2.1.

Cell Culture

Mouse bone marrow-derived macrophages (BMDM) were isolated from femurs of female, 6- to 12-week-old C57BL/6J mice and cultured as described²⁸ in medium containing 10% fetal bovine serum (FBS), $100\mathrm{U}/\mathrm{ml}$ penicillin, $100\mu \mathrm{g}/\mathrm{ml}$ streptomycin, and 10% media conditioned by Ltk-cells expressing recombinant MCSF ectopically. On day 7, $3\times {10}^5$ BMDMs were seeded on glass coverslips. The following day, cells were transferred to similar media containing either lipoprotein-depleted bovine serum (LPDS; Alfa Aesar) or normal FBS and then stimulated with cytokines. Lipopolysaccharides (LPS; Sigma) and interferon-$\gamma$ ($\mathrm{IFN}\text{-}\gamma$; R&D) were used at $10\mathrm{ng}/\mathrm{ml}$, and IL-4 and IL-13 (Biolegend) were used at $20\mathrm{ng}/\mathrm{ml}$. After 24 h of cytokine stimulation, media was replaced (with lipoprotein-depleted or normal media, per the experimental condition) and cells were exposed to 0 or $40\mu \mathrm{g}/\mathrm{ml}$ oxidized human low-density lipoprotein (Alfa Aesar) for the remaining duration of the experiment.

2.2.

Immunofluorescence Staining

Samples were fixed in 100% methanol and washed with 1% bovine serum albumin (BSA; MP Biomedical) in PBS. Cells were blocked with 5% normal donkey serum and stained for arginase-1 with a goat polyclonal primary antibody (1:50; Santa Cruz) and Alexa 594-conjugated donkey anti-goat IgG (1:500; Jackson ImmunoResearch). After washing in 1% BSA, cells were blocked a second time with 5% normal goat serum and stained for iNOS with a rabbit polyclonal primary antibody (1:50; Santa Cruz) and Alexa 488-conjugated goat anti-rabbit IgG (1:200; Jackson ImmunoResearch). Nuclei were visualized with Hoechst 33342.

2.3.

Fluorescence Lifetime Imaging Microscopy, Data Acquisition, and Processing

Fluorescence lifetime images were acquired with a Zeiss LSM 710 microscope (Carl Zeiss, Jena, Germany). A Titanium:Sapphire MaiTai laser (Spectra-Physics, Mountain View, California) with 100 fs pulses and 80 MHz repetition rate was used as a two-photon excitation source at 740 nm. The laser light was focused through a $40\times$, 1.2 N.A., water immersion objective (Carl Zeiss, Oberkochen, Germany). The autofluorescence was detected with a photomultiplier tube (H7422P-40, Hamamatsu, Japan) after passing through a bandpass $460/80\mathrm{nm}$ filter. A dichroic filter at 690 nm served to separate the excitation from the emission signals. Fluorescence lifetime data were acquired using an A320 FastFLIM FLIMbox (ISS, Champaign, Illinois). For each image, 60 frames were collected to obtain at least $100\text{photons}/\text{pixel}$ at 3 mW of power in the sample plane. The scan speed was $25.21\mu \mathrm{s}/\text{pixel}$, and the image size $256\times 256\text{pixels}$, which span across $154\mu \mathrm{m}$. The lifetime of rhodamine 110 was measured to calibrate the FLIM system, as it is established at 4 ns.

The SimFCS software, developed at the Laboratory of Fluorescence Dynamics (UC Irvine), was used to collect and process these FLIM data. As previously described,^23,29 the fluorescence intensity decay associated to each pixel of the FLIM image is mapped onto a two coordinates $(g,s)$ system, the so-called phasor plot. The phasor transformation is as follows:

2.4.

Statistical Analysis

Statistical significance between the phasors of the examined macrophage groups was determined by the Student’s $t$-test, and considered positive for $p<0.05$. The analysis was performed on 3 to 7 images per each group as indicated in each experiment. Cells from at least two separate mice were obtained, and a total of 12 to 32 cells were analyzed for each condition.

