with 2% v/v fetal bovine serum (FBS, VWR, USA). Red blood cells were lysed and the cell suspension was filtered through a fresh 70-µm filter. Cell density was determined through trypan blue exclusion. Where noted, B cells were isolated from bulk splenocytes by using a B cell enrichment kit (StemCell Technologies, USA) based on negative magnetic selection, according to manufacturer instructions. For all overnight cultures, cells were cultured at a concentration of 1×10⁶ cells/mL in “complete RPMI;” RPMI (Lonza, 16–167F) supplemented with 10 % FBS (VWR, Seradigm USDA approved, 89510–186) 1x L-glutamine (Gibco Life Technologies, 25030–081), 50 U/mL Pen/Strep (Gibco), 50 µM beta-mercaptoethanol (Gibco, 21985–023), 1 mM sodium pyruvate (Hyclone, GE USA), 1x non-essential amino acids (Hyclone, SH30598.01), and 20 mM HEPES (VWR, 97064–362) with 120 ng/mL IL-2 (Peprotech, USA).

Cell labelling.

To minimize nanoparticle aggregation, cellular labelling with nanoparticles was performed in an isotonic glucose solution without saline. Splenocytes were resuspended at 10×10⁶ cells/mL in a solution of 5% w/v D-glucose and 2% v/v FBS in ultra-pure water (water-glucose-serum solution, or 1x WGS; components from Thermo Fisher). A staining solution was prepared by mixing 3 parts nanoparticle stock solution (1 mg/mL in water), 2 parts 10x WGS, and 5 parts water. The staining solution was mixed in equal volumes with the cell suspension. This resulted in a 1x WGS solution that contained cells at 5×10⁶ cells/mL with nanoparticles at 0.15 mg/mL. Cells were incubated, protected from light, at room temperature for 30 min, then washed and resuspended at 1×10⁶ cells/mL in supplemented RPMI.

To label with Cell Tracker™ Blue CMF₂HC (4-chloromethyl-6,8-difluoro-7-hydroxycoumarin, Invitrogen), cells were resuspended at 1×10⁶ cells/mL in 1x PBS in the presence of 10 µM Cell Tracker™ Blue (CTB). Cells were incubated, protected from light, at 37 °C for 30 min, then washed and resuspended at 1×10⁶ cells/mL in supplemented RPMI.

Flow cytometry.

Cells were resuspended at 0.5 ×10⁶ cells/mL in 10 µg/mL anti-CD16/32 blocking antibody and incubated at 4 °C for 20 minutes. Antibody cocktail was added, and the cells were incubated for a further 30 minutes at 4 °C. Cells were then washed and resuspended at 0.5 ×10⁶ cells/mL and mixed with 5 µg/mL 7-AAD (AAT Bioquest). To stain bulk splenocytes, the antibody cocktail was comprised of antibodies for CD3, CD4, CD11c, and B220 in 1x PBS with 2% v/v FBS. Details on antibody reagents are provided in the Supplementary Methods. All flow cytometry data was collected on a Guava 12HT EasyCyte Cytometer (EMD Millipore, USA) using a 405 nm laser and 450/45 nm emission filter and analyzed using FCS Express 6.

B cell activation in vitro.

Isolated B cells were labeled with 5 µM CFSE (carboxyfluorescein diacetate succinimidyl ester, BD Biosciences) for 30 minutes at 37 °C, rinsed, and resuspended in supplemented RPMI media. B cells were phenotyped by flow cytometry as above, using anti-B220 (FITC). The cells were cultured at 1×10⁶ cells/mL for 48 hours with 0.2 µg/mL of IL-4 (Pepro Tech) and 10 µg/mL R848 (Invivogen) or with PBS control. Afterwards, the cells were removed from the plate and stained for flow cytometry as described above, using CD40 and CD80.

Ex vivo overlays.

