1. Cell Biology
  2. Physics of Living Systems
Download icon

Rod nuclear architecture determines contrast transmission of the retina and behavioral sensitivity in mice

  1. Kaushikaram Subramanian
  2. Martin Weigert
  3. Oliver Borsch
  4. Heike Petzold
  5. Alfonso Garcia-Ulloa
  6. Eugene W Myers
  7. Marius Ader
  8. Irina Solovei
  9. Moritz Kreysing  Is a corresponding author
  1. Max Planck Institute of Molecular Cell Biology and Genetics, Germany
  2. Center for Systems Biology Dresden, Germany
  3. Cluster of Excellence, Physics of Life, Technische Universität Dresden, Germany
  4. Technische Universität Dresden, Germany
  5. Biozentrum, Ludwig Maximilians Universität, Germany
Research Article
  • Cited 2
  • Views 1,395
  • Annotations
Cite this article as: eLife 2019;8:e49542 doi: 10.7554/eLife.49542

Abstract

Rod photoreceptors of nocturnal mammals display a striking inversion of nuclear architecture, which has been proposed as an evolutionary adaptation to dark environments. However, the nature of visual benefits and the underlying mechanisms remains unclear. It is widely assumed that improvements in nocturnal vision would depend on maximization of photon capture at the expense of image detail. Here, we show that retinal optical quality improves 2-fold during terminal development, and that this enhancement is caused by nuclear inversion. We further demonstrate that improved retinal contrast transmission, rather than photon-budget or resolution, enhances scotopic contrast sensitivity by 18–27%, and improves motion detection capabilities up to 10-fold in dim environments. Our findings therefore add functional significance to a prominent exception of nuclear organization and establish retinal contrast transmission as a decisive determinant of mammalian visual perception.

Introduction

The structure of the vertebrate retina requires light to pass through multiple cell layers prior to reaching the light-sensitive outer segments of the photoreceptors (Dowling, 1987). In nocturnal mammals, the increased density of rod photoreceptor cells demands a thicker (Němec et al., 2007; Peichl, 2005) rod nuclei-containing outer nuclear layer (ONL). For mice, where rods account for around 80% of all retinal cells (Hughes et al., 2017), this layer of photoreceptor nuclei is 55 ± 5 µm thick, thus creating an apparent paradox by acting as a more pronounced barrier for projected images prior to their detection (Figure 1A). Interestingly, rod nuclei are inverted in nocturnal mammals (Błaszczak et al., 2014; Falk et al., 2019; Kreysing et al., 2010; Solovei et al., 2009; Solovei et al., 2013) such that heterochromatin is detached from the nuclear envelope and found in the nuclear center, whereas euchromatin that has lower mass density (Imai et al., 2017) is re-located to the nuclear periphery. Given that this nuclear inversion is exclusive to nocturnal mammals and correlates with the light-focusing capabilities of isolated nuclei, it was proposed as an evolutionary adaptation to life under low-light conditions (Błaszczak et al., 2014; Kreysing et al., 2010; Solovei et al., 2009). However, the nature of any visual improvements that could arise from nuclear inversion remains unclear.

Figure 1 with 2 supplements see all
Light scattering by retinal nuclei reduces with chromocenter number during development.

(A) Longitudinal section showing the path of light through the mouse retina, including the rod nuclei dominated outer nuclear layer (ONL). Ganglion cell layer (GCL), inner nuclear layer (INL) and outer nuclear layer (ONL) and the inner and outer segments (IS and OS). (B1) (top) Downregulation of the lamina tether LBR (yellow) enables fusion of mobilized chromocenters and thereby an architectural inversion of mouse rod nuclei. (bottom) FISH images of rod nuclei stained with DAPI (blue) showing the dense chromocenters, LINE rich heterochromatin (H4K20me3, magenta) and SINE rich euchromatin (H3K4me3, green) (B2) DAPI section of WT mouse retina in comparison to a Rd1/Cpfl1-KO mouse retina showing the presence of only the inner retina. (B3) Quantification of image transmission shows that the inner retina alone (Rd1/Cpfl1-KO, N = 5) transmits approximately 50% more image detail than the full retina (N = 11), suggesting significant image degradation in the thick outer nuclear layer. (C) FACS scattering profiles comparing retinal neurons, cortical neurons and N2a neuroblastoma cells showing lower light scattering properties of retina neurons. (Inset) Volume-specific light scattering is significantly reduced in the retinal cell nuclei. (D, E) FACS scatter plot for isolated retinal nuclei from WT developmental stage week three pup (P25) and adult mice demonstrating stronger large angle scattering by the P25 nuclei. (F) Histogram of side scattering in adult and P25 retina depicting a higher side scattering for the developing retinal nuclei. (G) Sorting of developmentally maturing nuclei according to different side scattering signal. Insets show representative examples of Hoechst stained nuclei in the corresponding sort fractions. The rectangles represent sorting gates for microscopy analysis. (H) Quantification of reduced scattering with chromocenter number is sufficiently explained by a wave optical model of light scattering n = 38 nuclei. (Error bars in (H) show s.d.) Scale bars (A) - 10 μm. (B1), G - 5 μm, (B2) – 50 µm.

It is widely assumed that high-sensitivity vision depends on optimized photon capture (Schmucker and Schaeffel, 2004; Warrant and Locket, 2004) and often comes at the expense of image detail (Cronin et al., 2014; Warrant, 1999). Here, we show that nuclear inversion affects a different metric of vision, namely contrast sensitivity under low-light conditions. In particular, we experimentally show that nuclear inversion improves retinal contrast transmission, rather than photon capture or resolution. Advanced optical modelling and large-angle scattering measurements indicate that this enhanced contrast transfer emerges from previously coarse-grained (Błaszczak et al., 2014; Kreysing et al., 2010; Solovei et al., 2009) changes in nuclear granularity, namely a developmental reduction of chromocenter number (Figure 1B1). Moreover, genetic interventions to change chromocenter number in adult mice reduces contrast transmission through the retina, and compromise nocturnal contrast sensitivity accordingly. Our study therefore adds functional significance to nuclear inversion by establishing retinal contrast transmission as a decisive determinant of mammalian vision.

Results

Volume-specific light scattering from chromocenters

To test how the presence of densely packed rod nuclei in the light path affects the propagation of light through the retina, we compared transmission of micro-projected stripe images through freshly excised retinae of wild type (WT) (Figure 1AFigure 1B2 - left image) and Rd1/Cpfl1- KO mice (Chang et al., 2002), which lack all photoreceptors including the ONL (Figure 1B2 - right image). In the absence of photoreceptors and their nuclei, we observed 49% greater imaged detail (cut-off chosen at 50% residual contrast, Figure 1B3). Photoreceptor nuclei contain highly compacted and molecularly dense DNA with significant light-scattering potential (Drezek et al., 2003; Marina et al., 2012; Mourant et al., 2000), while photoreceptor segments have been described as image-preserving waveguides (Enoch, 1961). These findings suggest that light propagation in the mouse retina is significantly impacted, if not dominated, by the highly abundant rod nuclei of the ONL.

We then asked whether retinal cell somata are optically specialized with distinct light-scattering properties. We compared the light scattering by different cell types using high-throughput FACS (Feodorova et al., 2015) measurements. The suspensions of cells or papain-digested retinae were used to measure the cellular light scattering in the far-field using a commercial FACS set up. These measurements revealed that cells isolated from the mouse retina known to typically consist of ~80% rod photoreceptor cells (Hughes et al., 2017), scatter substantially less light than neurons of the brain and cultured neuroblastoma cells (Figure 1C). This trend is seen for forward-scattered light (measured in a narrow range around 0°) but is even more pronounced for side scattering (measured around 90 degrees, see supplementary methods for details), which reflects subcellular heterogeneity. Using forward scattering as a measure of cell size indicates that side scattering normalized by volume (volume-specific light scattering) is also noticeably lower in retinal cells (Figure 1C, inset). This suggests that retinal cells are indeed optically specialized, as they scatter less light for a given size. This unique property for the rod cells could stem from the unusually dense packing of the heterochromatin in the centre of their nuclei, which notably even excludes free GFP molecules (Figure 1—figure supplement 1B).

To determine when the low sideward light scattering characteristic of retinal nuclei emerges, we compared the scattering profile of retinal nuclei in P25 WT pups and WT adult (12 weeks) mice. We found little or no difference between forward light scattering (Figure 1D–E), as predicted by earlier models (Błaszczak et al., 2014; Kreysing et al., 2010; Nagelberg et al., 2017). In stark contrast however, sideward scattering, with a strong potential to diminish image contrast, was significantly reduced in adult retinal nuclei compared to the intermediate developmental stage (Figure 1F). Quantitative analysis of sorted nuclei from P25 retinae further revealed a monotonic relation between chromocenter number and sideward scattering signal (Figure 1G). In particular, those nuclei with the lowest number of chromocenters were found to scatter the least amount of light. In support of this experimental quantification, a wave-optical Mie model of light-scattering by refractive chromocenters closely reproduced the trend of light scattering reduction with chromocenter fusion (Figure 1H).

To establish whether rod nuclear inversion is required to cause the developmental reduction in light scattering, we used a transgenic mouse model (TG-LBR) in which heterochromatin remains anchored at the lamina which in turn prevents the complete fusion of chromocenters (Figure 2A1, A2 and B, Figure 1—figure supplement 2, Figure 1—figure supplement 1; Figure 2—figure supplement 1) (Solovei et al., 2013). FACS experiments of nuclei from TG-LBR retinae, in which > 70% of the nuclei are successfully arrested (Figure 2—figure supplement 2), revealed significantly increased light scattering (Figure 2C, Figure 1F). Specifically, the global maximum of the side-scattering was re-located precisely to the position that is characteristic of nuclei isolated from WT pups at P14, which possess a similar number of chromocenters as inversion arrested nuclei (compare Figure 2B,C, Figure 1F,G). Because inhibition of chromocenter fusion leads to specific increase in scattering, we conclude that the reduction of light scattering with chromocenter number is causal.

Figure 2 with 2 supplements see all
Developmental arrest of chromocenter fusion increases light scattering from rod nuclei in measurements and tissue simulations.

(A1) Schematic of the normal rod nuclear WT development and inversion arrested nuclei by LBR overexpression. (A2) EM images illustrating different electron densities in the euchromatic and heterochromatic phase underlying their refractive index (RI) differences (scalebar 5 μm). (mid-top) Immunostaining of overexpressed of LBR tethers (yellow), and high-density heterochromatin (DAPI, magenta). (mid- bottom) Heterochromatic chromocenters (DAPI, magenta) and euchromatin (H4K5ac, green) (B) Chromocenter number distribution in LBR overexpressing rod nuclei is drastically different from WT mice, and similar to a developing WT pup (P14). (C) Side scattering assessed by FACS for TG-LBR retina nuclei is higher than that of WT nuclei and comparable to that of a WT P14 nuclei with similar chromocenter numbers. Note the shift of peak value upon LBR overexpression. (D1) 3d RI distribution mapped onto anatomically faithful volumetric ONL images. WT inverted architecture (right, top) and early developmental state (left) (simulation). (D2) (top) Differential simulations of light propagation in the ONL, using same positions and shapes of about 1750 nuclei, but varying chromatin distributions. (bottom) Maximum projection illustrating greater proportions of scattered light (angles > 30 deg) in the ONL with multiple chromo-centered nuclei. (E) Quantitative analysis of this data. (F) Angle weighted volume-specific scattering strength for nuclei models evaluated by Mie scattering theory. (G) Excess scattering occurring in multi-chromocenter nuclei models. (H) Chromocenters scattering reconstituted in an emulsion of silica spheres in glycerol-water mixture. Pictograms reflect accurate number ratio of spheres.

Improved retinal contrast transmission

Next, we asked how nuclear substructure could affect the optical properties of the ONL. We first approached this via a simulation that built on recent advances in computational optics (Weigert et al., 2018). This allowed us to specifically change nuclear architecture, while leaving all other parameters, including the morphology and relative positioning of about 1750 two-photon mapped nuclei, unchanged (Figure 2D1, Videos 1 and 2) (Supplementary Methods).

Video 1
2 photon volumetric image of WT mouse retina.
Video 2
3D morphological models of ONL RI distribution used in light propagation simulations.

These simulations suggest that especially the large-angle scattering (cumulative scattering signal at angles > 30 deg) monotonically decreases when 10 chromocenters successfully fuse into one (Figure 2D2 and E). Physically, this effect of reduced scattering can be explained by a reduction of volume-specific scattering for weak scatterers in the size regime slightly above one wavelength of light, similar to scattering reduction techniques proposed for transparent sea animals (Johnsen, 2012) (Figure 2F,G). Furthermore, a minimal optical ONL model reconstituted from suspended beads of different size but same volume fraction (Supplementary Methods) illustrates how a decreased geometric scattering cross section after fusion leads to reduced scattering-induced veil that helps to prevent contrast losses (Figure 2H and inset). Taken together these data suggest that nuclear inversion might serve to preserve contrast in retinal transmitted images.

To experimentally quantify the optical quality of the retina with respect to nuclear architecture, we applied the concept of the modulation transfer function (MTF), a standard way to assess image quality of optical instruments (Boreman, 2001). Specifically, MTF indicates how much contrast is maintained in images of increasingly finer sinusoidal stripes (Figure 3—figure supplement 1B,C). We therefore devised an automated optical setup (Figure 3—figure supplement 1A) that allowed us to project video sequences of demagnified sinusoidal stripe patterns through freshly excised retinae and assess the retinal transmitted images for contrast loss. This custom built set-up mimics the optics of the mouse eye, in particular its f-number (Schmucker and Schaeffel, 2004), while circumventing changes of the optical apparatus in-vivo (Figure 3—figure supplement 1, Materials and methods).

Strikingly, we found that wildtype retinae improve contrast transmission throughout terminal development, with adult retinae showing consistently elevated MTFs compared to intermediate developmental stages (P14) in which rod nuclei still possess around five chromocenters (Figure 3A). In contrast to many lens-based optical systems, retinal MTFs have a long tail with non-zero residual contrast despite an initial rapid loss of contrast (a characteristic of scattering-dominated optical systems). The monotonic decay of retina-transmitted contrast indicates scattering-induced veil, rather than a frequency cut-off to be the cause of contrast loss (Figure 3A,B Figure 3—figure supplement 2A–D). Collected from >1300 high resolution images, this data reveals that, similar to the lens (Tkatchenko et al., 2010), the retina matures towards increasing optical quality during latest developmental stages, with chromocenter fusion as a putative mechanism of veil reduction.

Figure 3 with 2 supplements see all
Nuclear inversion improves retinal contrast transmission characteristics.

(A) Retinal contrast transmission increases during developmental stages of nuclear inversion, as experimentally revealed by measurements of retina-transmitted sinusoidal stripe patterns (modulation transfer functions). Developmental stage P12-14 (N = 18), compared to wildtype adult (N = 19 animals), note log scale. (B) These improvements in optical quality do not occur in retinae in which rod nuclei are transgenitically arrested in development and maintain 4–5 chromocenters. TG-LBR mouse (N = 18 animals) compared to WT reference (N = 19 animals), N = 1950 images in total. Mean + /- 95% CI. (C) Retinal contrast transmission at visually relevant spatial frequencies showcasing on an average ~49% and~37% better contrast transfer by the WT Adult retina (grey) in comparison to the WT-P14 pup (blue) and TG-LBR Adult (red) respectively. (D) The optical quality improvement of the retina (relative Strehl ratios), as caused by nuclear inversion, is two-fold (p=1.1880e-08 - WT adult vs WT pup, 3.4055e-08 - WT adult vs TG-LBR adult, 0.4761 - TG-LBR adult vs WT pup). (E1) Point spread function (PSF) for WT and LBR adult retinae by projection of 3 µm point light stimuli through the retina, N = 240 measurements in total six retinae. (E2) Intensity quantification along the white dotted line. Shaded region shows ±1 sd. Comparable resolution in transmitted images as assessed by the FWHM of the psf (inset). (E3) Near identical diffuse light transmission by both WT and TG-LBR retinae (n = 2 animals each, mean ± s.d.) (F) Intensity of a moving, retina-transmitted point stimuli for WT (black) and TG-LBR mouse (red). (G) Image-series of a cat approach as seen through the retina of mice, WT and transgenic genotype from various behaviorally relevant distances at the same vision limiting (arbitrarily chosen) signal to noise level. Consistent intensity differences of two or more color shades indicate significantly better predator detection potential for WT mice. Data magnified for clarity.

