Age-related differences in retinal function and structure in C57BL/6J and Thy1-YFPh mice

Age-related neuronal adaptations are known to help maintain function. This study aims to examine gross age-related in vivo retinal functional adaptations (using electroretinography) in young and middle aged C57BL/6J and Thy1-YFPh mice and to relate this to in vivo retinal structure (using optical coherence tomography). Elec-troretinography responses were generally larger in Thy1-YFPh mice than in C57BL/6J mice, with similar in vivo retinal layer thicknesses except for longer inner/outer photoreceptor segment in Thy1-YFPh mice. Relative to 3-month-old mice, 12-month-old mice showed reduced photoreceptor (C57BL/6J 84.0 ± 2.5 %; Thy1-YFPh 80.2 ± 5.2 %) and bipolar cell (C57BL/6J 75.6 ± 2.3 %; Thy1-YFPh 68.1 ± 5.5 %) function. There was relative preservation of ganglion cell function (C57BL/6J 79.7 ± 3.7 %; Thy1-YFPh 91.7 ± 5.0 %) with age, which was associated with increased b-wave (bipolar cell) sensitivities to light. Ganglion cell function was correlated with both b-wave amplitude and sensitivity. This study shows that there are normal age-related adaptations to preserve functional output. Different mouse strains may have varied age-related adaptation capacity and should be taken into consideration when examining age-related susceptibility to injury.


Introduction
In the normal aging eye, all ocular tissues from the cornea to optic nerve undergo age-related changes in structure that can impact their function (Grossniklaus et al., 2013).Aging is a major risk factor for many retinal conditions such as age-related macular degeneration and diabetic retinopathy, as well as glaucoma (Chader and Taylor, 2013;Wang et al., 2018).Increased incidence of these conditions often begins in the 4th decade, which coincides with many normal ocular age-related changes (Colijn et al., 2017;Tham et al., 2014).Transgenic mouse lines have become increasingly useful to study the process of normal aging as well as to understand how aging can impact the risk of ocular diseases.
The C57BL/6J mouse line has become widely used in the study of neurodegenerative and eye diseases; as a general-purpose strain or as a background strain for transgenic models.Previous studies have consistently shown that reduction in retinal photoreceptor and bipolar cell function measured using electroretinography (ERG) in C57BL/6J mice begins at 12 months of age (Ferdous et al., 2021;Gresh et al., 2003;Li et al., 2001;Wang et al., 2018;Williams and Jacobs, 2007).Structurally, some studies have reported age-related reductions in the number of outer retinal photoreceptor nuclei or in vivo thinning of the outer nuclear layer at 12 months of age (Cunea et al., 2014;Gresh et al., 2003;Wang et al., 2018), while others report no age-related change to inner nuclear layer thickness or the number or size of bipolar cells from 12 to 33 months of age (Liets et al., 2006;Trachimowicz et al., 1981).
Age-related differences to the inner retina in C57BL/6J mice have been less well documented.Histological studies report some evidence of ganglion cell loss from 12 months of age onwards (Danias et al., 2003).Samuel et al. (2011) reported that the total number of neurons were preserved, but ganglion cell morphology and density, as well as amacrine cell density, were altered and reduced with age.Such morphological changes would be expected to alter ganglion cell function, however quantification of ganglion cell function using ERG in normal aging mice has not been undertaken.Given that C57BL/6J mice are useful for inducible models of studying age-related eye diseases, it will be valuable to characterize how outer and inner retinal function and structure alter with age.
Transgenic mice have significantly furthered our understanding of Abbreviations: ERG, electroretinography; OCT, optical coherence tomography; YFP, yellow fluorescent protein; STR, scotopic threshold response; pSTR, positive scotopic threshold response.disease pathogenesis.Of particular interest in the study of neurodegenerative diseases are transgenic strains where neurons express fluorescent proteins under the control of Thy1 promoter, a glycoprotein highly expressed on the surface of neurons including retinal ganglion cells (Barnstable and Drager, 1984;Feng et al., 2000).Specifically, Thy1-YFPh mice, where less than 10 % of the total retinal ganglion cells fluoresce, allows us to study the extent of cell dendritic arbor (Feng et al., 2000;Iaboni et al., 2020;Oglesby et al., 2012) using ex vivo or in vivo imaging approaches to understand optic neuropathy, such as in inducible models of glaucoma or how treatment affects dendritic morphology (Feng et al., 2013;Johnson et al., 2016;Kalesnykas et al., 2012;Lee et al., 2014;Leung et al., 2011;Li et al., 2011;Smith et al., 2017).Thy1-YFPh mice are bred and backcrossed to C57BL/6J mice (Feng et al., 2000) and to date, no studies have reported abnormalities in retinal function or structure in these mice.Whether retinal function and structure of Thy1-YPFh mice change with age in the same way as C57BL/6J mice has yet to be investigated.
This study aimed to investigate how ERG parameters indicative of key retinal neuronal cell classes change with age in Thy1-YFPh mice, and how they compare to the more common C57BL/6J strain.In vivo retinal structures measured using optical coherence tomography (OCT) were also examined.We hypothesize that in vivo functional and structural differences with age are similar between Thy1-YFPh and C57B/6J mice.The most common "adult" age used in previous studies of 3month-old (human equivalent of 20 -30 years old) (Cunea et al., 2014;Li et al., 2001;Wang et al., 2018), was compared to older 12-or 16-month-old mice (human equivalent of 38 -47 years old) (Flurkey et al., 2007).Although this is not considered as 'aged' mice, it is a period where prevalence of age-related ocular conditions start to increase (Colijn et al., 2017;Tham et al., 2014), and experimental studies show evidence of retinal plasticity at 10-12 months of age (Berkowitz et al., 2014;Goodman et al., 2023) and age-related differences in how the eyes response to stress or injury (Kong et al., 2012;Lee et al., 2022).Thus, we examined this age group to investigate the early changes.

