Fluorescence Lifetime Imaging of the Ocular Fundus in Mice

Abstract

Purpose.: Fundus autofluorescence (AF) is characterized not only by its intensity or excitation and emission spectra but also by the lifetimes of the fluorophores. Fluorescence lifetime is influenced by the fluorophore's microenvironment and may provide information about the metabolic tissue state. We report quantitative and qualitative autofluorescence lifetime imaging of the ocular fundus in mice.

Methods.: A fluorescence lifetime imaging ophthalmoscope (FLIO) was used to measure fluorescence lifetimes of endogenous fluorophores in the murine retina. FLIO imaging was performed in 1-month-old C57BL/6, BALB/c, and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice. Measurements were repeated at monthly intervals over the course of 6 months. For correlation with structural changes, an optical coherence tomogram was acquired.

Results.: Fundus autofluorescence lifetime images were readily obtained in all mice. In the short spectral channel (498–560 nm), mean ± SEM AF lifetimes were 956 ± 15 picoseconds (ps) in C57BL/6; 801 ± 35 ps in BALB/c mice; and 882 ± 37 ps in C3A.Cg-Pde6b⁺Prph2^Rd2/J mice. In the long spectral channel (560–720 nm), mean ± SEM AF lifetimes were 298 ± 14 ps in C57BL/6 mice, 241 ± 10 ps in BALB/c mice, and 288 ± 8 ps in C3A.Cg-Pde6b⁺Prph2^Rd2/J mice. There was a general decrease in mean AF lifetimes with age.

Conclusions.: Although fluorescence lifetime values differ among mouse strains, we found little variance within the groups. Fundus autofluorescence lifetime imaging in mice may provide additional information for understanding retinal disease processes and may facilitate monitoring of therapeutic effects in preclinical studies.

Introduction

Fluorescence lifetime imaging microscopy (FLIM) is a technique where the mean fluorescence lifetime of a chromophore is measured at each pixel and displayed as contrast of a microscope image. Fluorescence lifetime is the time that a molecule spends to return to its ground state from its excited state and follows an exponential decay rate. The advantage of measuring the fluorescence lifetime of naturally occurring fluorescent molecules is that lifetimes are independent of the fluorophore's concentration or efficiency but strongly dependent on their respective protein binding within a given microenvironment. In FLIM, this phenomenon has been used for discriminating between free and protein-bound components of reduced nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD) both in vitro and in vivo.^1,2 In addition, Foerster resonance energy transfer (FRET) using FLIM, a mechanism describing energy transfer between two fluorophores, is widely used to investigate intra- and intermolecular interactions in living cells.³ Energy transfer from the donor molecule to the acceptor molecule will decrease the lifetime of the donor, and as such, FRET measurements can provide information about protein–protein interactions, protein-DNA interactions, and protein conformational changes and thus probe the environment of the fluorophore.^4,5 Furthermore, a recent report using FLIM has shown that oxidative stress increases the fluorescence lifetimes in retinal pigment epithelium (RPE) cells ex vivo.⁶ 

Although in vitro examination by FLIM permits molecular dissection on cellular and subcellular levels such as the RPE,⁶ only investigation of these processes within the complexity of the eye in the living animal can reveal the full range of pathophysiological changes that occur in retinal disease. Similar to FLIM, fundus autofluorescence lifetime imaging ophthalmoscopy (FLIO) measures fluorescence lifetimes in the retina but, in contrast to FLIM, on a macroscopic level. Use of FLIO has been reported in healthy subjects and in some case reports of patients.^7–10 Retinal imaging in rodents has become increasingly important in basic and preclinical research of retinal diseases. Longitudinal imaging is useful not only for tracking disease progression but also for monitoring the response of the disease process to potential therapies. Despite conventional retinal imaging modalities, such as optical coherence tomography (OCT)¹¹ and confocal scanning laser ophthalmoscopy (cSLO),^12,13 which are widely used for high-resolution in vivo assessment of retinal structures, an imaging method to monitor metabolic changes within rodent retina would provide additional information, especially in regard to disease mechanisms. 

