Marking cells with infrared fluorescent proteins to preserve photoresponsiveness in the retina

Abstract

Green fluorescent protein (GFP) and its derivatives are broadly used in biomedical experiments for labeling particular cells or molecules. In the mouse retina, the light (∼500 nm) used to excite GFP can also lead to photoreceptor bleaching (peak ∼500 nm), which diminishes photoreceptor-mediated synaptic transmission in the retinal network. To overcome this problem, we investigated the use of infrared fluorescent protein (iRFP) as a marker since it is excited by light in the near-infrared range that would not damage the photoresponsiveness of the retina. Initially, we tested iRFP expression in human embryonic kidney 293 (HEK293) cells to confirm that conventional fluorescence microscopy can detect iRFP fluorescence. We next introduced the iRFP plasmid into adeno-associated virus 2 (AAV-2) and injected the resulting AAV-2 solution into the intraocular space. Retinal neurons were found to successfully express iRFP three weeks post-injection. Light-evoked responses in iRFP-marked cells were assessed using patch clamping, and light sensitivity was found to be similar in iRFP-expressing cells and non–iRFP-expressing cells, an indication that iRFP expression and detection do not affect retinal light responsiveness. Taken together, our results suggest iRFP can be a new tool for vision research, allowing for single-cell recordings from an iRFP marked neuron using conventional fluorescence microscopy.

The retina is an ideal neural system with which to conduct in vitro physiological experiments because light, a natural stimulus, can be used to evoke synaptic transmission. Light-evoked synaptic responses can be recorded in retinal neurons when light sensitivity is preserved in photoreceptors. In the mouse retina, a widely used animal model for vision research, the rod and cone photoreceptors are sensitive to green light (wavelength ∼500 nm) and UV light (wavelength ∼360 nm) (1,2).

Transgenic mice expressing green fluorescent protein (GFP) are excellent models to identify particular neurons or molecules (3). The retinas of these mice allow researchers to visualize certain subsets of cells among the numerous retinal neuron types (4). These GFP-labeled mouse lines have also contributed to morphological studies (5,6). However, because the light (wavelength ∼500 nm) used to identify GFP-marked cells bleaches photoreceptors (Figure 1A), recording of retinal light responses is only possible by using two-photon laser microscopy, which preserves the light responsiveness of the photoreceptors (7,8). Although the two-photon microscope has become more popular, it is still an expensive state-of-the-art tool and carries a high maintenance cost.

Figure 1. Infrared fluorescent protein (iRFP) spectra and expression in HEK293 cells.

(A) iRFP excitation (red line) and emission (purple line) spectra and our filter (Cy5.5) wavelengths: excitation (Ex) range (red box), and emission (Em) range (purple box). The spectra of the UV cones and green (Gr) cones are represented in the blue- and green-shaded areas, respectively. (B) rpAAV-CMV-iRFP plasmid containing the iRFP DNA is inserted into the pAAV-CMV-MCS expression vector. (C) iRFP expression in HEK293 cells transfected with the iRFP (pShuttle-CMV-iRFP) plasmid after 48 h; differential interference contrast (DIC) image (left) and fluorescence image (right). (D) HEK293 cells transfected with iRFP (i.e., pAAV-CMV-iRFP). (E) HEK293 cells transfected with enhanced GFP (EGFP) (i.e., pcDNA3-EGFP).

Several research groups have recently developed infrared fluorescent protein (IFP or iRFP) (9,10). The excitation and emission spectra of IFP1.4 are 684 nm and 708 nm (9) whereas those of iRFP are 690 nm and 713 nm (10), respectively. Because these wavelengths are far from the spectrum range of photoreceptors (Figure 1A), iRFP may be an ideal tool for identifying retinal cells without interfering with the recording of photoreceptor-mediated light responses. For this study, we introduced a plasmid containing the iRFP sequence (GenBank ID, JN247409) into retinal neurons using an adeno-associated virus 2 (AAV-2) vector, which is widely used for gene transfection in the retina (11,12). Three weeks after the intravitreal injection, iRFP was expressed in retinal cells, and light-evoked synaptic responses were successfully recorded.

