Materials

Tris(hydroxymethyl)aminomethane (Tris), Guanidine-HCl, NaCl, KCl, CaCl2, MgCl2, DTT, EGTA, β-mercaptoethanol, NH4HCO3, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) (HEPES), Ames’ medium, ethanolamine, phosphatidylethanolamine, phosphatidylcholine, phosphatidylserine, cholesterol, acrylamide, Coomassie blue, cGMP, polyethyleneimine, sucrose, OCT, NH4Cl, citric acid, Triton X-100, Tween 20, Bovine Serum Albumin, chloramphenicol, cOmplete EDTA-free Protease Inhibitor Cocktail, paraformaldehyde, ketamine, xylazine, atropine, hydrocortisone and BaCl2 were purchased from Merck (Darmstadt, Germany).

DMEM, OptiMEM, penicillin, streptomycin, 2-(4-amidinophenyl)-1H-indole-6-carboxamidine (DAPI), Phosphate Saline Buffer (PBS), Fetal Bovine Serum (FBS), HBSS, glutamine, Normal Goat Serum, Normal Donkey Serum were purchased from ThermoFisher Scientific (Waltham, MA, USA).

Cloning, expression, and purification of GCAP1 variants

Human myristoylated WT-GCAP1 was expressed in E. coli BL21 (DE3) after co-transformation with pBB131 containing the cDNA of S. cerevisiae N-myristoyl transferase (yNMT) [62]. The cDNA for E111V variant was obtained by PCR using QuikChange II Site-Directed Mutagenesis kit (Agilent, Milan, Italy) as described in Ref [24], while the cDNA for His-tagged WT-GCAP1 (GCAP1His) was purchased from Genscript. Both variants were expressed and purified following the same protocol as for the WT [33], briefly consisting of: (i) denaturation of inclusion bodies with 6 M Guanidine-HCl; (ii) refolding by dialysis against 20 mM Tris–HCl pH 7.5, 150 mM NaCl, 7.2 mM β-mercaptoethanol, and a combination of (iii) Size Exclusion Chromatography (SEC, HiPrep 26/60 Sephacryl S-200 HR, GE Healthcare, Chicago, IL, USA) and (iv) Anionic Exchange Chromatography (AEC, HiPrep Q HP 16/10, GE Healthcare, Chicago, IL, USA). The purity of GCAP1 variants was assessed by 15% acrylamide SDS-PAGE, samples were either exchanged against PBS, aliquoted and frozen with liquid nitrogen, or exchanged against NH4HCO3, aliquoted and lyophilized. Protein samples were finally stored at -80 °C.

The three-dimensional structure of human GCAP1 was obtained by homology modeling using the structure of Ca2+-loaded chicken GCAP1 [63] following the procedure illustrated in Ref [18]. In silico mutagenesis of E111V variant was obtained according to the protocol detailed in Ref [24]. The structures presented in Fig. 1a and b were extracted from the last frame of 200 ns Molecular Dynamics simulations from Ref [24], whose settings and protocols for energy minimization, equilibration and production phases were elucidated in Refs [13, 15].

Electrophoretic mobility shift assay

WT-GCAP1 and E111V-GCAP1 were dissolved in 20 mM Tris–HCl pH 7.5, 150 mM KCl, 1 mM DTT at a concentration of 30 µM, incubated for 5 min at 25 °C with either 1 mM EGTA + 1.1 mM Mg2+ or 1 mM Mg2+ + 1 mM Ca2+, boiled, and run for 50 min at 200 V on a 15% acrylamide gel under denaturing conditions. Finally, protein bands were visualized by Coomassie blue staining.

Circular dichroism (CD) spectroscopy

The effects of ion binding and of the E111V substitution on the secondary and tertiary structure of GCAP1 were evaluated by CD spectroscopy using a J-710 spectropolarimeter (Jasco, Cremella, Italy) thermostated by a Peltier-type cell holder. Lyophilized proteins were dissolved in PBS pH 7.4 buffer at a concentration of 35 and 10 µM for near UV and far UV spectra, respectively. Five accumulations of each spectrum were recorded at 25 °C in the absence of ions (500 µM EGTA for near UV, 300 µM for far UV) and after serial additions of 1 mM Mg2+ and Ca2+ (1 mM for near UV, 600 µM for far UV, leading to a free Ca2+ concentration of 500 and 300 µM, respectively). All spectra were subtracted with that of the buffer, near UV spectra were also zeroed by subtracting the average ellipticity between 310 and 320 nm, where no signal was expected.