Next, we asked if developmental improvements in contrast transmission of the retina are indeed caused by chromocenter fusion. For this we used mice in which LBR-overexpression largely arrested chromocenter fusion, resulting in an elevated number of chromocenters in the adult animal, similar to P14 WT (Figure 2A2, B), without displaying any effect on other morphological characteristics (Figure 2—figure supplement 2). Strikingly, repeating MTF measurements on adult retinae of this inversion arrested mouse model (TG-LBR), we find near identical contrast attenuation characteristics as in the developing retina (compare Figure 3A and B). Thus, developmental improvements of retinal contrast transmission are indeed mediated by the inversion of rod nuclei. Notably, when we focus on the retinal transmission data within the spatial frequency regime that is relevant for mouse vision (Alam et al., 2015; Prusky et al., 2004; Prusky et al., 2000), (Figure 3C) it can readily be seen that the contrast transmission is up to 33% greater in WT compared to TG-LBR at frequencies ~ 0.28 cycles/deg. Equally, the contrast transmission in this behaviorally relevant regime also increases up to 45% in WT adult compared to the pups.

Frequently, the quality of image-forming optical systems is reported as a single parameter value called the Strehl ratio (Thibos et al., 2004). Since our image projection setup closely mimics the mouse eye, it allows meaningful comparisons of the Strehl ratios of retinae, by comparing the volumes under MTF curves. With regards to our MTF measurements, we that find the Strehl ratio (computed using the measurements in the spatial frequency range of 0–2 cycles/deg) of a fully developed retina is increased 2.00 ± 0.15 fold compared to that of pups (P14) in which chromocenters fusion was not completed, and similarly 1.91 ± 0.14 fold (ratio of means ± SEM) improved compared to TG-LBR adult retinae (p=3.4055e-08) in which chromocenter fusion was deliberately arrested (Figure 3D).

Since the Strehl ratio makes predictions for the peak intensity of a tissue-transmitted point stimulus, we analysed the effect of micro-projecting a point-like stimulus through the mouse retina (diameter here ~3μm, measurement constrained by outer segment spacing). We found that the resulting image at the back of the WT retina had a near two-fold (1.79 ± 0.38, mean ± SD) higher peak intensity compared to the TG-LBR retina (Figure 3 E1 & E2, N = 119, N = 121, measured in at a total of 6 animals). The measured resolution based on full-width half maximum (FWHM) of the PSF, however, did not show any differences (4.32 ± 2.38 μm, 3.75 ± 2.01 μm, mean ± SD for WT and TG-LBR retinae respectively, Figure 3 E2), especially no changes that could physiologically impact acuity, which is known to be significantly lower in mouse. In addition to independently corroborating our MTF measurements, these results emphasize that nuclear inversion enhances contrast transmission through the retina but is unlikely to benefit acuity. From a mechanistic point of view, these measurements indicate that contrast is lost due to the generation of image veil from side scattering, which overcasts attenuated, but otherwise unchanged signals. Accordingly, when comparing the integrated absolute transmission through rhodopsin-bleached retinae in dedicated experiments (Supplementary methods), we found near identical transmission values for WT and inversion arrested retinae (Figure 3 E3, TWT74 ± 8%, TLBR = 72 ± 5%, mean ± SD), which emphasizes that despite differential image signal, the overall photon arrival at the photoreceptor outer segments, remains unchanged.

An advantage of improved retinal contrast transmission is suggested when following the motion of individual (non-averaged) light stimuli that appear at considerably higher signal-to-noise ratios (Figure 3F) at the outer segments level. A putative visual advantage to appropriately scaled real-life examples, such as images of an approaching cat micro-projected through a mouse retina is illustrated in Figure 3G. Nuclear inversion results in cat images becoming visible considerably earlier compared to mice that lack nuclear inversion (0.70 vs 0.45 meters, at a given arbitrary noise threshold). These results suggest that nuclear inversion may offer enhanced visual competence that originates from improved contrast preservation in retinal images. More objective and established methodologies to test the impact of nuclear inversion for actual behavior is addressed in the next section.

Improved contrast sensitivity

To determine whether the improved retinal contrast transmission translates into improved visual perception, we carried out behavioral tests using Optomotor-reflex measurements (Figure 4A, Video 3). Specifically, we used a fully automated mouse tracking and data analysis pipeline (OptoDrum, Striatech, Germany) (Benkner et al., 2013) to compare the contrast sensitivities of adult WT mice and those with arrested nuclear architecture (TG-LBR). Firstly, contrast sensitivity assessed by the animal’s ability to detect moving stripes, did not differ significantly (p=0.5307, two sample t test) between the two genotypes at photopic light condition (70 Lux – the typical brightness of monitor). Transgenic and WT animals showed comparable visual sensitivity, as quantified by the area under the log-contrast sensitivity curve (AULC, Figure 4B, left) (Villegas et al., 2002). As nuclear adaptation is strongly correlated with nocturnal lifestyle (Solovei et al., 2009; Solovei et al., 2013), we adapted this set-up to assess contrast sensitivity under scotopic light conditions. At 20 mLux, which is the range of brightness in moonlight (Kyba et al., 2017), we again found comparable responses for coarse stimuli (wide large contrast stripes) suggesting equally functional rod-based vision in TG-LBR and WT mice (Figure 4B, right) without noticeable differences in absolute sensitivity. Furthermore, mice deficient of rhodopsin (Rho-/-) (Humphries et al., 1997; Jaissle et al., 2001) confirmed that visual behavior under the displayed scotopic conditions fully relies on the functionality of the rod pathway (Figure 4—figure supplement 1E).

Figure 4 with 3 supplements see all
Nuclear inversion improves contrast sensitivity in the dark.

(A) Illustration of the automated optomotor response experiment to assess the visual performance of mice, shown a 0.06 cycles/deg. (B) Photopic control condition and scotopic coarse stimulus (0.06cycles/deg) control showing no significant difference between WT and TG-LBR mice (p=0.5307, p=0.2842, t-test, Chi-square test). (C) Under scotopic conditions (20 and 2 mLux) the overall sensitivity of the WT mice is 22% and 29% higher than TG-LBR mice (area under log contrast sensitivity curve, AULC) mean+/-95% CI, (p=0.00081, 0.0047, two sample t-test). (D) Contrast sensitivity curves evaluated at 20mLux light intensity. Significant differences appear at angular sizes above 0.15 cycles/deg (p=0.038, two sample t-test). (E) Behavior differences are strongest for stimuli close to the visual threshold. Here the mice in possession of the inverted rod nuclei (WT) possess an up to 10 times higher sensitivity at intermediate contrasts (29% vs 3% correct response in 45–75% contrast range), and a six times reduced risk to miss a motion stimulus at high contrasts (10 vs 59% failure in detection, 0.26–0.3 cycles/deg), (p<0.0001, Chi-square test) mean ± s.d. (F) Rescue experiment demonstrating sufficiency of improved retinal contrast transmission to explain improved sensitivity. Adjusting the level of contrast at the photoreceptor level (by pre-compensation of differential contrast loss) restores sensitivity of TG-LBR mice. N indicates number of individual trials of 10 animals together for each mouse type.

Video 3
Behaving mouse in an Optomotor response set up.

When required to detect finer stripes, WT and TG-LBR mice displayed significant differences in their visual performance, specifically in contrast sensitivity (Figure 4C). At 20mLux we observed an 18% greater AULC for WT mice compared with TG-LBR mice (p=0.00081). At even lower light intensities (2mLux, comparable to a starry night), the difference in AULC values was even greater (ratio 27%, p=0.0047) albeit at lower absolute sensitivities, which agrees with reported values for WT mice (Alam et al., 2015; Prusky et al., 2000; Prusky et al., 2004). The most significant differences in the contrast sensitivity occur above 0.15 cycles/degree (Figure 4D). Especially, in the regime close to the visual acuity (0.26–0.30 cycles/degree), WT mice show up to 10 times (p<0.0001) greater positive response rates at intermediate contrasts (Figure 4E) compared to mice with inversion arrested rod nuclei. Moreover, at 90–100% contrasts, where WT mice approach a maximum responsiveness, we observed a near 6-fold reduced risk to miss a stimulus for WT compared to TG-LBR mice (false negative rates 11% WT, 59% TG-LBR).

Finally, we asked whether reduced visual sensitivity of mice lacking the inverted nuclear architecture can be sufficiently explained by inferior contrast transmission of the retina. Direct comparison of behavioral sensitivity with the MTF curves showed that vision mostly occurs in regions in which retinal contrast transmission is higher than 50% and substantial differences in MTFs occur. Specifically, the 18–27% difference in contrast sensitivity goes together with a 26% higher Strehl ratio in WT retinae when evaluated in the relevant frequency regime (0–0.36 cycles/degree).

This suggests that at low light levels, contrast sensitivity may be directly limited by contrast transmission through the retina, and that a reduction of contrast sensitivity in mice with non-inverted rod nuclei may be explained by increased contrast losses in the retina. Moreover, we did not observe unexpected side effects from the LBR overexpression, at level of retinal (Figure 2—figure supplement 2), ocular or lens anatomy (Figure 4—figure supplement 2), and non-limiting rod vision was normal (Figure 4B (Left)). Nevertheless, it is clear that the complexity of the eye does not permit an exhaustive comparison of all parameters that could potentially be affected by LBR overexpression, including subtle concentration changes in molecules relevant for phototransduction. So how can one rule out the possibility that the loss of sensitivity in LBR overexpressing mice is due to a loss of image contrast, rather than unspecific side effects?

To show that increased contrast sensitivity in WT mice is due to the increased contrast transmission of the retina, we designed a rescue experiment logically equivalent to rescue experiments that show specificity of molecular interventions. Frequently, one excludes nonspecific side effects of a molecular knock-down by rescuing the phenotype via the addition of the protein of interest (if possible a pathway-specific variant of this protein). To show that sensitivity is lost due to retinal loss of contrast, we performed an optical rescue experiment. For this, we first confirmed that contrast transmission through the inner retina is a linear process, with contrasts at the photoreceptor levels being proportional to contrasts in projected images (Figure 4—figure supplement 1B,C). We then adjusted the displayed contrasts in optomotor measurements to pre-compensate for higher contrast losses in the TG-LBR retina while also conserving image intensities. Strikingly, we found that with equal image contrast at the level of the photoreceptor segments, visual competence of LBR mice was rescued and becomes near identical to that of WT mice (Figure 4F). Thus, improved retinal contrast transmission is indeed sufficient to explain increased contrast sensitivity in mice.

Discussion

As an important determinant of fitness, animals evolved a wide range of visual adaptation to see in the dark (Nilsson, 2009; O'Carroll and Warrant, 2017; Thomas et al., 2017; Warrant and Nilsson, 2006; Warrant, 2017). Nocturnal vision is known to rely on highly efficient light capture, both at the level of the lens and photoreceptor outer segments, and often compromises spatio-temporal resolution by summation strategies of neuronal readout (Warrant, 1999; Warrant, 2017). Here we established nuclear inversion as a complementary strategy to maximize sensitivity under low light conditions. Centrally, we show that it is the direction into which light is scattered inside retinal tissue that translates into differential contrast sensitivity. Specifically, we find that the forward scattering characteristic of inverted nuclei (Kreysing et al., 2010; Solovei et al., 2009) mainly suppresses light scattering by nuclear substructure towards large angles, thus preventing image veil and contrast reduction resulting from it.

As mammalian eyes are evolutionarily multi-constrained systems, one could ask if nuclear inversion might also serve other functions beyond the improved contrast sensitivity that we have showed. Slightly reduced thickness of the ONL might translate into more efficient diffusion of nutrients, waste and signals. Similarly, the unusually large fraction of hetero-chromatin in rod photoreceptor cells (Wang et al., 2018), which might enable the small nuclear volume and/or more efficient fate-specific gene silencing (Becker et al., 2017; Hiler et al., 2015; Mattar et al., 2018; Wang et al., 2018), might hypothetically lead to architectural problems for the nucleus which could be circumvented by nuclear inversion. As an example, chromatin distribution is known to have the potential to modulate the mechanical properties of the nucleus (Kirby and Lammerding, 2018; Miroshnikova et al., 2017; Stephens et al., 2017) and LBR downregulation might even be required for shape changes enabling efficient packing of nuclei (Stephens et al., 2019). Lastly, although the size of the PSF is beyond the acuity limit of nocturnal vision, the increase in intensity of a point stimulus at the photoreceptor level could aid a thresholded or otherwise non-linear readout of rod cells, a long-standing hypothesis in the field of visual neuroscience (Barlow, 1956; Field and Rieke, 2002; Nelson, 2017) which was substantiated by the use of quantum-based single-photon sources (Tinsley et al., 2016).

While these additional functions of nuclear inversion currently remain speculations, it is worth reflecting about the relevance of the visual benefits demonstrated here for enhancing animal vision in general. Since our reported mechanism involves improvements in retinal image contrast rather than notable changes in photon transmission that could impact absolute sensitivity (Banks et al., 2015; Cronin et al., 2014; Nilsson, 2009; Warrant, 1999), one might ask why nuclear inversion as an adaptation is exclusive to nocturnal mammals. Wouldn’t improvements in retinal image contrast not also be beneficial for diurnal mammals? Firstly, the larger spacing of photoreceptor segments in the diurnal retina significantly reduces ONL thickness (Solovei et al., 2009; Sterling and Laughlin, 2015; Werner and Chalupa, 2004; Williams and Moody, 2004) and thereby the risk of scattering induced veil and loss of image contrast. Furthermore, as is well known from photography, shot-noise that accounts for image granularity (Barlow, 1956; de Vries, 1943; Rose, 1948) becomes less of a problem with increasing light levels. Million-fold higher light intensities during the day imply a higher safety margin from this noise floor (Warrant, 1999), (Figure 4—figure supplement 1F,G), as required for neural mechanisms of contrast enhancement to function (Artal et al., 2004; Flevaris and Murray, 2015; Hess et al., 1998; Shevell et al., 1992). Such compensatory mechanisms are also likely to explain why no behavioral differences are observed at elevated intensities and why augmented vision becomes pronounced only at low light levels. Last, but not least, our measurements show that, although nuclear inversion improves retinal contrast transmission via reduced image veil, resolution, the limiting factor for high acuity diurnal vision, remains largely unaffected. Besides a reduced need for inverted photoceptor nuclei in diurnal mammals, reduced efficiency of canonical DNA repair mechanisms (Frohns et al., 2014) in highly condensed chromocenters, could mean a significant disadvantage in the diurnal retina and susceptibility to stress and degeneration (Boudard et al., 2011; Dyer, 2016), could also mean a significantly higher cost for inverted nuclei in diurnal species, as their retinae are intrinsically strongly exposed to high-energy, ultra-violet photons.

