Single-Cell Resolution Imaging of Retinal Ganglion Cell Apoptosis In Vivo Using a Cell-Penetrating Caspase-Activatable Peptide Probe

Peptide probes for imaging retinal ganglion cell (RGC) apoptosis consist of a cell-penetrating peptide targeting moiety and a fluorophore-quencher pair flanking an effector caspase consensus sequence. Using ex vivo fluorescence imaging, we previously validated the capacity of these probes to identify apoptotic RGCs in cell culture and in an in vivo rat model of N-methyl- D-aspartate (NMDA)-induced neurotoxicity. Herein, using TcapQ488, a new probe designed and synthesized for compatibility with clinically-relevant imaging instruments, and real time imaging of a live rat RGC degeneration model, we fully characterized time- and dose-dependent probe activation, signal-to-noise ratios, and probe safety profiles in vivo. Adult rats received intravitreal injections of four NMDA concentrations followed by varying TcapQ488 doses. Fluorescence fundus imaging was performed sequentially in vivo using a confocal scanning laser ophthalmoscope and individual RGCs displaying activated probe were counted and analyzed. Rats also underwent electroretinography following intravitreal injection of probe. In vivo fluorescence fundus imaging revealed distinct single-cell probe activation as an indicator of RGC apoptosis induced by intravitreal NMDA injection that corresponded to the identical cells observed in retinal flat mounts of the same eye. Peak activation of probe in vivo was detected 12 hours post probe injection. Detectable fluorescent RGCs increased with increasing NMDA concentration; sensitivity of detection generally increased with increasing TcapQ488 dose until saturating at 0.387 nmol. Electroretinography following intravitreal injections of TcapQ488 showed no significant difference compared with control injections. We optimized the signal-to-noise ratio of a caspase-activatable cell penetrating peptide probe for quantitative non-invasive detection of RGC apoptosis in vivo. Full characterization of probe performance in this setting creates an important in vivo imaging standard for functional evaluation of future probe analogues and provides a basis for extending this strategy into glaucoma-specific animal models.


Introduction
Molecular imaging utilizing peptide probes provides a noninvasive means for interrogating cells in vivo for a specific intracellular biochemical event [1,2]. Activatable probes denote such events via enzyme-specific cleavage, resulting in a detectable signal [3,4]. Through an intramolecular optical quenching strategy, activatable probes with a fluorescent signal moiety remain optically silent until activation by the target enzyme. This probe design confers a high degree of specificity and maximizes the signal-to-noise ratio for optimal detection of events that may occur at a relatively low frequency. This strategy is particularly attractive for application to the eye, as optical imaging can be utilized to identify a fluorescent target using clinically available instruments.
Apoptosis occurs in both normal development and disease in a wide range of tissues and therefore is an attractive target for molecular imaging [5,6]. Glaucoma, a leading cause of blindness, is characterized by the selective degeneration of retinal ganglion cells (RGC) via apoptosis [7,8,9,10]. Apoptosis can be identified through the activity of the caspase family of proteases, which plays a central role in the enzymatic pathway leading to apoptotic cell death [11,12]. Previously, we described the design and validation in vitro of cell-penetrating peptide probes sensitive to activated effector caspase activity [3,13,14,15,16]. Intracellular delivery of these molecular imaging probes was enabled through conjugation to a cell-penetrating peptide sequence serving as a targeting moiety to facilitate rapid probe translocation into cells via endocytic pathways [17]. We demonstrated highly specific uptake by RGCs following intravitreal injection of fluorophore conjugated to our modified cell-penetrating peptide sequence [13], and subsequently utilized these probes in a rat model of NMDA-induced RGC degeneration to validate probe activity and localization ex vivo [18,19].
Using the rat NMDA model, herein we fully characterized probe activation in vivo using a confocal scanning laser ophthalmoscope (CSLO) and a clinically-relevant protocol. This approach enabled repetitive and sequential non-invasive fluorescence fundus imaging of individual animals to determine the temporal course of probe activation in vivo, the optimal kinetics of probe delivery, and a probe dose providing an optimal signal-to-noise ratio in this model. In addition, electroretinogram (ERG) studies were performed to rule out probe-related toxicity. These studies provide a rational basis for future structure-function probe evaluation and for extending these probes and strategies to more disease-specific models of RGC degeneration (i.e., glaucoma), both in rodents and primates. Activatable Cell-penetrating Peptide (TcapQ488)