Method Summary

Infrared fluorescent protein (iRFP) was expressed in retinal neurons by transfection using adeno-associated virus (AAV) vectors delivered via intravitreal injection. iRFP allowed cells to be visualized by a conventional fluorescence microscope without disrupting retinal photosensitivity.

Materials and methods

Plasmid and vector production

The pShuttle-CMV-iRFP DNA plasmid (plasmid number 31856; Addgene, Cambridge, MA) and the pcDNA3 enhanced GFP (EGFP) plasmid (plasmid number 13031; Addgene) were used for iRFP expression and EGFP expression, respectively, in human embryonic kidney 293 (HEK293) cells. The iRFP plasmid was also used for cloning the pAAV-iRFP DNA plasmid by using the pAAV-MCS expression vector (catalog number VPK-410; Cell Biolabs, San Diego, CA) (Figure 1B). The fragment of the iRFP plasmid containing BglII (5′ cloning site) and HindIII (3′ cloning site) sequences (∼1000 bp) was inserted into pAAV-MCS after the plasmid was digested with BamHI (5′) and HindIII (3′) (∼4600 bp). The pAAV-iRFP DNA plasmid was then amplified in Escherichia coli competent cells (MAX Efficiency DH5α cells; Life Technologies, Grand Island, NY) and purified by using the DNA Plasmid Maxi kit (Qiagen, Redwood City, CA). Insertion of the iRFP DNA fragment into the plasmid was verified by restriction digestion with EcoRI and HindIII. The AAV terminal repeats [inverted terminal repeats (ITRs)] are important for AAV replication and packaging. To confirm the existence of ITRs, SmaI digestion was performed. Our pAAV-iRFP construct had three SmaI recognition sequences. SmaI digestion produced 3 DNA fragments (2600, 1300, and 1500 bp), which were confirmed by gel electrophoresis (data not shown). The AAV-2 particles containing the pAAV-iRFP DNA plasmid were packaged and purified (Gene Therapy Center, University of North Carolina at Chapel Hill, Chapel Hill, NC). The production of viral particles was based on triple transfection of the HEK293 cells, and the recombinant AAV2-CMV-iRFP vector was purified by column chromatography.

Cell culture and gene transfection

The recombinant plasmid pAAV-CMV-iRFP was assessed using transfected HEK293 cells. The cells were grown in Dulbecco's modified Eagle's medium/minimum essential medium (DMEM/MEM) (Gibco-Life Technology, Grand Island, NY) with fetal bovine serum and a mixture of antibiotics (i.e., penicillin and streptomycin) in an incubator at 37°C and 5% CO₂ until the cells reached 70%–90% confluence; they were then trypsinized with 0.25% trypsin–ethylenediaminetetra-acetic acid (EDTA). The HEK293 cells (10⁴ cells/mL) were maintained in a 12-well plate, which contained a poly-D-lysine-coated coverslip in each well (Neuvitro, El Monte, CA). After 2 days of growth, the cells were then transfected using 5 μL Lipofectamine 2000 (Invitrogen-Life Technology) and 1 μg of pAAV-iRFP DNA with a 15 min incubation. As a positive control, transfection with the pcDNA3-EGFP plasmid (plasmid number 13031; Addgene) was done in the same manner. After 24 to 48 h of transfection, the HEK293 cells were viewed under a conventional fluorescence microscope (Slicescope Pro 2000; Scientifica Ltd, East Sussex, UK) that was equipped with a light-emitting diode (LED) module and a fluorescence filter set for iRFP expression. We used a 660 or 700 nm-peak LED for iRFP excitation and a 500 nm LED for GFP excitation. Fluorescence filters were Cy5.5 (Chroma Technology Corp., Bellows Falls, VT) (Figure 1A) for iRFP detection and FITC (Scientifica) for GFP detection.