Herein, we aimed to establish the method of in vivo fundus FLIO imaging in mice. FLIO imaging was subsequently used to characterize fluorescence lifetimes in different wild-type (WT) mice and in a mouse model of retinal degeneration. These data provide evidence that FLIO imaging is feasible in the rodent retina and that anatomical structures, such as retinal vessels, can be differentiated by their characteristic fluorescence lifetimes, using FLIO. Furthermore, we provide data for fluorescence lifetime characteristics in longitudinal measurements with increasing age. Ultimately this technique may be used in conjunction with fluorescence-conjugated probes for specific metabolic conditions like hypoxia or apoptosis to investigate disease mechanisms in vivo in the mouse model. 

Methods

Mice

Inbred C57BL/6 (WT inbred strain, pigmented), BALB/c (WT inbred strain, nonpigmented) and C3A.Cg-Pde6b⁺Prph2^Rd2/J (congenic mutant strain, pigmented, C57BL/6 background) mice were obtained from the breeding facility of the Department of Clinical Research, University Hospital of Bern. C3A.Cg-Pde6b⁺Prph2^Rd2/J mice feature a slow retinal degeneration.¹⁴ During experimentation, all mice were kept under a standard 12-h:12-h light–dark cycle with food and water available ad libitum. All animals were treated according to the Statement for the Use of Animals in Ophthalmic and Vision Research promulgated by the Association for Research in Vision and Ophthalmology and after governmental approval according to the Federal Swiss Regulations on Animal Welfare. 

Animal Preparation for Imaging Procedures

Mice were anesthetized by subcutaneous injection of 0.75 mg/kg medetomidine (Dormitor, 1 mg/mL; Pfizer, Zurich, Switzerland) and 45 mg/kg ketamine (Ketalar 50 mg/mL; Pfizer). 

After measurements were concluded, anesthesia was reversed by using 0.75 mg/kg atipamezole (Antisedan 5 mg/mL; Pfizer) at least half an hour after injection of the anesthetics, and subcutaneous injection of 300 μL of 0.9% NaCl was used to avoid dehydration. 

Pupils were dilated to a diameter of approximately 2 mm using tropicamide 0.5% and phenylephrine HCl 2.5% (ISPI, Bern, Switzerland). Methylcellulose (Methocel 2%; OmniVision, Neuhausen, Switzerland) diluted 1:1 with balanced salt solution (Alcon, Schaffhausen, Switzerland) was applied onto the cornea to prevent corneal desiccation and subsequent cataract formation. 

Image Acquisition

For acquisition of images, the mouse was placed on a platform mounted on the chin rest of the imaging device so that its eyes were positioned in the optic axis of the fluorescence lifetime imaging ophthalmoscope. 

In order to adapt the short axial length of the mouse eye (e.g., 3 mm), the optics of the FLIO system were adjusted by adding a 25-diopter (D) lens (f = 40/+25 D; Heidelberg Engineering, Heidelberg, Germany) in front of the FLIO aperture. Additional measurements to test whether results were influenced by the added optical lens or the methylcellulose did not reveal any significant differences. 

For each picture, a minimum of 700 to 1000 photons per pixel was recorded. This corresponded to an acquisition time of 2 to 4 minutes. 

Mice were measured first at 4 to 5 weeks of age and then measured monthly over a period of 6 months in order to assess whether lifetime values changed with increasing age of the animals. 

An OCT image (Heidelberg Retina Angiograph [HRA] Spectralis system; Heidelberg Engineering) and an autofluorescence intensity image were obtained at each measurement to monitor for retinal changes. For OCT imaging, a 78-D lens (Volk Optical, Inc., Mentor, OH, USA) was added in front of the OCT camera. In order to obtain a smooth surface of the mouse cornea for the OCT images, the eye was protected with methylcellulose and a contact lens (9-mm diameter; base curve, 7.2; power (F′) +4 diopters; Bausch & Lomb; OmniVision). 

Fluorescence Lifetime Imaging Ophthalmoscope

Fluorescence lifetime data of the murine retina were obtained using a fluorescence lifetime imaging ophthalmoscope, which was based on the HRA Spectralis system (Heidelberg Engineering) with preserved infrared and autofluorescence imaging functions.¹⁰ 

Retinal autofluorescence was excited by a 473-nm pulsed laser, raster scanning the central fundus with a repetition rate of 80 MHz. 