3.1.

Fluorescence Lifetime Imaging Microscopy of Free and Bound NADH in Unpolarized Macrophages

It is well established that macrophages stimulated with $\mathrm{IFN}\text{-}\gamma$ and LPS or with interleukins 4 and 13 (IL-4 and IL-13) render proinflammatory or prohealing polarization states, respectively.^7,30,31 We confirm this by immunofluorescence staining for the proinflammatory marker inducible nitric oxide synthase (iNOS, in green) and the prohealing marker arginase-1 (Arg1, in red) (Fig. 1). $\mathrm{IFN}\text{-}\gamma /\mathrm{LPS}$-polarized macrophages express higher levels of iNOS, while IL-4/IL-13-stimulated macrophages express predominantly Arg1, as expected. Nuclei were counterstained with Hoechst 33342, indicated in blue.

Fig. 1

Immunofluorescence staining of macrophage markers. Representative immunofluorescence images of macrophages polarized with (b) $\mathrm{IFN}\text{-}\gamma$ and LPS or (c) IL-4 and IL-13, or (a) untreated, and stained for the proinflammatory marker iNOS (in green), the prohealing marker arginase-1 (Arg1, in red), and the cell nuclei (Hoechst 33342, in blue). Macrophages polarized with $\mathrm{IFN}\text{-}\gamma$ and LPS express higher levels of iNOS, while macrophages polarized with IL-4 and IL-13 express predominantly Arg1. $\text{Scale bar}=25\mu \mathrm{m}$.

Before interrogating the polarized macrophages with FLIM, we first study NADH lifetime maps of unpolarized macrophages that we will use as the reference. Figure 2(a) shows an autofluorescence image of untreated macrophages, and Fig. 2(b) shows a map based on lifetime contrast. The color coding is derived from the phasor plot in Fig. 2(c). In this plot, the lifetime distributions of pure free NADH and NADH bound to lactate dehydrogenase (LDH) are indicated. The two distributions are found at opposite sides of the unit semicircle, also known as the universal circle,²³ which starts at point (1,0) for zero lifetime species, and continues on to point (0,0) for infinite lifetime species. Freely diffusing NADH has a lifetime of $\sim 0.4\mathrm{ns}$ and NADH bound to LDH has a longer lifetime of $\sim 3.4\mathrm{ns}$, the latter varies slightly with binding protein. Both lifetimes are characterized by a single exponential decay and therefore lie on the semicircle.³²

Fig. 2

NADH lifetime of unpolarized macrophages. (a) Autofluorescence intensity images of untreated macrophages. (b) Corresponding lifetime image with color code according to squares selected in (c): purple areas have a more bound NADH character, while yellow areas have a more free NADH character ($\text{scale bar}=25\mu \mathrm{m}$). (c) Phasor plots of NADH bound to LDH, free NADH, and the cells in image a/b (arrowhead). The line that joins the free with the bound NADH phasors, centered at 0.4 and 3.4 ns lifetimes, respectively, defines the metabolic trajectory. The cells’ phasor lies along this line, as a linear combination of the two lifetimes.

The two NADH states define the extremes of the so-called metabolic trajectory in the phasor plot.^33,34 In the cellular cytoplasm, we find both species at different ratios, which yield a lifetime distribution that falls along the metabolic trajectory [Fig. 2(c), arrowhead]. This lifetime distribution (or phasor) can be characterized as the linear combination of the free NADH and bound NADH phasors. The phasor approach to FLIM permits a rapid identification of changes in the lifetime distributions.²³ The metabolic state of the cell can be determined by the position of the phasor along the metabolic trajectory. We divide the lifetime distribution of untreated macrophages through the center of mass in two parts along the metabolic trajectory, so that the fraction of pixels in each section is $0.5\pm 10\%$. The left part contains the pixels that are closer to the bound NADH extreme of the metabolic trajectory (purple square), while the right part has those closer to the free NADH end (yellow square). In this two-color representation, each pixel in the lifetime image in Fig. 2(b) is presented as either purple or yellow. The color coding can be interpreted as a relative measure of bound versus free NADH: purple represents more bound character, a lower free to bound NADH ratio, and yellow represents less bound character, a higher free to bound NADH ratio. Using these analysis regions as a reference, we can examine how the phasors of the polarized macrophages move along the metabolic trajectory, and thus quantify the changes in the ratio of free to bound NADH for each group.