In conclusion, we showed that rod nuclear inversion is necessary and sufficient to explain optically enhanced contrast sensitivity in mice (Figure 4—figure supplement 3). Our work thereby adds functional significance to a prominent exception of nuclear organization and establishes retinal contrast transmission as a new determinant of mammalian fitness.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional
information
Strain, strain background (M. musculus)C57BL/6NRj, (WT)Janvier LabsColony maintained at biomedical facility of MPI-CBG
Strain, strain background (M. musculus)Tg(Nrl-EGFP)Kind Gift from Jung-Woong Kim (Anand Swaroop laboratory, Ophthalmology and
Visual Sciences, University of Michigan,
Ann Arbor).
(Akimoto et al., 2006)
Strain, strain background (M. musculus)ROSA26-eGFP-DTAKind gift from
Dr. Dieter Saur Klinikum rechts
der Isar, Technische
Universität München
(Ivanova et al., 2005)
Strain, strain background (M. musculus)TG-LBR (Nrl-Lbr)This paper, Dr. Irina Solovei, LMU Munich (Solovei et al., 2013)biomedical facility of MPI-CBGMaterials and methods
Tissue preparation for optical characterization and Flow cytometry
Strain, strain
background
(M. musculus)
Rd1/Cpfl1-KOAder Lab, CRTD Dresden,
TU Dresden
Animal facility
of CRTD
Materials and
methods
Strain, strain
background
(M. musculus)
Rho-/-(Humphries et al., 1997;
Jaissle et al., 2001)
Animal facility
of CRTD
Materials and
methods
Cell line (M. musculus)Neuro-2a
(Neuroblast cells)
DSMZACC-148;
RRID: CVCL_0470
Cell line maintained as per ATCC recommendations
Biological sample (M. musculus)RetinaThis paperbiomedical facility of MPI-CBG,Animal facility
of CRTD
Materials and methods
Tissue preparation for optical characterization and Flow cytometry
Biological sample
(M. musculus)
Brain sectionsThis paperbiomedical facility of MPI-CBGMaterials and methods
Flow cytometry
Antibodyanti-lamin B
(Goat, polyclonal)
Santa CruzSC-6217,
RRID: AB_648158
IF (1:50)
Antibodyanti-LBR (Guinea pig, polyclonal)Kind gift from Dr.H.Herrmann (DKFZ, Heidelberg)IF (1:50)
Antibodyanti- H4K5ac (Mouse monoclonal)Kind gift from Dr.H.Kimura (Tokyo Institute of Technology, Yokohama)Clone 4A7IF (1:100)
Sequence-based reagentFISH ProbesThis paper, Refer to methods for primer sequences.PCR primers(Solovei, 2010; Solovei et al., 2007).
Commercial assay or kitPapain dissociation system kitWorthington Biochemical CorporationPDS LK003150
Software, algorithmMie calculationsMATLAB Scriptomlc-mie(Mätzler, 2002)
Software, algorithmCalculation of MTFMATLABhttps://de.mathworks.com/products/matlab.htmlVersion 2017b, 2018b,
Software, algorithmRetinal light propagationbiobeambiobeam(Weigert et al., 2018)
Software, algorithmSPSSIBM, SPSSibm-spssVersion 25
Software, algorithmOptodrumStriatech GmbHStriatech
OtherAlexa555InvitrogenA31570;
RRID: AB_2536180
Fluorescent dyes
OtherAlexa 488InvitrogenA21202;
RRID: AB_141607
Fluorescent dyes
OtherHoechstThermo Scientific33342Fluorescent dyes
OtherVectashieldVector Laboratories, Inc, USACat. No. H-1000–10Antifade media
OtherAqua Poly-MountPolysciences, Inc, USACat. No. 18606–20Antifade media
OtherFACS tubesCorning Inc, USAREF 352054Falcon round bottom polystyrene
OtherResearch BeadsBD Biosciences655050BD FACSDiva CS and T
OtherSilica beadsWhitehouse ScientificMSS002, MSS004a
OtherND-1.2 filterRosco Laboratories Ince-color+ #299,

Retina sampling and preparation of cryo-sections

Request a detailed protocol

Wild type retinas were sampled from C57/BL6 mice. Eye balls of Nrl-GFP mice (Akimoto et al., 2006) were kindly provided by Jung-Woong Kim (Anand Swaroop laboratory, Ophthalmology and Visual Sciences, University of Michigan, Ann Arbor). Tissues from ROSA26-eGFP-DTA mice (Ivanova et al., 2005) were kindly provided by Dieter Saur (Klinikum rechts der Isar, Technische Universität München). The Rd1/Cpfl1-KO mice were maintained in the Animal facility of the CRTD, Dresden. Preparation of retina cryosections was performed according to protocol described earlier (Eberhart et al., 2013; Eberhart et al., 2012; Solovei, 2010). The enucleated eye balls were shortly washed with EtOH, punched with gauge 23 needle in the equatorial plane and fixed with 4% paraformaldehyde (PFA) (Carl Roth GmbH, Germany) in phosphate-buffered saline (PBS) solution for 3 hr. After fixation, samples were washed with PBS 3x 1 hr each, incubated in 10%, 20% and 30% sucrose in PBS at 4°C for 30 min in each concentration and left in 30% sucrose for overnight. The eyeballs were cut equatorially to remove the anterior parts, including cornea, lens and the vitreous, and eye cups were placed in a mold (Peel-A-Way Disposable Embedding Molds, Polysciences Inc) filled with tissue freezing medium (Jung tissue freezing medium, Leica Microsystems). Frozen blocks were prepared by either immersion of molds with tissues in freezing medium in a 100% ethanol bath precooled to −80°C, or by placing into a container filled with precooled to −70°C 2-methylbutane. After freezing, blocks were transferred to dry ice and then stored at −80°C. Cryosections with thickness of 14–20 μm were prepared using Leica Cryostat (Leica Microsystems) and collected on SuperFrost (Super Frost Ultra Plus, Roth, Germany) or StarFrost microscopic slides (StarFrost, Kisker Biotech GmbH and Co). After cutting, sections were immediately frozen and stored in at −80°C until use.

Immunostaining

Request a detailed protocol

Immunostaining was performed according to the protocol described in detail earlier (Eberhart et al., 2012; Eberhart et al., 2013). Prior to immunostaining, slides with cryosections were removed from −80°C freezer, allowed to thaw and dry at room temperature (RT) for 30 min and then re-hydrated in 10 mM sodium citrate buffer for 5 min. For the antigen retrieval, slides were transferred to a preheated to +80°C 10 mM sodium citrate buffer either for 5 min (H4K5ac) or for 25 min (lamin B and LBR staining). After brief rinsing in PBS at RT, slides were incubated with 0.5% Triton X100/PBS for 1 hr, and once more rinsed in PBS before application of antibodies. Primary and secondary antibodies were diluted in blocking solution [PBS with 0.1% Triton-X100, 1% bovine serum albumin (ICN Biomedicals GmbH) and 0.1% Saponin (SERVA)]. Incubation with antibodies was performed for 12–14 hr under glass chambers in humid dark boxes (Solovei, 2010; Solovei et al., 2007). Washings after incubation with antibodies were performed with PBS/0.05%Triton X-100, 3x 30 min at 37°C. Primary antibodies included anti-lamin B (Santa Cruz, SC-6217), anti-LBR (lamin B receptor; kindly donated by Harald Herrmann, German Cancer Research Center, Heidelberg), anti-H4K5ac (kindly donated by Hiroshi Kimura, Tokyo Institute of Technology, Yokohama). Secondary antibodies were anti-mouse IgG conjugated to Alexa555 (A31570, Invitrogen) and Alexa488 (A21202, Invitrogen). Nuclei were counterstained with DAPI or Hoechst added to the secondary antibody solution. After staining, the sections were mounted under a coverslip with Vectashield (Vector Laboratories, Inc, Burlingame, CA, USA) or Aqua Poly-Mount (Polysciences, Inc, USA) antifade media and sealed with nail polish.

For microscopic analysis of FACS sorted retinal nuclei, sorted nuclei were fixed with 4% PFA in PBS for 10 mins, stained with Hoechst 33342, washed 2x with PBS and mounted on slides under coverslips in antifade medium (see below). The imaging was performed on a confocal microscopy (Zeiss LSM 700 inverted) using a Zeiss 64x 1.4 oil objective.

FISH

Request a detailed protocol

FISH on cryosections was performed as described earlier (Solovei, 2010; Solovei et al., 2007). Probes for LINE, B1 and major satellite repeat (MSR) were generated by PCR using the following primers:

5’-GCCTCAGAACTGAACAAAGA and 5’-GCTCATAATGTTGTTCCACCT for LINE1;

5’-CACGCCTGTAATCCCAGC and 5’-AGACAGGGTTTCTCTGTA for B1;

5’-GCGAGAAAACTGAAAATCAC and 5’-TCAAGTCGTCAAGTGGATG for MSR.

Probes were dissolved in hybridization mixture (50% formamide, 10% dextran sulfate, 1xSSC) at a concentration of 10–20 ng/μl and hybridized to sections of mouse retina for 2 days. Post-hybridization washes included 2xSSC at +37°C (3x 30 min) and 0.1xSSC at +61°C (10 min). Sections were counterstained with DAPI and mounted as after immunostaining (see above).

Microscopy and image analysis

Request a detailed protocol

Single optical sections or stacks of optical sections were collected using either Zeiss LSM 700 or Leica TCS SP5 confocal microscopes equipped with Plan Apo 63x/1.4 NA oil immersion objective and lasers with excitation lines 405, 488, and 561 nm. Dedicated plugins in the ImageJ (Schindelin et al., 2012) program were used to compensate for axial chromatic shift between fluorochromes in confocal stacks, to create RGB stacks/images, and to arrange optical sections into galleries (Ronneberger et al., 2008)

To estimate the proportion of rods expressing LBR in retinas from TG-LBR mice, four stained cryosections from two homozygous mice were imaged. Not less than 12 image fields with pixel size of 100 nm were collected through each section. Scoring of LBR-positive and negative rods was performed in ImageJ using Cell Counter plugin. Number of chromocenters in TG-LBR rods and P14 WT pups was estimated in confocal stacks through retinas after FISH with major satellite repeat and lamin B immunostaining. Scoring of chromocenters in 210 and 65 nuclei of transgenic and P14 rods, respectively, was performed manually using ImageJ.

Electron microscopy

Request a detailed protocol

For electron microscopy, eyes of adult WT and TG-LBR mice were fixed by cardiac perfusion with a mixture of 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1M cacodilate buffer for 5 min. After eye enucleation, the eye-balls were further fixed in the same fixative for 1 hr and then postfixed with OsO4 in cacodilate buffer for 1.5 hr. Ultra-thin sections were stained with uranyl acetate and Reynolds lead citrate. Images were recorded with a megaview III camera (SIS) attached to a Philips EM 208 transmission electron microscope (FEI) operated at 70 keV.

Flow cytometry

Request a detailed protocol

FACS scattering analysis of retinal cells and sorts according to light scattering profiles were performed based on previously published methodologies (Feodorova et al., 2015). The Flow cytometric analysis was performed using FACS Aria Fusion (BD Biosciences) equipped with 488 nm laser and a 70 µm nozzle. For performance tracking and to ensure stability of the scattering signal, calibration beads from BD biosciences (BD FACSDiva CS and T Research Beads, 655050) were used. The scattering signal height vs width was then used to gate for singlet cell populations. Cell aggregates and debris were excluded for the data analysis. The papain dissociation system kit from Worthington Biochemical Corporation was used to digest the tissues. All the solutions for the digestion were prepared according to the manufacturer’s recommendation. The retinae from adult mice were gently and quickly isolated from enucleated fresh unfixed mouse eyeballs. 250 µl Papain digestion solution contained in a 2 ml Eppendorf tube was equilibrated (for 15–20 min) in 5% CO2. Two to four retinae were transferred to the equilibrated solution and incubated in a thermomixer at 37°C at 700 rpm. The tubes were periodically checked by visual inspection to ensure proper dissolution of the tissue. After 15–20 mins of incubation, the digest was added to a tube containing 15 µl DNAase-EBSS. The mixture was then mechanically agitated by pipetting the solution up and down 10 times with a 1 ml pipette until no tissue pieces are visible. After mechanical dissociation add to the mixture 400 µl ovomucoid-EBSS (10% v/v) a papain inhibitor to arrest further chemical dissociation.

For retinae from young mice (P14 and P25) digestion times were adapted from 20 to 10 min to compensate for a faster dissociation. For brain cortical cells, above described digestion was preceded by vibratome slicing of freshly obtained mouse brains.

Neuro-2a cells were trypsinized (0.05% Trypsin-EDTA, Thermo Fisher Scientific) and washed once with cold PBS. The relevant details of the cell lines used have been included in the Key resource table. The Neuro-2a (Mouse neuroblastoma) cells were obtained as frozen vials supplied and quality controlled by DSMZ, Germany (ACC-148; RRID: CVCL_0470). Cells were purchased in the year 2010, but were not long term cultured since then. Instead they had been stored in liquid nitrogen for the predominant amount of time and were only thawed days before the sorting experiment. In general cell culture facilities are regularly checked for bacterial infections, including mycoplasma infections, and there is no evidence suggesting an infection of the cells. Notably, cells were not recultured after FACS characterisation. In all cases the samples were filtered into Falcon round bottom polystyrene FACS tubes (Corning Inc, USA) using a 40 µm mesh cell strainer (FALCON, Corning Inc, USA) prior to FACS analysis.

For the calculation of the volume specific scattering, the side scattering area was normalized by volume of nuclei by taking the forward scattering area as a measure for size. Volume-specific scattering thus refers to the light scattering normalized by the amount of material, used to compare the light scattering by a material of given volume/mass but different size distribution.

Mie models of nuclei

Request a detailed protocol

The scattering intensity calculations for the multi-chromocenter-nuclei depicted in Figure 1H, were performed using Mie scattering models of spheres in a refractive index contrast of 2% (Kreysing et al., 2010), the reported contrast of refraction between heterochromatin and euchromatin. Mie calculations were implemented via a MATLAB script (Mätzler, 2002) that can be downloaded at the following link - https://omlc.org/software/mie/. The relevant parameters used were m_euchromatin/medium = 1.02, m_heterochromatin = 1.04 (Kreysing et al., 2010) which are refractive index of the euchromatin/medium and heterochromatin/particles respectively. The wavelength used was 500 nm and volume fraction vf = 0.3351. The diameter of particles used were in the range 0.9–4 µm. Relative scattering efficiencies for packed scatterers represented in Figure 3—figure supplement 2 (E) were calculated based on dependent scattering models (Twersky, 1978).

Micro projection setup

Request a detailed protocol

Ex-vivo retinal transmission measurements were carried out using a dedicated custom built, automated optical setup. This micro-projection setup (Figure 3—figure supplement 1 (A)) consisted of two distinct optical paths, one containing projection optics (functioned akin to the optics of the eye) that relayed images displayed by the projector LCD on to the image plane of the projection objective lens, and a second that recorded the retina transmitted images. The light source used (ML505L3, Thorlabs) had a spectrum close to that of the sensitivity of the rods ~ 510 nm. The objective lens (NA = 0.45, NPL Fluotar, Leitz, Germany) was chosen to closely match the f# number of the mouse eye (f#~1; Geng et al., 2011), with an added option to narrow the incident angular spectrum for absolute transmission measurements. The projected image on the retina was then collected via an imaging/efflux objective (Olympus U PlanApo 20x 0.75/inf corr) and recorded on an Andor Zyla-5.5 sCMOS camera.

Calculation of MTF

Request a detailed protocol

To quantify Modulation Transfer Functions (MTF) spatially extended sinusoidal stripe patterns of different spatial frequency were micro-projected using a custom optical setup (Figure 3—figure supplement 1A) and transmitted images were recorded. The MTF was calculated as the ratio of the contrast in the transmitted image and the projected image. With a customized digital projector setup, the implementation of the sinusoidal stripe projection became a straightforward analysis of the optical property for wide retinal regions (~625 µm x ~ 750 µm) (Figures. Figure 3—figure supplement 1A,C), Figure 3—figure supplement 2A-D). The projection of spatially extended images that display many periods is however also key to capture image veil, since scattering at large angles may reduce contrast not locally (from one peak into the neighboring minimum) but across multiple stripe periods of the test image. In industries MTF is predominantly used to assess various optical systems such as lens, cameras, displays etc. (Williams and Becklund, 1989). An advantage of the MTF approach over any spatial domain approach (i.e. PSF analysis) is that overall performance of a system with optical components in series can be conveniently described as a product of the MTFs of the individual components (Boreman, 2001). In particular, MTF describes the frequency domain performance of an optical system as a ratio of the contrasts in the output image to the input object as given below,

Contrast= Imax- IminImax+ Imin - I0,max- I0,minI0,max+ I0,min; MTFξ= Contrastimage(ξ)Contrastobject(ξ)

where, I is the image intensity and ξ is the spatial frequency (number of stripes per unit distance).

Practically, the raw images of the stripe patterns were first flat field corrected using Fiji (Schindelin et al., 2012) to ensure no global changes in contrasts affected further calculations. Each image was then processed using built in functions in MATLAB by taking an average along the direction orthogonal to the contrast modulation. The resulting one-dimensional sinusoidal intensity pattern was fit to a sine wave to extract Imax and Imin. (Figure 3—figure supplement 2A-D. Subsequently, the MTF was calculated according to the above formula. The MTF of the retina was then obtained by normalizing the measured MTF against the MTF of the optical setup alone. The differential readout of the transmitted image through the inversion arrested TG-LBR retina allows an explicit understanding of the optical impact of the inner retina and the outer nuclear layer architecture in relation to other ocular constituents, such as the lens and the reported optical properties of mouse eye in in vivo studies (Geng et al., 2011; de la Cera et al., 2006; van Oterendorp et al., 2011). As for the photoreceptors outer segments, their impact is minimal as they act as waveguides as described in previous ex vivo studies (Ohzu et al., 1972).Such an effect is also verified by our simulations.

Calculation of Strehl ratio

Request a detailed protocol

Strehl ratio (SR) is a commonly used single number estimate of the optical performance of a system that can also be used to evaluate the optical performance of ocular components (Marsack et al., 2004; Thibos et al., 2003). The SR in the spatial domain is formally defined to be the ratio of the peak intensities of a PSF to that of a diffraction limited PSF (Strehl, 1895). In terms of the frequency domain analysis, one can more accurately calculate the SR by taking the volume under the Optical Transfer Function, albeit for systems with negligible phase transfer properties (as planar tissues), the volume under the MTF suffices to calculate the SR. This way the SR was calculated for each biologically independent sample by taking area under the frequency weighted MTF curve along the spatial frequency (Figure 3—figure supplement 2F).