Murine lymph nodes were collected and sliced as previously reported.^52–54 Briefly, inguinal, axial and brachial lymph nodes were collected from male and female C57Bl/6 mice, embedded in 6% low melting point agarose (Lonza) and sliced 300-µm thick on a vibratome (Leica VT1000s, USA). Lymph node slices were immunostained with FITC anti-mouse B220 and Lyve-1 as previously reported.⁵⁵

CD3+ T cells were isolated by using a CD3 negative selection kit according to manufacturer instructions (StemCell Technologies, USA) from splenocytes sex-matched to the lymph node slices. T cells (1×10⁶ cells/mL) in 1x PBS were labelled by incubating with NHS-Rhodamine (1 μg/mL, Thermo Fisher) for 30 minutes at 37 °C. B cells were isolated and labeled with CFSE as above, mixed with labelled T cells and concentrated to 10×10⁶ cells/mL. The cell mixture was then overlaid onto immunostained lymph node slices for 1 hour at 37 °C. To remove excess cells, slices were incubated for at least 30 minutes in 1x PBS with gentle agitation at regular intervals.

Imaging.

Confocal microscopy was performed on a Nikon A1Rsi confocal upright microscope, using a 400 nm laser and 450/50 nm GaAsP detector. Images were captured with a 40x/0.45NA Plan Apo NIR WD objective. Widefield microscopy was performed on a Zeiss AxioZoom upright macroscope, using a Zeiss PlanNeoFluar Z 1x / 0.25 NA FWD 56mm objective, Zeiss Axiocam 506 mono camera, and HXP 200 C illuminator with metal halide lamp (Zeiss Microscopy, Germany). Images were collected with Zeiss filter sets 49 (Ex: 365, Em: 445/50), 38 HE (Ex: 470/40, Em: 525/50), 43 HE (Ex: 550/25, Em: 605/70), and 50 (Ex: 640/30, Em: 690/50). Image analysis was completed using ImageJ software 1.48v.⁵⁶

Results and Discussion

Synthesis of polymeric boron-based dyes

In order to generate polymeric nanoparticles with different surface chemistries, we synthesized an array of polymers based on the blue-emitting boron-based dye BF₂dbm (1–5 in Table 1, Figure 1A, and Scheme S1). We have previously shown that for BF₂dbmPLA, a molecular weight of ~10 kDa for the PLA corresponded to blue fluorescence,²⁹ so polymers in this size regime were targeted. The dye was prepared as either a primary alcohol (BF₂dbmOCH₂CH₂OH)³¹ or a phenol (BF₂dbmOH),⁵⁰ to act as an initiator or a coupler respectively (Figure S1). The initiator BF₂dbmOCH₂CH₂OH was used to grow BF₂dbmPLA (1), BF₂dbmPLLA (2), and BF₂dbmPCL (3) by a solvent-free, tin-catalyzed ring-opening polymerization.³¹ The BF₂dbmPCL product was further used as a macroinitiator for the ring-opening polymerization of lactide to prepare a block copolymer, BF₂dbmPCL-PLLA (4).⁵⁷ Finally, to generate the PEGylated material (5), BF₂dbmOH was coupled to PEG-PLLA via a Mitsunobu reaction.⁵⁸ The polymer molecular weights and polydispersities were determined by GPC and ¹H NMR spectroscopy (Table 1). The polydispersities were relatively low (Đ < 1.3) for all five samples, indicating the polymer chains were relatively uniform in chain length.

Table 1.

Polymer Synthesis Data

Polymer	#	Loadinga	Mnb (GPC)	Mwb (GPC)	Mnc (NMR)		Đb
BF2dbmPLA	1	1/100	12 300	13 000	15 200		1.06
BF2dbmPLLA	2	1/100	12 200	14 800	13 200		1.24
BF2dbmPCL	3	1/130	6 900	7 300	10 400		1.06
BF2dbmPCL-PLLA	4	1/100	7 700	9 000	14 100		1.17
BF2dbmPLLA-PEG	5	1/130	10 500	11 000	14 400		1.05

^aInitiator to lactide loading in the ring opening polymerization. For 1–3, BF₂dbmOCH₂CH₂OH was used as the initiator. For BF₂dbmPCL-PLLA (4), BF₂dbmPCL (3) was used as the initiator. For 5, PEG-OH (2 kDa) was used as the initiator and the dye (BF₂dbmOH) was added post polymerization. See Schemes S1 and S2 for reference.

^bAverage molecular weight (M_n, Da) and weighted molecular weight (M_w, Da) of dye-polymer conjugates determined by gel permeation chromatography (GPC) in THF. Đ = polydispersity index (M_w/M_n).

^cMolecular weight determined by ¹H NMR (PLA-H or PCL-CH₂ vs 4-Ar-H of the dye).

