A G86R mutation in the calcium-sensor protein GCAP1 alters regulation of retinal guanylyl cyclase and causes dominant cone-rod degeneration

The guanylyl cyclase-activating protein, GCAP1, activates photoreceptor membrane guanylyl cyclase (RetGC) in the light, when free Ca2+ concentrations decline, and decelerates the cyclase in the dark, when Ca2+ concentrations rise. Here, we report a novel mutation, G86R, in the GCAP1 (GUCA1A) gene in a family with a dominant retinopathy. The G86R substitution in a “hinge” region connecting EF-hand domains 2 and 3 in GCAP1 strongly interfered with its Ca2+-dependent activator-to-inhibitor conformational transition. The G86R-GCAP1 variant activated RetGC at low Ca2+ concentrations with higher affinity than did the WT GCAP1, but failed to decelerate the cyclase at the Ca2+ concentrations characteristic of dark-adapted photoreceptors. Ca2+-dependent increase in Trp94 fluorescence, indicative of the GCAP1 transition to its RetGC inhibiting state, was suppressed and shifted to a higher Ca2+ range. Conformational changes in G86R GCAP1 detectable by isothermal titration calorimetry (ITC) also became less sensitive to Ca2+, and the dose dependence of the G86R GCAP1–RetGC1 complex inhibition by retinal degeneration 3 (RD3) protein was shifted toward higher than normal concentrations. Our results indicate that the flexibility of the hinge region between EF-hands 2 and 3 is required for placing GCAP1-regulated Ca2+ sensitivity of the cyclase within the physiological range of intracellular Ca2+ at the expense of reducing GCAP1 affinity for the target enzyme. The disease-linked mutation of the hinge Gly86, leading to abnormally high affinity for the target enzyme and reduced Ca2+ sensitivity of GCAP1, is predicted to abnormally elevate cGMP production and Ca2+ influx in photoreceptors in the dark.

nant cone or cone-rod degenerations specifically suppress Ca 2ϩ sensitivity of RetGC1 isozyme in vivo (17,41,42), most often by directly altering GCAP1 EF-hand motifs and thus reducing metal binding in EF-hands 3 and 4 (reviewed in Ref. 21), but also indirectly, by affecting EF-hand 4 affinity for Ca 2ϩ via altering a "calcium-myristoyl tug" mechanism connecting the C-terminal EF-hand 4 with the N-terminal myristoyl group buried inside the N-proximal semiglobule of EF-hands 1 and 2 (1,43,44). Here we describe a new type of mutation in GCAP1, G86R, leading to a dominant retinopathy in humans. We present evidence that the flexible hinge Gly 86 connecting the semiglobules formed by the two pairs of EF-hands reduces GCAP1 affinity for RetGC1 but also adjusts the Ca 2ϩ sensitivity of the activator-to-inhibitor transition to the proper physiological range of RetGC1 regulation by Ca 2ϩ . We reason that the G86R mutation in GCAP1 would trigger photoreceptor death by abnormally elevating cGMP production rate in the dark.