PSF measurements

Request a detailed protocol

The point spread function (PSF) measurements were carried out using a 40 µm pinhole (P40H, Thorlabs) acting as a point light source, such that the demagnified point projected on the retina was of the size about 3 µm. Raw images were corrected for background by subtraction of a dark frame in FIJI. Resulting images were normalized with respect to the integral intensity in the field of view (~80 µm by ~ 80 µm), and the central region with an ROI of 40 µm by 40 µm was cropped, averaged and displayed in false color.

Diffuse transmission measurements

Request a detailed protocol

The measurements were carried out with the micro projection setup above such that a point source was projected through an effective NA of 0.05 on to the retina with a final size of ~30 µm diameter. The transmitted light was collected using an Olympus UPlanSApo 40x 1.25 NA silicone immersion objective lens and recorded on the camera. The fractional transmission of the samples was then calculated, after subtraction of a dark frame reference, based on the integrated intensity in the entire field of view compared to the intensity without the sample in place.

Hiding power

Request a detailed protocol

The angular-weighted integrated scattering intensity is also known as hiding power. Specifically, hiding power is represented as the product of the efficiency of scattering (Qsca) and the directional weightage component, otherwise known as the anisotropy factor (g) (Johnsen, 2012). The theoretical calculations based on Mie models presented in Figures(1H, 2 F-G) were done using a MATLAB script reported by Mätzler (2002) that can be downloaded at the following link - https://omlc.org/software/mie/. The relevant parameters used were m_euchromatin/medium = 1.02, m_heterochromatin = 1.04 (Kreysing et al., 2010) which are refractive index of the euchromatin/medium and heterochromatin/particles respectively. The wavelength used was 500 nm and volume fraction vf = 0.3351. The diameter of particles used were in the range 0.92–4 µm.

Optical reconstitution

Request a detailed protocol

Equal amounts (by weight) of silica beads of diameter 2 µm (MSS002) and 4.5 µm (MSS004a, lot obtained from Whitehouse Scientific) of RI 1.48 were dispersed in separate cuvettes containing glycerol-water mixture (RI = 1.43), the larger beads closely resembled the size of heterochromatin mass after chromocenter fusion, and the smaller beads corresponded to a ~ 12 chromocenter case nucleus (at a conserved total volume/mass). An edge was imaged through the two dispersions using a commercial mobile phone camera with a LED white lights acting as a light source.

Tissue preparation for optical characterization

Request a detailed protocol

Animals were sacrificed by cervical dislocation, and one eyeball immediately removed and opened in fresh environmentally oxygenated PBS. Next, the anterior of the eye, including the cornea and the lens, was fully removed. The retina was gently detached from the choroid, the optic nerve clipped and pulled out from the posterior cup. The retinal cup was placed on a 22 × 60 mm coverslip. Special attention was given to remove any residual vitreous humour sticking to the retina. While the retina remained floated in PBS radial incisions were made and the retinas were flattened on the coverslip by aspiring tiny amounts of the PBS. An appropriately flattened retina was mounted under a smaller coverslip in PBS. A 255 µm spacer was placed between the two coverslips under a stereomicroscope to prevent squeezing of the retina. Preparation were typically achieved in 2 min, and no retina was considered for measurement with a preparation time of more than 5 min. Optical measurements were done in an automated fashion with results in adult WT mice comparable to double pass experiments in vivo (Artal et al., 1998).

Behavioral assessment - Optomotor response

Request a detailed protocol

Visual behavioral response was assessed using a fully automated, monitor based optomotor drum setup obtained from Striatech (Striatech GmbH, Tübingen, Germany) and the experiments were conducted at the CRTD, TU Dresden, Germany. The optomotor setup was a closed box with four digital displays to simulate a rotating cylinder of stripe patterns. An opening above allowed the view of the animal via a camera. An independent computer-controlled software was used to track the mice on the platform. The presentation of the pattern and scoring of the movement tracking performance was done through a proprietary software program. Software details can be found in Benkner et al. (2013).

Age (5–6 months old) and gender matched mice from Wild type and TG-LBR (transgenic) mice were used for comparison of the behavior. The tests were performed under three different lighting conditions of 70 Lux (Photopic), 20 mLux and 2 mLux (Scotopic). Based on parameters reported previously in similar behavioral experiments (Umino et al., 2008), bar stripe patterns were presented at a speed of 15 deg/s for various spatial frequencies ranging from 0.01 to 0.44 cycles/deg in photopic condition. For the scotopic conditions, the stimulus was maintained at a constant temporal frequency of 0.73 Hz and spatial frequency in the range 0.01–0.3 cycles/deg. The temporal frequency here refers to the combination of spatial frequency (cyc/deg) and speed of movement in (deg/s), which gives an effective temporal frequency, namely the change of contrast at a given point on the screen, which was maintained constant at a particular temporal frequency (0.73 cyc/s or Hz). The contrast of the object displayed on the screen was in the range 100–2%. For determining the threshold contrast, display contrast was reduced in steps of 5% up to absolute contrasts of 10% and steps of 2% below 10% contrast. The sizes of stripes tested were (6, 8, 11, 22, 33, 44, 55, 66, 88, 95, 100, 106 cycles/360o). Each stimulus was presented for a total of 30–35 s in sets of 5 s each with a gap of 5 s between each presentation. The direction of rotation of the stripes was altered between left and right for each subsequent trial and chosen at random for trial-1. Once the threshold contrast was experimentally determined, for statistical analysis purpose, response values for contrasts above the threshold were designated to be ‘yes’ response and values below the threshold as ‘no’ response.

Nocturnal adaptation of behavioral testing setup

Request a detailed protocol

The ambient lighting of the test chamber for photopic condition was measured using a Lux meter (Testo 540). To reduce the lighting to scotopic levels appropriate ND filter sheets (ND 3.6, ND 4.8) were placed on the monitors. The ND filters were assembled by combining ND-1.2 filter sheets (e-color+ #299, Rosco Laboratories Inc). A custom made infrared light source was also installed to monitor and enable tracking of the animals on the platform under scotopic conditions. The Rho-/- mice were used as a control to ensure that the responses of the mice in the scotopic display test conditions purely relied on the rod visual pathway.

Modelling and simulation of light propagation

2 -photon mapping of ONL model

Request a detailed protocol

Wild-type C57BL/6J mice were sacrificed by cervical dislocation. Immediately, eyes were enucleated and then cut in half around the equator, discarding all components of the eye but the posterior eye-cup. Retina was peeled off from the eye-cup. The retinal isolation was performed in paraformaldehyde (PFA) 4% in phosphate-buffered saline (PBS) solution and then left suspended to complete fixation for 20 min. The sample was then transferred to a PBS solution at 4°C after fixation. The fixed sample was deposited inside a TEFLON container and embedded in low melting agarose. The agarose embedded sample was sectioned adapting the method described previously (Clérin et al., 2014). The resulting retinal cross sections were stained with Hoechst 33342 and then wet mounted in a 50% glycerol/PBS solution using a No. 1 cover slip (Corning Inc, USA). Imaging was performed with confocal microscope (LSM 780, Zeiss Germany) in two-photon mode, equipped with a tunable pulsed infrared laser (Chameleon Vision II, Coherent, US) (excitation wavelength 730 nm, Objective: Zeiss LCI Plan-Neofluar 63x/1.3). The acquired intensity image was of size 190 × 190 × 82 µm with pixel-sizes of 83 × 83 × 250 nm in lateral dimensions and in depth.

Image processing and segmentation of ONL model

Request a detailed protocol

To create a realistic refractive index map of packed nuclei within the ONL, the intensity image was first segmented into nuclei regions. To that end, a random forest classifier was trained via Fiji (Arganda-Carreras et al., 2017; Schindelin et al., 2012) to densely classify each pixel into background or foreground (nuclei). A watershed segmentation (van der Walt et al., 2014) was then applied on the probability map with manually generated seed points, resulting in 1758 individual nuclei instances. The refractive indices for these phases have been carefully estimated previously in single cell studies (Błaszczak et al., 2014; Kreysing et al., 2010; Schürmann et al., 2017; Solovei et al., 2009). Finally, the refractive index distribution inside each nuclei region was generated according to the two different models:

Inverted: Consisting of two refractive phases with n1 = 1.357 and n2 = 1.382, corresponding to euchromatin and heterochromatin, respectively. Each nuclei mask was split into shell and core regions of equal volume (via morphological shrinking operations on each mask), which were then assigned the respective refractive indices (n1 for shell, n2 for core).

Chromocenter: Here, 8–12 chromocenters were randomly picked within the nuclei mask and assigned points close to either the nuclei border or those chromocenters to the high refractive index phase (n2) until its joint volume reached half the full nuclei volume. The other points were then assigned the less dense refractive index n1.

The resulting refractive index distribution was then blurred in both cases with a small gaussian (sigma = 2 px) to create a smooth distribution. For both models it was furthermore ensured that both refractive phases occupied the same total volume.

Light propagation simulations and scattering

Request a detailed protocol

Light propagation through both ONL models was simulated with GPU-accelerated scalar beam propagation method (Weigert et al., 2018). A computational simulation grid of size (1024,1024,645) with pixel-size 83 nm was used and the propagation of a plane wave (wavelength 500 nm) through the different ONL refractive index distributions was simulated. The surrounding was assumed to have a refractive index of n0 = 1.33. The integrated side scattering cross sections were calculated from the angular spectrum as per previous reports (Weigert et al., 2018).

Relative contributions to MTFs from ONL & outer segments

Request a detailed protocol

In order to assess relative contribution of ONL and outer segments to the MTF of the retina dedicated simulations were carried out. These compared the scattering from chromocenters in an ONL (1 or 8 per nucleus), with outer segments that were simulated as cylinders that were 1.6 µm in diameter and 25 µm in length. The refractive index of the core of the outer segments was assumed to be 1.42 (Sidman, 1957). For the recorded simulations the scattering anisotropy factor and efficiency were extracted and converted into a frequency domain MTF data using appropriate theoretical models (Henyey and Greenstein, 1941; Wells, 1969). Results show that outer segments only have a negligible impact on the overall MTFs (Figure 3—figure supplement 2G), in agreement with previous experimental findings (Enoch, 1963) and models of the outer segment (Vohnsen, 2007; Vohnsen, 2014) acting as waveguides.

Measurements of ocular parameters

Request a detailed protocol

Freshly excised mouse eye balls were imaged under a Olympus stereo microscope SZX16 equipped with a Q-imaging camera. The lens was also imaged under dark field conditions to better visualize the lens periphery. From the recorded images, the ocular parameters - lengths of the eye along two orthogonal axes were measured manually using FIJI. For the lens, mean feret diameter from the contour of the lens periphery was measured as an average estimate of the size of the lens.

Data and materials availability

Request a detailed protocol

Data and specifications of simulations supporting the findings of this study are available via https://dx.doi.org/10.17617/3.3a. The biobeam software is available publicly from: https://maweigert.github.io/biobeam.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
    Retinal noise and absolute threshold
    1. HB Barlow
    (1956)
    Journal of the Optical Society of America 46:634–639.
    https://doi.org/10.1364/JOSA.46.000634
  8. 8
  9. 9
  10. 10
  11. 11
    Modulation Transfer Function in Optical and Electro-Optical Systems
    1. GD Boreman
    (2001)
    SPIE Press.
  12. 12
  13. 13
  14. 14
  15. 15
    Visual Ecology
    1. TW Cronin
    2. S Johnsen
    3. NJ Marshall
    4. EJ Warrant
    (2014)
    Princeton University Press.
  16. 16
  17. 17
  18. 18
    The Retina
    1. JE Dowling
    (1987)
    Harvard University Press.
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
    Evaluation of the rhodopsin knockout mouse as a model of pure cone function
    1. GB Jaissle
    2. CA May
    3. J Reinhard
    4. K Kohler
    5. S Fauser
    6. E Lütjen-Drecoll
    7. E Zrenner
    8. MW Seeliger
    (2001)
    Investigative Ophthalmology & Visual Science 42:506–513.
  39. 39
    The Optics of Life
    1. S Johnsen
    (2012)
    Princeton University Press.
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
    MATLAB functions for Mie scattering and absorption (Bern)
    1. C Mätzler
    (2002)
    MATLAB functions for Mie scattering and absorption (Bern).
  47. 47
  48. 48
  49. 49
  50. 50
    From Photon to Neuron
    1. P Nelson
    (2017)
    Princeton University Press.
  51. 51
    Subterranean Rodents: News From Underground
    1. P Němec
    2. P Cveková
    3. H Burda
    4. O Benada
    5. L Peichl
    (2007)
    129–160, Visual systems and the role of vision in subterranean rodents: Diversity of retinal properties and visual system designs, Subterranean Rodents: News From Underground, Berlin, Heidelberg, Springer.
  52. 52
    The evolution of eyes and visually guided behaviour
    1. D-E Nilsson
    (2009)
    Philosophical Transactions of the Royal Society B: Biological Sciences 364:2833–2847.
    https://doi.org/10.1098/rstb.2009.0083
  53. 53
    Vision in dim light: highlights and challenges
    1. DC O'Carroll
    2. EJ Warrant
    (2017)
    Philosophical Transactions of the Royal Society B: Biological Sciences 372:20160062.
    https://doi.org/10.1098/rstb.2016.0062
  54. 54
  55. 55
    Diversity of mammalian photoreceptor properties: adaptations to habitat and lifestyle?
    1. L Peichl
    (2005)
    The Anatomical Record Part A: Discoveries in Molecular, Cellular, and Evolutionary Biology 287:1001–1012.
    https://doi.org/10.1002/ar.a.20262
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
    Fluorescence in Situ Hybridization
    1. I Solovei
    (2010)
    71–82, Fluorescence in situ Hybridization (FISH) on Tissue Cryosections, Fluorescence in Situ Hybridization, Totowa, Humana Press.
  68. 68
  69. 69
  70. 70
  71. 71
    Principles of Neural Design
    1. P Sterling
    2. S Laughlin
    (2015)
    MIT Press.
  72. 72
    Aplanatische und fehlerhafte abbildung im fernrohr
    1. K Strehl
    (1895)
    Zeitschrift Für Instrumentenkunde pp. 362–370.
  73. 73
    Metrics of Optical Quality of the Eye
    1. LN Thibos
    2. X Hong
    3. A Bradley
    4. RA Applegate
    (2003)
    Citeseer.
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
    Modeling photoreceptor mosaic imaging as backscattering of light from multilayered discs
    1. B Vohnsen
    (2014)
    Frontiers in Optics.
  85. 85
  86. 86
  87. 87
  88. 88
  89. 89
    Invertebrate Vision
    1. E Warrant
    2. D-E Nilsson
    (2006)
    Cambridge University Press.
  90. 90
  91. 91
  92. 92
    The Visual Neurosciences
    1. JS Werner
    2. LM Chalupa
    (2004)
    MIT Press.
  93. 93
    Introduction to the Optical Transfer Function
    1. CS Williams
    2. OA Becklund
    (1989)
    SPIE Press.
  94. 94
    The Visual Neurosciences
    1. RW Williams
    2. SA Moody
    (2004)
    65–78, Developmental and Genetic Control of Cell Number in the Retina, The Visual Neurosciences, MIT Press.

Decision letter

  1. Jeremy Nathans
    Reviewing Editor; Johns Hopkins University School of Medicine, United States
  2. Ronald L Calabrese
    Senior Editor; Emory University, United States
  3. Austin Roorda
    Reviewer; University of California, Berkeley, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Thank you for submitting your article "Rod nuclear architecture determines contrast transmission of the retina and behavioral sensitivity in mice" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Ronald Calabrese as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Austin Roorda (Reviewer #1).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As you will see, all of the reviewers were impressed with the significance and thoroughness of your work. All three reviewers also had specific and useful comments for improving the manuscript. Among them are three general suggestions to which we would like to draw your attention:

1) The Materials and methods section needs substantial editing for clarity and detailed methods of how forward/side scatter measures were performed. A large portion of the manuscript depends on this analysis and it is imperative to include the details.

2) Include a rigorous description of which tests are more or less behaviorally relevant for mouse vision. Showing data for ~1 cycle/deg may exaggerate the biological benefit, as does modelling a ~3 micron point spread function at the retina. The cat face may be marginally relevant based on visual angle on the retina, although it has a certain charm for the non-expert reader.

3) We encourage you to discuss other possible interpretations of the data, including competing hypotheses for the role of nuclear inversion.