Animal studies

Animal protocols for all experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Wayne State University. Mice 3–4 weeks old (C57BL/6J strain, Jackson Laboratories, Bar Harbor, ME) were anesthetized by an intraperitoneal injection of ketamine (120 mg/kg) and xylazine (15 mg/kg). Intravitreal vector injection was performed under a dissecting microscope. A small hole was created in the sclera near the cornea with a 30-gauge needle (Figure 2A). The viral solution (3.8 × 10¹² vp/mL; 1.5 μL) was then injected into the intravitreal space using a Hamilton syringe with a 32-gauge blunt needle. Three weeks to 3 months after the viral injection, the animals were euthanized using CO₂ and pneumothorax, and their retinal tissue was used for examining iRFP expression and electrophysiological recordings.

Figure 2. Adeno-associated virus (AAV)-mediated expression of infrared fluorescent protein (iRFP) in retinal neurons.

(A) Diagram shows the site of the intravitreal injection. The retina is shown in yellow, and the box indicates the portion of the retina shown in panels B–H. (B) Differential interference contrast (DIC) image of the transverse retinal section 4 weeks after the AAV injection. The dotted lines indicate the inner plexiform layer (IPL). (C) Infrared fluorescence image of the retinal section in (B). The iRFP cells are detected only in the ganglion cell layer (GCL). (D) DIC image of a retinal section 5 weeks after the AAV injection. (E) Infrared fluorescence image of the retinal section in (D). The iRFP cells are visible in the GCL and in the inner nuclear layer (INL). The arrows indicate the iRFP labeled cells in the INL. (F) DIC image of a whole-mount retinal preparation 3 months after the AAV injection. The GCL cells are shown. (G) iRFP expression in the same layer. Ganglion cell somas and axon bundles can be observed. (H) At 50 μm below the inner surface of the retina, the inner nuclear cells were detected. The scale bar in (F) represents 50 μm and is applicable to all other panels in this figure.

Yellow fluorescent protein (YFP)-expressing ganglion cell layer (GCL) cells in the B6.Cg-Tg(Thy1-Clomeleon)1Gjau/J mouse strain (Jackson Laboratory) were used for the control recordings (2).

Retinal preparation

The AAV-injected mice were dark-adapted overnight, and their eyes were enucleated. The experimental techniques were similar to those described previously (13,14). The eyes were placed in a cooled oxygenated dissecting solution containing 115 mM NaCl, 2.5 mM KCl, 2.5 mM CaCl₂, 1.0 mM MgCl₂, 10 mM HEPES, and 28 mM glucose, and were adjusted to pH 7.4 with NaOH. Using a dissecting microscope, the corneas and the lenses were quickly removed, and the retinas were isolated. The whole-mount retinal preparation was mounted onto a plastic coverslip, and the iRFP expression was assessed.

For retinal slice preparations, the isolated retinal tissue was placed onto a piece of filter membrane (HABG01300, Millipore, Billerica, MA) and cut into 250-μm thick slices using a custom-made chopper. The slice preparation was used for observing iRFP expression and for the patch clamp study. Because of the uneven distribution of green-sensitive and UV-sensitive cones in the mouse retina (1,2), the dorsal half was used for recording green light-evoked responses, and the ventral half was used for recording UV light-evoked responses. The retinal slices were stored in an oxygenated dark box at room temperature. All procedures for dissection and patch clamp recordings were performed in a dark room using an infrared viewer (Nitemere, BE Meyers, WA). For some retinal preparations, biliverdin hydrochloride (25 μM, Number 30891; Sigma-Aldrich, St. Louis, MO) was added for 2 to 4 h to examine whether the intensity of iRFP expression was increased.

Whole-cell recordings and light-evoked response analysis

The iRFP expression was detected by viewing retinal slices with an upright microscope (Slicescope Pro 2000; Scientifica, East Essex, UK) equipped with a charge-coupled device (CCD) camera (Retiga-2000R; Q-Imaging, Surrey, BC, Canada). The real-time image was viewed with a computer monitor. An LED light at a peak of 660 or 700 nm was used to excite the iRFP (pE-2; CoolLED, Hampshire, UK) by using a Cy5.5 fluorescence cube (Figure 1A). A 500-nm-peak LED was also used to excite YFP with an FITC filter set. After iRFP or YFP-labeled cells were identified, the cell was marked as a target cell. Switching to the differential interference contrast (DIC) mode, whole-cell patch clamp recordings were performed from the target cell, which was verified by intracellular fluorescent dye injection with sulforhodamine B (Sigma-Aldrich). Control light-evoked excitatory postsynaptic potential (L-EPSP) recordings were performed with cells in AAV-injected mouse retinas. GCL cells were recorded using the regular infrared-DIC but without the infrared light. L-EPSPs in control cells were evoked by a green LED.