G86R GUCA1A causes dominant retinopathy
The patient evaluated at 62 years of age had no history of visual problems (other than myopia) until his 5th decade of life. He was then diagnosed as having a macular dystrophy. Best corrected visual acuities at age 56 were 20/60 and 20/200. Reading and distance vision decreased progressively over many years; and color discrimination became difficult. There was no photosensitivity and no peripheral field or night vision complaints. The patient did not report any general health issues and he was not taking medication with known retinotoxic effects. Visual acuities of the proband were 20/200 (with no significant refractive error) in both eyes. The clinical examination was within normal limits except for the retina. Maculae were granular and thinned in appearance. Peripheral retina, optic nerve, and vessel caliber were normal. Macular functional abnormalities were supported by retinal pigment epithelium (RPE) disease apparent on en face imaging (Results and Fig. S1) and retinal structural changes on cross-sectional imaging. Optical coherence tomography (OCT) (Fig. 1A) showed loss of central retinal lamination suggesting complete degeneration of rod and cone photoreceptors. By 10 o from the fovea, however, outer nuclear layer (ONL, where rod and cone photoreceptor nuclei are located, highlighted blue) thickness returned to normal. Beyond the perifoveal region, there was a mild thinning of both cone and rod outer segments extending across the width of the scan (Fig. 1, right panels). Illustrating these photoreceptor abnormalities are magnified images (lower panels). Shortened outer segments where phototransduction occurs is likely one of the contributors to the loss of visual function. Electrophysiological (Fig. 1B) and perceptual (45) (Results and Fig. S1) results suggested mild dysfunction across the retina of both rod and cone photoreceptor-driven function, and a severe loss of macular function.
The family history of the proband (Fig. 1C, III, 2) indicated 6 other members with visual symptoms, suggesting an autosomal dominant mode of inheritance, although there was no male-tomale transmission. DNA samples from the proband and two other affected family members (III, 3; and IV, 1) were studied. In the proband, a panel of genes associated with inherited retinal degenerations was screened using next-generation sequencing followed by confirmatory Sanger sequencing. A novel heterozygous missense variant, c.256G 3 C, in the GUCA1A gene was identified in the proband and in other affected family members by direct testing in the GUCA1A gene. In the disease-causing variant, Gly 86 , residue that connects the exiting helix of EFhand 2 and the entering helix of EF-hand 3, thus creating a hinge between the semiglobules of GCAP1 formed by two pairs of EF-hand domains (Fig. 2), was substituted by Arg.
The G86R mutation in GCAP1 reduces Ca 2؉ sensitivity of guanylyl cyclase regulation G86R GCAP1 effectively activated human RetGC1 expressed in HEK293 cells at low free Ca 2ϩ concentrations, similarly to the WT GCAP1 (Fig. 3, A and B), but failed to effectively decelerate the cyclase when free Ca 2ϩ concentrations rise. The [Ca] 1/2 values became shifted to 5-fold higher free Ca 2ϩ in both bovine and human orthologs of GCAP1. In the case of human GCAP1, the [Ca] 1/2 (mean Ϯ S.D.) for G86R increased to 0.94 Ϯ 0.09 M (n ϭ 4) from 0.14 Ϯ 0.06 M (n ϭ 3) in WT (p ϭ 0.0004, unpaired/unequal variance Student's t test). As a result, the RetGC1 remained stimulated by the G86R GCAP1 at free Ca 2ϩ concentrations that by far exceeded those found in darkadapted mammalian photoreceptors (17,47).
The affinities of GCAP1 to RetGC1 in both the Mg 2ϩ and Ca 2ϩ -liganded states are nearly equal ( Fig. 4 and Ref. 48). However, replacement of the hinge Gly 86 increased the GCAP1 apparent affinity for RetGC1 in the activator state of the mutant, K GCAPMg ϭ 0.34 Ϯ 0.063 M, n ϭ 4, versus 1.42 Ϯ 0.061 M, n ϭ 3, in WT (p Ͻ 0.0001) (Fig. 4, B and C), whereas the affinity of the Ca 2ϩ -liganded G86R was increased to a lesser extent: K GCAPCa ϭ 0.59 Ϯ 0.047 M versus 1.44 Ϯ 0.14 M in WT, n ϭ 4 (p ϭ 0.001) (Fig. 4C). The G86R GCAP1 gaining higher affinity for the cyclase in Mg 2ϩ -than in the Ca 2ϩ -liganded form, resulting in change of the K GCAPMg versus K GCAPCa , can account for a ϳ2-fold increase in [Ca] 1/2 ( Fig. 4 and Ref. 48). Nonetheless, this difference alone did not account for the Ͼ6-fold [Ca] 1/2 increase observed in Fig. 3B, thus suggesting that Gly 86 , despite its being not a part of the EF-hand motif per se, but a hinge between the two semiglobules of the molecule (Fig. 2) nonetheless affected GCAP1 affinity for Ca 2ϩ .

Activator-to-inhibitor transition in G86R GCAP1 requires higher Ca 2؉ concentrations
The dependence of the intrinsic GCAP1 Trp fluorescence on Ca 2ϩ (Fig. 5) is biphasic: first reduction and then increase of intensity, reflecting GCAP1 transition from a metal-free to a metal-bound state (49 -51). The "phase I" (decrease in fluorescence) reflects GCAP1 conversion to a partially Ca 2ϩ -or Mg 2ϩ -liganded "cyclase-activator" state, and "phase II" is the increase specifically in Trp 94 fluorescence, caused by Ca 2ϩ binding in EF-hand 4, which converts GCAP1 to its "cyclaseinhibitor" state (50,51). Phase I in the absence of Mg 2ϩ is less pronounced in human GCAP1 compared with the bovine GCAP1 (50,51), because although both have Trp 21 and Trp 94 , the human ortholog lacks Trp 51 contributing to the phase I amplitude. However, the two Ca 2ϩ -dependent phases of the cGMP synthesis and photoreceptor blindness fluorescence remained clearly identifiable in the human GCAP1. In G86R GCAP1, Ca 2ϩ -sensitivity of the fluorescence spectrum was markedly shifted toward higher range. The Ca 2ϩdependent phase II can be further isolated by eliminating phase I by first saturating GCAP1 with Mg 2ϩ (50,51). The resultant phase II (Fig. 5B) in G86R GCAP1 demonstrated a shift toward higher Ca 2ϩ concentrations and drastic reduction of the amplitude at saturating Ca 2ϩ compared with the WT. The Ca 2ϩspecific increase in fluorescence intensity at 200 M Ca 2ϩ versus 0 M Ca 2ϩ was lacking, 1.04-fold Ϯ 0.02 (S.D.), compared