Generation of blue fluorescent nanoparticles

Next, boron nanoparticles were fabricated by nanoprecipitation.¹¹ The polymer was dissolved in a water miscible solvent (DMF), which was added dropwise to DI water. Dialysis against water removed the organic solvent to yield the nanoparticles. Nanoparticle sizes (hydrodynamic radius; R_H) ranged from 37–61 nm (Table 2), in the range suitable for cellular uptake.^59,60 All of the nanoparticles have narrow size distributions with low polydispersity values (0.1 ~ 0.3) and negative zeta potentials.¹¹ As expected, all five fluorescence spectra were similar with only subtle differences in spectral features, with excitation in the UV range (366–397 nm) and blue emission (438–447 nm) (Figure 2, Table 2). For convenience, here and throughout this paper we refer to the nanoparticles formed from the dye-polymer conjugates simply by the name of the polymer. The second, low energy feature observed in the total emission spectra for PCL and PCL-PLLA is commonly observed for PCL-containing samples. This may be due to non-linear, long-range, dye-dye interactions, related to molar mass and matrix effects on the BF₂dbm-fluorophore, as previously described.⁶¹ These spectra make the dyes well suited for use on traditional DAPI filter sets, as well as for excitation via a UV or 405 nm laser by flow cytometry or confocal microscopy.

Table 2.

Nanoparticle Characterization Data

Nanoparticles	Radiusa (nm)	PDa	ζb (mV)	λmaxc (nm)	ελc (M−1cm−1)	λExd (nm)	λFd (nm)
PLA	61	0.13	− 28.062	398	32 100	381	439
PLLA	31	0.13	− 23.6	382	25 600	396	437
PCL	56	0.26	− 22.6	400	25 200	397	440
PCL-PLLA	51	0.12	− 36.9	399	21 100	381	438
PLLA-PEG	37	0.23	− 16.018	396	29 800	366	439

^aDetermined by dynamic light scattering, for nanoparticles at 200 μg/mL in water. Polydispersity (PD) is the standard deviation of the distribution. PD = (peak width/peak height)^2.

^bZeta potentials measured by Zetasizer.

^cAbsorption maximum, λ_max, and corresponding extinction coefficient at λ_max, ε_λ, for aqueous nanoparticle suspension (~ 50 μg/mL).

^dFluorescence excitation and emission maximum for aqueous nanoparticle suspension.

Figure 2:

Optical Properties of Boron Dye Nanoparticles. a) Images of freshly prepared nanoparticles suspended in water under UV illumination (λ_ex = 369 nm). b) Total excitation (solid lines) and emission spectra (dashed lines) of nanoparticle suspensions (λ_ex = 369 nm). Popular excitation sources including a DAPI filter set, 395 nm LED, and 405 nm laser all demonstrate good overlap with the spectra of the nanoparticles.

Physical and optical stability of nanoparticles

We assessed the physical and optical stability of the nanoparticles in aqueous environments suitable for bioimaging, at physiological temperature of 37 °C (Figure 3, Figures S2–S6). PLA nanoparticles show good shelf life with respect to molecular weight, size, absorption, and emission for months when stored at 4 °C in water (Figure S7).¹¹ At elevated temperature, all nanoparticles showed constant size in DI water over one week (Figure 3a). In PBS, however, nanoparticles containing PLA or PLLA polymers aggregated immediately, and the block copolymer PCL-PLLA aggregated after two days of incubation (Figure 3b). As an alternative to saline, we tested a solution of 5 % glucose in water, intended to match the osmotic pressure of the cells; 2 % serum was added to improve cell handling (water/glucose/serum solution). Compared with PBS, this solution greatly increased the stability of nanoparticles: PLA and PCL were stable for two days, and the other three particles maintained their sizes for over one week (Figure 3c). Surprisingly, complete cell culture media, RPMI supplemented with 10 % serum and nutrients (see Methods), better maintained nanoparticle sizes despite its high ionic strength, perhaps because of spontaneous protein coating that protected the nanoparticles against aggregation (Figure 3d). The minor fluctuation in radii observed in complete media, known as swelling and deswelling, could be caused by a concentration gradient of ions (e.g., Na⁺, Cl⁻, amino acids) between the nanoparticle matrix and the surroundings, leading to ion diffusion and osmosis.^63,64 Regardless, nanoparticle radii were maintained consistently.

