Blood-brain-barrier permeable fluorescent astrocyte probes

Abstract

The brain’s astrocytes play pivotal roles in brain homeostasis, but their contributions to neural circuit function are only beginning to be explored. In order to study astrocytes in their native environment, the field needs imaging probes that are cell-specific, non-toxic, and complementary to existing genetic techniques. We recently described a panel of cationic fluorophores that selectively label rodent and zebrafish astrocytes at low concentrations and short incubation times. Here, we expand on those first-generation astrocyte probes with a new class of brighter and more photostable dyes that allow for imaging across various wavelengths. We demonstrate these probes’ subcellular localization to the mitochondria and ability to cross the blood-brain barrier in zebrafish. This new class of astrocyte markers will aid researchers in visualizing and identifying subpopulations of astrocytes in both rodent and zebrafish model systems.

Astrocytes are the most abundant glial cell-type in the brain and have critical functions in every layer from the olfactory bulb to the cortices and midbrain.¹ They are active players in diverse brain processes like hormone metabolism, ion buffering, and neuronal signaling.^2,3 Given their important roles in brain function, it is not surprising that astrocytes are implicated in many diseases, including Alzheimer’s, Parkinson’s, glioblastoma, stroke, seizure, and depression.⁴ Accordingly, it is essential to be able to visualize and study these cells in their native environments so we can both better understand their biology and develop therapeutic interventions for astrocyte-associated disease.

While there exist a handful of astrocyte-specific genes whose promoters are utilized for genetic targeting and visualization (e.g., GFAP⁵, AldH1L1⁶, S100β^7,8), the process of creating transgenic organisms is time consuming and costly, and promoter-driven fluorescent reporters often fail to act as pan-astrocyte markers⁶ due to the heterogeneity of the astrocyte cell-type. Indeed, recent studies are beginning to uncover that astrocytes vary greatly between and within brain regions, expressing different genetic markers and performing unique functions.^9,10 In addition to the genetic markers used for studying astrocytes, researchers have recently explored the use of small molecule probes for astrocytes. By utilizing small molecule astrocyte-specific fluorophores like Sulforhodamine 101^11,12, Ala-Lys-Coumarin¹³, Cy-BASHY¹⁴, Rhodamine Q-MP¹⁵, and Rhodamine B-MP^16,17 researchers can complement the use of transgenic animals, image across a spectrum of visible and infrared wavelengths, and probe astrocytes in vivo.

We recently described a library of cationic dyes that leverage a pyridinium moiety to achieve astrocyte-specific labeling via astrocyte-resident organic cation transporters.^16,17 Diverse fluorophores including rhodamines and cyanines tagged with the pyridinium group produced strong labeling of mouse, rat, and zebrafish astrocytes, but were excluded from other glia and neurons.^15,16 Herein, we report next generation astrocyte probes based on the brighter, more photostable Janelia Fluor^® fluorophores^18–20 combined with an added pyridinium to drive astrocyte selectivity. We demonstrate the astrocyte-specificity of these fluorophores in cell culture and in vivo. We report labeling in cell culture and mitochondrial colocalization after only 20 minutes of incubation at low micromolar concentrations. Additionally, we demonstrate these probes’ ability to cross the blood brain barrier (BBB) in zebrafish in order to label their astrocyte targets after intravenous injection. The versatility in wavelength and structural modularity of this approach has the potential to uncover information regarding the heterogeneity of cells classified as astrocytes and provide a toolkit for efficient and easy live imaging without the need for specialized microscopy or invasive delivery methods.