cGMP synthesis and photoreceptor blindness
with the WT (1.51 Ϯ 0.09, n ϭ 3, p ϭ 0.0074, Student's t test), thus indicating that the activator-to-inhibitor conformational transition was restrained even at saturating free Ca 2ϩ .
To further isolate the phase II fluorescence component of the Ca 2ϩ -dependent activator-to-inhibitor transition in a human G86R GCAP1, phase I was eliminated using a W21F substitution (51). The remaining single Trp 94 fluorescence in human GCAP1 was measured in the absence of Mg 2ϩ (Fig. 6A), in the presence of near-physiological 0.9 mM free Mg 2ϩ (52) (Fig. 6B) or at saturating 9 mM free Mg 2ϩ (Fig. 6C). In all cases, the foldincrease of Ca 2ϩ -dependent Trp 94 fluorescence of G86R GCAP1, defining its transition to the state decelerating RetGC1 activity, became reduced: 1. The difference in the Ca 2ϩ -dependent transition to the cyclase-inhibitor state between the W21F mutants in G86R versus WT GCAP1 backbone (Fig. 6, D-F) was also more pronounced in the absence or at low physiological (52) concentrations of Mg 2ϩ . The respective [Ca] 1/2 ratios (Fig. 6G) at 0, 0.9, and 9 mM Mg 2ϩ were 17.7 Ϯ 1.47, 5.36 Ϯ 0.28, and 2.14 Ϯ 0.27 (p Ͻ 0.0001 for all pairs), indicating that the GCAP1 affinity for Mg 2ϩ in the EF-hand 4 also became reduced as a result of the mutation in the hinge region.

G86R mutation alters the thermodynamics of the Ca 2؉ -dependent conformational changes in GCAP1
Changes of the heat release by GCAP1 upon metal binding reflect a superposition of several processes including not only direct interaction of metal and protein, but conformational changes and reorganization of the protein-water shell as well (39,53,54) (Fig. 7). Therefore, the parameters for metal binding formally extracted from the ITC measurements describe the apparent metal affinity, but they do not match exactly those revealed by the intrinsic tryptophan fluorescence or RetGC1 regulation sensitivity (53). However, despite this limitation the comparison of the formal parameters extracted from ITC still reflects the change in the GCAP1 thermodynamics imparted by mutations (39,(53)(54). In case of G86R in the presence of 1 mM Mg 2ϩ (Fig. 7), the net heat release pattern in response to binding Ca 2ϩ changed from triphasic to biphasic and the endothermic component was more pronounced than in WT (Fig. 7, A and B). However, best fit of the data were obtained with a sequential binding model resulting in two apparent K D values for the WT and the G86R mutant (Fig. 7, A and B, and Table 1). Due to residual free Ca 2ϩ (90 -200 nM; see "Experimental procedures") in the presence of 20 -30 M GCAP1 variants, we assume partial or full saturation of a high affinity Ca 2ϩ -binding site in human GCAP1. The apparent affinity of Ca 2ϩ -binding was, however, different and became less in case of G86R (see values for K D 1 and K D 2 in Table 1). Parameters for Ca 2ϩ binding to GCAP1 variants in the absence of Mg 2ϩ exhibited a similar overall pattern (Table 1). In addition, G86R GCAP1 was more sensitive to the metal binding in Mg 2ϩ -dependent heat release tested by ITC (Fig. 7, C and D). Enthalpy changes were endothermic in both cases and best fit was obtained with a two-site model indicating conformational changes caused primarily by Mg 2ϩ binding in two EF-hands, occurring in a noncooperative manner.

Sensitivity of RetGC1-GCAP complex to inhibition by RD3
RD3 (retinal degeneration 3) protein inhibits both RetGCs basal activity and GCAP-stimulated activity (55,56). The ability of RD3 to suppress activation of RetGC1 is likely required for normal survival of photoreceptors, because rods and cones lacking RD3 degenerate much faster and more severely than those completely lacking both RetGC1 and RetGC2 isozymes, altogether (56). We found that activation of the cyclase by G86R GCAP1 becomes more resistant to inhibition by RD3 than the cyclase activated by WT RD3 (Fig. 8). The EC 50 rose to 17.1 Ϯ 2.8 nM, n ϭ 3, versus 1.7 Ϯ 0.4 nM, n ϭ 3, in WT (p ϭ 0.0007).