Dynamic light scattering (DLS)

The hydrodynamic diameter of Ca2+-loaded WT-GCAP1 and E111V-GCAP1 was estimated by DLS using a Zetasizer Nano-S (Malvern Instruments, Malvern, UK). Proteins were dissolved in PBS pH 7.4 at 42 µM concentration and filtered with a Whatman Anotop 10 filter (20 nm cutoff, GE Healthcare, Chicago, IL, USA) before starting the measurements. Samples were equilibrated for 2 min at 25 °C and for each variant at least 100 measurements were collected, each consisting of 13 runs.

Guanylate cyclase activity assay

GC1 enzymatic activity as a function of Ca2+ and GCAP1 concentration was measured after reconstituting WT-GCAP1 and E111V-GCAP1 with cell membranes of mGFP-GC1 cells (see below) previously extracted by lysis (10 mM HEPES pH 7.4, Protease Inhibitor Cocktail 1×, 1 mM DTT buffer) and 20 min centrifugation at 18,000×g, as previously described [30, 64, 65]. Cell membranes were resuspended in 50 mM HEPES pH 7.4, 50 mM KCl, 20 mM NaCl, 1 mM DTT and incubated with 5 µM GCAP1 variants at increasing [Ca2+] (< 19 nM to 1 mM, controlled by Ca2+-EGTA buffer solutions [66]) to estimate the Ca2+ concentration at which cGMP synthesis by GC1 was half-maximal (IC50). To estimate the GCAP1 concentration at which GC1 activation was half-maximal (EC50), cell membranes were reconstituted with increasing amounts of each GCAP1 variant (0–20 µM) at low Ca2+ (< 19 nM). Reported IC50 and EC50 values are represented as average ± standard deviation of 3 technical replicates. The statistical significance of the differences in IC50 and EC50 between WT-GCAP1 and E111V-GCAP1 was evaluated by means of two-tailed t tests (p value = 0.05).

Conjugation of CF640R-N-hydroxysuccinimide (NHS) ester with WT-GCAP1

Far-red fluorescent dye CF640R (Biotium, Fremont, CA, USA) was conjugated via NHS to WT-GCAP1 primary amines (Lys residues, Movie S1) according to the manufacturer protocol. Briefly, GCAP1 was diluted in PBS pH 7.4 and 1 mM DTT to a final concentration of 76 µM in a final volume of 900 µl; then the solution was added with 100 µl sodium bicarbonate 1 M pH 8.3 and 2 CF640R-NHS aliquots previously resuspended in 50 µl total DMSO. The mixture was then wrapped in aluminum and incubated in rotation at RT for 1 h. Unconjugated dye was removed by washing 4 times the protein solution (see Fig. S1b for representative spectra of the 4 flowthrough) with PBS pH 7.4 for 10 min at 4400 × g and 4 °C using an Amicon Ultra-4 concentrator with 3 kDa cutoff (Merck Millipore, Burlington, MA, USA). The degree of labelling (DOL = 1.96) was calculated as the ratio between the concentration of dye in the protein solution measured based on the absorbance at 642 nm (ε = 105.000 cm−1 M−1), and the concentration of protein calculated by considering the dilution factor and the retention of Amicon concentrators (95%, according to manufacturer instructions). The concentration of free-CF640R in the protein solution was calculated by measuring the absorbance at 642 nm of wash 4, which was < 1% with respect to protein concentration in all conjugation experiments. Unconjugated dye was blocked with 50 µl ethanolamine 1 M.

Fluorescence spectroscopy

The emission fluorescence spectrum of 2 µM GCAP1CF640R (645–680 nm) dissolved in PBS pH 7.4 was collected at 25 °C on a FP-750 spectrofluorometer (Jasco, Cremella, Italy) after excitation at 639 nm; the spectrum reported in Fig. 1h is an average of 3 accumulations after subtraction of the emission spectrum of the buffer in the same range.