Mice
All experimental procedures conducted adhered to the Australian code for the care and use of animals for scientific purposes set out by National Health and Medical Research Council and the Association for Research in Vision and Ophthalmology's Statement for the Use of Animals in Ophthalmic and Vision Research.Ethics approval was obtained from The Florey Institute of Neuroscience and Mental Health Animal Ethics Committee (17-028-UM).
Unrestricted access to standard lab chow (Barastoc mouse pellet, Ridley Corporation, Melbourne, Vic, Australia) and water was provided to all mice, housed at the University of Melbourne (Kenneth Myer Building, Parkville, Vic, Australia).The animal facility was well ventilated, maintained at 21 • C with room lighting kept on a 12-hour light/ dark cycle (lights on 7 am) and below 50 lux.

Electroretinography
Dark-adapted full-field ERG was performed to examine retinal function as described previously (Zhao et al., 2017), in both C57BL/6JArc (3-month-old, n = 46; 12-month-old, n = 37) and n = 32;n = 10;n = 16) mice.Mice were dark adapted overnight prior to ERG recording.All animal preparation was done using a dim red light in a darkened lightproof room for maximal retinal sensitivity to optimize ganglion cell mediated scotopic threshold response (STR) (Bui and Fortune, 2004;Saszik et al., 2002).
A pair of custom-made chlorided silver (99.9 % pure, A&E Metal Merchants, Sydney, NSW, Australia) connected to platinum leads (F-E2, Grass Technologies, West Warwick, RI, USA) formed the active (placed on the central cornea) and reference (placed just behind the equator) electrodes.A ground electrode (stainless-steel needle, F-E2, Grass Technologies) was inserted subcutaneously into the tail.Responses from both eyes were collected simultaneously within a Faraday cage (Photometric Solutions International, Huntingdale, Vic, Australia).
A series of luminous energies (-5.53 -2.07 log cd.s.m − 2 ), starting from the dimmest, were delivered by a pre-calibrated LED light source embedded into a Ganzfeld sphere (Photometric Solutions International, Huntingdale, Vic, Australia) to elicit signals for key retinal cell classes.At very dim luminous energies near absolute threshold, the positive scotopic threshold response (pSTR) amplitude and time provides a measure of ganglion cell response (Bui and Fortune, 2004;Saszik et al., 2002).The photoreceptor response (P3) was elicited using the brightest stimuli; with the initial electronegative component of the ERG waveform (the a-wave) modeled using a delayed-Gaussian function (Lamb and Pugh, 1992).This function returns (i) a saturated amplitude (RmP3, µV) indicating the number of functioning photoreceptors and (ii) a sensitivity (S, m 2 .cd− 1 .s− 3 ) measure of the efficiency of the phototransduction cascade.The positive deflection of the ERG waveform (the b-wave) is a composite of the negative photoreceptor response (P3) and the positive bipolar cell response (P2).Subtracting the modeled P3 from the raw ERG waveform returns the amplitude of each P2 waveform, which can be described by an agonist-receptor characteristics across the range luminous energies (Fulton and Hansen, 1988;Naka and Rushton, 1966).This function returns a maximum amplitude (V max, µV) indicating the combined bipolar cell response, and a sensitivity parameter (1/K, log cd − 1 .s− 1 .m 2 ).At the brightest intensity (2.07 log cd.s.m − 2 ), four flashes were delivered using an inter stimulus interval of 500 ms.The response to the first flash contained a mixed rod and cone response, and the responses elicited by subsequent flashes were from cones only as the stimulus was presented during the rod refractory period.