Clinical features of the disease caused by G86R GCAP1
The dominant retinopathy caused by G86R substitution presents a new example of a photoreceptor dystrophy linked to the mutations in GCAP1 ( Fig. 1 and Fig. S1). The clinical disease feature in common with previous studies of families with GUCA1A mutations coding for GCAP1 (28 -38) has been maculopathy with central retinal defects documented by functional and structural parameters (reviewed in Refs. 31 and 32).

cGMP synthesis and photoreceptor blindness
In addition to central cone disease, some patients in the previous studies also had cone photoreceptor dysfunction in extracentral retina and others have had cone as well as rod dysfunction. Of interest, a postmortem donor retina from a 75-year-old patient clinically diagnosed as having a cone degeneration by ERG criteria at age 45 years has been studied by histopathology; and the disease cause was a GUCA1A mutation (36,57). As in the present study, there was central retinal atrophy, widespread cone disease by cone ERG, and, despite a normal full-field rod ERG, rod inner and outer segment abnormalities beyond the macula. The clinical spectrum due to GUCA1A mutations thus includes macular degeneration, and retina-wide cone and rod dysfunction where cone disease may be equal to or greater than rod disease.

Molecular mechanisms of retinopathies caused by GCAP mutations
The majority of the known examples of retinopathies linked to the GUCA1A gene originate from mutations in GCAP1 EFhand motifs that disable Ca 2ϩ binding in EF-hands 3 and 4 (28 -40, 58, 59). Reduction of Ca 2ϩ affinity in these EF-hands leads to overly active production of cGMP at normal free Ca 2ϩ concentrations in the dark, increasing the fraction of the open cGMP-gated channels and accelerating the influx of Ca 2ϩ into photoreceptor outer segment (17,18,41). The overall Ca 2ϩ sensitivity of GCAP1 imparted by the intrinsic ability of their EF-hands to bind Ca 2ϩ or Mg 2ϩ is additionally fine-tuned by a "calcium-myristoyl tug" (1, 43), a structural link between the C-proximal EF-hand 4 and the N-terminal myristoyl residue