A 100-μm multimode optical detection fiber scanned and confocally filtered the emitted fluorescence. Fluorescence photons were detected by highly sensitive hybrid photon-counting detectors (Hybrid Photo Multiplier (HPM)-100-40; Becker & Hickl, Berlin, Germany) in a short (498–560 nm) and in a long (560–720 nm) spectral channel and registered by time-correlated single-photon counting (TCSPC) modules (SPC-150; Becker & Hickl). Simultaneously, a high-contrast confocal infrared reflection image was recorded by the system's eye tracking system. 

SPCImage version 4.6 software (Becker & Hickl) was used for further data analysis. Lifetime values were approximated using a bi-exponential decay model and a binning factor of 1 (Fig. 1A) as described previously.¹⁰ The goodness of fit was evaluated by reduced x² value with values close to 1 for an appropriate model and a random noise distribution. 

Figure 1

 

Fluorescence lifetime imaging of the ocular fundus in mice. (A1) Autofluorescence image and corresponding mean fluorescence lifetime image (Tm) of the retina of a BALB/c mouse (short spectral channel, color range: 800–1400 picoseconds [ps]). (A2) Representative lifetime decay curve of the indicated pixel (A1) for the short (above) and the long (below) spectral channel (red line: fit curve of decay trace, green line: instrument response function). Time between ticks = 500 ps. Term x² indicates the appropriateness of the exponential fit. (B) A standardized ETDRS grid was used for quantitative analysis of fluorescence lifetime values; C, center; IR, inner ring; OR, outer ring.

For further analysis, the mean fluorescence lifetimes, Tm, were calculated from the short and long lifetime components T1 and T2 and their respective amplitudes α1 and α2.  

 

 

A standard Early Treatment Diabetic Retinopathy Study (ETDRS) grid centered on the optic nerve was applied for circular analysis of mean fluorescence lifetimes by the FLIO reader (ARTORG Center for Biomedical Engineering Research, University of Bern, Switzerland) (Fig. 1B). The grid has circle diameters of 600 μm for the center (corresponding to approximately 3 disc diameters in mice),¹³ 1800 μm for the inner ring, and 3600 μm for the outer ring. Only averaged values from the inner ring (Fig. 1, IR) of the ETDRS grid were examined in this study. The central area with the optic nerve head (Fig. 1, C) and peripheral areas with image disparity (ETDRS grid outer ring [Fig. 1, OR]) were excluded from analysis. Measured lateral distances might not have been accurate because the ETDRS grid size was established for measurements in human retina, but axial OCT measurements seemed to be accurate.¹⁵ 

Statistical Analysis

Statistical analysis was performed using Prism software version 6 (GraphPad Software, Inc., La Jolla, CA, USA). Quantitative analysis of fluorescence lifetime values was performed by comparing the average values of the inner ring of the ETDRS grid. 

Mean lifetime values (±SEM) of the three mouse strains were compared using Mann-Whitney test (two-tailed, 95% confidence interval). Analysis was done for both spectral channels. P values less than 0.05 were considered statistically significant. 

Results

Fundus Autofluorescence Lifetime Values of the Murine Fundus

Fundus AF lifetime images from both spectral channels and corresponding autofluorescence images of the ocular fundus of C57BL/6, BALB/c, and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice are shown in Figure 2 with the respective color scales. 

Figure 2

 

Fluorescence lifetime images of the retina in three different mouse strains at 3 months of age. Representative fundus autofluorescence (AF) intensity (middle) and AF lifetime (FLIO) images (note, individual color scales for each image) in the short (left) and the long (right) spectral channel of a C57BL/6 mouse (A), a BALB/c mouse (B) and a C3A.Cg-Pde6b⁺Prph2^Rd2/J mouse (C).

Measured mean AF lifetimes (Tm) of the inner ring of the ETDRS grid for the three different strains of mice at the age of 12 weeks are shown in Figure 3 for the two distinct spectral channels: short (Fig. 3A) and long (Fig. 3B). 

Figure 3

 

Quantitative analysis of fluorescence lifetimes of the ocular fundus in three different mouse strains. Box plots of mean fluorescence lifetimes (Tm) of the inner ETDRS ring (see Fig. 1B) in 12-week-old C57BL/6, BALB/c, and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice for the short (A) and the long (B) spectral channel (box: median and 25th/75th percentiles; whiskers: 10th/90th percentiles; ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant; n = 8 mice for C57BL/6 and BALB/c and 14 mice for C3A.Cg-Pde6b⁺Prph2^Rd2/J mice; data were pooled from two independent experiments). (C) Fundus fluorescence lifetime images of C57BL/6, BALB/c, and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice in the same color scale (500–1000 ps; short spectral channel).