Reviewer #1:

The authors have expanded on their previous work to understand the functional significance of the `inversion' of the nuclear architecture in nocturnal mammals, specifically, in this case, the mouse. The conclusion is that the inverted nuclear structure minimizes side scattering, and facilitates forward scattering with a resulting benefit that higher contrast images can reach the photoreceptors, thereby improving contrast sensitivity.

While readers might have inferred this conclusion based on earlier papers by members of this team, the authors do a very nice job of confirming it by comparing contrast transmission and behavioral performance between wild-type mice and mice with a genetic modification (TG-LBR) that prevents the 'inversion' from taking place (these TG-LBR mice appear to be otherwise unaffected visually). The wild-type mice show better contrast sensitivity of a magnitude of 18% and 27% for scotopic (nighttime) light levels compared to the TG-LBR mice. Interestingly, the wild-type and TG-LBR mice behave similarly under photopic (daylight) conditions, which the authors sensibly attribute to a reduction in noise caused by high-photon flux.

The improvement in performance of 18-27% is modest, but not negligible. The authors show that these modest improvements in contrast sensitivity serve to increase detection probabilities many-fold at dim, near-threshold levels. Therefore, the functional advantages of the nuclear inversion are convincing.

The Materials and methods section is very sloppy and needs to be revised. Also, there are some details missing (eg how forward and side-scatter is measured). Otherwise, the paper is well-written, the science is solid, and it sheds new light on the fascinating process of retinal development.

1) Abstract: `…retinal optical quality improves 2-fold…'. The authors overstate the optical benefit by choosing to report on one metric, which was the ratio of the areas of the MTF between the wild type and TG-LBR mice. This is an odd choice, because most of the spatial frequencies used for this metric are seemingly irrelevant for mouse vision. It would be more appropriate for the authors to provide in the abstract numbers for the behavioral improvements (18-27%)

2) Abstract: there should be no hyphen in `contrast-transmission'. (here and throughout the document)

3) Introduction paragraph one: what does less-dense mean? Are the authors referring to refractive index, optical density or actual density?

4) Results paragraph two and three and subsection “Flow cytometry”: Since it is so critical for this paper, it would be helpful if the authors could briefly describe how forward- and side-scattering are measured rather than just providing a citation.

5) Results paragraph three: The definition of side-scatter is vague. Here the authors define it as narrow scattering at 90 deg, but later (eg in subsection “Improved retinal contrast transmission”) they define it as scattering at angles > 30 degrees. Also the authors need to define the axis labels `Forward Scattering Area' and `Side-Scattering Area' in figure 1.

6) Figure 1C (inset). What does Volume-specific scattering mean? This needs to be defined.

7) Figure 1G: What do the rectangles in Figure 1G represent? Are they just sketched in or do the dimensions have an important meaning.

8) Subsection “Improved retinal contrast transmission” –, Figure S5: The authors state that they mimic the mouse eye by using an optical system with a similar f-number. But in the next paragraph, they state that the MTFs '…do not display a strict resolution limit.' These are conflicting statements. The use of limited aperture in the system means that it will have its own MTF. The authors should show the optical system MTF in their plots on Figure 3.

9) In the same subsection: The initials T.V. should be deleted.

10) What range of spatial frequencies were used for these computations?

11) Subsection “Improved retinal contrast transmission”, Figure 3:D2 and D3, subsection “PSF measurements”: The intensity of the PSF in the figure is lower for the TG-LBR mouse across the entire displayed range of -20 to 20 microns. But the authors state that the integrated intensity is the same between the two when the PSF is integrated over an 80 x 80 micron area. I am very skeptical that the integrated intensity under the two curves in Figure 3:D2 will become equal.

12) Results section final paragraph: "This suggest…."

13) Discussion paragraph one: The lack of `nuclear inversion' in diurnal animals is intriguing and the authors make a very sensible suggestion that the ONL is significantly thinner in diurnal animals. However, that statement should be backed up by proper citations or, better yet, a table or a plot comparing ONL between nocturnal and diurnal animals.

14) Materials and methods: In general, this section is sloppily written with numerous typos, combinations of present and past tense – often in the same sentence – and unclear writing. There are numerous typos. The authors flip between the abbreviation SR and strehl ratio.

15) Subsection “Calculation of MTF”. How do the authors propose to use this technique to measure optical impact of outer segments? Note that ex vivo preparations are vulnerable to optical artifacts, especially the delicate optical properties of the retina.

16) Behavioral assessment: What does the temporal frequency mean? Was the stimulus flickering? Or moving, or both? This entire section is very poorly written.

17) Subsection “Image processing and segmentation of ONL model”: Why was this smoothing necessary? Were the final results different when they were not smoothed? Does the smoothing generate refractive index profiles that are more realistic?

18) Subsection “Relative contributions to MTFs from ONL and outer segments”: Replace OS with 'outer segment'

Perhaps the Matlab script mentioned in the text should be shared.

Reviewer #2:

Paper Summary:

The authors build on a body of literature that has identified the interesting phenomenon of "nuclear inversion" in nocturnal mammals. In this report, the authors test the hypothesis that the re-organization of euchromatin and heterochromatin within the nucleus of rod photoreceptor cells could serve to benefit nocturnal mammals by reducing scatter in the outer nuclear layer which is thick in rod-dominant mammals such as mice. An impressive set of data is collected in the report. The authors interpret their findings as supportive of a role of improved contrast sensitivity due to nuclear inversion which purportedly reduces optical scatter, and thereby improves the contrast ratio of images that must project through all retinal layers before striking the outer segments of rods.

The paper is thoughtfully composed and was generally a pleasure to read. The data set is impressive and authors are congratulated on a wholesome battery of tests that span in vitro preparation, phantom simulations, mouse behavioral testing, histology with immunolabeling and transgenic animals that support the general hypothesis. The major criticism for the report, however questions the very raison d'etre of the manuscript; "just how beneficial is this nuclear inversion to mouse visual performance?" While nuclear inversion is indeed a strange behavior of outer retinal cells (especially rods), it is unclear whether this is an epiphenomenon of some other function important to rods, or whether, as the authors would suggest, truly provides visual contrast benefit to the animal. The authors provide some evidence in support of this idea, but there are several misleading conclusions drawn from figures (especially Figure 3) which overstate the contrast benefit to mice by using simulations that are not behaviorally relevant.

Problem 1: Authors show the MTF improvement of contrast transmission when projecting sinusoidal patterns directly onto the retina. The differences in retinal contrast appear impressive in Figure 3AB. When comparing pups or TG-LBR mice (which also do not have nuclear inversion) to the adult WT mice that do have nuclear inversion, contrast transmission appears to increase. However the range of spatial frequencies tested are not generally thought to be behaviorally relevant to mice. Reports by Histed MH, Carvalho LA, Maunsell JH. (J Neurophysiol 2012, and corroborated by a multitude of other studies) suggest that maximum spatial frequency cutoff for the mouse is near 0.5 cyc/deg. This represents the very lowest of the tested spectrum in Figure 3AB. By those measures, roughly 2/3 of the data is behaviorally irrelevant to the normal mouse. When considering data from 0-0.5 cycles/degree, the effect is visually modest in comparison. Reviewer requests revision of the figure to reflect the improvement range to that closer of what is relevant to mouse visual behavior.

Problem 2: Projection of 3 micrometer PSF into the mouse retina (Figure 3D) is behaviorally irrelevant. Based on the literature that the authors cite (and more), Geng et al., Schmucker and Schaeffel, 2004 and others such as Remtulla and Hallett, 1985, a 3 micron PSF is a highly unnatural stimulus for the mouse retina because of spherical aberration, longitudinal chromatic aberration, transverse chromatic aberration, a constantly growing mouse eye and an optically thick retina. Anything less than a single-wavelength stimulus therefore would be impossible to naturally project at a 3 micrometer spot, and thus it is unclear why the authors are using this highly unnatural stimulus to model the PSF spread in Figure 3D.

Problem 3: Authors attempt to simulate the behavioral benefit to the mouse by a friendly example of what the mouse would "see" in an approaching cat by showing a phantom of the cat face. This is a fun example, but again represents a scenario that is unlikely due to the visual acuity of the mouse (adult or otherwise). If assumed that behavioral spatial frequency is limited to ~0.5 cyc/deg, there is little chance the mouse would visualize the cat eyes at any distance represented by Figure 3. The reviewer calculates that interpupillary distance of a typical house cat (which is assumed to be a biotypical natural predator of the mouse? certainly not a tiger!) is 36 mm (following Hughes, 1972 Vision Research). If we are generous and round this to 4cm, the subtended angle on the mouse retina will surely not render the eyes of the cat in such a way that the authors illustrate. At 4 meters, subtended angle is nearly 0.57degrees. At 2 meters, subtended angle is 1.14 degrees. Again, this far exceeds the reported visual acuity of the mouse and therefore the example is inappropriate, behaviorally irrelevant and is misleading to the general scientific audience. There would be no visual benefit to the mouse in these conditions even if nuclear inversion were found to benefit contrast transmission. Request removal of this figure.

Problem 4: Problems 1-3 are further compounded that the generous spatial frequency cutoff for the mouse is 0.5 cycles/deg for photopic conditions (Prusky et al., 2000; Histed et al., 2012). Spatial frequency tuning for the WT mouse is considerably worse under scotopic conditions which is the regime that stands to benefit from rod nuclear inversion (authors report this is a rod-dominated effect and cones generally do not show such behavior). Umino, Solessio and Barlow, 2008, show scotopic contrast sensitivity is even lower than photopic in the mouse. Behaviorally tested cutoff is near 0.2 cyc/. When this is projected back on to the data from Figure 3AB,D1,D2,E and F) the behavioral benefit in Figure 3 seem to be baseless.

Despite these shortcomings, the manuscript has merit. Problems 1-4 are somewhat mitigated by compelling data in Figure 4 which do show a slight benefit in WT mice (with nuclear inversion) vs LBR mice which presumably do not. Scientific audience is left to trust that TG-LBR mice have otherwise normal ocular behavior with the exception of high-chromocenter rod nuclei. Further description of the phenotype would convince skeptics further (including eye size and anterior optical media clarity which could also account for the result in Figure 4).

In the discussion, the authors do not provide enough latitude that other epiphenomenon and bioselection-driven reasons for nuclear inversion are possible. The manuscript would be stronger if such openings for these possibilities are explored further. The reader is left with the feeling that the problem is solved, which it is not. Data is provided to support a hypothesis.

Figure 4F not described in Figure 4 caption.

Supplementary data is appropriate and impressive.

Reviewer #3:

The work by Subramanian et al. demonstrates that the inversion of nuclear architecture in the rod photoreceptors of mice improves visual function in dim light conditions. The paper is very well-written and easy to follow. The work is of the highest quality and the well-thought-out experiments nicely support the conclusions. It was a fun paper to read!

The strength of the paper comes from using a range of approaches, whole animals, tissue histology, dissociated cells, excised retinas, in vitro model systems, and theoretical calculations to demonstrate not only that the inversion improves visual function, but also provide a clear mechanistic explanation. Very convincing!

https://doi.org/10.7554/eLife.49542.sa1

Author response

As you will see, all of the reviewers were impressed with the significance and thoroughness of your work. All three reviewers also had specific and useful comments for improving the manuscript. Among them are three general suggestions to which we would like to draw your attention:

1) The Materials and methods section needs substantial editing for clarity and detailed methods of how forward/side scatter measures were performed. A large portion of the manuscript depends on this analysis and it is imperative to include the details.

The Materials and methods section has been updated to provide clear and detailed descriptions, and to address the reviewers’ comments. A summary of the changes in the Materials and methods contained in the revised manuscript is provided below.

1) The description of the FACS experiments has been expanded as follows:

“Flow cytometry

FACS scattering analysis of retinal cells and sorts according to light scattering profiles were performed based on previously published methodologies (Feodorova et al.,

2015). […] Volume-specific scattering thus refers to the light scattering normalized by the amount of material, used to compare the light scattering by a material of given volume/mass but different size distribution.”

2) The Matlab scripts used for the scattering calculations and the parameters used have been added.

“Mie models of nuclei

The scattering intensity calculations for the multichromocenter-nuclei depicted in Figure. 1H, were performed using Mie scattering models of spheres in a refractive index contrast of 2% (Kreysing et al., 2010), the reported contrast of refraction between heterochromatin and euchromatin. […] Relative scattering efficiencies for packed scatterers represented in Figure 3—figure supplement 2E were calculated based on dependent scattering models (Twersky, 1978).”

3) The grammatical and typographic errors have been corrected.

4) The description of the behavioural studies has been revised and extended to include relevant details and revised for clarity:

“Behavioral assessment – Optomotor response

Visual behavioral response was assessed using a fully automated, monitor based optomotor drum setup obtained from Striatech (Striatech GmbH, Tübingen, Germany) and the experiments were conducted at the CRTD, TU Dresden, Germany.[…] Once the threshold contrast was experimentally determined, for statistical analysis purpose, response values for contrasts above the threshold were designated to be “yes” response and values below the threshold as “no” response.”

5) Based on requests by the reviewers to investigate differences in the ocular parameters between WT and TG-LBR mice, the Materials and methods section also now contains a description of these experiments:

“Measurements of ocular parameters

Freshly excised mouse eye balls were imaged under a Olympus stereo microscope SZX16 equipped with a Q-imaging camera. The lens was also imaged under dark field conditions to better visualize the lens periphery. From the recorded images, the ocular parameters – lengths of the eye along two orthogonal axes were measured manually using FIJI. For the lens, mean feret diameter from the contour of the lens periphery was measured as an average estimate of the size of the lens.”

2) Include a rigorous description of which tests are more or less behaviorally relevant for mouse vision. Showing data for ~1 cycle/deg may exaggerate the biological benefit, as does modeling a ~3 micron point spread function at the retina. The cat face may be marginally relevant based on visual angle on the retina, although it has a certain charm for the non-expert reader.

Thank you for raising these issues. We fully agree that the assessment of the optical properties of the retina partially exceeds the visual acuity of mice, particularly the ability to resolve moving stripe patterns under low light conditions. As we explain below in some detail, however, we think there is good reason to show some of the full data sets (i.e. the MTF curves). Nevertheless, we also agree that we could have done a better job of explaining their relevance. Also, other optical data was insufficiently motivated in the text (e.g. the point sources), and in one case required an adjustment in measurement conditions (the cat approach).

There are a variety of visual behavioural tests to assess the performance of the visual system (Pinto and Enroth- Cugell, 2014). Of these, the use of the optomotor response has been shown to best determine the differences in visual capabilities of mice taking into account the complete process of visual perception in the brain (Gasparini et al., 2019; Huberman and Niell, 2011). The similarity between the stimulus and pattern of images to assess the retinal optical quality and the behavioural sensitivity thus enables direct determination of the role of retinal optics in visual perception. The representation of the results and the rationale behind the optical experiments) have now been appropriately explained in the revised manuscript (see below for dedicated explanations on MTF, PSF and cat images).

Complete MTF representation:

We fully agree with the view that mice do not show behavioural responses to stripe patterns (especially sinusoidal ones) with frequencies above 0.5 cycle/deg, particularly not under low light conditions. We also agree that our analysis of the optical quality of the retinal tissue included contrast transmission at much higher spatial frequencies, the reasons for which are clarified later in this section. To start with, we included an additional panel focussed on the retinal image transmission capabilities in the visually relevant range for mouse (Panel C). To the same end, we recognised that a logarithmic scale is not very well suited to access the magnitude of the differential contrast transmission at spatial frequencies that were used during behavioural experiments. To make the retinal transmission data more accessible within this spatial frequency regime, we have now included a figure panel for low spatial frequency (coarser stripes) data in a linear scale (manuscript Figure 3C). We believe this will help readers to better understand which region of the MTF curves overlap with behaviour behavioural response in mice.

From this, it can readily be seen that the contrast transmission is about 33% greater in WT compared to TG-LBR at frequencies where the WT mouse has the greatest advantage over the TG-LBR mouse in our behavioural tests. Under these conditions, there is also a 45% greater contrast transmission than observed in a WT pup. The revised manuscript now has additional text describing the contrast loss in images in the regime relevant for mouse behavioural tests:

“Notably, when we focus on the retinal transmission data within the spatial frequency regime that is relevant for mouse vision (Alam et al., 2015; Prusky et al., 2004; 2000), (Figure 3C) it can readily be seen that the contrast transmission is up to 33% greater in WT compared to TG-LBR at frequencies ~0.28 cycles/deg. Equally, the contrast transmission in this behaviourally relevant regime also increases up to 45% in adult compared to the pups.”