cGMP synthesis and photoreceptor blindness
embedded in the semi-globule formed by the N-proximal EFhands 1 and 2 (Fig. 2). Leu 176 replacement by Phe, through the altered tug action, increases GCAP1 affinity for RetGC1 but also reduces the affinity of the EF-hand 4 for Ca 2ϩ and thus shifts Ca 2ϩ sensitivity of RetGC regulation outside the physiological range (43). The L176F substitution, originally used to change the calcium-myristoyl tug in GCAP1 in vitro (43), was soon after found in patients with a photoreceptor dystrophy (44), likely triggered by elevated cGMP production in the dark.
In contrast to the multiple mutations in GCAP1 affecting RetGC1 sensitivity to Ca 2ϩ causing cone-and cone-rod degenerations, a rare retinopathy accompanied by G157R mutation in GCAP2 (GUCA1B) affects primarily rods (60). The mutant GCAP2 is more prone to accumulating in photoreceptor inner segment (61), but its mechanistic significance in triggering photoreceptor death remains unclear.
Our present results show that not only EF-hands and calcium-myristoyl tug in GCAP1, but the hinge Gly 86 located between the two semiglobules formed by EF-hand pairs is critical for defining the normal Ca 2ϩ sensitivity in the whole-molecule structure of GCAP1. A dominant retinopathy (Fig. 1) occurs when the hinge is affected. The two semiglobules in GCAP1 connected by the Gly 86 (Fig. 2) can move relative to each other (51,62). Apparently, this flexibility contributes to interaction with the cyclase and Ca 2ϩ sensitivity of the target enzyme regulation, similarly to the calcium-myristoyl tug (43). Notably, the residues surrounding Gly 86 were implicated in forming the complex between GCAP1 and RetGC1 (46). The G86R substitution imparts an abnormally high affinity to the interface for the target enzyme at the expense of making GCAP1 Ca 2ϩ sensitivity abnormally low. The net outcome is a drastic increase of the cyclase activity at the free concentrations of Ca 2ϩ typical for the dark-adapted photoreceptors. Even if the shift in Ca 2ϩ sensitivity of the cyclase caused by the G86R GCAP1 is attenuated by competition with normal GCAP1 and GCAP2 (58), it should inevitably increase the cGMP synthesis rate, open an excessive number of the cGMP-gated channels, and increase the influx of Ca 2ϩ in dark-adapted rods and cones, like other retinal diseases triggered by similar dysregulation of the cyclase (17)(18)41).
Along with the calcium-myristoyl tug in GCAP1 (43), which is functionally and structurally distinct from the calcium-myristoyl switch prototypically observed in recoverin (63,64), the flexibility of the hinge region in the GCAP1 structure evidently provides the optimal trade-off between the affinity for the target and the sensitivity of its regulation by physiological intracellular Ca 2ϩ . EF-hand 1 (Fig. 2) has evolved in GCAPs to become the main part of the target-binding interface, instead of being a metal-binding domain (46,65). Three other EF-hands contribute differently to the cyclase regulation. EF-hands 2 and 3 bind Mg 2ϩ or Ca 2ϩ to properly present the target-binding interface and thus allow GCAP to dock with the cyclase (66, 67). The pair of the C-proximal EF hands 3 and 4 is the actual Ca 2ϩ sensor part of the molecule that enables the GCAP switch between the Mg 2ϩ -bound state in the light to the Ca 2ϩ -bound state in the dark and cause the activator-to-inhibitor transition of GCAP1 (51,53,66,67). This transition is affected by the blinding G86R mutation, such that GCAP1's preferential state now gains higher affinity for RetGC1, is more prone to remain in a Mg 2ϩ -bound state and less susceptible to undergo the transition to the Ca 2ϩ -bound inhibitor state (Figs. 3-7).
Notably, the overall susceptibility of G86R GCAP1 to converting into a Mg 2ϩ -liganded state in the ITC experiments increases (Fig. 7), in contrast to the Mg 2ϩ effect on the intrinsic Trp 94 fluorescence reflecting GCAP1 transition upon Mg 2ϩ to Ca 2ϩ replacement in EF-hand 4 (Fig. 6), which is more drastically affected at normal physiological Mg 2ϩ than at saturating Mg 2ϩ concentrations. This indicates that in EF-hand 4, responsible for the Ca 2ϩ -dependent fluorescence increase in Trp 94 in the neighboring EF-hand ␣-helix (51), the Mg 2ϩ binding affinity is reduced, whereas in the EF-hand 2 and/or -3 it becomes strengthen. The Mg 2ϩ binding in EF-2 and EF-3 is essential for the overall ability of GCAP1 to dock with the cyclase (67) and less critical for the GCAP1 switching between the activator and the inhibitor states, primarily driven by Ca 2ϩ binding in EF-4 (51, 66). The current ITC experiments are unable to identify the  (51). The data (mean Ϯ S.D., three independent measurements) in each case were normalized per fluorescence intensity at low Ca 2ϩ ; empirical fit. Note that in G86R GCAP1, the Ca 2ϩ dependence of fluorescence is shifted toward a higher range and the amplitude of the phase II is suppressed.

cGMP synthesis and photoreceptor blindness
EF-hand(s) that stronger binds Mg 2ϩ in G86R GCAP1, but the lack of inhibition of the cyclase at the normal physiological Ca 2ϩ concentrations and the Trp 94 fluorescence spectra argue that EF-hand 4 likely has a reduced affinity for both metals, consistent with the ITC titration detecting only two submillimolar-affinity Mg 2ϩ -binding sites in the G86R GCAP1. Physiological concentrations of Mg 2ϩ in the photoreceptor outer segment are near 1 mM (52), so the increase of Mg 2ϩ affinity in EF-hands other than EF-4 would unlikely affect RetGC regulation by G86R GCAP1, because their affinities even in WT are already high enough to maintain GCAP1 in the activator state. In contrast, the decrease in metal affinity of EF-4 is critical, preventing the proper activator-to-inhibitor transition in the dark.
Increased cGMP production in the dark is detrimental for photoreceptor viability. A body of evidence from in vivo studies using transgenic animal models also demonstrate that dysregu-lation of the negative Ca 2ϩ feedback on RetGC, either via altering the Ca 2ϩ -sensor properties of GCAPs or via preferentially increasing the affinity of the cyclase for the activator form of GCAP1 lead to severe retinal pathology (17,18,41,80). Our present study argues that triggering a blinding disease by GCAP1 can result by affecting its Ca 2ϩ sensitivity through a mechanistic reason different from directly disabling an EFhand or calcium-myristoyl tug in GCAP1. This adds a new distinct example to the array of diverse molecular mechanisms through which retinal guanylyl cyclase can become dysregulated by GCAP1 and cause dominant retinopathy.