In the short spectral channel, C57BL/6 mice showed significantly longer mean AF lifetimes Tm (956 ± 15 picoseconds [ps]) than BALB/c (801 ± 35 ps) and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice (883 ± 37 ps; P = 0.0006 and P = 0.0225, respectively). Lifetime Tm values of retinal AF of BALB/c mice were slightly shorter than those of C3A.Cg-Pde6b⁺Prph2^Rd2/J mice, but these differences were not significant (P = 0.3282). 

In the long spectral channel, the mean AF lifetimes were longer in C57BL/6 (298 ± 14 ps) and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice (288 ± 8 ps) than in BALB/c mice (241 ± 10 ps; P = 0.0379 and P = 0.0042, respectively). Similar lifetime values were measured in C57BL/6 and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice (P = 0.9135). 

Age Dependency of Fluorescence Lifetime Values of the Murine Fundus

In all mouse strains, fluorescence lifetime values were longest at the first measurement (4–5 weeks of age) in both spectral channels (Fig. 4). Over the course of 6 months, AF lifetimes decreased in all three mouse strains in both spectral channels. Whereas AF lifetimes decreased by approximately 40% and 60% in C57BL/6 and BALB/c mice, respectively, the decrease in C3A.Cg-Pde6b⁺Prph2^Rd2/J mice was less marked, with a 10% decrease over the investigated time course. 

Figure 4

 

Age dependency of retinal fluorescence lifetimes in three different mouse strains. Averaged mean lifetime values (Tm) ± SEM of the inner ETDRS ring in the short (A) and long (B) spectral channel. Measurements were performed monthly between months 1 and 7. Data shown are from one of two independent experiments. ▪ = C57BL/6 (n = 4); • = BALB/c (n = 4); ▴ = C3A.Cg-Pde6b⁺Prph2^Rd2/J (n = 7). (C1–C3) Same eye of a C3A.Cg-Pde6b⁺Prph2^Rd2/J mouse at three different measurement time points (color scale: 500–1400 ps; short spectral channel).

Fluorescence Lifetimes of Retinal Vessels

Mean fluorescence lifetime Tm values of retinal arteries and veins in C57BL/6, BALB/c, and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice were further analyzed in detail in the short spectral channel. For this, we analyzed sections of the retinal vessels on a defined standard circle within the inner ring of the ETDRS grid (Fig. 5A). Whereas the retinal veins displayed a clearly discernible AF lifetime from the surrounding retina, retinal arteries were not as well delimited from the retina. 

Figure 5

 

Fluorescence lifetimes of retinal vessels in three different mouse strains. (A) Representative autofluorescence (left) and mean fluorescence lifetime image (right, short spectral channel) of a C57BL/6 mouse representing the analysis method for quantitative analysis of fluorescence lifetimes of retinal arteries (red points) and the retinal veins (blue points) on a standardized circle within the inner ring of the ETDRS grid. Cleavages from indicated points were analyzed and compared. (B) Quantitative analysis of mean fluorescence lifetimes in retinal arteries and veins of 12-week-old C57BL/6, BALB/c, and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice. (box: median and 25th/75th percentiles; whiskers: 10th/90th percentiles; ***P < 0.001, **P < 0.01; ns = not significant; n = 8 mice for C57BL/6 and BALB/c and 14 mice for C3A.Cg-Pde6b⁺Prph2^Rd2/J mice; data were pooled from two independent experiments).

In the short spectral channel, retinal vessels had shorter mean AF lifetimes than the retina. Mean ± SEM AF lifetimes in retinal arteries were 817 ± 14 ps in C57BL/6 mice, 679 ± 25 ps in BALB/c mice, and 777 ± 15 ps in C3A.Cg-Pde6b⁺Prph2^Rd2/J mice. Retinal veins were readily identifiable and were characterized by a uniform mean AF lifetime with no significant differences among the three mouse strains (Table). The optic nerve head did not show specific AF lifetimes. Data for the long spectral channel were not analyzed any further because the vessels were not easily discernable in this channel (Table). 