Figure 3:

Nanoparticle (200 μg/mL) stability at 37 °C in various media. Nanoparticle size was monitored by DLS (left), and emission monitored by fluorimeter (right). For samples with extensive aggregation, particle size exceeds the DLS detection limit of 2000 nm and associated precipitation leads to a weak fluorescence signal that cannot be accurately detected, thus, data are not shown.

Despite their relative stability in terms of size, the fluorescence intensity of all particle types decreased significantly over time in all solvents at 37 °C, and the rate of decay was sensitive to the choice of solvent (Figure 3). Compared to other particles, PLA and PLLA particles retained the most fluorescence intensity over time, and compared to other buffers, they were the most stable in glucose/serum solution and in complete RMPI culture media (Figure 3). In terms of “color,” the peak emission wavelength was stable for PLA and PLLA particles (Figures S2, S3), and was blue-shifted by no more than 15 nm for PCL and PCL-PLLA nanoparticles (Figures S4, S5).

We hypothesized that the observed decay in fluorescence intensity was due to either hydrolysis of the polyester or degradation of the boron dye. Therefore, we tested polymer and fluorophore stability of the PLA and PCL nanoparticles after storage in water for 4 days at 37 °C. The major peaks in the GPC traces for Day 0 and Day 4 show considerable overlap, though a small high-molecular-weight shoulder became more pronounced by Day 4, suggesting increased aggregation, or possibly other side reactions over time, but not polymer degradation (Figure S8). Also, polydispersity index remained low (1.10), indicating little fluctuation in polymer molecular weight and therefore no evidence of polymer hydrolysis. Based on ¹H NMR spectra, the fluorophore in the PLA particle remained unchanged. In the PCL sample, resonances appeared at 6.76 ppm and 16.94 ppm after incubation, which are characteristic of the ArC(O)CH=C(OH)Ar proton and associated enol proton ArC(O)CH=C(OH) in boron-free dbmPCL, respectively. These results indicate chemical and optical stability for BF₂dbmPLA, but hydrolysis of BF₂ from the dbm binding site of PCL.¹¹ In the latter case, dye hydrolysis and the resulting reduction in dye concentration is expected to result in a blue-shifted emission,^29,32 consistent with our experimental results for the PCL nanoparticle. In summary, the physical and optical data suggested that nanoparticle stability in solution varies with the polymer chain and surrounding media. While it is not clear to what extent these results will predict the stability once internalized by a cell, we note that for short-term (hours) cellular imaging, boron nanoparticles have been widely and successfully used in many contexts. Longer-term stability in solution may be possible with additional polymer engineering.

Nanoparticle labelling of primary immune cells

With a good understanding of the particle chemistry and stability in solution, we next quantified the uptake of the nanoparticles by live cells. We excluded the racemic PLA particles from these studies, as they offered surface chemistry similar to PLLA but were less physically stable in solution. Primary murine splenocytes were incubated with each remaining type of nanoparticle, washed, and analyzed by flow cytometry (405 nm laser) immediately (day 0) or after 24 hr culture (day 1). For comparison, cells were labelled with Cell Tracker Blue (CTB), a commercially available, coumarin-derived, small molecule reagent that diffuses passively into the cell and binds covalently with thiol groups in the cytoplasm. Mixed splenocytes incubated with CTB were labelled with high efficiency, while a lower percentage of cells incubated with nanoparticles were labelled, indicating a more selective labelling mechanism (Figure 4a). PCL and PCL-PLLA nanoparticles each labelled ~ 50% of the mixed splenocyte culture, while PLLA nanoparticles labelled only 25% on average. Cellular phenotyping showed that the nanoparticles labelled CD11c-positive and B220-positive cells more efficiently than CD3-positive T cells (Figure 4b, Figure S9). This result suggests that the nanoparticles may label the cells through an active uptake mechanism, as both B cells and CD11c+ cells can act as antigen presenting cells and are more endocytic than T cells.⁶⁵ Indeed, the increased uptake of the hydrophobic PCL-containing particles compared to PLLA particles is consistent with improved internalization of more hydrophobic polymers⁴⁵ PEGylated nanoparticles (PLLA-PEG) labelled very few cells, consistent with PEG preventing binding and cellular internalization of nanoparticles.⁴⁹ Future work will explore the active uptake mechanism as a possible point of control; enabling the targeting of nanoparticles to different cellular subsets.