We chose to modify three Janelia Fluor^® probes with excitation wavelengths ranging from 503 to 585 nm to demonstrate targeted labeling of astrocytes using a subset of these next generation fluorophores. Following previously reported synthetic methods, we began with commercially available fluorescein for Janelia Fluor^® 549 and 503.^18,21 We then appended the methylpyridinium moiety that is responsible for astrocyte targeting via HATU amide coupling in DMF to obtain 549MP, 4a, and 503MP, 5a in 29% and 19% yield, respectively (Figure 1A). For the synthesis of 585MP, 6a, we followed previously reported synthetic methods to obtain the triflated carbofluorescein backbone over 8 steps.¹⁸ Our attempts at HATU coupling with the latter probe proved unsuccessful, probably due to its near-zero L-Z equilibrium constant, favoring the lactone-form and therefore limiting access to the carboxylic acid as a coupling handle.¹⁹ Structurally, these fluorescent probes differ either in their xanthene substituent or in their appended nitrogen-containing groups on the xanthene ring from our previously reported probe, Rhodamine B methylpyridinium (Rh_BMP)¹⁶. The reported photophysical properties of the second-generation parent fluorophores, JF503, 549, and 585 are superior in comparison to the parent fluorophore of our first probe, Rhodamine B, in their quantum yield and extinction coefficients¹⁹, which is what inspired us to make these fluorophores amenable to live-imaging in astrocytes.

Figure 1. Synthesis and cellular-specific uptake of methylpyridinium-modified fluorophores.

(A) General synthetic scheme for compounds 4a (549MP), 5a (503MP), and 6a (585MP) using HATU coupling or oxalyl chloride, respectively. (B–C) Compound 4a labels HEK293T cells expressing OCT1 (B), but not cells expressing control plasmid myc-eGFP (C) after 20 minutes of bathing at 1 μM. (D–E) Compounds 4a (D) and 6a (E) label primary mouse cortical astrocytes after 20 mins of bathing with 1 μM. Scale = 50 μm (B), 25 μm (D).

We sought to analyze the probes’ performance in cell culture considering their similar structure, yet superior quantum yield, extinction coefficient, and varying excitation/emission spectra which would make for a diverse tool-kit of brighter, more photostable^19,20 and spectrally-diverse astrocyte-targeted probes. We first evaluated 4a (549MP) in HEK293T cells with or without the overexpression of Organic Cation Transporter 1 (OCT1), which our previous studies have shown is sufficient for cellular uptake of the molecules^16,17. HEK293T cells transfected with OCT1 were successfully labeled with 4a (Figure 1B and S1), whereas HEK293T cells transfected with a GFP control did not take up the compound (Figure 1C and S1), suggesting that it is not cell permeable and is able to transit the OCT1 like its predecessors. We next evaluated 4a–6a for their ability to label primary astrocytes in culture. In these experiments, 4a and 6a successfully labeled astrocytes in primary cultures derived from mouse cortex at a working concentration of 1 µM (Figure 1D, E and S2). Despite its structural similarities, 5a did not show either specific or diffuse labeling in astrocytes, HEK293T cells, or primary cortical culture suggesting that it is neither astrocyte-targeted nor cell permeable (Figure S2 I–J). We have demonstrated previously that the presence of a +2 charge on the cargo molecule results in more robust targeting of astrocytes^16,17, which might explain the inability of the rhodol, 5a, to label these cells due to its overall +1 charge. On the other hand, 6a shows labeling of astrocytes while bearing a +2 charge and two 3,3-difluoroazetidine moieties at the xanthene ring, albeit less robust than the flagship 4a (Figure 1E and S2 E–F). The electronegative fluorine atoms may change the distribution of electron density, which may decrease uptake by the organic cation transporter responsible for astrocyte labeling.

Next, we evaluated the cytotoxicity of compounds 4a, 5a, and 6a using the colorimetric MTT cell viability assay. At a concentration of 1 μM and exposure time of 20 min, the conditions we employ for cultured astrocyte labeling, we observed 90–100% cell viability for all fluorophores in HEK293T cells expressing OCT1 (Figure S3).¹⁶ In the most extreme case, viability in the MTT assay drops to approximately 85% for 6a-treated cells when we increase the concentration to 10 μM, 10-fold above the concentrations used for cell labeling, suggesting the probes have minimal toxic effects in mammalian cells (Figure S3 A).