Whole-cell recordings were performed at the resting membrane potential in Ames' medium buffered with NaHCO₃ (294 mOsm) (Sigma-Aldrich) at 33°C. Ames' medium was continuously bubbled with 95% O₂ and 5% CO₂, and the pH was 7.4 at 33°C. The intracellular solution contained the following: 111 mM K-gluconate, 1.0 mM CaCl₂, 10 mM HEPES, 1.1 mM EGTA, 10 mM NaCl, 1.0 mM MgCl₂, 5 mM ATP-Mg, and 1.0 mM GTP-Na, and was adjusted to a pH of 7.2 with KOH (269 mOsm). Liquid junction potentials were corrected after each recording. Electrodes were produced by pulling borosilicate glass (1B150F-4; World Precision Instruments, Sarasota, FL) with a P1000 Micropipette Puller (Sutter Instruments, Novato, CA) and had resistances of 5 to 10 MΩ. Clampex software and a MultiClamp 700B amplifier (both from Molecular Devices, Sunnyvale, CA) were used to generate waveforms, acquire data, and control LED light stimuli. The data were digitized and stored on a personal computer using a data acquisition system (Axon Digidata 1440A; Molecular Devices). Responses were filtered at 2 kHz with the 4-pole Bessel filter on the MultiClamp 700B and sampled at 2 to 5 kHz.

Light stimuli were generated using the pE-2 system (CoolLED), which was controlled by the Clampex software. A 500-nm LED light was projected directly onto photoreceptors in the slice preparation from the AAV mouse through a 60× objective lens. Light at 500 and 360 nm was projected onto the slice preparation from the YFP-labeled mice. Because ganglion cells and photoreceptors cannot be observed together with a 60× objective lens, we moved the microscope view from the GCL to the photoreceptor layer after configuring the patch clamp with a ganglion cell. This method enabled us to illuminate photoreceptors in the vicinity of the recorded cells. Step L-EPSPs were recorded in the dark-adapted conditions.

To examine how quickly iRFP bleaches, we captured time-lapse images of the iRFP-expressing cells in a whole-mount retinal preparation (Figure 3). The cells were exposed continuously to infrared light and iRFP images were captured every minute. The arbitrary units of fluorescence intensity were measured and then normalized to the maximum intensity.

Figure 3. Infrared fluorescent protein (iRFP) bleaches slowly.

The iRFP-expressing cells were exposed to the infrared light continuously. The fluorescence of the iRFP was plotted as a function of time. The red open circles and error bars show the mean and SEM of the black dots.

The peak amplitude (mV) and charge transfer (Q) of the L-EPSPs were measured using a Clampfit data analysis module (Molecular Devices). Due to the spike generation in many ganglion cells, we used a charge transfer for further analysis. We normalized the charge transfer to its maximum value in each cell and plotted it as a function of light intensity (Figure 4, C–E). For each recording, the light intensity–response relationship was fit to the Hill equation as follows: Y =ax^b / (L₅₀^b + x^b), in which a is the maximum response, b is the slope factor, and L₅₀ is the light intensity at half-maximum response (Sigma Plot; Systat Software, Inc. San Jose, CA). The luminance evoking a half-maximal response (L₅₀) and the slope factors were determined from the fit the mean and standard error of the mean (SEM) were calculated, and the average curve was plotted in the same graph. An unpaired t-test was used to determine whether each parameter was significantly different between the iRFP-expressing cells and the non-iRFP-expressing cells. The values are presented as the mean ± SEM. Differences were considered significant if P < 0.05 (two-tailed).