Competition of the G86R GCAP1 with RD3 becomes more effective
RD3 protein (68, 69) is required for effective accumulation of RetGC in the outer segment (70,71) and also inhibits RetGC

cGMP synthesis and photoreceptor blindness
activation by GCAPs (55)(56). The inhibitory function of RD3, which likely occurs in the inner segment (71), is essential for the survival of photoreceptors, because rods and cones lacking the RD3-dependent inhibition of RetGC degenerate much faster and more severely than those completely lacking RetGC itself (56). The cyclase activated by G86R GCAP1 resists the inhibition by RD3 more effectively (Fig. 8), because the mutant GCAP1 now has higher affinity for the cyclase (Fig. 3). A similar effect was previously observed in the case of CORD6-linked mutation in Arg 838 of RetGC1, which increases RetGC affinity for GCAP1 (18,80). According to a hypothesis that is currently under investigation, inhibition of RetGC1 by RD3 is required for preventing premature activation of the cyclase by GCAPs, whereas in transit from the inner segment (55,56,69). It is conceivable that the G86R GCAP1 outcompetes RD3 and   , concentration of RD3 in assay, [RD3] 1/2 , concentration of RD3 causing 50% inhibition, and H, Hill coefficient. B, the fractional inhibition of the RetGC1-GCAP1 complex by RD3. The data from A were normalized per maximal activity of the cyclase in each case. Note a nearly 10-fold, increase in [RD3] 1/2 for G86R GCAP1. The overall higher activity of the cyclase in the presence of the G86R GCAP1 at the 1.5 M GCAP1 is due to the higher than WT affinity of the mutant for the cyclase (see Fig. 4).

cGMP synthesis and photoreceptor blindness
therefore prematurely activates the cyclase while in the photoreceptor inner segment. This, in addition to the abnormal cGMP production in the outer segment, would lead to aberrant cGMP production in the inner segment and aggravate the progression of the G86R GCAP1-produced retinopathy.

Clinical studies
Human studies-A family with a multigeneration history of visual loss suggesting an autosomal dominant inheritance was studied. The proband was examined with clinical, electrophysiological, psychophysical, and imaging tests, and records were obtained from other affected members. Informed consent was obtained and procedures followed the Declaration of Helsinki guidelines and were approved by the institutional review board.
Electroretinography (ERG)-Rod, mixed rod-cone, and cone full-field ERGs were performed according to published protocols (72). In brief, bipolar Burian-Allen contact lens electrodes were used with an Espion system (Diagnosys, Lowell, MA). Under dark-adapted conditions, rod photoreceptor driven function was probed with dim blue Ϫ1.6 log scot-cd s m Ϫ2 flashes, and combined output of rod and cone photoreceptors was probed with white ϩ1.2 log scot-cd s m Ϫ2 flashes. Under light-adapted conditions, cone photoreceptor function was isolated with white 0.8 log phot-cd s m Ϫ2 flashes at stimulation rates of 1 and 30 Hz using white adapting backgrounds of 1.5 and 0.8 log phot-cd m Ϫ2 , respectively.
Psychophysical testing-Kinetic visual fields were performed with a Goldmann perimeter using two target sizes (V-4e, 1.72°a nd I-4e, 0.11°). Imaging-En face images were obtained using a confocal scanning laser ophthalmoscope (Spectralis HRA, Heidelberg Engineering, Heidelberg, Germany) to determine RPE integrity. Near-IR reduced-illuminance autofluorescence images (NIR-RAFI) and short-wavelength reduced-illuminance autofluorescence imaging (SW-RAFI) were acquired using methods previously reported (73,74). For NIR-RAFI acquisition, excitation was with 790-nm NIR light (100% laser power) and the detector sensitivity was held invariant at 105%; for SW-RAFI, excitation at 488 nm was used at 25% laser power and 105% detector sensitivity. The automatic normalization feature was turned off and imaging was obtained using the automatic real time feature, which averages multiple frames to improve the signal to noise ratio. Cross-sectional retinal imaging was performed with optical coherence tomography (RTVue-100; Optovue Inc., Fremont, CA, and ultrahigh resolution SDOCT Bi-; Kowa Company, Ltd., Tokyo, Japan). Overlapping horizontal line scans were used to create a profile along the horizontal meridian covering eccentricities up to 30 o in temporal and nasal directions. Segmentation analysis was performed using custom programs (Matlab 9.1; MathWork, Natick, MA) based on signal feature assignments as previously published (75).

