Loss of Gq/11 Genes Does Not Abolish Melanopsin Phototransduction

In mammals, a subset of retinal ganglion cells (RGCs) expresses the photopigment melanopsin, which renders them intrinsically photosensitive (ipRGCs). These ipRGCs mediate various non-image-forming visual functions such as circadian photoentrainment and the pupillary light reflex (PLR). Melanopsin phototransduction begins with activation of a heterotrimeric G protein of unknown identity. Several studies of melanopsin phototransduction have implicated a G-protein of the Gq/11 family, which consists of Gna11, Gna14, Gnaq and Gna15, in melanopsin-evoked depolarization. However, the exact identity of the Gq/11 gene involved in this process has remained elusive. Additionally, whether Gq/11 G-proteins are necessary for melanopsin phototransduction in vivo has not yet been examined. We show here that the majority of ipRGCs express both Gna11 and Gna14, but neither Gnaq nor Gna15. Animals lacking the melanopsin protein have well-characterized deficits in the PLR and circadian behaviors, and we therefore examined these non-imaging forming visual functions in a variety of single and double mutants for Gq/11 family members. All Gq/11 mutant animals exhibited PLR and circadian behaviors indistinguishable from WT. In addition, we show persistence of ipRGC light-evoked responses in Gna11−/−; Gna14−/− retinas using multielectrode array recordings. These results demonstrate that Gq, G11, G14, or G15 alone or in combination are not necessary for melanopsin-based phototransduction, and suggest that ipRGCs may be able to utilize a Gq/11-independent phototransduction cascade in vivo.


Introduction
Intrinsically photosensitive retinal ganglion cells (ipRGCs) comprise a distinct subset of retinal ganglion cells (RGCs) and express the photopigment melanopsin (Opn4) [1]. ipRGCs constitute the sole conduit of light information from the retina to non-image forming visual centers in the brain and are responsible for driving a variety of behaviors [2,3]. These behaviors include circadian photoentrainment, which is the process by which the circadian clock is aligned to the environmental light-dark cycle, and the pupillary light reflex (PLR), in which the area of the pupil changes in response to changes in light intensity.
Despite the well-established role for ipRGCs and melanopsin in the regulation of non-image forming visual functions, little is known about the molecular components of melanopsin phototransduction. Previous research has suggested that ipRGCs likely utilize a phototransduction pathway similar to that used in Drosophila microvillar photoreceptors [1,4], in which the activated opsin stimulates a G q/11 protein. In Drosophila, the a-subunit of the G q/11 protein activates phospholipase C-b (PLC-b), which results in the opening of TRP and TRPL channels allowing Na + and Ca 2+ to flow into the cell resulting in depolarization of the rhabdomere in response to light [5,6].
Homologs of the components of the Drosophila phototransduction pathway are found in mice. Specifically, there are four G q/11 genes (Gnaq, Gna11, Gna14, and Gna15), four Plc-b genes (Plc-b1 -4), and seven Trpc channel genes (Trpc1-7). The tandemly duplicated Gna15 and Gna11 genes are linked to mouse chromosome 10 [7,8], and Gnaq and Gna14 colocalize to mouse chromosome 19 [9]. To date, there have been several electrophysiological studies implicating G q/11 , Plc-b, and TrpC genes in ipRGC phototransduction [4,10,11]. However, there have been no functional studies investigating the identity of the G q/11 protein utilized by melanopsin in vivo or any studies of the effects of the loss of any presumptive ipRGC phototransduction genes on behavior. In this study, we sought to determine the identity(ies) of the G q/11 protein(s) utilized for melanopsin phototransduction in vivo.
We performed single-cell RT-PCR on individual ipRGCs to determine which of the genes were expressed in ipRGCs and if there was heterogeneity in their expression among the ipRGC population. Similar to previous studies, we found that the majority of ipRGCs express both Gna11 and Gna14, but not Gnaq or Gna15. Since loss of the melanopsin protein results in well-characterized deficits in the pupillary light reflex and circadian behaviors, we examined these non-imaging forming visual functions in Gna11 2/2 ; Gna14 2/2 (Gna11; Gna14 DKO) mice and Gnaq flx/flx ; Gna11 2/2 ; Opn4 Cre/+ (Gnaq; Gna11 DKO) mice as well as several single G q/11 gene knockouts [9,[12][13][14]. All genotypes examined exhibited non-image forming visual functions indistinguishable from WT. Furthermore, multielectrode array recordings revealed no deficits in ipRGC light responses in Gna11; Gna14 DKO animals. Contrary to previous work, this study indicates that ipRGCs may be able to utilize a G q/11 -independent phototransduction cascade in vivo.

