To determine the pharmacological characteristics of the new MI agents targeting TSPO that we are studying, we required a cell line that expresses relatively high levels of TSPO. Using qPCR, we found that DBT cells, a highly malignant mouse astrocytoma cell line 
, express high levels of TSPO mRNA compared to two other mouse astrocytoma cell lines, D30 and D1A cells () 
. By comparison, levels of TSPO mRNA in DBT cells were also higher than in healthy mouse brain, mouse microglia in primary culture and BV-2 cells (a mouse microglia cell line) () 
. High TSPO expression in DBT cells was confirmed by filtration radioligand binding assays using [3
H]-PK 11195 (we detected 14 pmol of TSPO per
gram of protein, , ). Increasing concentrations of PK 11195 and DAA1106 competed for [3
H]-PK 11195 binding with Ki
of 2.0 and 0.2 nM, respectively (), values that are well within the range of what has been reported 
. These results show that DBT cells express relatively high levels of TSPO and therefore constitute a reliable cell model to determine the pharmacological characteristics of agents targeting TSPO with nanomolar affinity.
qPCR determination of TSPO mRNA levels in cells and tissue.
Ability of NIR-conPK and NIR-6T to compete for [3H]-PK 11195 binding to DBT cell homogenates.
Kd and Bmax of [3H]-PK 11195 binding using homogenates prepared from cytokine-treated DBT cells.
Using DBT cell homogenates, we then tested whether NIR-conPK and NIR-6T compete for [3H]-PK 11195 binding, and found that NIR-conPK was relatively ineffective (with 10 µM displacing only ~15% of [3H]-PK 11195 binding) (), whereas NIR-6T competed for [3H]-PK 11195 binding with a Ki of 412 nM (). These results show that the addition of an NIR moiety to con-PK greatly reduces its affinity at TSPO (to the point that it does not significantly bind to this protein anymore). Furthermore, the addition of an NIR moiety to 6T reduces its affinity by ~2000 fold, but this MI agent still binds significantly to TSPO.
Radioligand binding performed on cell homogenate represents the gold standard for the pharmacological characterization of the binding of ligands to their targets 
.Yet evidence shows that the local microenvironments found in intact cells, which include local ionic concentrations and protein-protein interactions, greatly influences the pharmacological characteristics with which ligands bind to their targets 
. This prompted us to test whether NIR-conPK and NIR-6T bind to TSPO in intact DBT cells. To do so, we developed an assay to measure specific binding in intact cells by quantifying NIR fluorescence, and varied the parameters required to optimize specific binding based on what had been described for radioligand binding assay performed in intact cells 
. Specifically, we cultured DBT cells in 96 well plates, and optimized growing and total binding conditions by varying cell density, culture media, presence or absence of serum, type of plates, coatings plates with extracellular matrices, pre-incubation times and challenge versus
competition experiments. We opted for plating the cells at ~80% confluence at the time of experimentation, in defined media void of serum and in optical black-sided well plates, all of which resulted in the least non-specific binding to the plate, reduced light scattering and gave the greatest percent of specific binding. Thus, we incubated the cells under those conditions with either NIR-conPK or NIR-6T, rinsed the cells once (to remove unbound MI agents) and then quantified the amount of MI agent that had remained bound to intact cells with a LI-COR Odyssey® Infrared Imaging system 
The first parameter that we sought to determine was the amount of time required for these MI agents to reach binding equilibrium with intact cells. Total binding of NIR-conPK to DBT cells started to plateau after 60 min, while no such equilibrium was reached with NIR-6T, even after 4 hrs (). Second, when using a 60 min incubation time, we verified that increasing concentration of both NIR-conPK and NIR-6T resulted in linear increases in fluorescence, and indeed found a linear correlation (Fig. S2
). These results show that both MI agents bind to intact DBT cells in a time- and concentration-dependent manner, although NIR-6T does not reach equilibrium, even after several hrs.
NIR-conPK and NIR-6T binding in intact DBT cells.