Animals
TcapQ488 was produced and purified using methods as previously described [17,18]. This activatable peptide probe consists of an all D-amino acid modified Tat cell-penetrating peptide, an L-amino acid effector caspase recognition sequence (DEVD), a quencher (QSY7), and a fluorophore (Alexa Fluor-488). This particular fluorophore was chosen for compatibility with clinically-available fluorescent fundus imaging instruments, which are designed for use with either fluorescein or indocyanine green. Upon cleavage of the effector caspase recognition sequence and subsequent loss of fluorescent quenching, fluorescence from the retained intracellular fluorophore is detectable via fluorescence imaging. Stock solutions of purified peptides were formulated in milliQ water at various concentrations and stored at 220uC.

Intravitreal Injections
All probe, PBS, and NMDA intravitreal injections were performed as described [18,19]. Briefly, rats were anesthetized by isoflurane inhalation or intraperitoneal injection (0.10 ml/ 100 g) of a cocktail containing 2 ml ketamine (100 mg/ml), 1.5 ml xylazine (20 mg/ml) and 0.5 ml isotonic sterile saline. Intravitreal injections were performed under a microscope with a U-100 29 gauge, K inch length insulin needle and a co-axial 33 gauge, 1 inch length blunt metal needle on a 2.5 ml Hamilton syringe (Hamilton Company, NV). The eye was first punctured by the 29 gauge needle at approximately 1 mm behind the limbus on the nasal side of the eye, aiming toward the optic nerve head. The 33 gauge needle on the 2.5 ml Hamilton syringe was then inserted into the eye through the puncture to deliver injectate. Caution was taken to deliver injectate slowly, over approximately 3 minutes. Erythromycin ointment was applied topically after each intraocular injection.
In experimental groups, eyes were pretreated by injecting 2 ml of 2.5, 12.5, 25 or 40 mM (corresponding to 5, 25, 50 and 80 nmol, respectively) NMDA (Sigma, St. Louis, MO) prepared in 0.01 M phosphate buffered saline (PBS, pH 7.4). Twenty-four hours later, 2.5 ml of 0.039, 0.078, 0.125, 0.155 or 0.310 mM TcapQ488 in PBS were injected into the vitreous. In control experiments, eyes were pretreated with 2 ml of 0.01 M PBS instead of NMDA and followed by 2.5 ml TcapQ488 injection. Animals with visible lens damage, vitreous or retinal hemorrhage or retinal detachment were excluded from analysis, although this was a rare occurrence.

In Vivo Imaging
A CSLO (Retinal Angiograph II, Heidelberg Engineering, Inc., Germany) equipped with a 55u field of view (FOV) lens was used for in vivo image acquisition. Rat retinas were typically imaged at 4, 12, 24, 48 and 72 hours post probe injection. Rats were positioned on a platform attached to the CSLO chin rest. A custom-made PMMA contact lens (Cantor & Nissel Limited, Brackley, UK) was placed on the eye for all imaging to protect the corneal surface and improve image quality. Pupils were dilated with one drop of Tropicamide ophthalmic solution USP 1% (Bausch & Lomb, Tampa, FL). The CSLO image was focused on the retinal nerve fiber layer under infrared reflection mode (IR, diode laser at 820 nm) at 50% intensity to center the optic disc in the image ( Figure 1A). Fluorescent angiograph mode (FA, blue solid laser at 488 nm with 500 nm barrier filter) was then used to acquire activated TcapQ488 fluorescence signal of the RGC layer ( Figure 1B). All CSLO images were recorded as an average of 100 aligned frames to obtain a single low-noise, high contrast image.

Ex Vivo Imaging
Retinal flat mounts were prepared and fluorescence microscopy performed as described previously [18].

Image Analysis and Cell Counting
Images used for data analysis/cell counting were centered at the optic disc. Imaging parameters, including gain, were kept constant during image acquisition. Fluorescence signals were counted manually using Image J software [20] based on fluorescence intensity and sharpness. Manual counting enabled consistent and accurate analysis throughout cell counting. Quantitation by manual counting was confirmed in a subset of animals using ''Find Maxima'', an automated cell counting program in ImageJ (http://rsb.info.nih.gov/ij); noise tolerance was set at 17, edge and center (optic disc) maxima were excluded from the analysis field (See Figure S1).