Results
Gna11 and Gna14 are expressed in ipRGCs Previous reports have shown that G q/11 genes are expressed in ipRGCs. However, there has been disagreement regarding which G q/11 genes are actually expressed, with one study reporting heterogeneous expression of each of the four G q/11 genes and another reporting primarily Gna14 and some Gna11 expression [4,15]. We therefore sought to definitively identify which G q/11 genes are expressed in ipRGCs. We isolated individual ipRGCs by dissociating retinas of Opn4 Cre/+ Z/EG mice, in which ipRGCs ipRGCs are labeled with GFP, and picking individual ipRGCs with a microneedle. We specifically chose to utilize retinas from P1 and P4 mice since there is GFP labeling of some cones in adult Opn4 Cre/+ Z/EG mice [16]. By RT-PCR, we confirmed that the 32 isolated cells expressed melanopsin ( Figure 1A, F-H), and then screened those 32 melanopsin-expressing cells for the four G q/11 genes ( Figure 1B-H). 23 of the 32 ipRGCs expressed both Gna11 and Gna14, and an additional 6 cells expressed either Gna11 or Gna14 ( Figure 1F-H). Neither Gnaq nor Gna15 were detected in any of the melanopsin-expressing cells, and 3 melanopsinexpressing cells had no detectable levels of any G q/11 gene ( Figure 1F-H).
We also recorded wheel-running activity in 4-6 month old WT (n = 14), MKO (n = 9), Gna15 KO (n = 7), Gnaq; Gna11 DKO (n = 8), and Gna11; Gna14 DKO (n = 7) mice to measure the daily activity rhythms of these mice ( Figure 3). We conducted these measurements under three different conditions: a 12:12 light/dark cycle, constant darkness, and constant light. We also administered a 15-minute light pulse in constant darkness to determine the amplitude of the light-evoked circadian phase shifts in each mouse line. All genotypes were able to photoentrain to the 12 Figure 3A, D). In contrast, Gna15 KOs (24.9960.28 hours) and Gna11; Gna14 DKOs (24.6660.42 hours) were indistinguishable from WT mice. While Gnaq; Gna11 DKO animals (24.3460.5 hours) did show a significantly shorter period than WT animals, the period was still significantly longer than MKO animals ( Figure 3A, D).

ipRGC light responses persist in Gna11; Gna14 double knockouts
The lack of behavioral deficits in G q/11 mutant animals led us to examine whether melanopsin phototransduction is perturbed at the cellular level in these lines. We therefore examined the light responses of ipRGCs in isolated retinas of WT and Gna11; Gna14 DKO mice using multielectrode array (MEA) recordings. We recorded from retinas of postnatal day 3 mice, since it has been shown that outer retinal signaling to ganglion cells is not present at early postnatal ages [21], and thus ipRGCs constitute the only light-responsive ganglion cells at this age [22,23]. Nonetheless, to guarantee that all detected light responses were from ipRGCs, we included a cocktail of synaptic blockers in the Ames' medium to inhibit any glutamatergic, GABAergic, and glycinergic signaling to ipRGCs. Additionally, we included cholinergic blockers to minimize interference from retinal waves present at this developmental stage [24]. Retinas were dark adapted for at least 20 min and then stimulated with diffuse, uniform light of both low (7610 12 photons/cm 2 ? sec) and high light intensity (7610 13 photons/cm 2 ? sec) for 60 sec at 480 nm, the peak wavelength for melanopsin activation [25,26]. We also stimulated the retinas with bright white light (267 mW/cm 2 ). The retinas were allowed to readapt to dark for 5 min between stimulations. Figure 4A shows representative voltage traces of typical ipRGCs in WT and Gna11; Gna14 DKO mice at both low and high light intensity. We found that Gna11; Gna14 DKO ipRGCs were indistinguishable from the WT controls. ipRGCs in both WT and Gna11; Gna14 DKO mice responded to increasing light intensities with increased spiking ( Figure 4B) that reached maximum levels several seconds following light onset. After light offset, ipRGCs continued to spike for as long as 20 seconds ( Figure 4A, C, D). These slow dynamics are consistent with previous descriptions of melanopsin-dependent light responses [23,[27][28][29]. These data show that despite the fact that Gna11 and Gna14 were the only G q/11 genes expressed in ipRGCs, they are not required for melanopsin phototransduction.
Other G q/11 genes are up-regulated in single and double G q/11 knockouts Since Gna11 and Gna14 were the only G q/11 genes detected in ipRGCs, we were surprised that Gna11; Gna14 DKO mice did not recapitulate any of the phenotypes observed in melanopsin knockout animals. To test whether removal of one or two G q/11 genes results in upregulation of other G q/11 family members, we performed quantitative RT-PCR on RNA extracted from the retinas of mutant mice ( Figure 5). We measured the mRNA levels of Gnaq, Gna11, Gna14, and Gna15 in WT, Gna14 KO, Gna15 KO, Gnaq; Gna11 DKO, and Gna11; Gna14 DKO mice. We found that all animals had levels of Gnaq mRNA that were indistinguishable from WT ( Figure 5A). It is important to note that in Gnaq; Gna11 DKOs, Gnaq is conditionally knocked-out in ipRGCs (Gnaq flx/flx ; Gna11 2/2 ; Opn4 Cre/+ ); therefore, we did not expect a significant reduction in whole retinal Gnaq mRNA in these mutants. Gna14 KOs and Gna15 KOs exhibited normal levels of Gna11 mRNA, while, Gnaq; Gna11 DKOs, and Gna11; Gna14 DKOs had undetectable levels ( Figure 5B). Levels of Gna14 mRNA were reduced in Gna14 KOs and Gna11; Gna14 DKOs, but increased in   as indicated by the reduced mRNA levels, the mutant transcript is degraded. Gna15 mRNA was undetectable in Gna15 KOs, but increased in Gna14 KOs, Gnaq; Gna11 DKOs, and Gna11; Gna14 DKOs ( Figure 5D). These data indicate there is upregulation of other G q/11 genes in some G q/11 knockout lines; however, it remains unknown whether the upregulation occurs in ipRGCs and if such upregulation would be sufficient to drive melanopsin phototransduction.