Liposome preparation

LPs were prepared by hydrating a thin lipid film of the same composition as photoreceptors rod outer segment membranes [67] (phosphatidylethanolamine, phosphatidylcholine, phosphatidylserine, and cholesterol at a molar ratio of 40:40:15:5) previously mixed in chloroform and dried in a speed-vac concentrator. Four mg of lipid film were hydrated with 1 ml PBS pH 7.4, vortexed for 30 min at room temperature, sonicated for 15 min in a water bath on ice and extruded 20 times through a 200 nm polycarbonate filter (Whatman, Maidstone, UK). The encapsulation of CF640R, WT-GCAP1, E111V-GCAP1, GCAP1His, or GCAP1CF640R in LPs was achieved by dissolving the molecule to be loaded in PBS before lipid film hydration. Unencapsulated molecules were removed by washing at least 4 times the LPs suspensions with PBS pH 7.4 for 20 min at 4 °C and 5000×g using an Amicon Ultra-4 concentrator with 100 kDa cutoff (Merck Millipore, Burlington, MA, USA). The degree of encapsulation was calculated by subtracting from the total mass of the molecule to be encapsulated that present in the flow-through and was found to be higher than 75% in all LP preparations. The efficient separation of non-encapsulated proteins was assessed by measuring protein concentration of the flowthrough of the 4 washing steps, similarly to what was done for CF640R. The concentration of non-encapsulated protein in LP suspensions was estimated from the concentration of protein in the last washing step and was found to be < 7% of the encapsulated protein.

Nanoparticle tracking analysis (NTA)

The concentration and size of LP suspensions were measured at 25 °C by means of NTA on a NanoSight (Malvern Instruments, Malvern, UK) by recording 3 videos of 1 min each at 25 fps by setting 20 µl/min flow rate; camera level and detection threshold were automatically optimized for each measurement to maximize the signal-to-noise ratio. LP size reported in Fig. S2 and LP concentration reported in Table S1 represent the average ± standard error of 3 technical replicates.

Fluorescence imaging of gel-immobilized liposomes

Stock suspensions of LPs, either filled with free-CF640R or empty, were diluted 1:400 v/v in 0.5% low gelling temperature agarose in Ames’ medium at 37 °C. A thin film was polymerized over a pure agarose meniscus in a Petri dish and covered with Ames’ medium. 3D image stacks were acquired with a 63x/0.9NA water immersion objective and a CCD camera (DFC350 FX, Leica Microsystems, Milan, Italy) in an upright widefield fluorescence microscope (DM LFSA, Leica Microsystems, Milan, Italy) using a Cy5 filterset (49,006; Chroma, Olching, Germany). Stacks were deconvolved and max projected along the z-axis using Fiji/ImageJ as detailed in Ref [68].

Generation of cGFP-GC1 and mGFP-GC1 stable HEK293 cell lines

HEK293 cells were cultured in DMEM medium supplemented with fetal bovine serum (10%, v/v), penicillin (100 units/ml) and streptomycin (100 μg/ml) at 37 °C in humidified atmosphere with 5% CO2. Cells (6.25 × 105) were seeded in 6-well plates in DMEM medium and grown overnight; the next day cell medium was replaced with OptiMEM reduced serum medium and cells were transfected using polyethyleneimine (PEI) as transfection reagent and 2 different vectors to obtain eGFP-expressing stable cell lines: (i) pIRES encoding for eGFP and human GC1 under the same promoter, thus resulting in a cytosolic fluorescence (cGFP), and (ii) pcDNA3.1 + N-eGFP encoding for GC1-eGFP fusion protein for localizing fluorescence on the membrane (mGFP). DNA (2.5 µg) was mixed dropwise to 10 µl PEI solution at a concentration of 1 µg/µl (DNA:PEI ratio of 1:5 w/w), added dropwise to 500 µl of pre-warmed OptiMEM, mixed and incubated 30 min at room temperature to allow DNA-PEI polyplex formation. Polyplexes were finally added dropwise to each well and the plate was incubated overnight at 37 °C and 5% CO2. The next day, OptiMEM medium was replaced with DMEM and 48 h after transfection eGFP positive cells were selected using geneticin (500 µg/ml).

Live-cell imaging

Cells (8 × 104) were seeded in 4-well chambers (Ibidi, Graefelfing, Germany) in DMEM medium; two days later the medium was replaced with OptiMEM reduced serum medium, then cells were incubated with 100 µl LP suspension per well (containing each ~ 0.4 mg lipid) and monitored in live-cell imaging. Experiments with fluorescently labelled GCAP1CF640R were performed taking care of incubating the cells with the same nominal concentration of protein encapsulated in the LP aqueous core.

Live-cell imaging was performed using TCS-SP5 Inverted Confocal Microscope (Leica Microsystems, Milan, Italy) equipped with temperature and CO2 controller and motorized stage that provides precise and automated acquisition of multiple fields of view. Images were collected simultaneously on different points of the sample immediately after cell-LP incubation and at 30 min interval for 24 h or 48 h total acquisition time. Images were captured after 488 nm and 633 nm laser excitation with a 40× objective (1.2 NA oil immersion) and further analyzed by Imaris 9.8 software (Oxford Instruments, Abingdon-on-Thames, UK). The fluorescence intensity profiles of mGFP and LP-GCAP1CF640R reported in Fig. S4 were collected along the line across the cell shown in the insets using ImageJ.