Electroretinography (ERG) Protocol
Rats were dark-adapted overnight, prepared for recording under dim red illumination, and anesthetized initially by subcutaneous injection of ketamine/xylazine cocktail (0.10 ml/ 100 g) containing 2 ml ketamine (100 mg/ml), 1.5 ml xylazine (20 mg/ml) and 0.5 ml isotonic sterile saline and supplemented as necessary at 1/4 of the original dose. Pupils were dilated with 1.0% atropine sulfate ophthalmic solution (Bausch & Lomb, Tampa, FL). During testing, a heating pad controlled by a rectal temperature probe maintained body temperature between 36.5uC and 37.5uC. Each rat was positioned with its head just inside the opening of a Model 2503D Ganzfeld stimulator (LKC technologies, Gaithersburg, MD). ERGs were recorded using a platinum loop electrode (2.0 mm in diameter) positioned in a drop of 0.5% atropine sulfate in 1.25% hydroxypropyl methylcellulose (GO-NAK; Akorn Inc., Buffalo Grove, IL) on the corneal surface of the eye to be tested. A reference needle electrode was inserted under the skin at the vertex of the skull and a ground electrode was inserted under the skin of the back or tail. Flash ERG measurements were performed with a UTAS-E3000 Visual Electrodiagnostic System using EM for Windows software (LKC Technologies, Inc., Gaithersburg, MD). The stimulus (trial) consisted of a brief full field flash (10 ms), either in darkness or in the presence of dim (29 cd/mm) or bright (198 cd/mm) background illumination after 10 minutes of light adaptation. The initiation of the flash was taken as time zero. The response was recorded over 250 ms plus 25 ms of pre-trial baseline. Responses from several trials were averaged and the interval between flash trials was adjusted so that responses did not decrease in amplitude over the trial series of for each step. The log light intensity (log [ cd*s / m2 ]) was calculated based on the manufacturer's calibrations. To reduce noise interference, a 60 Hz digital notch filter was used for weak responses (routinely for intensities below 2 2 log; as needed for brighter intensities).
Rats underwent ERG testing pre-injection, at 1 week and at 2 months post intravitreal injection of TcapQ488. Probe concentrations of 0.078 mM and 0.155 mM in 2.5 ml (0.193 and 0.387 nmol, respectively) were examined. Control rats underwent PBS injection only. All conditions were tested in triplicate.

Statistical Analyses
Data were analyzed using SAS version 9.3 (Base SASH 9.3, Procedures Guide: Statistical Procedures, SAS Institute, Inc., Cary, NC, 2011). All p-values were two-tailed. A p-value ,0.05 was considered significant. An analysis of variance was performed on the log transformed probe activation data to determine the effect of eye, light intensity, probe dose, and NMDA concentration on probe activation. Simple effects tests were done to determine significant effects of probe dose at each NMDA concentration and vice versa. For ERG data, a repeated measures analysis of variance was used to determine the effect of eye, probe, and time on B-wave amplitude. Significant interactions were followed-up with simple effects tests for probe conditions and time periods. Significant simple effects tests were further investigated using pair-wise comparisons.

Correspondence of In Vivo and Ex Vivo Imaging of Probe Activation
Strong fluorescent signals were observed with in vivo imaging in the RGC layer of eyes of living rats pretreated with NMDA followed by TcapQ488 ( Figure 1B). The fluorescence was cellspecific and punctate, consistent with our previous ex vivo observations that fluorescence was restricted to the cell bodies of individual RGCs containing activated probe. Several eyes underwent in vivo imaging followed by histological assessment of retina flat mounts ex vivo to directly compare in vivo fluorescence fundus images with ex vivo flat mount fluorescence microscopy images of the same retina. Retinal flat mounts displayed distinct fluorescent signals in individual cell bodies, consistent with probe activation in a specific subset of cells. Comparison of in vivo and ex vivo images revealed excellent correspondence of the same individual fluorescent cell bodies ( Figure 2). As documented previously [18], vertical retinal sections confirmed that the vast majority of fluorescent cell bodies, with the exception of a few cells in the inner nuclear layer, were located within the RGC layer (data not shown). Spread of fluorescence to the optic nerve suggested that at least the cleavage product, in this case Alexa Fluor 488, may enter axonal transport pathways in RGCs.