Figure 4:

Choosing a polymer formula for labelling primary immune cells. (a) Mixed primary splenocytes were labelled with Cell Tracker Blue (CTB), boron-based polymer nanoparticles, or no label. Uptake was measured immediately by flow cytometry. One-way ANOVA with comparisons of each group to CTB control. (b) Uptake of PCL-PLLA by cell type as determined by flow cytometry. More phagocytic cell types were more readily labelled by nanoparticles. (c,d) Stability of cellular labelling in mixed splenocytes after overnight culture as measured by flow cytometry both in terms of fraction of cells labelled (c) and intensity (d). (e,f) Viability of labelled and unlabeled cells in mixed splenocytes (e) and isolated B cells (f) as determined by flow cytometry after overnight culture. Viability is defined as 7-AAD negative. Two-way ANOVA with multiple comparisons. Each dot indicates one biological replicate. **** p<0.0001, *p=0.0148, ns denotes p > 0.05

Next, we tested the stability of cellular labelling. The fluorescence excitation and emission spectra immediately after cellular uptake (i.e. after 30 min incubation) were comparable to nanoparticles in solution (Figure S10). Labelled cells were also monitored after 24 hours. Cell Tracker Blue, like other cytoplasmic dyes, often suffers from a significant decrease in fluorescence intensity after the first 24 hours as unbound dye diffuses out of the cell. Indeed, we observed that the average CTB intensity per cell dropped by two-fold overnight, while there was no decrease in the percentage of CTB-labelled cells (Figure 4c/d). Interestingly the nanoparticle labelled cells showed an opposite effect. The percentage of cells that were labelled with nanoparticles decreased to 75%, on average, of its initial value after overnight culture, while the mean intensity per labelled cell did not (Figure 4c/d). These data suggest that the nanoparticles may be exported by a fraction of the cells. The relative stability of nanoparticle intensity per cell during this time frame compared to CTB labelled populations may be a point of control when transitioning this technology for long-term tracking.

Finally, we addressed whether labelling with nanoparticles affected cellular viability. Average viability in mixed splenocytes was unchanged by labelling with nanoparticles or CTB, except for a 20% reduction in viability of PLLA-labelled splenocytes after 24 hours of culture (Figure 4e). Thus, the nanoparticle-labelling process is compatible with maintaining high viability in primary splenocyte cultures. Based on these data, we selected the PCL-PLLA co-block polymer nanoparticles for further testing, based on its stability in solution, high initial labelling efficiency of the cell types of interest, and cytocompatibility.

Nanoparticle labelling does not affect immune cell function

Having demonstrated that splenocyte viability was not affected by nanoparticle labelling, it was essential to determine whether labelling affected relevant cellular functions. Given their prominent role in adaptive immunity, high frequency in the splenocyte culture, and high rate of staining, we focused on the effect of labelling on B cells in particular. B cells act as “antibody factories,” and upon stimulation proliferate to begin the process of producing the most effective antibody against the pathogen, generating a strong adaptive response. We have shown that labelling with nanoparticles did not affect the viability of B cells in a mixed splenocyte culture (Figure 4f) and so to test the response to stimulation, we used the small molecule R848, which acts through Toll-like receptor 7 (TLR7) and results in proliferation, antibody production, and up-regulation of surface activation markers such as CD40 and CD80.⁶⁶ In preliminary tests, R848 elicited a stronger response in unlabeled B cells than other common stimuli at matched doses (Figure S11).

Purified B cell populations were labelled with either CTB or the PCL-PLLA nanoparticle and stimulated with R848. Proliferation was tracked by CFSE staining, the brightness of which is reduced with each cycle of cellular proliferation. In the absence of R848 stimulation, nanoparticle labelling did not induce B cell proliferation (Figure 5a–b) or activation marker expression (Figure 5c–d, Figure S12), indicating that the particles did not activate the cells on their own. After 48-hr R848 treatment, neither the Cell Tracker Blue nor the nanoparticles suppressed the proliferation and upregulation of activation markers on purified B cells (Figure 5). B cells labelled with PCL-PLLA nanoparticles did proliferate at a higher frequency than the unlabeled cells, but did not exhibit an increase in activation marker expression, indicating that any synergy of the nanoparticle loading with R848 stimulation was modest. In separate experiments, the ability of CD3+ T cells to respond to a non-antigen-specific stimulation was also uncompromised by nanoparticle labelling (Figure S13). Overall these data show that the labelling of primary splenocytes with PCL-PLLA nanoparticles does not compromise immune function.