We hope this modification allows a better comparison of the MTF curves with the behavioural data and provide a convincing argument of the causal relationship between them. Nevertheless, we would also like to explain our rationale to show the complete data rather than just a subset. Owing to the broader interest of an interdisciplinary readers and the broader implications of the retinal tissue as an “optical model tissue”, the complete presentation of the findings seems justified to us. In particular:

a) An MTF curve provides information on the mechanism of the contrast reduction. In particular, our data shows that there is no frequency cut off, which speaks for a contrast reduction due to image veil caused by high-angle scattering. We think that showing the full data here will help readers to interpret the scattering characteristics of the rod nuclei and the outer nuclear layer described in the sections of the manuscript preceding the results on retinal image transmission.

b) The long tail of transmitted frequencies that we show further explains why, despite a rapid reduction of contrast with increasing frequencies, the photoreceptor mosaic can be reconstructed in an OCT measurement. Although not explained in our paper in great detail, we think that the display of the full MTF data addresses two very different aspects on the topic of retinal optics: (i) differential transmission at low and mid-range frequencies due to nuclear inversion, (ii) the ability to reconstruct rod mosaic images at much higher resolution than relevant for vision (depending on confocal detection of scattered photons at very high frequencies). OCT imaging is clearly not the topic of this paper, but in our experience, ophthalmologists welcome this side reference.

c) Furthermore, MTFs and Strehl ratios are some of the canonical metrics to quantify an optical system. Especially the quantification of Strehl ratios is one of the most common and comprehensive measures of an optical system performance. Thus, we think from an optical point of view it is imperative to include the measurable contrast transfer information from all the contributing spatial frequencies.

PSF measurements:

Regarding the transmission of a 3μm PSF, we fully agree that the improved transmission cannot be completely related to any functional consequences, and we clarify that such an inference was not intended in the manuscript. Furthermore, we agree with the reviewers that the mouse lens is not likely to be of sufficient quality to project such a sharp image that the differential peak intensities, which we read out as the Strehl ratio, would be of physiological relevance. At the same time, we feel, it is still worthwhile presenting this data because:

i) increased the peak intensity are predicted by the volume under the MTF curve, and we think it showing this consistency further increases some readers confidence in our data.

ii) As stated in the manuscript, this data is suitable to show that the resolution of retinal transmission is not strongly affected, thereby ruling out a different and alternative optical explanation for the occurrence of nuclear inversion.

iii) Lastly, knowing the aberrations of the lens, the additional explicit knowledge of the PSF of the retina might be very useful for improved theoretical predictions of image quality at the level of photoreceptor cells. In other words, it is still widely thought that the retina is completely transparent, and that image quality is mostly determined by the lens. We think showing the PSF of the retina alone will also help to refine this over-simplified picture.

To make sure there are no misinterpretations, we modified the relevant passage of the manuscript accordingly by adding:

“The measured resolution based on full-width half maximum (FWHM) of the PSF, however, remained virtually unchanged did not show any differences (4.32 ± 2.38 𝜇m, 3.75 ± 2.01 𝜇m, mean ± SD for WT and TG-LBR retinae respectively, Figure 3 E2), especially no changes that could physiologically impact acuity, which is known to be significantly lower in mouse. In addition to independently corroborating our MTF measurements, these results emphasize that nuclear inversion enhances contrast transmission through the retina but is unlikely to benefit acuity. From a mechanistic point of view, these measurements indicate that contrast is lost due to the generation of image veil from side scattering, which overcasts attenuated, but otherwise unchanged signals.”

To conclude, we would like to reiterate that we are aware that the increase in retinal optical quality is substantially greater than the integrated sensitivity (albeit not greater that the maximum sensitivity increments at critical conditions). However, for the reasons listed above, we would like to retain the physical description of the tissue as an optical medium as presented.

Cat image:

We thank the reviewers for bringing up the significance of this data. We included the data to bridge the gap between the artificial sinusoidal stripe patterns and real-life images. The motivation was simply to illustrate that also real-life images are transmitted at increased contrast. In search of an image for this, we thought a cat might be appropriately illustrative and we generally received very positive feedback when showing this to colleagues in the field. However, the reviewers are completely correct that the choice of distances was not very well backed up by existing knowledge about mouse visual capabilities, especially acuity cut-off under low light conditions.

We therefore repeated the analysis at frequencies below 0.5 cycles/degree (namely 0.1, 0.2 and 0.3 cycles/ degree, revised manuscript Figure 3G). According to the literature provided by the reviewers, this shifts the assay into a behaviourally relevant regime. New results show a comparable advantage at distance below 1 meter. The only difference in the new measurements is that intensities below distance of approximately 0.5m saturate for WT mice, which however does not take away from the original idea of improved predator images at the level of the photoreceptors. Although our study does not address whether a mouse first sees or smells or hears a cat (or about the abundance of domesticated cats during mouse eye evolution), it does fulfil the purpose of illustrating the effects with a real-life image. The experiments with the approach of a predator by projection of a cat face have now been updated for the visually relevant distances. The section describing these results have also been modified for clarity and relevance of these measurements:

“An advantage of improved retinal contrast transmission is suggested when following the motion of individual (nonaveraged) light stimuli that appear at considerably higher signal-to-noise ratios (Figure. 3F) at the outer segments level. A putative visual advantage to appropriately scaled real-life examples, such as images of an approaching cat micro-projected through a mouse retina is illustrated in Figure 3G. Nuclear inversion results in cat images becoming visible considerably earlier compared to mice that lack nuclear inversion (0.70 vs 0.45 meters, at a given arbitrary noise threshold). These results suggest that nuclear inversion may offer enhanced visual competence that originates from improved contrast preservation in retinal images. More objective and established methodologies to test the impact of nuclear inversion for actual behaviour is addressed in the next section.”

3) We encourage you to discuss other possible interpretations of the data, including competing hypotheses for the role of nuclear inversion.

The reviewers are right, of course, that we cannot exclude other functions of nuclear inversion. Mammalian, eyes are amongst the most complex organs we know, and definitely the retina and nuclear architecture therein are multi-constrained systems. This raises various relevant questions:

i) Does the improved contrast sensitivity stem from optical change downstream of nuclear inversion, or could there be other reason for the differential sensitivity?

ii) Why aren’t photoreceptor nuclei of diurnal animals inverted?

iii) Despite the direct link between enhanced optical properties and improved sensitivity, does nuclear inversion also have other functions?

We have touched upon question (i) already in the existing manuscript with a rescue experiment in which we demonstrated sufficiency of increased retina contrast transmission to explain improved contrast sensitivity. We would like to apologise that the relevant figure cation that described the results of the rescue experiments were inadvertently cropped in the manuscript that we submitted for review. We think that this missing caption might have prevented reviewers from engaging with these rescue experiments. Furthermore, we have performed additional control experiments as suggested by the reviewers and establish that the ocular anatomy of the WT and TG-LBR mice are not significantly different and do not contribute to any possible changes in the behavioural sensitivities. We have provided this data in supplementary figures. We now include the 3 missing lines of figure caption (manuscript Figure 4F), and we also expanded the paragraph explaining the rationale behind these rescue experiments (underlined explanations were added):

“Finally, we asked whether reduced visual sensitivity of mice lacking the inverted nuclear architecture can be sufficiently explained by inferior contrast transmission of the retina. […] Thus, improved retinal contrast transmission is indeed sufficient to explain increased contrast sensitivity in mice.”

Question (ii) had already been addressed by explaining that there would be reduced benefit, for which we provide a detailed discussion and illustration in Figure 4—figure supplement 1. Some additional arguments have been presented in this regard in the revised manuscript:

“Besides a reduced need for inverted photoceptor nuclei in diurnal mammals, reduced efficiency of canonical DNA repair mechanisms (Frohns et al., 2014) in highly condensed chromocenters, could mean a significant disadvantage in the diurnal retina and susceptibility to stress and degeneration (Boudard et al., 2011; Dyer, 2016), could also mean a significantly higher cost for inverted nuclei in diurnal species, as their retinae are intrinsically strongly exposed to high-energy, ultra-violet photons. In conclusion, we showed that rod nuclear inversion is necessary and sufficient to explain optically enhanced contrast sensitivity in mice. Our work thereby adds functional significance to a prominent exception of nuclear organization and establishes retinal contrast transmission as a new determinant of mammalian fitness.”

Question (iii) is a very valid concern that we have now addressed. For this, we have expanded the Discussion with the following additional paragraph to discuss other interpretations and competing hypotheses.

“As mammalian eyes are evolutionarily multi-constrained systems, one could ask if nuclear inversion might also serve other functions beyond the improved contrast sensitivity that we have showed. […] While these additional functions of nuclear inversion currently remain speculations, it is worth reflecting about the relevance of the visual benefits demonstrated here for enhancing animal vision in general.”

Reviewer #1:

The authors have expanded on their previous work to understand the functional significance of the `inversion' of the nuclear architecture in nocturnal mammals, specifically, in this case, the mouse. The conclusion is that the inverted nuclear structure minimizes side scattering, and facilitates forward scattering with a resulting benefit that higher contrast images can reach the photoreceptors, thereby improving contrast sensitivity.

While readers might have inferred this conclusion based on earlier papers by members of this team, the authors do a very nice job of confirming it by comparing contrast transmission and behavioral performance between wild-type mice and mice with a genetic modification (TG-LBR) that prevents the 'inversion' from taking place (these TG-LBR mice appear to be otherwise unaffected visually). The wild-type mice show better contrast sensitivity of a magnitude of 18% and 27% for scotopic (nighttime) light levels compared to the TG-LBR mice. Interestingly, the wild-type and TG-LBR mice behave similarly under photopic (daylight) conditions, which the authors sensibly attribute to a reduction in noise caused by high-photon flux.

The improvement in performance of 18-27% is modest, but not negligible. The authors show that these modest improvements in contrast sensitivity serve to increase detection probabilities many-fold at dim, near-threshold levels. Therefore, the functional advantages of the nuclear inversion are convincing.

The Materials and methods section is very sloppy and needs to be revised. Also, there are some details missing (eg how forward and side-scatter is measured). Otherwise, the paper is well-written, the science is solid, and it sheds new light on the fascinating process of retinal development.

We apologise for numerous errors and partially lacking description of the methods. The Materials and methods section has now been updated to provide clear and detailed descriptions, and to address the reviewers’ concerns. A summary of the changes in the Materials and methods section is provided in response to the editors point 1.

1) Abstract: `…retinal optical quality improves 2-fold…'. The authors overstate the optical benefit by choosing to report on one metric, which was the ratio of the areas of the MTF between the wild type and TG-LBR mice. This is an odd choice, because most of the spatial frequencies used for this metric are seemingly irrelevant for mouse vision. It would be more appropriate for the authors to provide in the abstract numbers for the behavioral improvements (18-27%)

We agree that the chosen metric of optical improvements is not straight forward connected to the reported improvement of contrast sensitivity. However, we think that these findings on the overall improvement in the optical quality of retinal tissue are important and meaningful to the audience interested in tissue optics and microscopy. The similarity of the methods and test patterns used in the assessment of the tissue transmission and behaviour may indeed help draw correspondences between the improvements based on contrast and measures of image detail. The more general question of stimulus detectability (unlike resolvability) may take into account the retina’s ability to transmit higher spatial frequencies. Moreover, 18-27% integrated sensitivity improvement equally refers to one among many possible metrics; the response rates, for instance, in visually challenging regimes is shown to be enhanced by up to 10x (manuscript Figure 4E). Keeping these points in mind, we feel that reporting the numbers related to the independent observation of an optical quality improvement in a living heterogeneous tissue such as the retina, based on the most comprehensive metric (Strehl ratio), seems justified in the Abstract. To avoid any misleading implications for mouse vision, the numbers for behavioural improvements have now been added to the revised Abstract:

“Rod photoreceptors of nocturnal mammals display a striking inversion of nuclear architecture, which has been proposed as an evolutionary adaptation to dark environments. […] Our findings therefore add functional significance to a prominent exception of nuclear organization and establish retinal contrast transmission as a decisive determinant of mammalian visual perception.”

2) Abstract: there should be no hyphen in `contrast-transmission'. (here and throughout the document)

Thank you. This has been corrected in the Abstract. Corrections have been appropriately included in the revised manuscript.

3) Introduction paragraph one: what does less-dense mean? Are the authors referring to refractive index, optical density or actual density?

This is a good point. The density here refers to the actual mass density of chromatin, that are known also for other cell types (Imai et al., 2017). It is these increased mass densities that cause the RI differences earlier reported by the lead authors. The corresponding sentence in the revised manuscript has been modified to avoid any confusion.

“Interestingly, rod nuclei are inverted in nocturnal mammals (Błaszczak et al., 2014; Kreysing et al., 2010; Solovei et al., 2009; 2013), such that heterochromatin is detached from the nuclear envelope and found in the nuclear center, whereas euchromatin that has lower mass density (Imai et al., 2017) is re-located to the nuclear

periphery.”

Triggered by the reviewer’s remark, we noticed that we forgot to point at a related finding, which is that the heterochromatin core is of such high density that it even excludes free GFP molecules (Figure 1—figure supplement 1) The relevant description is now included in the Results section:

“This suggests that retinal cells are indeed optically specialized, as they scatter less light for a given size. This unique property for the rod cells could stem from the unusually dense packing of the heterochromatin in the centre of their nuclei, which notably even excludes free GFP molecules (Figure1—figure supplement 1B).”

4) Results paragraph two and three and subsection “Flow cytometry”: Since it is so critical for this paper, it would be helpful if the authors could briefly describe how forward- and side-scattering are measured rather than just providing a citation.

The details of the FACS measurements have been elaborated as indicated below and an extended description of the protocol has been added to the Materials and methods. Kindly also refer to our response to editors point 1.

“We then asked whether retinal cell somata are optically specialized with distinct light-scattering properties. […] Using forward scattering as a measure of cell size indicates that side scattering normalized by volume (volume-specific light scattering) is also noticeably lower in retinal cells (Figure 1C, inset).”

and

“In stark contrast however, sideward scattering, with a strong potential to diminish image contrast, was significantly reduced in adult retinal nuclei compared to the intermediate developmental stage (Figure 1F). Quantitative analysis of sorted nuclei from P25 retinae further revealed a monotonic relationship between chromocenter number and sideward scattering signal (Figure 1G). In particular, those nuclei with the lowest number of chromocenters were found to scatter the least amount of light.”

5) Results paragraph three: The definition of side-scatter is vague. Here the authors define it as narrow scattering at 90 deg, but later (eg in subsection “Improved retinal contrast transmission”) they define it as scattering at angles > 30 degrees. Also the authors need to define the axis labels `Forward Scattering Area' and `Side-Scattering Area' in Figure 1.

Thank you for pointing out the ambiguity. The terms forward and sideward scattering and the angles associated with their measurements are presented as defined in the commercial FACS system. Their definitions are now included in the text. Please refer to the response to reviewer1 comment 4 above. To avoid confusion, the side scattering term has been now changed to “large-angle scattering” in the section describing the results from simulations. We used simulations for extrapolation of angular light scattering that could not be captured in FACS measurements.

“These simulations suggest that especially the large-angle scattering (cumulative scattering signal at angles >30 deg) monotonically decreases when 10 chromocenters successfully fuse into one (Figure 2D2, 2E).”

6) Figure 1C (inset). What does Volume-specific scattering mean? This needs to be defined.

Volume-specific light scattering is the amount of light scattering per unit volume. The definition of the term has now been provided in the supplementary text.

“For the calculation of the volume specific scattering, the side scattering area was normalized by volume of nuclei by taking the forward scattering area as a measure for size. Volume-specific scattering thus refers to the light scattering normalized by the amount of material, used to compare the light scattering by a material of given volume/mass but different size distribution.”

7) Figure 1G: What do the rectangles in Figure 1G represent? Are they just sketched in or do the dimensions have an important meaning.

The description of Figure 1G is modified accordingly to include the following explanation.

“The rectangles represent sorting gates for microscopy analysis.”

8) Subsection “Improved retinal contrast transmission”, Figure S5: The authors state that they mimic the mouse eye by using an optical system with a similar f-number. But in the next paragraph, they state that the MTFs '…do not display a strict resolution limit.' These are conflicting statements.

We are grateful to the reviewer for pointing this out. As we clarify in the next point the MTF of the measurement system is there, and it has been taken in to account. Noteworthy however, compared to relevant frequencies transmitted by the tissue, the cut off frequency of the measurement system is extremely high, such that the measurements comfortably accommodate the MTF with its long tail. This statement was intended to convey the observation of a long-tailed MTF of the retina with non-zero residual contrast (a characteristic of MTF in scattering-dominated systems). We now have modified the statement for clarity:

“In contrast to many lens-based optical systems, retinal MTFs have a long tail with non-zero residual contrast despite an initial rapid loss of contrast (a characteristic of scattering-dominated optical systems). The monotonic decay of retina-transmitted contrast indicates scattering-induced veil rather than a frequency cut-off to be the cause of contrast loss…”

The use of limited aperture in the system means that it will have its own MTF. The authors should show the optical system MTF in their plots on Figure 3.