Probe Activation as a Function of Time
Following pre-treatment with NMDA and TcapQ488 injection, serial in vivo fluorescence fundus imaging in individual animals using the CSLO enabled clear delineation of the spatial and temporal pattern of probe activation (Figures 3, 4). Animals were routinely imaged immediately before probe injection and five times post-probe injection (at 4, 12, 24, 48 and 72 hours) to determine the kinetics of probe activation in vivo and the optimal time point for detecting probe activation. TcapQ488 activation was evident 4 hours post probe injection (Figures 3, 4). Notably, stronger fluorescent signal was noted near the injection site, consistent with a subtle gradient effect reflecting diffusion of both probe and NMDA throughout the vitreous from the injection site. Figure 4A shows the time course of fluorescent probe activation determined from analysis of the images from Figure 3B-F. Probe activation increased significantly in the first 12 hours, remaining relatively steady out to 72 hours. As expected, probe activation signals were dependent on probe dose and NMDA concentration, but this kinetic pattern was remarkably consistent across all probe dose and NMDA concentration combinations ( Figure S2). Figure 4B represents a normalized analysis of probe activation kinetics over the first 72 hours based on pooled data across the study. From this kinetic analysis, the 12 hour time point was utilized to quantify probe activation across all NMDA and probe concentrations.

Probe Activation as a Function of NMDA Concentration
We previously determined the frequency of probe activation at one single time point across multiple NMDA concentrations through ex vivo analyses of retinal flat mounts [18]. Using the same model and NMDA concentrations (2.5, 12.5, 25, 40 mM) with in vivo imaging, we confirmed the NMDA concentration-response relationship with probe activation at multiple time points postprobe injection ( Figure 5). Control injections of PBS were also performed to determine the background level of probe activation in the absence of NMDA excitotoxicity. As expected, NMDA-

Probe Signal as a Function of Probe Dose
To determine the probe concentration that maximized RGC apoptosis detection while minimizing background probe signal, five TcapQ488 doses (0.097, 0.193, 0.313, 0.387 and 0.775 nmol) were examined at each standard NMDA concentration (Figures 6,  7, S2). TcapQ488 activation across probe doses for 25 mM NMDA is shown in detail in Figure 6. As TcapQ488 dosage increased, the probe activation frequency also increased, suggesting that the sensitivity of TcapQ488 in detecting apoptosis using our in vivo imaging strategy was dependent on probe dose. Figure 7 shows probe activation results for each combination of NMDA concentration and TcapQ488 dosage at the 12 hour time point. There were significant differences in activation frequency across probe doses at all NMDA concentrations (2.5 mM, p = 0.02; 12.5 mM, p = 0.002; 25 mM, p,0.001; 40 mM, p,0.001). Probe activation generally increased as probe dose increased for each NMDA concentration. The relationship between probe dose and activation, however, began to saturate at higher probe doses indicating that further increases would likely be of limited benefit (Figure 7). Consistent with this, pair-wise tests revealed no significant difference in probe activation between probe doses of 0.387 nmol and 0.775 nmol at all NMDA concentrations. In addition, pair-wise comparisons of consecutive probe doses higher  Figure S2 shows a comprehensive 3-dimensional plot of probe activation as a function of NMDA concentration and probe dose at all 5 time points.
The frequency of ''background'' probe activation, as indicated by PBS-only control injections, also gradually increased with increasing probe doses and there was a significant difference found across probe doses in the PBS condition (p,0.0001). This relationship was not as linear as that seen at each NMDA concentration (Figure 7). Instead, probe activation frequency in the PBS condition remained relatively low until probe dose was increased from 0.313 nmol to 0.387 nmol with pair-wise testing revealing a significant increase (p = 0.013).
To better evaluate which probe concentration resulted in the best signal-to-noise ratio at the NMDA concentrations examined, data from Figure 7 were reanalyzed as fold-increase over PBS across NMDA and TcapQ488 dose combinations at 12 hours post-probe  injection ( Figure 8). Consistent with the findings that TcapQ488 activation began to saturate at probe doses higher than 0.313 nmol and ''background'' probe activation also increased significantly at doses higher than 0.313 nmol, a probe dose of 0.313 nmol exhibited the best signal-to-noise ratio, linearity, and the largest dynamic range across NMDA concentrations. For lower NMDA concentrations (2.5 and 12.5 mM), which resulted in less frequent probe activation, the signal-to-noise ratio was optimized at the lowest probe dose examined (0.097 nmol), but the overall dynamic range was reduced.