We next sought to evaluate the subcellular distribution of compounds 4a and 6a in labeled cells. In previous reports with Rh_BMP, and here with compounds 4a and 6a (Figure 1D, E, and S2), we observed a punctate labeling pattern in primary mouse and zebrafish astrocytes¹⁶. We also observed a similar subcellular distribution in HEK293T and HeLa cells transfected with OCT1 (Figure 1B, C, and S1). This punctate labeling suggested organelle-specific rather than non-specific or cytosolic subcellular localization. We hypothesized that punctate labeling was mitochondrial, since the pattern is consistent with mitochondrial labeling in cultured cells and there are several small molecule probes for mitochondria that possess a positive charge (Rhodamine 123 and.tetramethylrosamine²²). To evaluate mitochondrial labeling, we performed a colocalization experiment with MitoTracker^® and compounds 4a and 6a. After labeling mouse primary cortical astrocytes with 1 µM of the compounds for 20 min, we observed a clear colocalization of 4a and our first-generation probe, Rh_BMP with the mitochondrial dye, MitoTracker^® via confocal microscopy (Figure 2 and S4). This result suggests that fluorophores of varying excitation/emission patterns can be modified in order to target the mitochondria of astrocytes. Both 4a and our first-generation probe, Rh_BMP strongly label the mitochondria of mouse cortical astrocytes, while 6a does not appear as robust (Figure 2 and S4). The percentage of pixels of 4a, 6a, and Rh_BMP that colocalize with MitoTracker^® are 90%, 95%, and 93%, respectively, as calculated by the Mander’s colocalization coefficient (MCC) (Figure S4)^23,24. Moving forward, we decided to focus our efforts on the flagship molecule 4a since it produced the most robust labeling of astrocytes and their mitochondria. Taken together, these probes can be utilized for imaging studies where transgenic organisms may not be available or where visualization of both astrocytes and their mitochondria is desired. These results also give insight into the types of molecules that mitochondria are likely to take up, and imply similar modifications researchers might make in order to target other small molecule cargo to astrocytes.

Figure 2. Fluorescent probes colocalize to mitochondria in mouse primary cortical astrocytes.

After 20 mins incubation at 1 μM, 4a (A–C), Rh_BMP (D–F), and 6a (G–I) colocalize with MitoTracker in mouse primary cortical astrocytes. Scale = 25 μm.

After confirming that the dye was astrocyte-specific and observing mitochondrial-targeted labeling, we sought to move in vivo to assay its performance in a whole organism model system. Zebrafish are frequently used in brain-imaging studies due to their ability to produce a large number of embryos quickly and their transparency in larval stages after exposure to the drug N-Phenyl-2-thiourea (PTU). It is well-known that zebrafish contain cells homologous to mammalian astrocytes that express both GFAP and regulate neuronal signaling.^25–27 Upon direct intracerebroventricular injections of our flagship molecule, 4a, we see clear colocalization with the GFAP+ astrocytes of 3–5 days post fertilization (dpf) larval zebrafish, corroborating our cell culture results (Figure 3 and S5). Importantly, no overlap is seen between the probe and neurons, oligodendrocytes, or microglia in this whole-organism system (Figure 3H–J and S5), suggesting that 4a is indeed astrocyte specific in both mammalian cell culture and in the larval zebrafish brain.

Figure 3.