Table

 

Fluorescence Lifetime Values of Murine Retinas and Human Fundus

Channel	Retina	Artery	Vein
Short spectral channel (498–560 nm)
C57BL/6	957 ± 15	817 ± 14	627 ± 32
BALB/c	801 ± 35	679 ± 25	597 ± 26
Human	255 ± 6	415 ± 10	412 ± 7
Long spectral channel (560–720 nm)
C57BL/6	298 ± 14	318 ± 6	279 ± 5
BALB/c	241 ± 10	263 ± 3	246 ± 8
Human	282 ± 4	348 ± 4	349 ± 4

 Data show a comparison among mean fluorescence lifetime values (Tm, ps) of retinal areas of C57BL/6 and BALB/c mice and human fundus. Mean values ± SEM are listed for the retina, retinal arteries, and retinal veins in the short and the long spectral channels.

By analyzing the individual lifetime components T1 and T2 (Equation 1) and the corresponding amplitudes, areas of the murine retina, the retinal vessels and the vessel walls were clearly identifiable (Fig. 6). The short decay component T1 was for the most part responsible for the mean AF lifetime Tm, with amplitudes of 60% to 90%. Retinal vessels showed the shortest lifetime component T1, with the highest amplitude a1. 

Figure 6

 

Distribution histogram of lifetime components. Representative image of the retina of a C57BL/6 mouse (A1) fluorescence lifetime image (short spectral channel); (A2) autofluorescence image (A3) infrared image. (B–D) Selected areas of the murine fundus displayed in detail: neuronal retina (B), retinal vessels (C), vessel walls (D). (B1, C1, D1) Representative histogram of the short decay component T1 (x-axis, picoseconds [ps]) compared with the long component T2 (y-axis, ps). Areas of the retina (solid ellipse [B1]), the retinal vessels (dotted ellipse [C1]), and the vessel walls (dashed ellipse [D1]) are marked. (B2, C2, D2) Distribution of the respective lifetime spots marked in 1 and 3. (B3, C3, D3) Representative histogram of the short decay component T1 (x-axis, ps) compared with the corresponding amplitude α1 (y-axis, %). The distribution areas of the retina (solid ellipse), the retinal vessels (dotted ellipse), and the vessel walls (dashed rhombus) are marked.

Fluorescence Lifetimes and Retinal Thickness

Next, we analyzed whether fluorescence lifetimes were influenced by the retinal thickness. Therefore, mean total retinal thickness in the inner ring of the ETDRS grid measured by OCT was analyzed for all three mouse strains (see Supplementary Fig. S1). When the mice were 3 months of age, the retinal thicknesses were 246 ± 2 μm in C57BL/6 mice and 238 ± 4 μm in BALB/c mice and were thinnest in C3A.Cg-Pde6b⁺Prph2^Rd2/J mice, at 193 ± 1 μm. These values were similar to those of previously published data.^16–18 The differences in retinal thicknesses between C57BL/6 and BALB/c mice were not significant (P = 0.071), whereas the retina was significantly thinner in C3A.Cg-Pde6b⁺Prph2^Rd2/J mice at all time points (P < 0.0001; Supplementary Fig. S1D). 

Whereas retinal thicknesses remained stable in C57BL/6 and BALB/c mice over the course of 6 months, retinal thickness decreased by 20% from 208 ± 2 μm at the age of 1 month to 165 ± 2 μm at the age of 7 months in C3A.Cg-Pde6b⁺Prph2^Rd2/J mice. 

We did not find any significant correlation between retinal thickness and mean fluorescence lifetimes for the three mouse strains. 

Histology of Mouse Retina

For each mouse strain, histological sections were prepared after the experimental course of 6 months at the mouse age of 7 months and stained with hematoxylin and eosin (Supplementary Fig. S1). C57BL/6 and BALB/c mice showed identical retinal structures with the only difference in the pigmentation of the RPE (detailed sections are shown in Supplementary Fig. S1E). In C3A.Cg-Pde6b⁺Prph2^Rd2/J mice at this age, the photoreceptor layer was missing, the outer nuclear layer was clearly rarified, and the outer plexiform layer was much attenuated compared to the retinal layer structure in C57BL/6 mice. 