Optical coherence tomography
Retinal OCT was taken for each eye, centered at the optic disc (volume scan of 30 • x 25 • , 8.0 mm x 6.7 mm, 768 A/B-scan, 121 B-scans, Heidelberg Engineering GmbH, Heidelberg, Germany) as described previously (Lee et al., 2022).The average of both eyes was calculated in C57/BL6JArc (3-month-old, n = 46; 12-month-old, n = 42) and Thy1-YFPh (3-month-old, n = 44; 12-month-old, n = 34) mice and Thy1-YFPh mice.The retinal layers were segmented automatically using the in-built segmentation algorithm (HEYEX v6.16.2).The ganglion cell layer in mice is a single cell layer and is difficult to segment, as such in this study the ganglion cell layer and the inner plexiform layer were combined and presented as ganglion cellinner plexiform layers.
P.Y. Lee and B.V. Bui

Data analysis and statistics
The results of both eyes were averaged within an individual animal.All ERG parameters were normally distributed when tested with either the Shapiro-Wilk or Kolmogorov-Smirnov normality tests.A two-way ANOVA was conducted in Prism 8 (GraphPad Software, San Diego, CA, USA) to compare the group averaged ERG parameters between 3and 12-month-old C57BL/6J and Thy1-YFPh mice.Sidak's multiple comparison test was conducted when there was a significant age-strain interaction effect.To compare the age effects of ERG parameters in Thy1-YFPh mice, a one-way ANOVA was conducted in Prism 8. Some OCT parameters were not normally distributed, thus generalized linear model was used to analyze the results (IBM SPSS Statistics for Windows, Version 29.0.0.0,Armonk, NY: IBM Corp, USA).All group data were expressed as group mean ± 95 % confidence interval.
To model the relationship between ERG components, multiple linear regressions were fitted for each mouse strain in Minitab 19 (Minitab, LLC, State College, PA, USA).

Age-related comparison of retinal function
Fig. 1 shows the group averaged full-field ERG waveforms in C57BL/ 6J (Fig. 1A) and Thy1-YFPh (Fig. 1B) mice, as well as their corresponding group averaged parameters (Fig. 1C-H).Traces toward the bottom of the columns were elicited with the dimmest flashes to return the scotopic threshold response (STR), which is characterized by positive (pSTR) and negative (nSTR) components.Rod ON-bipolar cell responses dominate the ERG component at moderate luminous energies (-2.75 and -1.61 log cd.s.m − 2 ).At brighter light levels, the murine ERG signal comprises of rod and some cone photoreceptor negative a-wave, followed by rod and cone (ON and OFF) bipolar cell (b-wave).Cone responses (upper most traces) in mice have a small b-wave that is about a quarter the size of the mixed full field response.In general, responses from older eyes were smaller than those from younger eyes, which was evident at all luminous energies.

Further comparison of aging in Thy1-YFPh mice
Table 1 summarizes the ERG parameters and one-way ANOVA results comparing the three Thy1-YFPh age groups, revealing significant differences for all parameters.Post-hoc analyses showed that there was a significant decline in photoreceptor amplitude in older eyes compared to younger eyes (RmP3, 12 m vs 3 m, p = 0.001; 16 m vs 3 m, p < 0.001).However, the decline in photoreceptor sensitivity was only evident in 16-month-old eyes compared to 3-month-old eyes (log S, p = 0.003).
Similarly, there was a significant decline in bipolar cell amplitude in the older eyes compared to younger eyes (V max , 12 m vs 3 m, p < 0.001; 16 m vs 3 m, p < 0.001).Interestingly, there was a significant increase in bipolar cell sensitivity to light in 12-month-old Thy1-YFPh eyes compared to 3-month-old eyes (1/K, p = 0.003).This was followed by a downward trend in 16-month-old eyes compared with 12-month-old eyes (Fig. 1F).Ganglion cell pSTR amplitude was not reduced at 12 months of age, and only became significantly reduced when 16-monthold eyes were compared to 3-month-old eyes (pSTR, p = 0.001).In a similar manner, ganglion cell response timing was slower in 16-monthold compared to 3-month-old mice (pSTR time, p = 0.002), with no significant difference noted between 3-and 12-month-old Thy1-YFPh mice.