Injection of flagship molecule 4a (549MP) directly into the brain ventricle of 3–5 dpf larval zebrafish results in astrocyte uptake and mitochondrial colocalization. (A) Schematic of a left-facing zebrafish indicating site of ventricle injection (2.3 nL of 100 μM solution). (B–D) Grayscale image of 4a distribution (B), GFAP+ astrocytes (C), and pseudo-colored overlay of both channels (D) showing colocalization between 4a and astrocytes in the telencephalon. (E–G) Grayscale image of 4a labeling (E), GFAP+ astrocytes (F), and pseudo-colored overlay (G) of both channels showing overlap in the hindbrain region. Colocalization is not seen between neurons (H), microglia (I), or oligodendrocytes (J) in the hindbrain. (K) After co-injection of both 4a (600μM) and MitoTracker (400μM), colocalization is seen in the telencephalon. (L–N) Inset from (K) showing 4a labeling (L), MitoTracker labeling (M), and a pseudo-colored overlay (N), indicating overlap of the two fluorophores. (O) Spontaneous and photic-evoked larval swim behavior assay. Line graph showing average movement count of 6 dpf larval zebrafish at 18 hpi of 4a (n = 10), 6a (n = 10), Rh_BMP (n = 10), and a water control (n = 10). Scale = 25 μm (B, E).

To assay for high level behavior effects of these molecules, we used a spontaneous and light-evoked locomotion paradigm where spontaneous movement is monitored in zebrafish larvae at 6 dpf. Larvae acclimate to a behavior chamber for 20 mins in the light and are then recorded for 15 mins of spontaneous movement followed by removal of illumination, which elicits a visual motor response, a stereotypical zebrafish behavior.^28,29 This allows for the quantitative assessment of the molecules’ effects on normal zebrafish behavior. All larvae exhibit the typical visual motor response, demonstrated by a temporary increase in movement at 15 mins when light switches to dark (Figure 3O). We see no significant difference in movement count, duration, and distance between larvae injected with our probes versus a water control at 18 hours post-injection (hpi) in the acclimation and dark periods (Figure S5 Q, S, and S6 B, D, F, H). We do see a small increase in movement count during the light-period, between 4a and control (Figure S5 R) in addition to increases in the duration of movements during the light period between all probes and the control (Figure S6 C). Taken together, fish injected with the fluorescent probes do not show deficits in their locomotor behavior. Additionally, we see no significant difference in additional parameters including distance traveled (Figure S6 E–H), swim length, and swim speed (Figure S7), suggesting that fish injected with our probes do not show behavioral deficits in their overall movement patterns.

Since we had seen mitochondrial colocalization (Figure 2) in primary cortical astrocytes, we sought to assay the probes cellular compartmentalization in vivo, as well. Accordingly, when both MitoTracker and 4a are co-injected into wild-type (WT-AB) fish, strong colocalization is seen in the telencephalon, optic tectum, and hindbrain regions (Figure 3K and S9 A–F). Furthermore, when MitoTracker, alone, is injected into GFAP-GFP+ fish, we see that it does indeed label GFAP+ cells, of course non-exclusively, since it is meant to label all cells’ mitochondria (Figure S9 S–X). Undeterred by our unexpected in vitro results with 5a and 6a, we decided to assess their performance in vivo. The rhodol 5a produces labeling (Figure S8 B–C) and mitochondrial colocalization (Figure S9 M–O), especially in the telencephalon and hindbrain, albeit less robust and at a higher injection concentration of 1 mM. Moreover, 6a labels astrocytes (Figure S8 D–I) over neurons and other glial cells (Figure S8 J–R) after a 1 mM injection, and is also phagocytosed by microglia as demonstrated by its engulfment by cells marked by GFP under control of the mpeg1 microglial-specific-promoter (Figure S8 M–O). Moreover, we show that 6a colocalizes with mitochondria in hindbrain (Figure S9 P–R) and our first-generation probe, Rh_BMP, exhibits strong mitochondrial colocalization in both telencephalon (Figure S9 G–I) and optic tectum (Figure S9 J–L). Taken together, these results suggest that our molecules specifically label the mitochondria of astrocytes in zebrafish in addition to mammalian cells in culture.