3.2.

Fluorescence Lifetime Imaging Microscopy Identifies Different Lifetime Distributions in Polarized Macrophages

Using this approach, we next examine NADH lifetime distributions of polarized macrophages. Figure 3(a) shows an image with lifetime contrast obtained from unpolarized cells, and the accompanying color scheme is defined in the phasor plot of Fig. 3(b). As before, the cells display different color regions, corresponding to relative changes in the fluorescence lifetime, although the shape of the overall lifetime distribution may vary slightly from analysis of different cell populations. Using the same definition of the lifetime contributions as in Fig. 3(b), Fig. 3(c) shows a lifetime image of cells treated with $\mathrm{IFN}\text{-}\gamma$ and LPS, corresponding to proinflammatory macrophages. Compared to untreated cells, the $\mathrm{IFN}\text{-}\gamma /\mathrm{LPS}$ cells have a phasor slightly shifted toward the free NADH end of the trajectory [Fig. 3(d)], which translates into larger yellow areas in the cells [Fig. 3(c)]. This is expected, as $\mathrm{IFN}\text{-}\gamma /\mathrm{LPS}$ macrophages turn on glycolytic pathways for energy generation.³⁵ In contrast, cells treated with IL-4 and IL-13 have a higher contribution from bound NADH, as indicated in purple, as shown in Fig. 3(e). The phasor plot in Fig. 3(f) is now shifted toward the direction of pure bound NADH, with effectively longer lifetimes. This observation is consistent with the notion that alternatively activated macrophages favor oxidative phosphorylation,³⁶ which requires protein-bound NADH.¹⁹ $\mathrm{IL}\text{-}4/\mathrm{IL}\text{-}13$-treated macrophages show a markedly elongated profile compared to $\mathrm{IFN}\text{-}\gamma /\mathrm{LPS}$ and unpolarized macrophages, this is expected as macrophage elongation has been previously related to an alternatively activated phenotype.¹⁶

Fig. 3

Polarized macrophages have different lifetime distributions. (a) Untreated macrophages lifetime map ($\text{scale bar}=25\mu \mathrm{m}$), and (b) corresponding phasor plot. The purple and yellow squares select the areas in the cells with more bound NADH character, and more free NADH character, respectively, and are kept constant throughout the entire analysis. (c) Inflammatory macrophages, polarized with $\mathrm{IFN}\text{-}\gamma$ and LPS, and (d) corresponding phasor plot, shifted toward a markedly free NADH character. (e) Healing macrophages, polarized with IL-4 and IL-13, and (f) corresponding phasor plot, shifted toward the bound NADH side of the plot. (g) Average fraction of pixels in each square, normalized to the total number of pixels in both squares, for the three macrophage cultures, the error bars correspond to the standard deviation. $*p<0.05$, Student $t$-test. (h) Immunofluorescence images of the untreated macrophage culture, (i) the $\mathrm{IFN}\text{-}\gamma$- and LPS-polarized culture, and (j) the $\mathrm{IL}\text{-}4/\mathrm{IL}\text{-}13$-polarized culture stained for iNOS (green), Arg1 (red), and nuclei using Hoechst 33342 (blue). $\text{Scale bar}=50\mu \mathrm{m}$.