Relative age-related changes in ERG component amplitudes
To understand how the responses of each retinal cell class differ between strains and with age, response parameters from each animal was expressed relative to the 3-month-old group average of its own strain (Fig. 2).Photoreceptor responses had declined to 84.0 ± 2.5 % in C57BL/6J mice and was similarly reduced to 80.2 ± 5.2 % in Thy1-YFPh mice by 12 months of age.Consistent with this reduction in photoreceptor input into downstream neurons, we observed an agerelated reduction in bipolar cell (75.6 ± 2.3 %) and ganglion cell (79.7 ± 3.7 %) responses that were similar in magnitudes in 12-monthold C57BL/6J mice.However, in Thy1-YFPh mice, although bipolar cell function was reduced to 68.1 ± 5.5 % compared to 3-month-old animals, there was a relative preservation of ganglion cell function (91.7 ± 5.0 %).A significant interaction effect (two-way ANOVA, F 2,90 = 7.34, p = 0.001) showed that the pSTR was relatively preserved in Thy1-YFPh mice compared to C57BL/6J mice at 12 months of age.However, at 16 months of age, this preservation of the pSTR in Thy1-YFPh mice was not significant, with a general decline in responses of the 3 cell classes with age (Fig. 2B).

Relationship between serially generated ERG components
As the components of the ERG waveforms are generated in a series (i.e., outer retinal differences are expected to affect inner retinal responses), it is worth considering if aging changes the relationships between ERG parameters, and if such differences vary between strains.Fig. 3 shows the relationship between photoreceptor and bipolar cell amplitudes and sensitivities in both strains of mice for all eyes measured.There was a significant linear relationship between photoreceptor amplitude and its corresponding bipolar cell amplitude in 3-month-old C57BL/6J (Fig. 3A, slope 1.37, R 2 = 0.87, p < 0.001) and Thy1-YFPh (Fig. 3B, slope 1.49, R 2 = 0.91, p < 0.001) mice.Whilst smaller amplitudes were expected, the slope became significantly flatter with age (C57BL/6J, Fig. 3A, p = 0.024; Thy1-YFPh, Fig. 3B, p = 0.003), which means for a given drop in photoreceptor amplitude, there was less decline (or more preservation) of bipolar cell amplitude in older mice compared with younger mice.
Similarly, there was a significant linear relationship between photoreceptor and bipolar cell sensitivities to light in both strains (C57BL/6J, Fig. 3C, slope 1.56, R 2 = 0.63, p < 0.001; Thy1-YFPh, Fig. 3D, slope 1.02, R 2 = 0.56, p < 0.001).In 12-month-old C57BL/6J mice, there was a significant flattening of the slope compared to younger eyes (Fig. 3C, 1.00 vs 1.56, p = 0.002).This suggests that there was a change in gain relationship in bipolar cell sensitivity with age, that is, for a given reduction in photoreceptor sensitivity, there was some preservation of bipolar cell sensitivity.
Multiple linear regression showed that photoreceptor amplitude alone provided a good estimate of the bipolar cell amplitude in both strains (Table 2).Adding photoreceptor sensitivity to the regression did not help to further explain variability in bipolar cell amplitude.
Fig. 4 explores the relationship between the ganglion cell generated pSTR and its input as indicated by bipolar cell amplitude and sensitivity, and how these change with age in both strains.Fig. 4A shows that in C57BL/6J mice, there was a significant linear relationship between bipolar cell and ganglion cell amplitudes (3-month-old, slope 0.02, R 2 = 0.33, p < 0.001; 12-month-old, slope 0.02, R 2 = 0.40, p < 0.001).However, there was no significant slope difference between 3-and 12month-old C57BL/6J mice (Fig. 4A, p = 0.931).
Multiple linear regression showed that in both strains, bipolar cell amplitude and sensitivity together provided a better estimate of the ganglion cell pSTR amplitude compared to either the amplitude or the   sensitivity alone (Table 3).

Age-related differences in retinal structure
Fig. 5 summarizes the thickness of key retinal layers in 3-and 12month-old C57BL/6J and Thy1-YFPh mice measured using OCT.Generalized linear models showed that there was a small but significant thickening in all layers with age except the inner nuclear layer in both strains.Between strains, there was photoreceptor inner and outer segments were thicker in Thy1-YFPh mice compared to C57BL/6J mice (Fig. 5, p = 0.033).There was also a significant strain and age interaction in the ganglion cellinner plexiform layers (Fig. 5C, p = 0.016).
To examine the relationship between retinal layers and the corresponding ERG parameter of each neuronal cell class, we correlated ERG and OCT in a subgroup of C57BL/6J mice, where no significant correlations were noted in both 3-and 12-month-old C57BL/6J mice (Supplementary Fig. 1).

