Systematic spatio-temporal mapping reveals divergent cell death pathways in three mouse models of hereditary retinal degeneration

Calcium (Ca2+) dysregulation has been linked to neuronal cell death, including in hereditary retinal degeneration. Ca2+ dysregulation is thought to cause rod and cone photoreceptor cell death. Spatial and temporal heterogeneities in retinal disease models have hampered validation of this hypothesis. We examined the role of Ca2+ in photoreceptor degeneration, assessing the activation pattern of Ca2+-dependent calpain proteases, generating spatio-temporal maps of the entire retina in the cpfl1 mouse model for primary cone degeneration, and in the rd1 and rd10 models for primary rod degeneration. We used Gaussian process models to distinguish the temporal sequences of degenerative molecular processes from other variability sources. In the rd1 and rd10 models, spatio-temporal pattern of increased calpain activity matched the progression of primary rod degeneration. High calpain activity coincided with activation of the calpain-2 isoform but not with calpain-1, suggesting differential roles for both calpain isoforms. Primary rod loss was linked to upregulation of apoptosis-inducing factor (AIF), although only a minute fraction of cells showed activity of the apoptotic marker caspase-3. After primary rod degeneration concluded, caspase-3 activation appeared in cones, suggesting apoptosis as the dominant mechanism for secondary cone loss. Gaussian process models highlighted calpain activity as a key event during primary rod photoreceptor cell death. Our data suggests a causal link between Ca2+ dysregulation and primary, non-apoptotic degeneration of photoreceptors and a role for apoptosis in secondary degeneration of cones, highlighting the importance of the spatial and temporal location of key molecular events, which may guide the evaluation of new therapies.

whereas the cpfl1 line suffers from a mutation in the homologous, cone-specific Pde6c gene. To align our experiments with earlier studies of mouse photoreceptor degeneration, we first 2 3 7 compared changes in outer nuclear layer (ONL) thickness over the first post-natal month 2 3 8 between the mutants and wt mice (Fig. 2). The number of photoreceptor rows dramatically 2 3 9 decreased in post-P12 rd1 and in rd10 after P18, illustrating the difference in onset of 2 4 0 photoreceptor cell death in these models (Arango-Gonzalez et al., 2014). The rd10 2 4 1 degeneration is less aggressive than rd1 as it takes longer for complete ablation of the ONL in 2 4 2 rd10 than in rd1. In contrast, in the cone degeneration cpfl1 mutant the ONL thickness  We next aimed at elucidating the role of Ca 2+ dysregulation in photoreceptor cell death. To proteases, which can be considered as surrogate markers for Ca² dysregulation (Croall & 2 5 0 11 Ersfeld, 2007;Goll et al., 2003). In rd1 retina, the number of ONL cells showing increased 2 5 1 calpain activity was significantly higher compared to wt and cpfl1, at all time-points, with a 2 5 2 peak of activity at P12 (Fig. 3A, D; for detailed statistics of all comparisons, see Table 2).

5 3
Calpain activity in rd10 ONL was also significantly increased compared to wt and cpfl1 (at 2 5 4 P18, 24, and 30), peaking around P18 (Fig. 3B, D). In cpfl1, calpain activity tended to be 2 5 5 higher than in wt, yet, without a clear peak within the inspected time frame (Fig. 3C, D). Overall, rd1 retinae displayed the highest levels of calpain activity. These data are in line with other studies (Arango-Gonzalez et al., 2014;Paquet-Durand et al., 2006) that showed 2 5 8 increases in calpain activity at specific time-points in models of RP and achromatopsia.

5 9
Activation of calpain-2, but not calpain-1, increases during disease progression We then investigated whether the high levels of calpain activity observed in the ONL could be 2 6 1 attributed to specific calpain isoforms. We focussed on calpain-1 and calpain-2, which have 2 6 2 been reported to be ubiquitously expressed in all mammalian cells and to play opposite roles 2 6 3 in neurodegeneration (Chen et al., 2007;Goñi-Oliver et al., 2007). We hypothesized that 2 6 4 calpain-2 may contribute to the observed increased calpain activity, as it requires a high 2 6 5 [Ca 2+ ] that is considered beyond the physiological range of photoreceptors (Goll, 1995). Using antibodies recognizing the activated proteases, we found significantly increased 2 6 7 numbers of calpain-2 positive cells in both rd1 and rd10 ONL, when compared to cpfl1, as 2 6 8 well as to wt controls (Fig. 4A, B, D; for all statistics, see Table 2). The peaks appeared to 2 6 9 coincide with those for the calpain assay, namely at P12 (rd1) and P18 (rd10). In cpfl1, the 2 7 0 number of calpain-2 positive cells was slightly increased over the wt level (Fig. 4C, D), 2 7 1 however, without a clear peak. To our surprise, a significant increase in calpain-1 positive cell 2 7 2 numbers from wt level was observed in rd10 ( Fig. 5A-C). Calpain-1 labelling increased over 2 7 3 background levels around P18 and was even more pronounced at the end of the observed time 2 7 4 window (P30) (Fig. 5D). Next to rd10, only rd1 showed a detectable (but not significant) 2 7 5 increase in calpain-1 positive cell counts.