Discussion
In this study, we provide the first investigation of the melanopsin phototransduction pathway in vivo. We determined that genetic inactivation of the G q/11 proteins that are normally expressed in ipRGCs does not abolish melanopsin-dependent behaviors or electrophysiological responses. Specifically, we found that no tested G q/11 knockout line exhibited the behavioral deficits observed in melanopsin knockout mice. All tested G q/11 mutant lines exhibited circadian behaviors and pupillary light reflexes that were indistinguishable from WT mice. Additionally, using singlecell RT-PCR for G q/11 genes in ipRGCs, we found only expression of Gna11 and Gna14, often expressed together. However, using multielectrode array we detected no changes in intrinsic light responses of ipRGCs in Gna11; Gna14 DKO compared to WT controls.
Previous reports have shown expression of G q/11 genes in ipRGCs although there were inconsistencies as to which G q/11 genes were detected [4,15]. Specifically, in Graham et al. the authors used single cell RT-PCR and determined that expression of all four G q/11 genes can be detected in ipRGCs, although expression was heterogeneous among the cells sampled and Gna14 was detected in the majority of cells [4]. Siegert et al. examined ipRGCs as a population and reported expression of Gna11 and Gna14 [15], which is consistent with our findings here. However, neither of these studies investigated the function of ipRGCs in the absence of any of these specific genes and in fact Siegert et al. observed the expression of other heterotrimeric G proteins [15].
Electrophysiological investigations of ipRGC phototransduction have supported the involvement of the G q/11 pathway. Specifically, Xue et al. showed that melanopsin phototransduction is substantially reduced in the absence of Plc-b4 [10]. In agreement with work from Perez-Leighton et al., Xue and co workers additionally showed that loss of both Trpc6 and 7 virtually abolished the melanopsin-dependent photoresponse suggesting that Trpc6 and 7 function in a combinatorial fashion [10,11]. Since G q/11 family members are defined based on their ability to activate PLC, it is reasonable to predict that if Plc-b is a critical component of melanopsin phototransduction then there must also be a member of G q/11 family involved. This prediction was supported with the use of pharmacological inhibitors of the G q/11 family on dissociated ipRGCs [4]. However, here, we show that in vivo, mice mutant for G q/11 family members do not exhibit the behavioral deficits indicative of a loss of melanopsin-dependent light responses.
Several possibilities exist to explain these discrepancies. One is that G q/11 signaling is not required for melanopsin phototransduc-  [15] and thus melanopsin could activate a G i or G o protein, as has been observed in vitro [30], the dissociation of which could result in the beta/gamma subunit activating PLC-b4 as has been observed with PLC-b1 and 3 [31]. Another possibility is that there is compensatory upregulation from other remaining G q/11 family members in the tested mutant lines. Our data supports this possibility since Gna14 and Gna15 were upregulated in Gnaq; Gna11 DKOs. Also, Gna15 was upregulated in Gna14 knockouts; although, the increase in Gna15 expression was not significant in Gna11; Gna14 DKOs. However, our qRT-PCR experiments were performed on whole retinal RNA, and expression of Gna15 has not consistently been reported in ipRGCs. Thus, it is unknown whether there is ectopic expression of Gna15 in ipRGCs in G q/11 knockout lines. Whether other G q/11 family members are upregulated in the conventional G q knockout lines could be investigated by creating a mouse line that has all four G q/11 genes knocked-out in ipRGCs. Due the fact that G q/11 genes exist as two closely linked pairs on two single chromosomes, this quadruple knockout will require creation of a new mutant line in which the linked genes are knockout together. This mouse line would definitively reveal the contribution of the G q/11 class alpha subunits to the melanopsin phototransduction cascade.
Additionally, it remains possible that Gna11 and Gna14 are required for the activation PLC-b4 and TRPC6/7, but this pathway is not required for normal ipRGC-mediated behavior. In support of this idea, a small residual light-activated current exist in Plc-b4 2/2 and Trpc6/7 2/2 ipRGCs [10]. Importantly, voltage recordings were not performed in these mutants. Therefore, it remains possible that this small residual current is sufficient to drive spiking in ipRGCs, which then drives normal non-imageforming visual behaviors. To test this, behavioral assays need to be performed on Plc-b4 2/2 and Trpc6/7 2/2 mice.
It is important to note that ipRGCs are not a homogeneous population and ipRGC subtypes (termed M1-M5) have stereotyped yet distinct electrophysiological light responses. Thus, it is possible there is variability in the components of the melanopsin phototransduction cascade among ipRGC subtypes. The study showing that ipRGCs have a severe reduction in their intrinsic light responses in mouse lines mutant for Trpc6 and -7 channel genes and Plc-b4 [10] only examined the M1 ipRGC subtype, and in Trpc6 mutant mice, both M1 and M2 ipRGCs show some deficits in melanopsin-dependent light responses [11]. While M1 ipRGCs are the predominant subtype mediating circadian behaviors, non-M1 ipRGCs may contribute to the PLR [16]. It remains unknown whether the intrinsic responses of other ipRGC subtypes are affected in Trpc6 and Trpc7 double knockouts or in Plc-b4 knockouts. Because we picked single cells for RT-PCR at a developmental time, we could not be certain whether we were picking M1 or non-M1 ipRGCs. A careful analysis of the