Again, this is a very good and insightful point. Although not very prominently addressed, this data was already present in the first submission (see manuscript Figure 3—figure supplement 1C). The MTF of the optical set up is now included in a separate panel in Figure 3—figure supplement 2H. The authors feel that inclusion of the MTF measurement of the optical set up (which has already been accounted for in the retinal MTF calculations, kindly see Materials and methods) in the main Figure 3 would be a technical detail that does not directly impact the interpretation of the biological message conveyed. The overall optical quality of the microscope set up is an order of magnitude better than that of the retinae.

9) In the same subsection: The initials T.V. should be deleted.

We apologise for the oversight. This was a citation that was not properly included. It has been fixed now.

10): What range of spatial frequencies were used for these computations?

The range of spatial frequencies used is 0-2 cycles/deg. The relevant lines in the revised manuscript have been modified to include this information:

“With regards to our MTF measurements, we that find the Strehl ratio (computed using measurements in the spatial frequency range of 0-2 cycles/deg) of a fully developed retina is increased 2.00 ± 0.15-fold compared to that of pups (P14) in which chromocenter fusion was not completed, and similarly 1.91 ± 0.14-fold (ratio of means ± SEM) improved compared to TG-LBR adult retinae (p = 3.4055e-08) in which chromocenter fusion was deliberately arrested (Figure. 3D).”

Please also refer to earlier section “Complete MTF representation" in response to the editor for an explanation that justifies the depiction and complete measurement of MTF for frequencies beyond the behaviourally relevant regime.

11) Subsection “Improved retinal contrast transmission”, Figure 3:D2 and D3, subsection “PSF measurements”: The intensity of the PSF in the figure is lower for the TG-LBR mouse across the entire displayed range of -20 to 20 microns. But the authors state that the integrated intensity is the same between the two when the PSF is integrated over an 80 x 80 micron area. I am very skeptical that the integrated intensity under the two curves in Figure 3:D2 will become equal.

Thank you for raising the concern. We agree that it feels counterintuitive that the intensities in the displayed ROI for the TG-LBR result in the same integral intensities in both cases. The ambiguity stemmed from the different sizes of ROI chosen for display and the ROI chosen for the normalization of the PSF curves to ensure same total integral intensities. Normalization was done over a window of 80x80 μm, whereas we show only the central 40x40 μm of the image. The description is therefore modified for clarity and reads as follows in the revised manuscript:

“PSF measurements

The point spread function (PSF) measurements were carried out using a 40 μm pinhole (P40H, Thorlabs) acting as a point light source, such that the demagnified point projected on the retina was of the size about 3 μm. Raw images were corrected for background by subtraction of a dark frame in FIJI. Resulting images were normalized with respect to the integral intensity in the field of view (~80 μm x ~80 μm), and the central region with an ROI of 40 μm by 40 μm was cropped, averaged and displayed in false color.”

Furthermore, please find in Author response image 1 a representation of the PSF in a larger field of view. When going beyond the 20 μm ROI of the PSF intensity images, the local intensities of the TG-LBR retinal PSF is equal or higher than the WT retinal PSF (inset in Author response image 1). Given the higher radial weight, this makes the integrated intensities converge to the same value. To further clarify this aspect, the ensquared energy of the PSFs have been provided for an ROI ~525 μm x ~525 μm, showcasing the total intensities in both retinal types asymptotes to the same value. Notably, at a center-to-edge distance of ~40 μm (ROI of ~80 μm x ~80 μm, which we choose to prevent the collection of too much dark current), the integrated intensities reach ~97% and ~93% of the final values in WT and TG-LBR retina, respectively. Thus, our earlier normalization in the smaller ROI does not affect the Strehl Ratio measures by more than 4%. Notably, if we would re-normalize the data to take into account the full field of view (but potentially including also more camera noise), the difference between the peaks would even be higher, as the TG-LBR peak would be 4% lower. For reference, the results of the encircled energy have been included in the supplementary information of the revised manuscript (Figure 3—figure supplement 2I).

Author response image 1
Point spread function of the retinae.

PSF intensities of retina transmitted point source images. Inset showing region where the intensities of the TG-LBR retina is higher than that of WT retina such the total normalized intensities in both cases is the same.

12) Results section final paragraph: "This suggest…."

Sorry, we corrected this.

13) Discussion paragraph one: The lack of `nuclear inversion' in diurnal animals is intriguing and the authors make a very sensible suggestion that the ONL is significantly thinner in diurnal animals. However, that statement should be backed up by proper citations or, better yet, a table or a plot comparing ONL between nocturnal and diurnal animals.

We agree that this information was poorly accessible from this section. References appeared for a similar statement in the introduction. They have been re-added here with the following additional references:

1) Solovei, I., Kreysing, M., Lanctôt, C., Kösem, S., Peichl, L., Cremer, T., Guck, J., and Joffe, B. (2009). Nuclear architecture of rod photoreceptor cells adapts to vision in mammalian evolution. Cell 137,356–368.

2) Werner, J.S., and Chalupa, L.M. (2004). The Visual Neurosciences (MIT Press).

3) Williams, R.W., LM, S.M., JS, W., 2003 (2003). Developmental and genetic control of cell number in the retina.

4) Sterling, P., and Laughlin, S. (2015). Principles of Neural Design (MIT Press).

“Wouldn’t improvements in retinal image contrast not also be beneficial for diurnal mammals? Firstly, the larger spacing of photoreceptor segments in the diurnal retina significantly reduces ONL thickness (Solovei et al., 2009, Sterling and Laughlin, 2015; Werner and Chalupa, 2004; Williams and Moody, 2003) and thereby the risk of scattering induced veil and loss of image contrast.”

14) Materials and methods: In general, this section is sloppily written with numerous typos, combinations of present and past tense – often in the same sentence – and unclear writing. There are numerous typos. The authors flip between the abbreviation SR and strehl ratio.

We apologise for numerous typographical errors in this section and thank the reviewer for pointing them out. These have been rectified. A summary of the changes in the Materials and methods section has been provided in response to editor point 1.

15) Subsection “Calculation of MTF”. How do the authors propose to use this technique to measure optical impact of outer segments? Note that ex vivo preparations are vulnerable to optical artifacts, especially the delicate optical properties of the retina.

This is a good point. We think we were not completely clear in our descriptions here. What we meant to say in the relevant part of the Materials and methods section is that genetically changing the optics of the ONL enables a better understanding where scattering occurs in the retina. This can also be compared to the performance of the lens, as reported by other studies. For optical contributions of the outer segments we mostly referred to the literature, and we would like to clarify that it is not straight forward to experimentally confirm the contribution of the segment layers by our method, and we don’t see highest need for it as our experiments include the direct comparison of 2 retina phenotypes with outer segments present.

To further investigate the relative contribution of inner retina vs segment layers, simulations were used, but we only really touched on these aspects. We present in Author response image 2 more details of our simulations. The literature suggests that the outer segments have limited impact on MTF (they have wave guiding properties, (Vohnsen, 2007; 2014)). Gaining optical access to pure OS structures is admittedly is not trivial. Therefore, we adopted the simulation strategy to exclusively look at light wave propagation in the OS, which gave similar results to those of previous studies (Vohnsen, 2007; 2014) (Author response image 2).

Author response image 2
Modelling light propagation in mouse rod OS.

(Top panels) RI distribution in the rod OS assumed at a contrast of 6-8% and OS cylinder diameter assumed 1.6um and 25um long. (Bottom) Intensity distribution of a plane wave propagating longitudinally in the cylindrical outer segment structures indicating a guiding effect. Simulations performed using Biobeam (Weigert et al., 2018).

These simulations were then used to present a conservative estimate of the impact on the MTF according to previous models and show that it is low (Figure 3—figure supplement 2G). Results show that outer segments only have a negligible impact on the overall MTFs (Figure 3—figure supplement 2G), in agreement with previous experimental findings (Enoch, 1963) and models of the outer segment (Vohnsen, 2007; 2014) acting as waveguides.

Moreover, this issue applies to a young mouse, where outer segments are not fully developed. If the outer segments were to have an impact, with the growth of the outer segments, the loss of contrast would be only more rapid. However, we show an improvement in the optical properties of the retina as it develops. Furthermore, we also show the same response in the adult transgenic retinae in comparison to the retina of a mouse pup. This also implies that only ONL has an optical impact. Together, these findings gave us confidence that the fragile outer segments did not bias our results. To ensure that a misleading idea is not conveyed to the readers that the effect of the fragile outer segments could be assessed with the optical set up, the relevant lines in the Materials and methods have been modified as follows:

“The differential readout of the transmitted image through the inversion arrested TG-LBR retina allows an explicit understanding of the optical impact of the inner retina and the outer nuclear layer architecture in relation to other ocular constituents, such as the lens and the reported optical properties of mouse eye in in-vivo studies (Geng et al., 2011; la Cera et al., 2006; van Oterendorp et al., 2011). As for the photoreceptors outer segments, their impact is minimal as they act as waveguides as described in previous ex vivo studies (Ohzu et al., 1972). Such an effect is also verified by our simulations.”

16) Behavioral assessment: What does the temporal frequency mean? Was the stimulus flickering? Or moving, or both? This entire section is very poorly written.

We are grateful to the reviewer for raising this concern and we apologise for any ambiguity in the description. The stimulus was not flickering (meaning there were no temporal changes in light intensity) The description of the method has been updated for clarity:

“The temporal frequency here refers to the combination of spatial frequency (cyc/deg) and speed of movement in (deg/s), which gives an effective temporal frequency, namely the change of contrast at a given point on the screen which was maintained constant at a particular temporal frequency (0.73 cyc/s or Hz).”

Also, kindly refer to our response to the editors point 1above for a revised version of the behavioural experimental methods.

17) Subsection “Image processing and segmentation of ONL model”: Why was this smoothing necessary? Were the final results different when they were not smoothed? Does the smoothing generate refractive index profiles that are more realistic?

As the reviewer correctly assumed, the reason we applied some smoothing was to remove sharp transitions that we considered to be biologically less realistic. Also, sharp boundaries frequently may increase simulation artefacts. But we did not represent in detail how much this smoothing would change results. To address the effect of smoothing, we reran our simulations without any additional blurring step. Although the overall scattering cross section did slightly change, as expected (since the refractive index map changes), the relative reduction in side scattering for the inverted case compared to the conventional architecture was even found to be even slightly increased (65% vs 58%). Hence the smoothing leads to conservative results, and the gain of nuclear inversion might be even slightly stronger than inferred by our simulations. See Author response figure 3.

Author response image 3
Differential simulations of light propagation in the ONL, illustrating differences between the use of a refractive index maps with (A) and without (B) blurring of the refractive index distributions.

18) Subsection “Relative contributions to MTFs from ONL and outer segments”: Replace OS with 'outer segment'

Thank you. This has been fixed.

Perhaps the Matlab script mentioned in the text should be shared.

The MATLAB script that we used has been cited and can be obtained from the link within this publication. The Materials and methods section has also been updated with parameters used for calculations:

“Mie calculations were implemented via a modified MATLAB script (Mätzler, 2002) that can be downloaded at the following link – https://omlc.org/software/mie/. The relevant parameters used are m_euchromatin/medium=1.02, m_heterochromatin=1.04, which are refractive index of the euchromatin/medium and heterochromatin/particles, respectively. The wavelength used was 500nm and volume fraction vf = 0.3351. The diameter of particles used was in the range 0.92-2 𝜇m.”

Reviewer #2:

Paper Summary:

The authors build on a body of literature that has identified the interesting phenomenon of "nuclear inversion" in nocturnal mammals. In this report, the authors test the hypothesis that the re-organization of euchromatin and heterochromatin within the nucleus of rod photoreceptor cells could serve to benefit nocturnal mammals by reducing scatter in the outer nuclear layer which is thick in rod-dominant mammals such as mice. An impressive set of data is collected in the report. The authors interpret their findings as supportive of a role of improved contrast sensitivity due to nuclear inversion which purportedly reduces optical scatter, and thereby improves the contrast ratio of images that must project through all retinal layers before striking the outer segments of rods.

The paper is thoughtfully composed and was generally a pleasure to read. The data set is impressive and authors are congratulated on a wholesome battery of tests that span in vitro preparation, phantom simulations, mouse behavioral testing, histology with immunolabeling and transgenic animals that support the general hypothesis.

Thank you for summarising the research findings, acknowledging our motivation to study the optics of the retina and appreciating our interdisciplinary approach to address the long-standing hypothesis of the role of nuclear inversion in nocturnal mammals.

The major criticism for the report, however questions the very raison d'etre of the manuscript; "just how beneficial is this nuclear inversion to mouse visual performance?" While nuclear inversion is indeed a strange behavior of outer retinal cells (especially rods), it is unclear whether this is an epiphenomenon of some other function important to rods, or whether, as the authors would suggest, truly provides visual contrast benefit to the animal. The authors provide some evidence in support of this idea, but there are several misleading conclusions drawn from figures (especially Figure 3) which overstate the contrast benefit to mice by using simulations that are not behaviorally relevant.

We appreciate the points raised by the reviewer. We agree that particularly in manuscript Figure 3 we probed spatial frequencies beyond the acuity limit of mice. We agree with the reviewer that our reasoning for this was not very well communicated. Below, we provide a more detailed explanation in cases where we still think it is justified to extend the optical analyses to these high frequencies, or we have updated measurements and additional panels showing the results for visually relevant regime. We furthermore explain why we think that the main conclusion of the manuscript remains valid and apologize for a missing part of a figure (manuscript Figure 4F) caption that might have made the interpretation of the data difficult.

Problem 1: Authors show the MTF improvement of contrast transmission when projecting sinusoidal patterns directly onto the retina. The differences in retinal contrast appear impressive in Figure 3AB. When comparing pups or TG-LBR mice (which also do not have nuclear inversion) to the adult WT mice that do have nuclear inversion, contrast transmission appears to increase. However the range of spatial frequencies tested are not generally thought to be behaviorally relevant to mice. Reports by Histed MH, Carvalho LA, Maunsell JH. (J Neurophysiol 2012, and corroborated by a multitude of other studies) suggest that maximum spatial frequency cutoff for the mouse is near 0.5 cyc/deg. This represents the very lowest of the tested spectrum in Figure 3AB. By those measures, roughly 2/3 of the data is behaviorally irrelevant to the normal mouse. When considering data from 0-0.5 cycles/degree, the effect is visually modest in comparison. Reviewer requests revision of the figure to reflect the improvement range to that closer of what is relevant to mouse visual behavior.

Thank you for the remark. We are aware that some of the frequencies that we probed go beyond the behaviourally relevant regime, particularly under low light conditions. We further acknowledge that presentation of data in the original submission was not well suited (log-scale) to access the visual benefits in the behaviour relevant regime. However, data presented as an integrated optical quality measure in the relevant spatial frequencies (0 – 0.3 cycles per degree) had already been provided in manuscript Figure 3—figure supplement 2D. We appreciate the reviewer’s point about showing more behaviourally relevant data here, and we have therefore added a panel illustrating the contrast-transmission characteristics for behaviourally relevant spatial frequencies. Now, manuscript Figure 3(C) depicts the MTF transmission in the visually relevant spatial frequencies. Also, we made the data more accessible by using a linear, rather than logarithmic scale. This data clearly shows that in the relevant frequency regime, contrast is increased by 33-45% which is similar in magnitude to the behavioural benefit demonstrated in terms of contrast sensitivity improvements which is in the range 18-27%. The experiments with the approach of a predator by projection of a cat face have now been updated for the visually relevant distances. Kindly refer to the sections “Complete MTF representation” and “Cat Image for further detailed answer to these concerns.

Given the interdisciplinary nature of the work and the broad audience, we would like to keep the other MTF and psf figure representations intact. The results demonstrating a 2-fold increase in optical image transmission capabilities are important from an optical perspective (Strehl ratio is likely the best-established metric for optical performance) and furthermore reveal the magnitude of optical penetration that could be achieved in living tissues, if their optical properties could be controlled (our future research direction). As such, we believe that these data are of interest to the wider biological microscopy community and should be retained in the manuscript.

Problem 2: Projection of 3 micrometer PSF into the mouse retina (Figure 3D) is behaviorally irrelevant. Based on the literature that the authors cite (and more), Geng et al., Schmucker and Schaeffel 2004 and others such as Remtulla and Hallett (1985), a 3 micron PSF is a highly unnatural stimulus for the mouse retina because of spherical aberration, longitudinal chromatic aberration, transverse chromatic aberration, a constantly growing mouse eye and an optically thick retina. Anything less than a single-wavelength stimulus therefore would be impossible to naturally project at a 3 micrometer spot, and thus it is unclear why the authors are using this highly unnatural stimulus to model the PSF spread in Figure 3D.