ERG Assessment for Toxicity
ERG analysis consisted of dark-adapted A-wave amplitudes, dark-and light-adapted B-wave amplitudes, and implicit time measurements. We performed ERG measurements on eyes receiving intravitreal injections of PBS or TcapQ488 to assess for possible retinal toxicity related to probe injection. Since rods are the dominant photoreceptor subtype in rodents [21], only dark-adapted B-wave data are presented (Figure 9). Comparison between pre-injection, 1 week post-injection, and 2 months postinjection was performed for each probe condition (PBS, 0.193 nmol TcapQ488, 0.387 nmol TcapQ488) ( Figure 9A-C). Statistical analysis of mean B-wave amplitudes revealed a significant interaction between the three probe conditions and three time points (p,0.001). There were no significant differences, however, among probe conditions when tested at each time point (pre-injection, p = 0.63; 7 days, p = 0.27; 2 months, p = 0.11). The dark-adapted B-wave showed a mild diminution in amplitude for each condition at 1 week compared with pre-injection, which reached significance for PBS (p = 0.010) and 0.387 nmol probe (p,0.0001), but not for 0.193 nmol probe (p = 0.35). This decrease continued to be statistically significant at 2 months for both PBS (p,0.0001) and 0.387 nmol probe (p,0.0001), while 0.193 nmol probe actually showed a significant increase versus pre-injection (p = 0.004) ( Figure 9E). Taken together, these data were consistent with minimal outer retinal toxicity as a function of probe concentration, particularly as the effect appears attributable to the intravitreal injection itself (PBS) rather than probe exposure.