Ethics Statement
All protocols, animal housing, and treatment conditions were approved by the Johns Hopkins University Animal Care and Use Committee (IACUC).

Single Cell RT-PCR
Single ipRGCs were isolated from Opn4 Cre/+ ; Z/EG mice following the protocol described in [32]. Reverse transcription of the RNA from single cells from P1 and P4 retinas, and amplification of the cDNA was performed as described in [32]. The following primers were designed to amplify from the 39 end of the transcript and used to detect phototransduction components in the resulting amplified cDNA obtained from single ipRGCs:

Wheel Running Behavior
Mice were placed in cages with a 4.5-inch running wheel, and their activity was monitored with VitalView software (Mini Mitter), and cages were changed at least every 2 weeks.

Quantification of circadian behavior
All free-running periods were calculated with ClockLab (Actimetrics) using the onsets of activity on days 10-17 of constant darkness similar to [3]. Phase shifts were calculated similar to [3] and described as follows: an onset for the day after the light pulse was predicted based on the onsets of the previous 7 days. Phase shifts were then determined based on the difference between the predicted onset and the shifted onset on the day after the light pulse. For all animals, the free-running period in constant light was measured with ClockLab (Actimetrics) using the onsets of activity on days 3-10 of constant light. Some animals (2 WT, 1 Gna15 KO, 2 Gnaq; Gna11 DKOs, 1 Gna11; Gna14 DKO, and 1 MKO) reduced their activity so much that an accurate period could not be measured and they were thus excluded.

Q-RT-PCR
Retinas were dissected from WT, Gna14 KO, Gna15 KO, Gnaq; Gna11 DKO, and Gna11; Gna14 DKO (N = 3 mice for each; 2 retinas per RNA sample). RNA was extracted from the retinas using an RNeasy mini kit (Qiagen; cat# 74106), and reverse transcription was performed using a RETROscript kit (Life Technology; cat # AM1710) and random hexamer primers. Quantitative PCR on the resulting cDNA was performed with SYBR Green PCR Master Mix (Fermentas, cat# K0221), samples were analyzed in duplicate, and the levels were normalized to 18S RNA. The following primers were used: Gnaq