In an effort to quantify the metabolic state of the cells, we compute the fraction of pixels in each image that fall into the predefined lifetime categories (indicated here by the purple and yellow squares in the phasor plot). Figure 3(g) shows the results of averaging the fraction of pixels (fop) over three images for the unpolarized cells, with a total of 14 cells from two different mice, and over seven images for the polarized groups, that contained a total of 21 and 32 cells for each inflammatory and healing macrophages, respectively, obtained from three separate animals. Untreated macrophages show roughly equal amounts of short and longer lifetimes, as defined by the chosen purple (${\mathrm{fop}}_{\text{lower free to bound NADH ratio}}=0.49\pm 0.03$) and yellow (${\mathrm{fop}}_{\text{higher free to bound NADH ratio}}=0.51\pm 0.03$) subgroups [Figs. 3(a) and 3(b)]. However, macrophages polarized with $\mathrm{IFN}\text{-}\gamma$ and LPS have a significant increase in the fraction of yellow pixels (${\mathrm{fop}}_{\text{higher free to bound NADH ratio}}=0.83\pm 0.06$ versus ${\mathrm{fop}}_{\text{lower free to bound NADH ratio}}=0.17\pm 0.06$), those that report an increase in glycolytic metabolism. These cells are depicted mainly as yellow in Fig. 3(c). Macrophages treated with IL-4 and IL-13 have the opposite behavior, the fraction of pixels within the purple square (${\mathrm{fop}}_{\text{lower free to bound NADH ratio}}=0.65\pm 0.12$) is significantly larger than those in the yellow square (${\mathrm{fop}}_{\text{higher free to bound NADH ratio}}=0.35\pm 0.12$). As a result, the cells are colored predominantly in purple in Fig. 3(e). After FLIM imaging, the cells were fixed and stained to examine their expression levels of proinflammatory and prohealing markers [Figs. 3(h)–3(j)]. The stained cells confirm the correlation between NADH metabolism and macrophage phenotype, as $\mathrm{IFN}\text{-}\gamma /\mathrm{LPS}$-treated cells have higher expression of iNOS [Fig. 3(i)] and Arg1 expression predominates in $\mathrm{IL}\text{-}4/\mathrm{IL}\text{-}13$-treated cells [Fig. 3(j)].

In addition to evaluating the aggregate response of a population of cells with the overall lifetime distribution, the phasor approach to FLIM also permits the visualization of intracellular features with different lifetime properties. More specifically, the nuclear regions in the cells tend to have a higher free NADH concentration (as shown in yellow) in the lifetime images [white arrowheads in Figs. 3(a), 3(c), 3(e)], whereas the cellular cytoplasm displays a higher concentration of bound NADH (as shown in purple). A shift toward free NADH in the nuclei has been previously reported using this same method to identify early stages of cellular differentiation.³⁷

In addition to a rapid visualization method, the phasor approach can be used to classify the cells according to their average individual phasor value. In that regard, Fig. 4 shows the aggregate phasor distribution of untreated and polarized macrophages, and locates the position of single cells analyzed within the distribution. The purple and yellow squares are kept constant and divide the zoomed-in area in shaded purple and yellow regions for reference. We find two clear groups, the $\mathrm{IL}\text{-}4/\mathrm{IL}\text{-}13$-polarized macrophages (purple diamonds) fall within the purple square, with higher contribution of protein-bound NADH, and the $\mathrm{IFN}\text{-}\gamma /\mathrm{LPS}$-polarized macrophages (yellow squares) fall into the yellow area with lower lifetimes, assigned to higher contribution of free NADH. Unpolarized macrophages stretch across the areas, with a skewed preference toward the left side of the plot, indicative of an oxidative metabolism basal state.^14,35

Fig. 4

Cell phasor plots. The phasor plot agglutinates the lifetime distributions of the cells in the untreated and the polarized macrophages. The purple and yellow squares are located in the same position for the entire analysis. The insert zooms into the core of the phasor to show the average phasor point of each cell analyzed across separate macrophage cultures. The shaded areas align with the squares position.

3.3.