Figure 5:

Labelling of primary splenocyte B cells did not affect response to activation. (a) Representative histograms of CFSE intensity as measured by flow cytometry. Proliferation was measured as the percentage of cells with reduced CFSE intensity compared to unstimulated cells. (b) Quantification of proliferated cells. Stimulation with R848 resulted in increased proliferation in all labelling schemes. (c) Percentage of live CD40 high cells. (d) Percentage of live CD80 high cells. Stimulation with R848 resulted in higher activation marker percentages in both cases. Two-way ANOVA with multiple comparisons. **** p<0.0001, *** p=0.0003, ns denotes p>0.05.

Nanoparticle labelling is intracellular and bright

For cell tracking applications, labelling with the nanoparticles must be intracellular, as any extracellular particles or aggregates could easily be washed away during handling. Therefore, we determined the distribution of PCL-PLLA nanoparticles in or on labelled splenocytes via confocal microscopy (400 nm laser excitation). Cell membranes were marked with anti-CD45, a surface-bound pan-lymphocyte marker, to visually determine whether each cell was unlabeled, labelled in the cytoplasm, labelled extracellularly (outer surface of membrane), or labelled on the membrane itself (Figure 6a/b). This method of data collection leaves some ambiguity as to whether the “membrane”-associated nanoparticles are in contact with the inner or outer leaflet of the cell membrane or penetrate entirely. As expected for a cytoplasmic dye, all CTB signal was detected within the bounds of the cell membrane. Nanoparticle labelling was either cytoplasmic or associated with the membrane, and none of the latter cells had signal from the nanoparticles protruding out into the surrounding media. Thus, the majority of the signal from the PCL-PLLA nanoparticles is cytoplasmic and is useful for potentially tracking cells. We noted in this experiment that the CTB-labelled cells appeared dimmer than the nanoparticle labelled cells. This difference in brightness is consistent with the poor overlap between CTB and the 400 nm excitation source (Figure S14), though it does contrast with results seen earlier by flow cytometry, likely due to the difference in sensitivity between detectors on the two instruments.

Figure 6:

Imaging PCL-PLLA labelled splenocytes. (a) Quantification of staining location as determined by confocal microscopy (400 nm excitation). (b) Representative images of labelling location. Cells were co-labelled with AF488 anti-CD45 antibody (green) to determine membrane location. Brightness and contrast differs between images to account for different excitation efficiencies. Scale bar is 10 μm. (c) Quantification of intensity using widefield microscopy (360 nm excitation) of labelled and unlabeled cells. PCL-PLLA labelled cells were on average 1.43 times brighter than CTB labelled cells. Each dot represents one sample averaged over 200+ cells. Ordinary one-way ANOVA with multiple comparisons. ** p=0.0029, *** p=0.0001, n.s. denotes p>0.05 (d) Representative images of cells labelled with CTB and PCL-PLLA nanoparticles with identical brightness and contrast settings. Cells were co-labelled with Cell Trace Far Red to locate cells efficiently (purple outlines). Large aggregates of PCL-PLLA particles were detected in this labelling scheme (arrowhead). Scale bar is 10 μm.

The original goal was to develop a very bright blue labelling system for fluorescence microscopy, as many blue dyes suffer from low quantum yields.⁶ Thus, we measured the fluorescence intensity of labelled cells when imaged by widefield microscopy under standard UV excitation (360/60 nm filter). The intensity of cells labelled with PCL-PLLA nanoparticles was 1.43-fold brighter than cells labelled with CTB. In fact, CTB-labelled cells were not significantly brighter than unlabeled cells under these imaging conditions (Figure 6c). In this experiment, we did note large aggregates of nanoparticles (arrow head in Figure 6d) outside of the cells, which were difficult to remove. We speculate that this aggregation arose from the experimental procedure, in which cells were resuspended in 1x PBS for dual-labelling with Cell Trace Far Red after labelling with the blue reagent; 1x PBS causes free nanoparticles to aggregate quickly. Further experimentation is needed to determine the best approach to minimize extracellular aggregate formation during cellular manipulation, for example by keeping cells in protein-rich media.