Figure 4. Light-evoked excitatory postsynaptic potentials (L-EPSPs) recorded in infrared fluorescent protein (iRFP)-expressing cells were similar to L-EPSPs in non-labeled cells.

(A) In the slice preparation, patch clamp recordings were conducted in an iRFP-marked cell (indicated by the blue arrow), shown in a DIC image (upper) and in a fluorescent image (lower). (B) Representative L-EPSPs from an iRFP-expressing cell. Step green light stimuli (1 s) were applied at the indicated intensities. Increasing the light intensity evoked and increased L-EPSPs. The scale bar indicates 5 mV for all panels. (C) Light intensity-response curves from the iRFP-labeled GCL cells. Each black line shows the normalized L-EPSPs from one cell. Each line was fit with an equation (see the Methods section), and the L₅₀ values were averaged. The average line and SEM of the L₅₀ values is plotted in red (L₅₀ = 1.6 × 10⁴ ± 4200; slope factor = 3.3 ± 0.4, n = 10). (D) Light intensity-response curves from non-iRFP-expressing cells in AAV-injected mice. The average curve is plotted in blue (L₅₀ = 1.2 × 10⁴ ± 7300; slope factor = 2.5 ± 0.4, n = 10) (P = 0.68 for L₅₀, P = 0.12 for the slope factor, between iRFP and non-iRFP cells, unpaired t-test). (E) Light intensity-response curves from YFP-labeled GCL cells. The L-EPSPs were evoked by UV light because the green cones were bleached.

Results and discussion

Expression of iRFP in HEK293 cells

To examine the transfection rate and brightness of iRFP with a conventional fluorescence microscope, we transfected the iRFP gene incorporated into a conventional pShuttle-CMV promoter plasmid into HEK293 cells. After 24 to 48 h of transfection, iRFP was expressed in HEK293 cells (Figure 1C). iRFP expression was detected in many HEK293 cells with our system, a conventional fluorescence microscope equipped with a CCD camera. We then constructed an AAV-iRFP vector plasmid in preparation for AAV infection in mouse retinal tissue (Figure 1B). We tested if this recombinant pAAV-CMV-iRFP plasmid expresses iRFP using HEK293 cells. After 24 h of transfection, iRFP was expressed (Figure 1D).

For the control, we expressed EGFP, a standard fluorescent marker, in HEK293 cells (Figure 1E) and compared transfection rates and brightness. The transfection rate for iRFP (38.5 ± 3.8%; n = 9) was higher than that for EGFP (25.4 ± 3.4%; n = 3) (P < 0.05, unpaired t-test). The brightness was estimated by the light intensity required to excite these markers. For viewing the EGFP-expressing cells, the intensity was ∼700 mW/cm²; for viewing iRFP-expressing cells, the intensity was ∼400 mW/cm². Taken together, iRFP's gene transfection rate and fluorescence brightness are even better than EGFP's.

Expression of iRFP in retinal cells

To express iRFP in retinal neurons in vivo, we constructed an AAV-iRFP vector for AAV infection in mouse retinal tissue. We used the AAV-2 vector as a vehicle to deliver the iRFP gene into mice retinal neurons (11,15). The AAV vector was produced by the University of North Carolina Gene Therapy Center Vector Core.

The AAV vector with iRFP was injected into the eyes of the experimental mice (concentration: 3.8 × 10¹² vp/mL) (Figure 2A). We injected AAV into 17 mice in total. 3 to 4 weeks after intravitreal injection, the mice were euthanized, and retinal preparations were made in oxygenated HEPES-buffered solution. Using a conventional fluorescence microscope, iRFP was clearly observed in the ganglion layer cells (Figure 2, B and C). The iRFP expression was observed all over the retina, including the central and peripheral regions. At 3 to 4 weeks after the intravitreal injection, iRFP expression was observed only in cells located in the GCL, which is next to the vitreous cavity (Figure 2C). The iRFP expression was rarely observed in the outer layer neurons. However, 5 weeks to 3 months after injection, iRFP expression was clearly observed in the inner nuclear layer cells (INL) (Figure 2, D and E, 5 weeks; Figure 2, F–H, 3 months). Amacrine cells, rod and cone bipolar cells, and Muller cells were identified in the INL. Horizontal cells may be labeled; however, this is a less common cell type (∼3%) (16) and could not be clearly identified.