Discussion

Imaging of autofluorescence intensity of endogenous retinal fluorophores has found many clinical applications, and recently, several reports in mice have been published.^12,19,20 However, endogenous fluorophores can be characterized not only by their intensity but also by their decay time, or lifetime. Because fluorescence emission is proportional to the number of molecules in the first excited state, exponential decay can be reconstructed using TCSPC. Because energy is transferred to the microenvironment during the decay of the fluorophore, the energy transfer rate is influenced by the concentration of ions, the oxygen concentration, the pH value, or the binding to proteins in a cell.^21,22 The diagnostic potential of lifetime-based imaging has been repeatedly demonstrated by distinguishing lifetime differences between normal and diseased tissues.^23–26 

Herein, we report fluorescence lifetime characteristics of the murine fundus in vivo in two commonly used laboratory mouse strains, C57BL/6 and BALB/c. Furthermore, we present AF lifetime data in a mouse model of retinal degeneration (C3A.Cg-Pde6b⁺Prph2^Rd2/J). 

C57BL/6 and BALB/c mice showed little variance within the groups and were clearly distinguishable by their AF lifetime values. Thereby, the C57BL/6 animals displayed longer lifetime values. On the other hand, C3A.Cg-Pde6b⁺Prph2^Rd2/J mice showed greater variance in lifetime values, especially in the short spectral channel. This variance is likely to mirror the course of retinal degeneration. These mice showed degeneration of the outer nuclear layers of the retina, primarily the rod cells, followed by the cone cells and finally loss of all retinal layers including the pigment epithelial cells due to two retinal degeneration mutations (Pde6brd1 and RdsRd2). C3A.Cg-Pde6b⁺Prph2^Rd2/J mice were originally bred on a C57BL/6 background, which might explain the similar lifetime values in the long spectral channel. 

There are likely to be many components which contribute to the measured autofluorescence intensities and lifetimes in the retina.⁷ In a mixture of fluorophores, the observed mean fluorescence decay time will most likely be dominated by the predominant fluorophore. Lipofuscin is believed to be the dominant fluorophore in fundus AF in humans,²⁷ although this may not be the case in mice.^12,19,20 Interestingly, very little difference was found between the distribution of the lifetimes in the short and the long wavelength channels, suggesting that one group of closely associated fluorophores dominates the entire emission spectrum. 

We can only speculate about the difference in AF lifetimes between C57BL/6 and BALB/c strains in both the short and long spectral channel. The most obvious difference would seem to be their degree of pigmentation among identical histological morphology. In published reports, fluorescence decay lifetimes of melanin are described with 1200 ns, with a broad lifetime range.²⁸ This is in accordance with our measured data, whereby C57BL/6 mice showed the longest lifetimes with mean values at approximately 950 ps. However, other reports^6,9 have shown that melanin might have significantly shorter lifetimes, and this would not explain the longer lifetimes in C57BL/6 mice. Further studies using FLIM have shown that hemoglobin is characterized by very short AF lifetimes.²⁹ It may be that, at least in BALB/c mice, the main AF lifetime signal derives from the choroid, where the high content of blood and erythrocytes may lead to shorter AF lifetimes in these nonpigmented mice. 

It is interesting, and reassuring, that clearly identifiable structures like the retinal veins feature comparable lifetime values independent of the mouse strain. These consistent values point toward similar tissue composition of retinal veins in all three mouse types. In contrast to the veins, significant differences of AF lifetimes were measured in retinal arteries. This is most likely due to the smaller caliber of retinal arteries, making it difficult to demarcate the vessel outlines. This hypothesis is further supported by the finding that the lifetime values in arteries is closely associated with the ones measured in the retina due to crossover of fluorescence. It is therefore likely that AF lifetime values in retinal arteries reflect a mixture of retinal and arterial AF lifetimes. 

These findings may have important implications as this technique advances research in various systemic diseases affecting or associated with blood vessels. In this context, FLIM has been shown to reliably identify plaque formations with high content of either collagen or lipids in atherosclerotic vessels.^30,31 After further refinements to characterize AF lifetimes of retinal arteries, FLIO may offer a tool for studying atherosclerosis in vivo and an excellent readout for therapeutic effects. 