NADH Lifetime Detected by FLIM is a Robust Indicator of Macrophage Glycolytic State

Given that macrophages are involved in the progression of cardiovascular disease, a condition in which the cellular microenvironment contains significantly more lipid, we were interested to know whether or not the presence of lipid influenced the polarization of macrophages as determined by FLIM. We interrogated polarized macrophages with low (lipoprotein deficient serum, LPDS, based medium) or excess (oxidized low-density lipoprotein, oxLDL, containing medium) lipid. None of the cells in these conditions appeared to accumulate significant lipid droplets (data not shown). We find the same results: inflammatory macrophages (cells polarized with $\mathrm{IFN}\text{-}\gamma$ and LPS) have a significantly larger fraction of glycolytic pixels in both cases; whereas healing macrophages (cells polarized with $\mathrm{IL}\text{-}4+\mathrm{IL}\text{-}13$) significantly bind more NADH to protein to undergo oxidative phosphorylation than inflammatory cells do [Fig. 5(i)], independent of the lipid environment. The fraction of pixels quantified in [Fig. 5(i)] are the average 5 to 7 images that contained a total of 12 to 23 macrophages from two separate mice. $\mathrm{IFN}\text{-}\gamma /\mathrm{LPS}$-stimulated macrophages treated with LPDS ($N=6$ images, 12 cells) have in average a ${\mathrm{fop}}_{\text{lower free to bound NADH ratio}}=0.27\pm 0.14$ and ${\mathrm{fop}}_{\text{higher free to bound NADH ratio}}=0.73\pm 0.14$, while $\mathrm{IL}\text{-}4/\mathrm{IL}13$-stimulated macrophages treated with LPDS ($N=5$ images, 23 cells) have in average a ${\mathrm{fop}}_{\text{lower free to bound NADH ratio}}=0.57\pm 0.05$ and ${\mathrm{fop}}_{\text{higher free to bound NADH ratio}}=0.43\pm 0.05$. Macrophages stimulated with oxLDL followed the same trend, the average fraction of pixels and standard deviation for $\mathrm{IFN}\text{-}\gamma /\mathrm{LPS}$-stimulated macrophages ($N=7$ images, 15 cells) are ${\mathrm{fop}}_{\text{lower free to bound NADH ratio}}=0.24\pm 0.10$ and ${\mathrm{fop}}_{\text{higher free to bound NADH ratio}}=0.76\pm 0.10$, and for $\mathrm{IL}\text{-}4/\mathrm{IL}13$-stimulated macrophages ($N=5$ images, 18 cells) are ${\mathrm{fop}}_{\text{lower free to bound NADH ratio}}=0.61\pm 0.09$ and ${\mathrm{fop}}_{\text{higher free to bound NADH ratio}}=0.39\pm 0.09$. Once again this is visually translated at the cellular level by yellow in the $\mathrm{IFN}\text{-}\gamma /\mathrm{LPS}$ cases [Figs. 5(a) and 5(c)], and purple in the $\mathrm{IL}\text{-}4/\mathrm{IL}13$ cases [Figs. 5(b) and 5(d)]. We further confirmed by immunofluorescence staining that cells in these conditions maintained their phenotypic markers, as shown in Figs. 5(e)–5(h). Thus, this method appears robust to alterations in the presence of different lipid microenvironments.

Fig. 5

Macrophages under altered lipid uptake conditions. Macrophages polarized with $\mathrm{IFN}\text{-}\gamma$ and LPS and macrophages polarized with IL-4 and IL-13 treated with LPDS and with oxLDL maintain their glycolytic phenotype. (a–d) Lifetime maps ($\text{scale bar}=25\mu \mathrm{m}$) and (e–h) immunofluorescence images of the corresponding macrophage groups stained for iNOS (green), Arg1 (red), and nuclei with Hoechst 33342 (blue). $\text{Scale bar}=50\mu \mathrm{m}$. (i) Average fraction of pixels in the yellow and purple squares, normalized to the total fraction of pixels, for each group of macrophage culture, the error bars correspond to the standard deviation. $*p<0.05$, Student $t$-test.