To determine how much of this total binding of NIR-conPK and NIR-6T to intact cells is due to their specific binding to TSPO, we pre-treated intact DBT cells with either PK 11195 (10 µM) or DAA1106 (10 µM) to saturate TSPO, and assessed whether this pre-treatment reduces the amount of NIR-conPK and NIR-6T bound to intact cells. Remarkably, we found that the total binding of NIR-6T was not affected by either PK 11195 or DAA1106 (), whereas the total binding of NIR-conPK was reduced by ~25% by PK 11195 and ~50% by DAA1106 (), suggesting that NIR-conPK exhibits some amount of specific binding. Increasing the concentration of NIR-conPK yield a Kd of 120 nM when using an excess of PK 11195 () and a Kd of 85 nM when using an excess of DAA1106 (). There are two possible interpretations of these results. One is that the pharmacological characteristics of NIR-conPK and NIR-6T binding to TSPO are opposite when using either cell homogenates or intact cells, with only NIR-6T binding to TSPO in cell homogenates, while only NIR-conPK binds to TSPO in intact cells. Another interpretation is that neither NIR-6T or NIR-conPK bind to TSPO in intact cells, and NIR-conPK binds to a protein distinct from TSPO that exhibits a pharmacological profile closely related to that of TSPO (since its binding is competed by both PK 11195 and DAA1106).
To distinguish between these two possibilities, we sought to determine if the amount of specific binding measured for NIR-conPK correlates with TPSO expression. To perform these experiments, we used primary microglia in culture, because a positive correlation between immune cell activation and increases in TSPO expression has been reported 
. Thus we treated primary microglia in culture with TNFα, IFNγ or IFNγ/TNFα, all of which activate microglia toward a pro-inflammatory phenotype, or with TGFβ2, which inactivates microglia. Twenty four hrs later, we quantified the amount of specific binding obtained with NIR-conPK, as well as the expression levels of TSPO (by both qPCR and western blot). We found that IFNγ led to a significant increase in the amount of specific binding exhibited by NIR-conPK in intact microglia, whereas TNFα, TNFα+IFNγ and TGFβ2 did not (). Importantly, the increase in NIR-conPK specific binding was not associated with a change in TSPO protein and mRNA ( and Fig. S3A
). Please note that the treatment of microglia with these cytokines did not change cell density (Fig. S3C
). Thus, changes in NIR-conPK specific binding do not correlate with changes in TSPO expression, and are likely due to the change in expression of a distinct protein.
No correlation between changes in the specific binding of NIR-conPK and TSPO expression.
We then performed similar experiments in DBT cells and also found no correlation, since here IFNγ+TNFα and TGFβ2 treatments led to a significant decrease in the amount of NIR-conPK specific binding in intact DBT cells (), without concomitant change in TSPO expression or cell density ( and Fig. S3B
,D). In this case, because of the ease of harvesting large amounts of DBT cells compared to primary microglial cells, we also performed radioligand binding experiments. shows that cytokine treatment did not significantly change the amount of TSPO (Bmax), nor did it affect the affinity of PK 11195 at TSPO (Ki). This later point is important because such a change in affinity in response to cytokines has been reported and could have accounted for the increase in NIR-conPK specific binding following cytokine treatment 
). Together these results show that the cytokine-induced increases in NIR-conPK specific binding do not correlate with changes in either TSPO expression or the affinity of PK 11195 at TSPO, reinforcing the conclusion that this novel MI agent binds to a protein distinct from TSPO.
In the final part of our study, we sought to better understand the neuroimmune function played by the activated microglia expressing more protein targeted by NIR-conPK. To do so, we sought to correlate increases in NIR-conPK specific binding with increases in immune effectors released by microglia. Specifically, under basal conditions, primary microglia in culture released 5 immune effectors (out of the 24 that we measured), namely IL-1α, IL-6, monocyte chemoattractant protein 1 (MCP-1), keratinocyte chemoattractant (KC) and RANTES (also referred to as chemokine (C-C motif) ligand 5, CCL5) ( and Table S1
). When treating primary microglia with IFNγ, TNFα, IFNγ+TNFα or TGFβ2 for 24 hrs, we found that these stimuli lead to differential increases in the release of IL-1α, IL-6, and RANTES, and did not affect the release of KC and MCP-1 ( and Table S1
). Remarkably, only the cytokine-induced release of IL-1α correlated with the cytokine-induced increase in NIR-conPK specific binding ( and ). These results suggest that only a particular subtype of microglia activation profile (here typified by IL-1α release and induced by IFNγ) is associated with an increase in the protein targeted by NIR-conPK.
Cytokine and chemokine release by activated microglia.