Discussion
Our study showed that there were general reductions in photoreceptor, bipolar cell and ganglion cell amplitudes in both 12-month-old C57BL/6J and Thy1-YFPh mice compared to their younger counterparts (Fig. 1).We also found that between 3 and 12 months of age photoreceptor sensitivity was stable, but there was an increase in bipolar cell sensitivity to light with age in both strains.All neuronal responses were generally larger in Thy1-YFPh mice than C57BL/6J mice of the same age.Finally, there was greater age-related preservation of the ganglion cell-mediated pSTR responses in Thy1-YFPh compared with C57BL/6J mice (Fig. 2).
The reduction in photoreceptor amplitude with age may be associated with anatomical findings of photoreceptor cell loss (Gresh et al., 2003;Wang et al., 2018), or a decrease in photoreceptor layer thickness in mouse eyes aged 12-18 months (Ferdous et al., 2021).However, some studies reported no age-related decline in photoreceptor density or cell loss in 12-month-old (Li et al., 2001) or 33-month-old (Trachimowicz et al., 1981) mice.Our in vivo measurement using OCT showed a small thickening of the outer nuclear layer and photoreceptor inner/outer segments (Fig. 5).Why a thickening was observed was unclear.Previous studies have shown reduction in aquaporins channels with age, and absence of aquaporin-4 has the potential to impact outer nuclear layer thickness (Ortak et al., 2013;Yuan et al., 2009).Further studies are required to investigate the age-related changes in aquaporins and how they relate to outer nuclear layer thickness.Given these values fall close to the limits of the axial resolution (3.87 µm), these data suggest that there was little gross age-related structural loss in 12-month-old  mice, which is supported by ex vivo quantification of outer nuclear layer thickness (Supplementary Fig. 2).It is worth noting that the small sample size in the ex vivo quantification of outer nuclear layer thicknesses in the C57BL/6J mice may contribute to the discrepancy in thickness measured using OCT.Further study using larger sample sizes and/or using hematoxylin-eosin staining is warranted for better quantification of retinal cross section morphology.Given that there was a 16.0 % and 19.8 % reduction in photoreceptor function in C57BL/6J and Thy1-YFPh mice, respectively, one may have expected a reduction in outer retinal thickness, indicative of fewer photoreceptors or shorter outer segments.The absence of gross thinning might suggest that murine photoreceptors become less efficient with age.
As there was no age-related loss in photoreceptor sensitivity, it appears that the amplification stages of the phototransduction cascade (rhodopsin activation of transducin and phosphodiesterase hydrolysis of cGMP) had similar efficiency between 3 and 12 months of age.In the absence of gross neuronal loss, photoreceptors may have smaller responses via a range of age-related mechanisms including dysregulation of cellular maintenance and repair processes (e.g.autophagy, DNA repair capacity), as well as reduced energy supply due to impaired mitochondrial metabolism (Boya et al., 2016;Corso-Diaz et al., 2020;Nadal-Nicolas et al., 2018b;Parapuram et al., 2010;Wang et al., 2010Wang et al., , 2018)).Age-related reductions in sodium potassium ATPase activity and cGMP basal levels have been noted in other neural tissues and sensory organs (Ding et al., 2018;Poehlman et al., 1993;Scavone et al., 2005).Loss of sodium potassium ATPase was shown to impact visual function and cause age-dependent photoreceptor degeneration (Luan et al., 2014).
Histological studies have shown reduced inner nuclear layer thickness with age in rats (Katz and Robison, 1986;Weisse, 1995).In contrast, our study did not show any significant age-related differences in gross inner nuclear layer thickness (Fig. 5 and Supplementary Fig. 2).High resolution imaging showed sprouting/dendritic growth of rod bipolar cells extending through the outer plexiform layer into the outer nuclear layer in old C57BL/6J mice, suggestive of synaptic changes with age (Liets et al., 2006;Samuel et al., 2011;Terzibasi et al., 2009), which may not be picked up with in vivo OCT imaging.Synaptic remodelling may reflect plasticity bipolar cells to maintain inner retinal responses as the retina ages.
Consistent with this possibility, despite the age-related reduction in the ERG bipolar cell b-wave amplitude, we found a significant increase in bipolar cell sensitivity between 3 and 12 months of age (Fig. 1).This increase was not evident in 16-month-old Thy1-YFPh mice (Fig. 1).Why compensatory mechanisms decline between 12 and 16 months of age is unclear.Nadal-Nicolas et al. (2018a) reported a reduction in the estimated total number of cells in inner nuclear layer between 6 and 15 months of age.Thus it may be that from 15 months old onwards loss of cells overwhelms compensatory mechanisms resulting in attenuation of both bipolar cell amplitudes and sensitivity.Charng et al. (2011) also observed an increased bipolar cell sensitivity in older albino rats (18-vs 3-month-old).A potential mechanism accounting for an increase in bipolar cell sensitivity may be changes in the interactions between bipolar cells and inner retinal inhibitory pathways.There have been reports of a reduction in the peak amplitude of amacrine cell-driven oscillatory potentials in aged rats (>18 months old) compared to 4-month-old rats (Katano et al., 2001).With age, there was a 21 % reduction in the number of a subset of amacrine cells (tyrosine hydroxylase-immunoreactive) in 24-month-old compared to 3-month-old rats (Roufail and Rees, 1997).The densities of vGlut3-positive and ChAT-positive starburst amacrine cells in the inner nuclear and ganglion cell layers were reported to be reduced in old (24 -28 months) compared to young (3 -5 months) C57BL/6J mice (Samuel et al., 2011).How amacrine cells numbers and their interactions with bipolar cells change with age require further investigation.