Discussion
Currently, there are several strategies for diagnosing and monitoring human glaucoma, including intraocular pressure measurement, perimetric assessment of the visual field, examination of the optic disc, and assessment of the optic disc and nerve fiber layer by clinically available imaging devices. These   approaches for assessing RGCs are limited primarily to identifying RGC loss and subsequent changes to the field of vision, optic disc, or nerve fiber layer and therefore fail to provide real-time information regarding the status of RGCs which may reside in early injury or pre-death states. Thus, there is a need to improve our ability to monitor the course of RGC injury in glaucoma to assess therapy over time with a more facile marker of disease progression. Molecular imaging with activatable probes provides one such method and has potential to improve our ability to detect and monitor apoptotic processes in RGCs over time, guiding therapeutic choices in the management of this disease. The ability to monitor RGC status without the delays inherent in current clinical approaches should serve to reduce the duration and expense of glaucoma clinical trials and may accelerate the evaluation of potential neuroprotective agents in both animal models and human trials.
Use of a cell-penetrating activatable probe for in vivo imaging of RGC physiology has not been previously reported. Herein, TcapQ488 detects effector caspase activity signifying apoptotic cascade activation within a select subset of RGCs in rat retina. These results confirm our previous ex vivo analysis of retinal flat mounts using the same NMDA model in rats [18]. Direct comparison of images in vivo and flat mounts ex vivo from the same animal showed precise correlation of fluorescence in individual RGCs, confirming that the focal fluorescence seen on in vivo imaging faithfully represents our previous findings. Also as noted previously, the frequency of cells displaying probe activation in vivo showed the expected dose-response relationship with NMDA. This finding was now confirmed across multiple probe doses and was consistent at each concentration studied.
In vivo imaging enables repeated analysis of probe activation in a single animal across multiple time points post-intravitreal injection. Using this strategy, we identified a representative time point, 12 hours post injection, at which maximal or near maximal probe activation occurred. This time point was consistent across probe and NMDA concentrations and was therefore used for most analyses. Additional analyses will be required to determine Figure 8. Signal-to-noise ratio across NMDA concentrations and TcapQ488 probe doses. TcapQ488 activation data from Figure 7 was normalized to PBS pre-treatment (i.e., NMDA = 0) for each probe dose to reflect relative ''signal-to-noise.'' For lower NMDA concentrations (2.5 and 12.5 mM), the ratio was highest at 0.097 nmol probe, while for higher NMDA concentrations (25 and 40 mM), the ratio was similar from probe doses 0.097 to 0.313 nmol. Data represent mean normalized labeled cells 6 Error Propagation. doi:10.1371/journal.pone.0088855.g008 whether the same cells showed evidence of probe activation throughout the first 72 hours post-probe injection, or whether the totals represent a flux between ongoing probe activation and eventual loss of fluorescence in individual RGCs as apoptosis progresses to completion. The half-life of unactivated probe and activated probe fluorescence, and the possibility of uptake by phagocytic cells also require additional study. Given that our approach results in distinct single-cell fluorescence, it should be possible in future studies to focus on the fate of individual cells throughout the duration of fluorescence resulting from probe activation. This will be critical to understanding whether probe activation will persist in the presence of a repeated or ongoing insult, as might occur with elevated intraocular pressure, or whether repeated injections of probe might be informative.
Rigorous 3-dimensional titration curves of probe activation were generated based on NMDA concentration, TcapQ488 dosage, and time. This analysis was needed to determine probe concentrations which would maximize the signal-to-noise ratio, which is critically important for eventual application of this strategy in human glaucoma. The use of the lowest probe concentration possible should reduce the risk, if any, of potential probe-related retinal toxicity. As anticipated, detection sensitivity increased with increasing probe dose, but only to a point. Increasing TcapQ488 dosage above 0.313 nmol provided minimal additional benefits in terms of sensitivity and may increase any potential probe-related toxicity. Our signal-to-noise analysis indicated by the ratio of fluorescent labeled cells at each NMDA and probe dose to that seen in the PBS-only condition, varied with each of these parameters. As discussed previously [22], the relative frequency of RGC apoptosis is expected to be much less in human glaucoma than that observed at the higher concentrations of NMDA utilized in this analysis. As such, the sensitivity to detect apoptosis at a relatively low concentration of NMDA (i.e., 5 nmol) may be of greater importance than that seen at higher concentrations. In our analysis, the signal-to-noise ratio was maximized for the two lowest NMDA conditions at a much lower probe dose. This suggests that for a low apoptosis frequency, the decreased ''noise'' offered by a lower dose of probe outweighs the benefits of increased sensitivity and signal amplification seen at higher doses.
One distinct advantage of an activatable peptide probe is that activation and subsequent fluorescence signal should not occur until intracellular uptake and cleavage by the specific target enzyme has been initiated. Ideally there should be no probe activation in the PBS only condition. However, evidence of probe activation was detected in the absence of NMDA at the higher probe doses. Several possible mechanisms apply. TcapQ488 activation could reflect a baseline level of RGC apoptosis which is occurring in the rat retina, although this would be expected to be relatively infrequent. Probe activation could result from intravitreal injection-related RGC apoptosis regardless of the nature of the injectate, being related to the small increase in intraocular pressure resulting from even the small volumes (2 ml) utilized in this study [23]. In either case, the frequency of probe activation in the PBS-only condition would be expected to be stable across all probe concentrations examined. Since there was a clear increase in probe activation in the PBS-only condition going from 0.313 nmol probe to 0.387 nmol, these two factors would not fully explain our findings. Probe-associated fluorescence could occur through a concentration-dependent non-specific dequenching of fluorescence not involving true probe activation. Probe activation could also result from probe-related toxicity causing apoptosis in a subset of RGCs, resulting in probe activation. Assuming the vitreous volume of an adult rat eye to be ,56 ml [24] and the probe volume for the 0.313 nmol dose, the effective concentration of probe in the vitreous was , 5.5 mM. While below the previously determined LD 50 of 1062 mM from in vitro experiments with a similar probe [3], a subset of cells may show toxicity at this dose. Either mechanism could account for the fact that probe activation under the PBS-only condition was at least partly probe dose-dependent. Alternatively, the ''background'' probe activation might reflect a combination of both probe dosedependent and -independent effects.