GCAP1 expression and purification
Two orthologs of myristoylated GCAP1 for biochemical and Trp fluorescence analysis, a bovine (D6S variant) and a human (E6S variant), were expressed from pET11d vector (Novagen/ Calbiochem) in a BL21(DE3)pLysS Escherichia coli strain (Novagen/Calbiochem) harboring a pBB131 plasmid coding for yeast N-myristoyltransferase and purified using previously published procedures (51) modified as follows. Cells were typically grown in 2.0 liters of a standard LB medium (Thermo Fisher Scientific) containing 50 g/ml of kanamycin and 100 g/ml of ampicillin to reach A 600 0.6 -0.7. Free myristic acid (Sigma) was added from a concentrated ethanol solution to the suspension of bacterial cells to a final concentration of 100 g/ml, 30 min prior to the induction with 0.5 mM isopropyl ␤-D-thiogalactopyranoside (Research Products International). Three hours after the induction, the bacterial pellet was harvested by centrifugation at 8,000 ϫ g for 20 min at 4°C and frozen in Ϫ70°C. The thawed pellet was resuspended in 100 ml of 10 mM Tris-HCl (pH 7.5) containing 2 mM EGTA and 14 mM 2-mercaptoethanol, and the cells were disrupted by ultrasonication. The expressed GCAP1 in the insoluble fraction of the inclusion bodies was collected by centrifugation at 20,000 ϫ g for 20 min, 4°C, extracted from the pellet by homogenization in 30 mM Tris-HCl (pH 7.5) containing 2 mM EGTA, 14 mM 2-mercaptoethanol, 2 mM MgCl 2 , and 8 M Sigma Ultra urea for 30 min at 4°C, and first dialyzed at 4°C for 3-4 h against 2.0 liters of 10 mM Tris-HCl buffer (pH 7.5) containing 0.5 mM EGTA, 2 mM MgCl 2 , and 14 mM 2-mercaptoethanol, and then overnight against 2.0 liters of 10 mM Tris-HCl buffer (pH 7.5) containing 0.1 mM EGTA, 2 mM MgCl 2 , and 14 mM 2-mercaptoethanol. The insoluble material was removed by centrifugation at 20,000 ϫ g for 20 min, 4°C. The concentration of Tris-HCl buffer (pH 7.5) in the supernatant was adjusted to 50 mM and CaCl 2 was added to a final concentration of 10 mM and kept for 20 min at room temperature. The precipitate was removed by centrifugation at 20,000 ϫ g for 20 min, 4°C. Supernatant was collected and, after adding NaCl to 1 M and DTT to 5 mM, applied on a 1.6 ϫ 5.0-cm butyl-Sepharose Fast Flow column (GE Healthcare) pre-equilibrated with 20 mM Tris-HCl (pH 7.5) containing 1.0 M NaCl. The column was washed with ϳ10 volumes of the same buffer and GCAP1 was eluted with 5 mM Tris-HCl (pH 7.5) and concentrated to 5 ml using a Amicon Ultra-15 (10,000 MWCO) centrifugal filter (Thermo Fisher Scientific). Concentrated solution was centrifuged at 200,000 ϫ g for 10 min, 4°C, in a Beckman Optima TLX centrifuge and chromatographed on a GE Healthcare Sephacryl S-100 column (2.6 ϫ 60 cm) pre-equilibrated with 20 mM Tris-HCl (pH 7.5), 100 mM NaCl. The main peak containing GCAP1 was collected and EDTA was added to 2 mM to remove Ca 2ϩ bound to GCAP1. We observed that using Chelex resin was not sufficient to remove all bound Ca 2ϩ from GCAP1, whereas preincubation with EDTA yielded metal-free GCAP1. The excess EDTA was then removed by 4 cycles of 20-fold concentration/dilution in 10 mM Tris-HCl (pH 7.5) containing 30 M EDTA using Amicon Ultra-15 (10,000 MWCO) centrifugal filter. We observed that even in solutions prepared using reagents containing Ͻ5 ppb Ca 2ϩ and 18 megohms cm water, Ca 2ϩ contaminations could reach as high as 1 M. Because GCAP1 has high affinity for Ca 2ϩ , even these low levels of contamination are sufficient to skew measurement of the Ca 2ϩ binding. Hence, the 30 M EDTA was used to prevent GCAP1 from rebinding Ca 2ϩ from the solutions during the concentration/dilution cycles. The cGMP synthesis and photoreceptor blindness final concentration of GCAP1 in stock solution was typically 300 -350 M (Ͼ10-fold higher than that of EDTA), and the corrections for the presence of the trace amounts of EDTA were made in all subsequent experiments. Concentrated protein was frozen in small aliquots and stored at Ϫ70°C. The purity of GCAP1 preparations estimated by SDS gel electrophoresis was Ն95%.