Anatomically, amacrine cells appear to show less age-related changes in dendritic morphology than bipolar cells (Samuel et al., 2011).In older murine retina, bipolar cell dendrites sprouted along with horizontal cell dendritic sprouting, whereas bipolar cell axons remained largely unchanged (Samuel et al., 2011).This raises the possibility that an age-related increase in bipolar cell sensitivity might also arise from the outer plexiform layer.Indeed, the gain relationship between photoreceptors and bipolar cells is readily modified by background light levels, an effect that appears to be mediated by horizontal cell interactions with bipolar cells via GABAergic channels that modify calcium influx and can change the gain relationship between photoreceptors and bipolar cells (Snellman et al., 2008).Further studies are needed to better understand the role of age-related differences in the outer and inner plexiform layers in increasing b-wave sensitivity.
Reductions in ganglion cell function with age may reflect direct changes to ganglion cells, or a reduction in input from photoreceptors and/or bipolar cells.Given the age-related reduction in the photoreceptor a-wave amplitude (at 12 months 16.0 % smaller than 3-monthold, Fig. 2A), we expected to see a proportional decline in the b-wave and in turn the pSTR.This was observed in C57BL/6J mice for the bwave (24.4 % smaller than 3-month-old), although the pSTR was a little less affected (20.3 % smaller than 3-month-old).However, preservation of the inner retinal response was striking in 12-month-old Thy1-YFPh mice, where the pSTR was only reduced by 9.3 %, whilst the bipolar cell response for the same eyes was reduced by 31.9 % (and photoreceptors down by 19.8 %) relative to 3-month-old mice (Fig. 2A).This relative preservation of the pSTR may be accounted for by the observed increase in bipolar cell sensitivity to light.As shown in Fig. 1F, the increase in bipolar cell sensitivity between 3 and 12 months of age was similar in C57BL/6J (0.13 log cd − 1 .s− 1 .m 2 ) and Thy1-YFPh (0.16 log cd − 1 .s− 1 .m 2 ) mice.Although similar in magnitude, this surprisingly did not lead to the same pSTR preservation, which was much larger in Thy1-YFPh mice.This suggests that the inner retinal compensation for reduced outer retinal input is different or proceeds at a different rate with aging in the two strains of mice.
Although Thy1-YFPh mice have a background strain of C57BL/6J mice, the introduction of fluorescent proteins may not be entirely inert.Comley et al. (2011) showed an upregulation in half of all the cell stress genes in Thy1-YFP mice compared to wild type.Evidence of altered physiology has been noted in other strains with a range of transgenic mice showing varying rates of aging (Koks et al., 2016;R. Yuan et al., 2009).Differences in retinal responses and how they change with age have been noted in different strains of rats of close genetic makeup raised under the same conditions (Charng et al., 2011;Heiduschka and Schraermeyer, 2008).Why aging might be accelerated in some strains is not clear, however understanding the genetic determinants of age-related differences would be insightful.It is not clear whether the preservation of inner retinal function seen here in Thy1-YFPh mice represented accelerated or delayed aging relative to C57BL/6J mice.As inner retinal function was less reduced, one might suggest that Thy1-YFPh mice retina show signs of delayed aging.
Studies of age-related differences in inner retinal structure have returned conflicting results.Some studies reported an age-related reduction in retinal ganglion cell numbers (Danias et al., 2003;Katz and Robison, 1986;Neufeld and Gachie, 2003;Weisse, 1995), while others have found stable cell ganglion cell numbers in aged rodent retinae (Harman et al., 2003;Nadal-Nicolas et al., 2018a).We did not observe any significant change in the density of ganglion cell nuclei in a subset of eyes from both strains of both ages (Supplementary Fig. 2).
Furthermore, others have shown thinning of total retinal, retinal ganglion cell and inner plexiform layers in retinal cross sections of old mice compared to young mice, however, the thickness relative to the total retinal thickness was unaltered with age (Samuel et al., 2011).Whole mouse eyes imaged using magnetic resonance imaging did not reveal any changes to total retinal thickness in C57BL/6J mice or mice with C57BL/6J background aged 18 months compared to 2 months (Berkowitz et al., 2014(Berkowitz et al., , 2017)).Previous in vivo studies have shown progressive thinning of total retinal and nerve fibre layer thicknesses in older rats, from 6 months of age (Jiang et al., 2018;Nadal-Nicolas et al., 2018a).Shariati et al. (2015) reported a modest ganglion cell complex thinning of 1.3 μm measured using in vivo OCT in 12-month-old compared to 3-month-old C57BL/6J mice.In contrast to this, we found a small thickening with age in both nerve fibre layer (1.34 μm in C57BL/6J and 1.28 μm in Thy1-YFPh mice) and ganglion cellinner plexiform layers (0.04 μm in C57BL/6J and 0.96 μm in Thy1-YFPh mice) (Fig. 5).It is worth noting that these differences are small and well within the axial resolution limits of spectral domain OCT.Also, these differences did not arise from a degradation of image quality with age.
Using the image quality index during OCT acquisition as an indication of clarity of ocular media, there was no significant strain differences in our study (two-way ANOVA, F 1,328 = 0.02, p = 0.888; 3-month-old, Thy1-YFPh 36.6 ± 0.4 vs C57BL/6J 36.8 ± 0.4; 12-month-old, Thy1-YFPh 33.3 ± 0.5 vs C57BL/6J 33.2 ± 0.5).Nevertheless, differences in thickness findings between our study and previous studies may be due to the variation between devices, the retinal region quantified, and the specific retinal layer segmentation approaches employed by the different devices.