Fluorescence microscopy of mouse retinas following ex vivo incubation

All animal experiments made use of adult C57Bl/6 J mice of both sexes. These were reared at around 22 °C in small groups with the addition of environmental enrichment items, a 12 h day/12 h night cycle, ad libitum food and water. As in previous studies by our group, and in accordance with authorized protocols, dark adapted mice were deeply anesthetized with ketamine (80 mg/kg) + xylazine (5 mg/kg) and their retinas extracted through a corneal incision in room temperature Ames' medium under dim red light. This approach avoided even brief exposure of the tissue to anoxic conditions, which could affect protein and/or liposome uptake. Animals were then immediately sacrificed with an overdose of anesthetic. After removing the vitreous each retina was placed, freely floating, in a plastic well containing incubation solution (1–2 ml depending on the experiment), and the wells inserted in an airtight box with a water layer at the bottom and a 95%O2/5%CO2 atmosphere. Incubation solutions consisted in the test suspension/solution diluted in Ames’ medium, taking care of reaching virtually the same final concentration for each suspension. The box was left floating in a water bath at 37 °C. After the prescribed time the retinas were returned to room temperature Ames' medium, made to adhere to black filter paper (AABP02500; Merck, Burlington, MA, USA) with gentle suction and, optionally, sliced at 250 µm thickness with a manual tissue chopper. Image stacks were acquired as described for the imaging of gel-immobilized LPs, with 4x/0.1NA air, 20x/0.5NA and 40x/0.8NA water immersion objectives. Excitation was provided by an Hg lamp preheated to achieve stable output. Stacks were lightly deconvolved (Richardson-Lucy algorithm, 10 iterations) and a single image obtained by averaging along the z-axis a few adjacent slices of the stack, in all cases chosen well below the cut surface. The borders of retinal layers were identified by imaging the same tissue volume in the near IR (Fig. S9). Cones were identified based on their characteristic location and morphology, leveraging our experience with their intracellular staining. Identical acquisition parameters were used when comparing retinas treated with different incubation solutions.

Immunofluorescence experiments with mouse retinas following ex vivo incubation

Mice were anesthetized with isoflurane, euthanized via cervical dislocation and their retinas extracted through a corneal incision in room temperature DMEM medium supplemented with FBS (25%, v/v), HBSS (25% v/v), glutamine (1% v/v), penicillin (100 units/ml) and streptomycin (100 μg/ml). After 30 min incubation at 37 °C and 5% CO2, tissues were incubated with 180 µl PBS, 180 µl of 100 µM free-GCAP1His or 180 µl of 4.5 nM LP-GCAP1His (containing the same number of GCAP1His molecules in the aqueous core as compared to the free protein solution case) for 30 min, 4 h 30 min, and 24 h, and finally washed 3 times with PBS.

Retina sections were then fixed for 40 min in 10% formalin in PBS buffer, washed 3 times with PBS, incubated with 10%, 20% and 30% sucrose for 1 h each at RT, and kept overnight at 4 °C. The next day samples were incubated at RT for 1 h with OCT compound: 30% sucrose at a 1:1 ratio and processed for cryo-sectioning at – 14 °C.

Sections (14 µm thickness) were fixed for 5 min with paraformaldehyde, washed 3 times with PBS, incubated with 0.1% Triton X-100 in PBS for 1 h at RT, washed 3 times with PBS, incubated with ammonium chloride for 20 min, and washed 5 times with PBS.

Sections were incubated overnight at RT with mouse anti-His primary antibody (1:1000 dilution, SouthernBiotech, Birmingham, AL, USA) and PNA (1:250 dilution, Molecular Probes, Eugene, OR, USA) in blocking solution (5% Normal Goat Serum, 1% Bovine Serum Albumin, 0.3% Triton X-100 in PBS). The following day samples were washed 3 times with PBS and incubated with an Alexa Fluor 647-conjugated goat anti-mouse secondary antibody (1:1000 dilution, Invitrogen, Waltham, MA, USA). Cell nuclei were stained with a 1:1000 DAPI dilution in PBS; slides were coverslipped with Dako fluorescence mounting medium (Agilent, Milan, Italy). Sections were visualized using TCS-SP5 Inverted Confocal Microscope (Leica Microsystems, Milan, Italy), images were captured after 405 nm and 633 nm laser excitation with a 63× objective (1.2 NA oil immersion) and further analyzed by Imaris 9.8 software (Oxford Instruments, Abingdon-on-Thames, UK).