To move our probe toward future clinical usage, physiological evidence of potential probe toxicity was assayed by electroretinogram (ERG). Full-field ERG has been used in both humans and lower animals as an indicator of retinal function and ERG changes can provide evidence of retinal injury or toxicity, particularly to the outer retina [25,26]. Under conditions of dark adaptation, the A-wave is generally associated with rod photoreceptor activity while the B-wave is associated with a combination of Muller and bipolar cell layer activity [27]. While these results do not specifically address RGC or inner retinal toxicity, they do provide a means for assessing potential overall retinal toxicity related to the intravitreal probe injection, particularly at higher probe doses. Overall, there was little evidence in our ERG analyses to suggest probe toxicity. Maximal signal-to-noise ratios and linearity occurred at 0.313 nmol of probe with minor toxicity observed both by potential autoactivation and B-wave imaging, but 0.193 nmol still yield nearly identical signal-to-noise ratios (Figure 8), minimal autoactivation, and no observable toxicity via B-wave imaging.
Several molecular strategies for apoptosis detection have been developed. One approach exploits the high affinity of annexin V for phosphatidylserine [28]. This phospholipid is located primarily in the inner leaflet of the plasma membrane of normal cells and is externalized to the outer leaflet early in the commitment to apoptosis [29]. Optical imaging with annexin V-Alexa Fluor 488 has been applied to the retina [30,31]. While of considerable promise, a fundamental criticism of imaging with annexin V is that the polypeptide binds to phosphatidylserine regardless of whether the phospholipid is exposed on the external leaflet or, if accessible, on the internal leaflet. Cells that undergo necrosis develop defects in the plasma membrane that enable large molecular weight molecules like annexin V access to phosphatidylserine residing in the inner leaflet and thus, in the absence of a second marker of membrane integrity (e.g., propidium iodide), these labeled polypeptides do not distinguish apoptosis from necrosis. This limits diagnostic specificity of the technique by any imaging modality. Other limiting factors for the general use of annexin V include rapid serum half-life of the polypeptide, poor penetration of the blood-brain barrier (and the blood-retinal barrier), pharmacokinetic complexities related to size, potential for immunogenicity, and low signal-to-noise ratios for imaging applications [32,33,34].
Modular probe design as utilized in our approach represents a distinct advantage. Any or all probe components -targeting sequence, quencher, fluorophore, and cleavage site -may be altered to achieve a desired result. Given the preferential uptake by RGCs of our cell-penetrating peptides as previously demonstrated [13] and the potential to conjugate them to diverse moieties [17], may render these CPPs useful for delivering other molecular imaging probes to RGCs. Alternatively, further modifications to these CPPs may result in probes with enhanced uptake by other retinal cell types. While identification of CPPs which cross the blood-retinal barrier is desirable, intravitreal injection of retinal therapeutics has become increasingly commonplace in recent years with a low incidence of adverse events [35] Similarly, while TcapQ probes are designed to detect enzymes integral to the apoptotic cascade, it is clear that this platform will accommodate specific peptide sequences recognized by other proteases, kinases, and enzyme activities. As such, our current ''proof of concept'' experiments should provide the basis for a broad class of novel imaging agents with applications throughout ophthalmology. These applications are not limited to imaging, as ''theranostic'' probes may ultimately be designed with both imaging and therapeutic capabilities. Identification of important structurefunction relationships is an ongoing process facilitated by our ability to evaluate next generation probes in both well-established in vitro [14,16,19] and in vivo (NMDA) models.
The NMDA model employed here is ideal for both the initial validation and future structure-function analysis of these molecular imaging probes. The model is well-established, titratable, highly reproducible, has a relatively short turnover time, and apoptosis can be largely limited to RGCs. As we move towards further preclinical validation of this and future generation probes, a more disease-specific animal model of RGC degeneration resulting from elevated intraocular pressure and more closely mimicking human glaucoma in terms of the time course and frequency of RGC apoptosis is desirable. Several such models exist in rodents and primates and will provide a means for further translation of this approach towards human application [36]. Our current results will guide our choice of probe concentration and imaging intervals going forward into more advanced glaucoma-specific models. Figure S1 Comparison of manual fluorescent RGC cell counting versus automated cell enumeration using ImageJ. Images at 4, 12, 24, 48, and 72 hours post probe injection were analyzed by manually counting fluorescent cells using Image J software. Fluorescent cell counts were based on fluorescence intensity and sharpness in a group of animals (n = 6) treated with 25 nmol NMDA and 0.313 nmol TcapQ488. The same images were analyzed using the automated counting program ''Find Maxima'' in ImageJ (http://rsb.info.nih.gov/ij). Noise tolerance was set to 17, edge and center (optic disc) maxima were excluded from the analysis field. Fluorescent cell numbers derived from each method are plotted as a function of probe injection time. Data represent mean 6 SD. (TIF)