Overall, there were no clear gross age-related structural differences that would account for the observed functional differences.This disconnect between structure and function in aging has been noted previously (Bissig et al., 2013;Spear, 1993) and is also supported by our findings that there were no significant correlations between in vivo functional and structural measures in C57BL/6J mice regardless of age (Supplementary Fig. 1).There is a possibility that age-related elongation of the eye may affect ERG outcomes.An age-related increase in eye size would lead to a reduction in effective retinal illuminance which would equally affect all ERG components, particularly those that are mediated by the radially oriented neurons (i.e., photoreceptors and bipolar cells) (Chou et al., 2011;Frishman and Wang, 2011).Given that the STR was less affected than outer retinal components in our study and previous studies have shown similar eye weights and axial lengths in C57BL/6J mice aged between 2 and 6 months and 12-18 months (Chou et al., 2011;Ferdous et al., 2021), these issues are less likely to account for the current observations.More subtle dendritic or synaptic differences may be present to make the pSTR larger than expected given the age-related reduction of the awave and b-wave in 12-month-old mice, particularly those of the Thy1-YFPh strain (Lee et al., 2022).Light responses from retinal ganglion cells are mediated by their excitatory and inhibitory synaptic inputs from bipolar cells and amacrine cells (Pang et al., 2003;Sagdullaev et al., 2011), however how this interaction alters with age is unclear.Even though the dendritic arbours of retinal ganglion cells and synaptic density in the inner plexiform layer have been shown to decrease with age, single cell electrophysiological recordings of older ganglion cells in C57BL/6J mice suggest preservation of function (Samuel et al., 2011).The intrinsic excitability of retinal ganglion cells showed changes in firing behaviours during development (Myhr et al., 2001;Sernagor et al., 2001).Whether this changes with age to maintain the overall function of the older ganglion cells require further investigation.
We observed a small but significant thickening of ganglion cellinner plexiform layers between 3-and 12-month-old Thy1-YFPh mice (Fig. 5).The age-related thickening of this combined layer contradicts findings from previous studies as discussed above (Samuel et al., 2011;Shariati et al., 2015).However, our finding that the age-related thickening was more apparent in Thy1-YFPh mice compared to C57BL/6J mice may suggest that the Thy1-YFPh mice are capable of more remodelling in these layers, perhaps an adaptation to preserve the ganglion cell function to overcome the age-related decline in function in the outer retina.This inner retinal gain appeared to be on a downward trend by 16 months of age (Fig. 2).
Interestingly, not only did we find that the Thy1-YFPh mice age differently to the C57BL/6J mice, but they also appeared to have larger functional responses (Fig. 1).Previous reports suggested that Thy1-YFP mice had ERG waveforms similar to C57BL/6J mice (Hung et al., 2016;Li et al., 2019).However, in those previous studies the two strains were not directly compared.We found that at 3 months of age, the photoreceptor a-wave, bipolar cell b-wave and ganglion cell pSTR responses were larger (22.8 %, 32.3 % and 60.9 %, respectively) in Thy1-YFPh compared to C57BL/6J mice.When we compared retinal layer thicknesses, the difference between the two strains of mice were subtle.Indeed, the only significant difference between the two strains was slightly longer photoreceptor inner/outer segments in Thy1-YFPh (thicker layer) compared to C57BL/6J mice (Fig. 5G).Whether this could potentially capture more light photons and thus elicit a larger ERG P.Y. Lee and B.V. Bui response is unknown.Despite gross structural similarities, Thy1-YFPh eyes not only produce a larger signal but are also substantially more sensitive to light.Slightly longer photoreceptors may account for larger and more sensitive ERG P3 responses (difference of 0.17 log m 2 .cd− 1 .s − 3 , 3-month-old).The sensitivity difference between strains is even greater for bipolar cell responses (difference of 0.37 log cd − 1 .s− 1 .m 2 , 3-month-old).These differences may contribute to the different gain relationships between photoreceptors and bipolar cells in the two strains.
Previous studies employing the full-field ERG have reported declines in a-wave and b-wave amplitudes, which are indicative of outer retinal functional decline with age.The current work provides a measure of ganglion cell function in addition to quantification of upstream input from photoreceptors and bipolar cells.This approach raises the possibility that there is preservation of ganglion cell function in older eyes may well account for the preservation of visual performance to ages of 18 months or more in C57BL/6J mice, despite reductions in outer retinal mitochondrial activity at these ages (Kam and Jeffery, 2015).This study has a number of limitations worthy of mention.Whilst the findings from this study provide insights into the relationship between the different neuronal responses in the retina and how they alter with age, further examination linking the ERG to visual performance, for example visual evoked potentials and/or optokinetic response is warranted (Berkowitz et al., 2014;Sato and Adachi-Usami, 2003;Sugita et al., 2020).Moreover, emerging in vivo imaging approaches such as mitochondrial uncoupler 2,4 dintirophenol OCT and magnetic resonance imaging (Berkowitz et al., 2020) or optoretinography (Ma et al., 2021) will likely be useful tools to uncover the role of mitochondrial activity in the retina.
It is important to note that the C57BL/6J strain lacks nicotinamide nucleotide transhydrgenase (Nnt) which has been shown to show glucose intolerance (Toye et al., 2005).The absence of Nnt may influence the way in which the C57BL/6J eye ages as evidenced by findings that C57BL/6J mice show different responses in age-related responses to caloric restriction (Forster et al., 2003;Li et al., 2012).Whether this might account for our finding that between 3 and 12 months of age C57BL/6J mice tended to show less relative preservation of ganglion cell pSTR requires further investigation.Thus consideration of strain genetics is important in studies of aging.
The sample sizes of 12-and 16-month-old Thy1-YFPh mice were relatively small compared to the 3-month-old Thy1-YFPh cohort, as well as 12-month-old C57BL/6J mice.This means that there is less certainty in the correlations between ERG parameters in the older Thy1-YFPh cohort compared with other groups.Nonetheless, we found a significant age-related decline in neuronal responses and a significant difference in correlation slopes.In addition, we did not have ERG data on 16to 18-month-old C57BL/6J mice, nor structural data on 16-month-old Thy1-YFPh mice.Having this would allow a better understanding of how the retina ages past 12 months of age and facilitate a comparison of these responses to the Thy1-YFPh mice.