RetGC1 expression and activity assay
Human recombinant RetGC1 was expressed from a modified Invitrogen pRCCMV vector in HEK293 cells transfected using a calcium-phosphate precipitation method and the membrane fraction containing the expressed cyclase was purified as previously described (48). The guanylyl cyclase activity was assayed as previously described in detail (48,56). Briefly, the assay mixture (25 l) containing HEK293 membranes, 30 mM MOPS-KOH (pH 7.2), 60 mM KCl, 4 mM NaCl, 1 mM DTT, 2 mM Ca 2ϩ /Mg 2ϩ /EGTA buffers, 0.9 mM free Mg 2ϩ , 0.3 mM ATP, 4 mM cGMP, 1 mM GTP, and 1 Ci of [␣-32 P]GTP, 100 M zaprinast and dipyridamole, and 10 mM creatine phosphate, 0.5 unit of creatine phosphokinase (Sigma) was incubated at 30°C for 30 min and the reaction was stopped by heat inactivation at 95°C for 2 min. The resultant [ 32 P]cGMP product was separated by TLC using fluorescently-backed polyethyleneiminecellulose plates (Merck) developed in 0.2 M LiCl, eluted with 2 M LiCl, and the radioactivity was counted using liquid scintillation. Ca 2ϩ /EGTA buffers maintaining variable-free Ca 2ϩ concentrations at 0.9 mM physiological for the photoreceptors (52) free Mg 2ϩ were prepared using Tsien and Pozzan method (76) and verified by fluorescent indicator dyes as previously described in detail (51). Data fit and statistical analysis (Student's t test) was performed using Synergy Kaleidagraph software.

Protein fluorescence spectroscopy
The intrinsic Trp fluorescence of GCAP1 was assayed as previously described in detail (51). In brief, a 332-nm Trp fluorescence emission intensity of GCAP1 in solution containing 100 mM MOPS/KOH (pH 7.2), 40 mM KCl, 1 mM EGTA, and specified concentrations of MgCl 2 was recorded at 23°C ( ex ϭ 290 nm). Small aliquots of concentrated CaCl 2 were added to obtain the required free Ca 2ϩ concentrations calculated according to the method of Brooks and Stoney (78), utilizing the algorithm of Marks and Maxfield (79). Data fit and statistical analysis (Student's t test) was performed using Synergy Kaleidagraph software.

GCAP1 mutagenesis
The mutations were introduced into GCAP1 cDNA by PCR following conventional "splicing-by-overlap extension" procedure utilizing a high-fidelity Thermo Scientific PhusionFlash polymerase. The mutated cDNA was inserted into the NcoI/ BamHI sites of a pET11d vector, downstream from T7 promoter, as described previously (46) and verified by automated Sanger sequencing.

RD3 expression and purification
Recombinant human RD3 was expressed from a pET11d vector in a BL21(DE3)Codon Plus E. coli strain (Stratagene/Agilent Technologies) induced by isopropyl ␤-D-thiogalactopyranoside, extracted from the inclusion bodies and purified by salt precipitation and dialysis as previously described in detail (55,56).

ITC experiments
ITC experiments with hGCAP1 and G86R mutant were performed on a VP-ITC from MicroCal (Northhampton, MA). Briefly, purified myristoylated calcium-free (no EGTA) WT GCAP1 and G86R mutant were present in the recording cell in titration buffer (20 mM Hepes, pH 7.4, 60 mM KCl, 4 mM NaCl) at 24 M and were titrated with 5 l of 0.5 mM CaCl 2 stock solution at T ϭ 25°C (50 injections, each 5-l). The titration buffer was decalcified using a self-packed gravity flow Chelex 100 column (Bio-Rad). The remaining Ca 2ϩ concentration was determined by a BAPTA absorbance assay and was found to range between 90 nM in the presence of 24 M WT and 200 nM in the presence of 30 M G86R mutant (no EGTA used). All buffers were filtered (0.22 m) and degassed twice immediately before use. Three independent repetitions were made for each titration set. Protein samples for each repetition were obtained from two to three separate expressions and three separate purification performances. Reference injections of Ca 2ϩ into decalcified buffer were performed without any protein, and the reference was subtracted in each experiment. Each titration was analyzed by applying a model implemented in the Origin software (MicroCal) assuming two Ca 2ϩ -binding sites, or two Mg 2ϩ -binding sites. The best fitting results were used to obtain dissociation constants K D,app and enthalpy changes (⌬H).