We next sought to evaluate these astrocyte-targeted fluorophores’ blood-brain barrier (BBB) permeability by injecting them systemically through the pericardial vein and assessing labeling in the CNS. Zebrafish are the smallest vertebrate model with a functional BBB and studies suggest histological and ultrastructural similarities between rodent and zebrafish BBB, making zebrafish a good model for predicting small molecule BBB permeability.^30,31 The zebrafish BBB begins functioning as early as 2.5 dpf, contains mature endothelial cells, pericytes, and contacting glia by 5 dpf, and a more complex double-layer vessel structure by 10 dpf.^30,32,33 We injected 2.3 nL of a 1 mM solution of 4a–6a and Rh_BMP, which resulted in rapid distribution of the compounds throughout the vasculature of 7 dpf larval zebrafish (Supplemental Movie 1). After allowing the fish to recover for 3 hours we observed robust labeling of 4a in the brain, indicating the molecules’ ability to cross the BBB in this whole organism system, despite its size and charge (Figure 4B–D and S10 A–F). Additionally, Rh_BMP, 5a, and 6a are able to cross the BBB to label their astrocyte counterparts (Figure 4E–H and S10 G–S). Conversely, the unconjugated JF549^® and structurally similar MitoTracker^® Red FM are both unable to label the brain at the same concentration of 1 mM, showing the BBB is indeed intact, for similarly sized molecules (Figure 4I–J and S10 T–X). This suggests that systemic delivery of these fluorophores might be a reasonable, minimally-invasive route of administration when moving forward into mammals like rodents in order to visualize astrocytes for further live studies since they are able to cross the BBB. This also raises the question of if chemical modification with the methyl-pyridinium moiety is sufficient to permit uptake of other small molecule cargo across the BBB which could be useful for targeted delivery of drugs, for example, to the brain.

Figure 4. Fluorescent probes cross the BBB and label astrocytes after systemic introduction into 7 dpf GFAP-GFP zebrafish.

(A) Schematic showing injection location of the fluorophores at 2.3 nL of a 1 mM solution. (B–D) Crossing of the BBB and colocalization of 4a (B) with astrocytes (C) in the hindbrain region of the fish is seen 3 hpi. (D) Overlay of (B) and (C). (E–G) colocalization is seen between 6a (E) and astrocytes (F) in the telencephalon region of the fish. (G) Overlay of (E) and (F). (H) Similarly, first generation probe, Rh_BMP crosses the BBB to label astrocytes in the telencephalon. The control molecule, MitoTracker, is seen in the vasculature, but not in the brain, in both telencephalon (I) and dorsal to optic tectum (J) region. Arrowheads indicate vasculature labeling outside of the brain. Scale = 25 μm.

In summary, we have synthesized new astrocyte-specific fluorophores and have shown that our small library selectively marks mitochondria and has the ability to traverse the BBB in zebrafish. We anticipate expanding the utility of this targeting method to additional fluorophores that span the visible and infrared-light spectrum so long as they can be modified to achieve a permanent +2 charge containing the methyl-pyridinium astrocyte-targeting moiety. Additionally, these experimental results suggest there is potential to add this targeting moiety to other biologically relevant molecules like transcription activators, drugs, or calcium indicators for targeted delivery to astrocytes, or where an increase in BBB permeability is desired, since delivery of molecules to the brain is a central challenge faced in drug-delivery. Traceless delivery of small molecules that can be uncaged using light or enzyme stimuli is of particular interest, resulting in spatiotemporal control of fluorescent imaging, drug delivery, or transcription activation. It is also of use to researchers to classify subpopulations of astrocytes considering their heterogenous nature, in order to inform the diverse functions of these cells. Elucidating this sub-population and obtaining more information about the mechanisms of transport into astrocytes is of particular interest, since it can help guide further cargo diversification.

ABBREVIATIONS

BBB

blood-brain-barrier

CNS

central nervous system

DMF

dimethylformamide

dpf

days post-fertilization

GFAP

glial fibrillary acidic protein

GFP

green fluorescent protein

hpi

hours-post injection

OCT

Organic Cation Transporter

HATU

1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxid hexafluorophosphate

MCC

Mander’s colocalization coefficient

mM

millimolar

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide

PTU

phenylthiourea

μM

micromolar

Funding Statement

NIH T32GM136572, Boehm fellowship to DAC and NG.