Conclusions
This study detailed the in vivo functional and structural retinal changes in 2 strains of mice of the same background, C57BL/6J and Thy1-YFPh mice, and how these differed between the two strains.In general, there were larger ERG responses in the Thy1-YFPh compared to C57BL/6J mice.In vivo structure measured using OCT was largely similar between the 2 strains, except for evidence of slightly longer inner/outer photoreceptor segments in Thy1-YFPh mice.At 12 months of age, there was evidence of functional decline in photoreceptor and bipolar cell responses but increased bipolar cell sensitivities in both mouse strains.There was also a relative preservation of inner retinal function up to 12 months of age, especially in the Thy1-YFPh cohort.These data provide evidence of greater age-related retinal plasticity in Thy1-YFPh compared with C57BL/6J mice.Importantly, these data highlight the possibility that different strains of mice have differing capacity to compensate for cell loss and may show varying susceptibility to injury.Understanding the age-related differences between strains may help us understand why different strains of mice can show different susceptibility to injury.

Table 2
Summary of multiple linear regression fitting for bipolar cell amplitude (V max ) and photoreceptor amplitude (RmP3) and sensitivity (S).

Table 3
Summary of multiple linear regression fitting for ganglion cell amplitude (pSTR) and bipolar cell amplitude (V max ) and sensitivity (K).