Compared to the human retina outside the fovea, with mean AF lifetime values of 200 to 300 ps in the short spectral channel, lifetimes in the murine retina were considerably longer, with mean values of 800 to 950 ps (Table). In the long spectral channel, lifetime values were in the same range, with mean values between 250 and 300 ps.¹⁰ The longer lifetimes in the short spectral channel in mice may be caused by lens-specific influences or smaller pupil diameter.¹⁰ In addition, the retinas of human and mice cannot be directly compared as the murine retina has a larger rod-to-cone ratio than the human retina.^32,33 In human retinas, lifetimes were longer in more peripheral retinal areas where the amount of rods is greater than in the central macula with cone dominance. This might indicate cell type-specific fluorescence lifetime characteristics. 

In the human fundus, retinal vessels featured longer lifetimes than the surrounding retinal tissue, with mean values of 410 ps in the short and 350 ps in the long spectral channel (Table; Dysli C, Wolf S, Zinkernagel MS, unpublished data, 2014). However, retinal vessels in mice had clearly shorter lifetimes than the surrounding retinas, with a mean lifetime of 600 ps for the retinal veins. 

Another finding that merits further discussion is that AF lifetime values decreased constantly with age in C3A.Cg-Pde6b⁺Prph2^Rd2/J mice as well as in C57BL/6 and BALB/c mice. Whereas the mean retinal thickness in C3A.Cg-Pde6b⁺Prph2^Rd2/J mice decreased with progressive retinal degeneration and age, the thicknesses in C57BL/6 and BALB/c mice remained constant over time. Therefore, retinal AF lifetimes do not seem to be directly related to retinal thickness in the murine fundus, at least in the latter two strains. In C3A.Cg-Pde6b⁺Prph2^Rd2/J mice, we found a positive correlation between retinal thickness measurements with AF lifetimes and increasing age. The decrease in AF lifetimes over the course of 6 months is quite puzzling as we have previously shown that in humans, AF lifetimes increase with age.¹⁰ Retinal pigment epithelium lipofuscin as the major fluorophore in the retina has been shown to increase with age in humans^34–38 as well as in mice.^20,39 Initially measured lifetimes in young mice are much longer than in humans, and therefore the accumulation of lipofuscin may lead to a decrease in general lifetimes of the murine retina toward the lifetime of lipofuscin, whereas the AF lifetimes increase with age in humans. Therefore, AF lifetimes in the retina of humans and mice might approach similar levels with increasing age and accumulation of lipofuscin. 

The influence of lipofuscin, however, may be attenuated by other fluorophores such as melanin. This may explain the more pronounced decrease of AF lifetimes with age in BALB/c mice compared with C57BL/6 mice, as the lack of melanin in BALB/c mice may intensify the influence of lipofuscin. 

Fluorescence lifetime imaging in ocular fundus of mice is associated with specific limitations. The ocular size of the mouse eye is considerably smaller than the human eye. As the optical system of the FLIO device is designed for confocal measurement in human eyes, application in the murine eye is associated with image disparity toward the retinal periphery due to the higher convexity of the mouse eye. For this reason, only the inner retinal ring with sparing of the optic nerve was analyzed in this study. Additionally, the spatial resolution of the mouse retinal FLIO images might be limited by the optic aberration of the mouse eye. Also, the excitation laser and the wavelengths of the detection channels of the device are predefined and cannot be adapted. 

Conclusions

Fluorescence lifetime imaging with a fluorescence lifetime imaging ophthalmoscope is applicable to in vivo measurement of lifetimes of natural retinal fluorophores in the murine fundus. Different mouse models (C57BL/6, BALB/c, and C3A.Cg-Pde6b⁺Prph2^Rd2/J mice) showed strain-specific lifetime values within the retina, with little variance within the groups. Our findings serve as the foundation for further in vivo research of different metabolic conditions in various mouse models. 

Acknowledgments

The authors thank Monika Kilchenmann and Federica Bisignani for technical assistance. The authors acknowledge the facilities and scientific and technical assistance of the Department for Clinical Research of the University of Bern. 

Disclosure: C. Dysli, None; M. Dysli, None; V. Enzmann, None; S. Wolf, Allergan (C), Bayer (C), Novartis (C), Optos (C), Heidelberg (F); M.S. Zinkernagel, Novartis (C), Bayer (C), Allergan (C), Heidelberg (F) 

Supported by a grant from the Swiss National Science Foundation (SNSF; #320030_156019).