Nanoparticle labelled cells are brightly visible against tissue autofluorescence

When working with typical blue fluorophores, their low fluorescence intensity is especially challenging to detect against tissue, a complex matrix that exhibits high background.⁶ We hypothesized that the bright-blue fluorescent nanoparticles would not suffer from this limitation. We tested the ability to detect nanoparticle-labelled cells embedded in living tissue with widefield microscopy, by labelling purified B cells and overlaying them onto live slices of murine lymph node as a model system.⁵⁵ As in the in vitro imaging above, the PCL-PLLA nanoparticle labelled cells were significantly brighter than both unlabeled and CTB labelled cells (Figure 7a) and easily detectable against the autofluorescence of the tissue.

Figure 7:

Imaging PCL-PLLA labelled cells in tissue. (a) Quantification of labelled cells against tissue background. Nanoparticle labelled cells were significantly brighter after background subtraction compared to both CTB and unlabeled cells. Dotted line represents the average intensity of tissue autofluorescence in the DAPI filter. Two-way ANOVA with multiple comparisons. **** p<0.0001, ns p>0.05. (b) Image of representative stained tissue slice (3x mag). The live tissue was immunostained with FITC anti-B220 (white), and eFluor660 anti-Lyve-1 (false-colored pink) to show B cell follicles and lymphatics. (c) View of a single B cell follicle within the tissue slice (10x mag). Here we can visualize overlaid B cells labelled with PCL-PLLA nanoparticles (false-colored green) and T cells labelled with NHS-Rhodamine (false-colored blue). (d) Individual channel data from (c). Some nanoparticle aggregates align well with the Lyve-1 signal (dashed ovals).

The nanoparticles allowed us to capture bright signal within the blue channel of the widefield fluorescence microscope (360/60 nm excitation), freeing the remaining three channels for additional fluorophores to monitor multiple cell populations and tissue structures. As a proof-of-principle, we labelled purified B cells with the PCL-PLLA nanoparticles, labelled purified T cells with a rhodamine-based cytoplasmic dye (NHS-rhodamine), and overlaid these cells on a slice of lymph node tissue that was live immunostained for B220 (B cell zones) and Lyve-1 (lymphatic vessels) (Figure 7b–d). With this multi-color staining approach, we could visualize individual B cell follicles and the mixture of B cells (green) and T cells (blue) on the tissue. The fluorescent puncta from nanoparticle-labelled cells were smaller than the diameter of the cytoplasmic-labelled cells, since the particles do not fill the entire cytoplasm. Interestingly we also observed a collection of small signals in the nanoparticle channel that aligned with the lymphatic staining (Figure 7d, dashed oval). We speculate that this signal comes from free nanoparticles or aggregates coming into contact with the endocytic cells that line the lymphatic vessels in the lymph node.⁶⁷ This limitation should be taken into consideration when translating this technology in the future; it could also be taken advantage of to label endocytic cells in live tissue slices. Overall, we are able to easily visualize all four fluorophores simultaneously by widefield microscopy, using a set of four standard fluorescence filter sets.

Conclusions

This paper described the fabrication and the characterization of a family of five bright-blue fluorescent boron-based polymeric nanoparticles. The nanoparticles exhibited high excitation and emission efficiency at wavelengths that match well with a variety of traditional sources for microscopy and flow cytometry, including a 405 nm laser, 395 nm LED and 360/60 nm excitation filter. Given nanoparticle stability was best in glucose solution without saline at 37 °C, this medium was selected for the cell labeling process, after which labelled cells were resuspended in culture medium. Primary splenocytes, particularly the phagocytic cell types, were well labelled by these particles and remained brightly fluorescent overnight with no detectable impact on viability or ability to respond to stimulation. PCL-PLLA co-block polymer-based nanoparticles had the highest labelling efficiency and cellular retention over time. Nanoparticle labelling was cytoplasmic and is significantly brighter than cells labelled with Cell Tracker Blue, a commercially available cytoplasmic blue dye, when imaged by both confocal and widefield microscopy. Cells labelled with PCL-PLLA were easily distinguished against tissue autofluorescence, which enabled straightforward four-color imaging. In the future, we will use this approach to image B and T cell interactions within ex vivo lymph node samples and map where these interactions occur. Broadly, these bright blue fluorescent nanoparticles expand the toolbox for cellular labelling and tracking in multi-color imaging experiments and may find many applications in a variety of tissues. Potential future improvements include varying the surface chemistry of the nanoparticles to enhance uptake by non-phagocytic cells, and adding functional groups such as succinimidyl esters or maleimides to increase dye retention within the cells.