To test how quickly iRFP bleaches, iRFP-expressing GCL cells in the whole-mount retinal preparation were exposed to infrared light continuously. iRFP images were captured and the intensity was measured. The plot in Figure 3 shows that iRFP bleached slowly.

Using our system, iRFP expression was clear and bright. However, a previous study suggested that iRFP fluorescence intensity was dim and attributable to the intrinsic properties of iRFP (9). Because biliverdin (BV) increased iRFP fluorescence in that study (9), we tested if BV could increase the brightness of iRFP by incubating retinal preparations with BV (25 μM) for 2 to 4 h. No improvement in iRFP fluorescence was seen (119.5 ± 38% arbitrary fluorescence) in the presence of BV compared with the absence of BV (P = 0.4 between the 2 conditions; n = 4 samples for each condition).

Light-evoked synaptic responses in iRFP-expressing cells

We tested if iRFP-expressing cells could be useful for retinal physiological studies. For this experiment, we used retinal slice preparations, which are easier for targeting cells with a patch clamp pipette. We conducted whole-cell recordings in an iRFP-expressing ganglion cell that was detected by infrared illumination (Figure 4A) and evoked light responses with green light stimuli (500 nm). The L-EPSPs were successfully recorded at the resting membrane potential (Figure 4, B and C) (n = 10). The light sensitivity (L₅₀) of the L-EPSPs varied among GCL cells. This is most likely due to the existence of ∼15 distinct GCL subtypes and their diverse rod and cone dominances (17,18). For the control experiment, we used non–iRFP-labeled GCL cells from AAV-injected mice. The L-EPSPs were evoked by green light at a similar intensity range and gave a variety of responses, which were not statistically different from the L-EPSPs from the iRFP-expressing cells (n = 10; P = 0.12 for the slope factor; P = 0.69 for L₅₀; unpaired two-tailed t-test) (Figure 4D).

Another method exists for recording light responses from fluorescent marker-expressing cells using conventional fluorescence microscopy. If green light is used to find a GFP-expressing cell, mid-wavelength photoreceptors are bleached. However, the mouse retina also has UV cones that express UV-sensitive opsin protein (peak wavelength ∼360 nm) (Figure 1A) (1). The light sensitivity of UV opsin is similar to or higher than the sensitivity of green opsin (19). Therefore, UV light responsiveness might be preserved in these cells after exposure to green illumination. We used YFP-expressing cells (see Methods section), whose excitation wavelength is close to that for EGFP (514 nm for YFP vs. 484 nm for EGFP). We recorded the L-EPSPs evoked by UV light in YFP-expressing ganglion cells (n = 4) (Figure 4E). The light sensitivity (L₅₀) was not different between the iRFP-expressing cells (green-evoked L-EPSPs) and the YFP-expressing cells (UV-evoked L-EPSPs) (P = 0.53, unpaired t-test), although the slope factor was significantly smaller in the YFP cells (P < 0 01, unpaired t-test). Although we tested only four cells for this condition, the light sensitivities of all four cells were within the same range as that of iRFP cells (Figure 4, D and E), Taken together, the UV light sensitivity in the YFP cells was preserved even after green light exposure. However, green light sensitivity for these YFP cells was 10⁵ times less (data not shown). In this respect, using iRFP may be more beneficial because green photoreceptors are still highly sensitive to light.

In conclusion, we successfully expressed iRFP in retinal neurons using AAV-2-mediated delivery and recorded light-evoked synaptic responses in iRFP-expressing neurons. The iRFP fluorescence was comparable to that of EGFP, and the light sensitivity in iRFP cells was similar to that in the control cells. These results demonstrate that iRFP is a novel, noninvasive marker for retinal physiological research that can be used with a conventional fluorescence microscope with a CCD capturing system.