7 6
Taken together, we found differential patterns of calpain isoform activation: While calpain-2 2 7 7 was strongly activated in rd1 and rd10 (and much less so in cpfl1), calpain-1 was more 2 7 8 strongly activated in rd10 compared to the other lines. Moreover, the calpain-2 expression 2 7 9 peak coincides with the peak of calpain activity assay data and precedes that of calpain-1. To compare the activity and expression of calpain in the ONL with the incidence of cell death, 2 8 2 we used the TUNEL assay, which labels nick-ends in a cell's DNA. While TUNEL is an does not discriminate between different cell death mechanism(s) (Kraupp et al., 1995). In  Table 2). A slight increase in TUNEL-positive cell number was also seen in the cpfl1 mouse positive ONL cell numbers (P12 in rd1; P24 in rd10) coincided temporally with those of 2 9 0 calpain activity and calpain-2 labelling. involving calpain activation (Arango-Gonzalez et al., 2014;Doonan et al., 2005). To 2 9 6 distinguish between these two proposed pathways, we used antibodies for cleaved caspase-3, 2 9 7 13 a marker protein for apoptosis (Mazumder, Plesca, & Almasan, 2008), and AIF, a marker for statistics, see Table 2) as those of TUNEL and calpain activity. In rd1 retina, a significant   Table 2). Significantly higher numbers of caspase-3 positive cells were seen in rd1 retina analysed for rd1 cones expressing caspase-3, we found that at P24 a substantial number of  In the experiments so far, we noticed that the distribution of data points for a certain time-  peaks of calpain-2, calpain activity, and TUNEL staining appeared to coincide (~P12), followed by smaller peaks of caspase-3 (P24) and AIF (P30 or later). In rd10, the broader 3 3 5 "peaks" of calpain-2, calpain activity, and TUNEL also coincided, yet they were more spread 3 3 6 out (from P18 to P30) and -as opposed to rd1 -associated with minor elevations of AIF-and 3 3 7 calpain-1, but less so with caspase-3.

8
To more quantitatively identify the time-points at which each cell death markers peaked in   in the rd1 moose (Noell, 1958). Cell death patterns seen in the rd10 retina are less well-4 1 0 known but may begin in the mid-periphery and work simultaneously towards and away from 4 1 1 the optic nerve (Barhoum et al., 2008;Strettoi & Pignatelli, 2000). In our examination of the 4 1 2 rd10 mouse, a centre to peripheral gradient was indeed seen for calpain activity, calpain-2, 4 1 3 and TUNEL, although this gradient was not as distinct as that in the rd1 mouse. This could be 4 1 4 due to time-point selection or to the fact that the degeneration seen in the rd10 mouse is less 4 1 5 synchronous than that in the rd1 mouse. regulation or activation was detectable, as seen in the individual heat maps (Fig. 11-16). This 18 is probably due to the low number of cones in the mouse retina and the stochastic nature of 4 2 0 the cell death seen in all mouse models of retinal dystrophies (Clarke et al., 2000). It is also 4 2 1 possible that due to differences in the phototransduction machinery cones are more resilient to  One of the confounding factors that complicate studies into the causative cell death calpain-1 dependent protection (as discussed above).

8
The rd10 retina also displayed a more marked up-regulation of AIF compared to rd1. Since AIF is a mitochondrial protein, and photoreceptor mitochondria in the inner segments grow 4 5 0 and mature with post-natal age, it is conceivable that this apparent increase in AIF is entirely 4 5 1 due to the later onset of rd10 degeneration. If correct, then AIF may also be associated with 4 5 2 rd1 degeneration, as has been suggested before (Sanges et al., 2006), but would be more 4 5 3 difficult to detect because at the onset of rd1 degeneration its expression was much lower.

5 4
Taken together, the cell death mechanisms seen in primary rod degeneration, in both rd1 and pathways. Conversely, calpain-1 activation may play a protective role, however, a role which 4 5 8 ultimately is not strong enough to save photoreceptors from mutation-induced degeneration. Here, we adopted a "standard" experimental strategy which emphasises the use of large temporal resolution. Where the temporal precision is critical, it would be preferable to adopt a 4 6 5 proactive strategy, using GP models to pre-select the optimal time-points for particular hypotheses (Chaloner & Verdinelli, 1995;Lindley, 1956;Pillow, 2016). This should make it 4 6 7 20 possible to infer more accurately the non-linear progression of each marker over time, with 4 6 8 fewer samples than would be needed in a predetermined or purely random approach.