We discussed this in detail in response to the editors’ questions (see section PSF measurements). In brief: The PSF measurements were used as an independent confirmation of the MTF measurements (especially their volumes). Additionally, we think it is interesting to characterize the point spread performance of the retina, a heterogeneous biological tissue, as an independent optical element of the eye. The PSF measurements also complement the frequency domain MTF measurements. At this juncture, we would like to clarify the superiority of the frequency domain MTF analysis to pin point the improvements imparted by the retina in visual behaviourally relevant spatial frequencies. To reiterate and clarify the quantification of the “point spread”, we agree that the point stimulus as projected by the ocular system of the mouse will indeed be aberrated. However, the PSF spread shown in the manuscript Figure 3 does not relate to the PSF of the ocular system of the mouse as projected on to the retina. The PSF measurement serves as an independent confirmation of the improved optical quality according to canonical measure ‘Strehl ratio’. We did not intend to convey that such a point source is an ecologically relevant stimuli, either with or without aberrations by the lens. These points have been made clear in the revised manuscript. For more details kindly refer to the section on PSF measurementsin response to the editors, which also justifies the presentation of individual non-averaged point source.

Problem 3: Authors attempt to simulate the behavioral benefit to the mouse by a friendly example of what the mouse would "see" in an approaching cat by showing a phantom of the cat face. This is a fun example, but again represents a scenario that is unlikely due to the visual acuity of the mouse (adult or otherwise). If assumed that behavioral spatial frequency is limited to ~0.5 cyc/deg, there is little chance the mouse would visualize the cat eyes at any distance represented by Figure 3. The reviewer calculates that interpupillary distance of a typical house cat (which is assumed to be a biotypical natural predator of the mouse? certainly not a tiger!) is 36 mm (following Hughes, 1972 Vision Research). If we are generous and round this to 4cm, the subtended angle on the mouse retina will surely not render the eyes of the cat in such a way that the authors illustrate. At 4 meters, subtended angle is nearly 0.57degrees. At 2 meters, subtended angle is 1.14 degrees. Again, this far exceeds the reported visual acuity of the mouse and therefore the example is inappropriate, behaviorally irrelevant and is misleading to the general scientific audience. There would be no visual benefit to the mouse in these conditions even if nuclear inversion were found to benefit contrast transmission. Request removal of this figure.

This is a valid point that was also raised by other reviewers. Thank you for the detailed calculations pointing out that the illustration may not be relevant to mouse visual behaviour. The experiments with the approach of a predator by projection of a cat face have now been updated to show visually relevant distances. In brief, visual acuity only determines the resolvability of features. This does not provide information about the presence or absence of a stimulus. We however have taken the reviewers’ valid concerns into account and repeated the experiments of the approach of a cat for visually relevant distances. New results show a comparable advantage at distance below 1 meter (70cm, 45cm and 25cm). The figure has been revised accordingly. Kindly refer to the section on “Cat Image”in response to the editor for more details.

Problem 4: Problems 1-3 are further compounded that the generous spatial frequency cutoff for the mouse is 0.5 cycles/deg for photopic conditions (Prusky et al., 2000; Histed et al. 2012). Spatial frequency tuning for the WT mouse is considerably worse under scotopic conditions which is the regime that stands to benefit from rod nuclear inversion (authors report this is a rod-dominated effect and cones generally do not show such behavior). Umino, Solessio and Barlow, 2008, show scotopic contrast sensitivity is even lower than photopic in the mouse. Behaviorally tested cutoff is near 0.2 cyc/. When this is projected back on to the data from Figure 3AB,D1,D2,E and F) the behavioral benefit in Figure 3 seem to be baseless.

Again, referring to sections Complete MTF representationt, he authors would like to point out that the differences in the contrast transmission by the retina at ~0.28cycles/degree is around 33-45% which is similar in magnitude to the behavioural benefit demonstrated in terms of contrast sensitivity improvements which is in the range 18-27%. We hope this substantiates the causal relationship between the retinal contrast transmission and the differences in the behavioural contrast sensitivity.

For the cat image, please refer to our answers above. With respect to the individual point sources, we think it is worth showing them because i) they complement intensity-averaged PSF measurement, and make the amount of fluctuations better accessible, ii) we would like to point out that detectability and resolvability are not the same. As in microscopy, the inability to resolve a point source like fluorophore, does not mean that it cannot be detected. iii) adding to this, it is a rather well-known phenomenon that intensity distribution may impact detectability (which explains stars can be better seen with glasses, although they remain unresolvable to the eye), iv) more recently the strong evidence for non-linear / thresholded detection has been demonstrated for the human eye (Tinsley et al., 2016).

Despite these shortcomings, the manuscript has merit. Problems 1-4 are somewhat mitigated by compelling data in Figure 4 which do show a slight benefit in WT mice (with nuclear inversion) vs LBR mice which presumably do not. Scientific audience is left to trust that TG-LBR mice have otherwise normal ocular behavior with the exception of high-chromocenter rod nuclei. Further description of the phenotype would convince skeptics further (including eye size and anterior optical media clarity which could also account for the result in Figure 4).

Thank you for these suggestions.

Before addressing the concerns of the reviewer, we wish to highlight and expand upon the already existing control experiments in the manuscript providing some evidence that the TG-LBR mice have otherwise normal visual behaviour except for their arrested nuclear inversion in rod photoreceptors.

a) There is no difference in diurnal vision (manuscript Figure 4B, Left).

b) Similar sensitivities in the lowest spatial frequencies (manuscript Figure 4B, Right, manuscript Figure 4D) where differential retinal-contrast transmission was also comparable (manuscript Figure 3B, 3C) indicate an uncompromised downstream neural visual pathway that is responsible to low spatial frequency roll off of the CSF.

c) Comparable absolute transmission in the retina (manuscript Figure E3) indicates the absence of any non-physiological state that could lead to protein aggregates inducing isotropic scattering and alter photon transmission.

Furthermore, we now provide experimental data comparing the ocular parameters of the WT and TG-LBR mice. The axial lengths of the eye and the diameter of the lens are not significantly different between the two phenotypes. This further confirms that the differences in contrast sensitivities close to the visual acuity of mice arises solely from the differences in retinal-contrast transmission. These measurements are now reported in Figure 4—figure supplement 2.

Nevertheless, the complexity of the visual pathway is too high to rule out all thinkable side effects. For instance, a 30% down-regulation of a sparse neuro-receptor that even deep sequencing would have little chance to pick up, could also be affecting visual behaviour. Indeed, the question remains whether the modest contrast loss in the retina is sufficient to explain the approx. 20% improvement in contrast sensitivity. In the existing manuscript, however, we already provide evidence that the Strehl ratio improvements in region of sensitivity (0 – 0.36 cycles / degree) is 24% ± 14%, which is similar in magnitude to the integrated sensitivity improvements seen in behaviour (18-27%). Specifically, at 0.28 cylces / degree where behavioural benefits are strongest, nuclear inversion improves the optical quality of the retina 33%-45%. But vision remains a highly non-linear process, and the reviewer is right that knowledge about differential contrast transmission alone might not be sufficient to explain the differential sensitivity.

How can we then show than that the retinal contrast loss is sufficient to account for the loss of sensitivity under low light conditions? The answer is a recueexperiment. (see also our response to point 3 raised by the editor). Perturbations often come with a risk of unspecific side effects. To relate an observed phenotype and a protein hypothesised to be involved in a particular cellular pathway, a candidate protein may be knocked down via RNAi interference. To show specificity, a recue should be performed by interfering with the pathway as close to the final phenotypic read out as possible. Our optical rescue experiment is analogous to a recue in an RNAi interference experiment. We did not stop by showing that sensitivity is increased, as the reviewer incorrectly stated. In addition to showing an enhancement in contrast sensitivity, we designed and performed such a rescue experiment to explicitly show that the contrast loss in the retina is sufficient to account for the loss of sensitivity.

As explained in the text (see paragraph beginning “Finally, we asked whether reduced visual sensitivity of mice lacking the inverted nuclear architecture can be sufficiently explained by inferior contrast transmission of the retina”), we designed the rescue experiment such that we would restore only the optical consequence of the TG-LBR phenotype. This is included in the last panel (F) of manuscript Figure 4 and Figure 4—figure supplement 2A-C, and although described in detail in the manuscript, unfortunately in the submitted version, the figure legend got cropped. We apologise for the inadvertent omission. We wish to thus elaborate on our motivation to perform these experiments.

First, the linearity of optical systems dictates that contrast losses are always relative, and independent of absolute contrasts. Because this is the case, we can describe the optical part of the vision systems as a serial system of contrast modulators, starting with the lens, and ending just before the light sensitive segments. This linear description implies that we can compensate the optical consequences of inversion arrest in rod nuclei. That is, we can pre-compensate the contrast such that photoreceptor outer segments of the TG-LBR mouse experience the same contrast level as in the WT retina. When we did this experiment, we found that under these conditions visual sensitivity was restored (manuscript Figure 4 F).

The beauty of this rescue furthermore is that it is highly specific, because very far downstream in the visual pathway, and as such does not bear significant risks of restoring the retina from LBR transgene side effects. Clearly, we cannot rule out other side effects, for which however we observed no evidence at level of retinal (Figure 2—figure supplement 1), ocular or lens anatomy (Figure 4—figure supplement 2), and non-limiting rod vision was normal (Figure 4B (Left)). Our rescue experiments clearly show that the extra loss of retinal contrast transmission downstream of LBR expression and inversion arrest is sufficient to explain the reduced sensitivity in rod inversion arrested mice. We hope that these explanations bring clarity to the relevant description of the rescue experiments in the manuscript that was unfortunately not assessed by the reviewers.

Given these arguments, we maintain our previous conclusion that a gain in contrast transmission is sufficient to explain the gain in sensitivity. To make sure this important point is properly conveyed, we extend the relevant part in the paper. At the same time, the reviewers are right that we cannot fully exclude that there are also other functions of nuclear inversion. (see also our response to point 3 raised by the editor).

In the discussion, the authors do not provide enough latitude that other epiphenomenon and bioselection-driven reasons for nuclear inversion are possible. The manuscript would be stronger if such openings for these possibilities are explored further. The reader is left with the feeling that the problem is solved, which it is not. Data is provided to support a hypothesis.

This is likely the only point where we disagree with any of the reviewers. We refute the view that nuclear inversion, and if we understand the reviewer correctly also the reduced retinal light scattering, could be an epiphenomenon that does not explain the differential sensitivity of mice. We believe that we provide strong, unambiguous evidence for a causal relationship with the occurrence of nuclear inversion. Most centrally, as we explained above, the rescue experiment shows that improved retinal contrast transmission is sufficient to explain improved contrast sensitivity. However, we agree that we could have been more through in explaining how we arrived at our conclusion and openly discuss alternative explanations. We have accordingly addressed open questions and potential additional benefits of rod photoreceptor nuclear inversion. Please refer to our response to the editor Point (3).

Figure 4F not described in Figure 4 caption.

Thank you. We apologize for the inadvertent omission of the figure caption. We are sorry that this information was not accessible and might have made it difficult to follow our sufficiency argument. This has been rectified. The corresponding caption now reads as follows:

“(F)Rescue experiment demonstrating sufficiency of improved retinal contrast transmission to explain improved sensitivity. Adjusting the level of contrast at the photoreceptor level (by pre-compensation of differential contrast loss) restores sensitivity of TG-LBR mice. N indicates number of individual trials of 10 animals together for each mouse type.”

https://doi.org/10.7554/eLife.49542.sa2

Article and author information

Author details

  1. Kaushikaram Subramanian

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. Center for Systems Biology Dresden, Dresden, Germany
    3. Cluster of Excellence, Physics of Life, Technische Universität Dresden, Dresden, Germany
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9050-5672
  2. Martin Weigert

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. Center for Systems Biology Dresden, Dresden, Germany
    3. Cluster of Excellence, Physics of Life, Technische Universität Dresden, Dresden, Germany
    Contribution
    Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
  3. Oliver Borsch

    Center for Regenerative Therapies Dresden, Technische Universität Dresden, Dresden, Germany
    Contribution
    Software, Validation, Investigation, Methodology
    Competing interests
    No competing interests declared
  4. Heike Petzold

    Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    Contribution
    Data curation, Validation, Investigation, Methodology
    Competing interests
    No competing interests declared
  5. Alfonso Garcia-Ulloa

    Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  6. Eugene W Myers

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. Center for Systems Biology Dresden, Dresden, Germany
    3. Cluster of Excellence, Physics of Life, Technische Universität Dresden, Dresden, Germany
    4. Department of Computer Science, Technische Universität Dresden, Dresden, Germany
    Contribution
    Software, Supervision, Funding acquisition, Validation, Methodology, Project administration
    Competing interests
    No competing interests declared
  7. Marius Ader

    Center for Regenerative Therapies Dresden, Technische Universität Dresden, Dresden, Germany
    Contribution
    Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Validation, Project administration
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9467-7677
  8. Irina Solovei

    Biozentrum, Ludwig Maximilians Universität, München, Germany
    Contribution
    Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Project administration
    Competing interests
    No competing interests declared
  9. Moritz Kreysing

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. Center for Systems Biology Dresden, Dresden, Germany
    3. Cluster of Excellence, Physics of Life, Technische Universität Dresden, Dresden, Germany
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Visualization, Methodology, Project administration
    For correspondence
    kreysing@mpi-cbg.de
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7432-3871

Funding

Max-Planck-Gesellschaft

  • Kaushikaram Subramanian
  • Martin Weigert
  • Heike Petzold
  • Alfonso Garcia-Ulloa
  • Eugene W Myers
  • Moritz Kreysing

Technische Universität Dresden

  • Oliver Borsch
  • Marius Ader

Deutsche Forschungsgemeinschaft (AD375/6-1)

  • Oliver Borsch
  • Marius Ader

Bundesministerium für Bildung und Forschung (031L0044)

  • Kaushikaram Subramanian
  • Eugene W Myers
  • Moritz Kreysing

Deutsche Forschungsgemeinschaft (SO1054/3)

  • Irina Solovei

Deutsche Forschungsgemeinschaft (FZT111)

  • Oliver Borsch
  • Marius Ader

Deutsche Forschungsgemeinschaft (EXC68)

  • Oliver Borsch
  • Marius Ader

Deutsche Forschungsgemeinschaft (SFB1064)

  • Irina Solovei

European Research Council (853619)

  • Kaushikaram Subramanian
  • Moritz Kreysing

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We would like to acknowledge the help from the following facilities at MPI-CBG - biomedical services, transgenic core, DNA sequencing, cell technology and FACS, light-microscopy during multiple phases of the project. The authors thank Caren Norden, Jochen Guck as well as Thomas Cremer, Jayaram K Iyer, Elisabeth Knust, Iain Patten, Andreas Reichenbach and Marino Zerial for helpful discussion and comments. The authors thank OptoDrum, Striatech GmbH for assistance in adapting the optomotor setup (OptoDrum) to scotopic light levels. 

Ethics

Animal experimentation: All animal studies were performed in accordance with European and German animal welfare legislation (Tierschutzgesetz), the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and the NIH Guide for the care and use of laboratory work in strict pathogen-free conditions in the animal facilities of the Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany and the Center for Regenerative Therapies Dresden, Germany. Protocols were approved by the Institutional Animal Welfare Officer (Tierschutzbeauftragter) and the ethics committee of the TU Dresden. Necessary licenses 24-9168.24-9/2012-1, DD24.1-5131/451/8 and TVV 16/2018 (DD24-5131/354/19) were obtained from the regional Ethical Commission for Animal Experimentation of Dresden, Germany (Tierversuchskommission, Landesdirektion Sachsen).

Senior Editor

  1. Ronald L Calabrese, Emory University, United States

Reviewing Editor

  1. Jeremy Nathans, Johns Hopkins University School of Medicine, United States

Reviewer

  1. Austin Roorda, University of California, Berkeley, United States

Publication history

  1. Received: June 20, 2019
  2. Accepted: December 11, 2019
  3. Accepted Manuscript published: December 11, 2019 (version 1)
  4. Version of Record published: January 21, 2020 (version 2)

Copyright

© 2019, Subramanian et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,395
    Page views
  • 210
    Downloads
  • 2
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Cell Biology
    2. Immunology and Inflammation
    Emma J Fenech et al.
    Research Article
    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Senem Aykul et al.
    Research Article Updated