Intravitreal injections

Mice were first anesthetized with ketamine (80 mg/kg) + xylazine (5 mg/kg), followed by application of eyedrops containing atropine and chloramphenicol (1%) + hydrocortisone (0.5%). Intravitreal injections were made under a stereomicroscope and dim blue light as follows: (i) a hole was made in the cornea near the ora serrata with the tip of a 31G insulin needle; (ii) glass micropipettes with a broken tip, connected to a 25 µl syringe (Hamilton, Reno, NV, USA) via PE tubing filled with mineral oil (330,779; Merck, Burlington, MA, USA), were front loaded with 2 µl of solution; (iii) the micropipette was inserted in the hole and the entire volume slowly injected in the vitreous. Mice were returned to their cages and allowed to recover in a paper blanket. After 20–24 h we performed retinal dissection, slicing and imaging as described for ex vivo incubations.

Long duration ex vivo ERG recordings

ERG experiments were made in a custom designed incubation and recording chamber [38]. Retina pairs were isolated as described for ex vivo incubations, made to adhere to white filter paper (SMWP02500; Merck, Burlington, MA, USA) and placed at the bottom of two adjacent plastic wells, containing 2 ml/retina of 40 µM AP4 (0101; Tocris, Milan, Italy) in Ames' medium. Retinas were centered on a hole leading to the anode, while the cathode was in the chamber itself. In some experiments we dispensed with the filter paper and used instead small transparent cups to immobilize the retinas (Fig. 1b in Ref [38]). Both electrodes were silver chloride wires inserted in an agar bridge. The well assembly was placed on an aluminum platform covered with a layer of water, inside a sealed incubation chamber purged with 95% O2/5% CO2. The temperature of the platform was actively controlled with a custom apparatus [69]. Small diameter PTFE tubing, leading from inside the wells to syringes residing outside the chamber, allowed injection and mixing of test solutions (100 µl/retina) into the wells during the recordings with minimal perturbation. Immediately above the wells, attached to the lid of the chamber, a LED (505 nm; ND filters) delivered the same flash sequence every 15 or 30 min: (ph/µm2|no. of flash repetitions) 3.98|12, 8.27|10, 18.9|8, 50.5|6, 151|6, 510|4, 1660|3. Transretinal potentials were amplified by 5000, filtered in the band DC-100 Hz, digitized at 5 kHz and acquired with pClamp 9 (Molecular Devices, San Jose, CA, USA). Electrophysiological records were analyzed in Axograph X with automated custom scripts. i50 was determined by fitting a Hill function to a plot of response amplitudes measured 90–130 ms after the flash (Fig. S7). This range minimized the contribution of the very slow glial response and gave parameter values close to those in BaCl2 [38]. TTP@i50 was estimated as the weighted average of the TTPs of the two flash responses straddling i50 (10 Hz Gaussian filtered records). Two rounds of normalization were applied to these raw data, as follows. We assumed that the two retinas, being from the same animal, behaved identically except for (i) an initial stabilization phase due to slight variations in their isolation and manipulation, and (ii) a scaling factor in their steady state light sensitivity due to small differences in their orientation in the recording chamber. We first normalized the two sets of raw values over their respective pre-treatment levels. We then removed any trends common to both retinas by dividing the normalized values of the treated retina by those of the control. We were thus left with a single time series that shows the net effect of the tested drug (Fig. 6, red lines). Treated and control retina positions in the wells were alternated from animal to animal to cancel out any environmental biases. In a limited number of tests (Fig. 6h) BaCl2 was injected in the wells using the syringe system after an initial stabilization period and stirred to obtain a final concentration of 50 µM. In a subset of the experiments, we included HyClone PenStrep (SV30010; Cytiva, Breisgau, Germany) in the incubation medium at 1% vol/vol, which enabled us to prolong our recordings up to 18 h.

Statistics

Statistical analyses were performed with the open-source software JASP 0.16 (jasp-stats.org; RRID:SCR_015823), Kaleidagraph 5 and Excel (Microsoft, Redmond, WA, USA). For ERG recordings visual representation of the population effect of a tested drug (Fig. 6) were given by the confidence interval of the Hodges-Lehmann estimator [38]. Statistical significance was estimated by the following parametric and non-parametric tests as mentioned in the text: two tailed t-test, paired Wilcoxon signed-rank, one sample Wilcoxon signed-rank.