Originally, in ANF-RGC, the entire stretch beyond the “KHD” was termed as CCD (Chang et al., 1989; Chinkers and Garbers, 1989; Chinkers et al., 1989; Lowe et al., 1989; Duda et al., 1991). However, discovery of the ROS-GC subfamily changed this picture. There, CCD was followed by an extra C-Terminal Extension (CTE) sequence (Goraczniak et al., 1994) where the targeted sites of its two Ca²⁺ sensors, GCAP2 and S100B, were present (Duda et al., 2002, 2005). Thus, in the structural paradigm shift the CCD of ROS-GC is sandwiched between the N-terminal SHD and C-terminal CTE domains (Figure 1C); in this environment, CCD is able to transduce three types of phototransduction linked [Ca²⁺] signals: one, generated upstream by GCAP1, second and third generated downstream by GCAP2 and S100B (Figure 1; Duda et al., 2016). With an additional clue that within the CCD resides the targeted site of another Ca²⁺ sensor, neurocalcin δ (NCδ) (Kumar et al., 1999; Venkataraman et al., 2008) (Figure 1B), the signal transduction role of CCD expanded. CCD was no longer merely a translational center of all effector signals, it was also the regulatory element, meaning, it is a complex Ca²⁺-signaling translating element.

Guided by these cues, the boundary and the newer functional roles of CCD were reexamined. The recombinant ROS-GC1 fragment, M⁷³³-K¹⁰⁵⁴ was chosen for detailed analysis (Venkataraman et al., 2008). The key findings were that: (i) CCD represents the G⁸¹⁷-Y⁹⁶⁵ segment of the guanylate cyclase; (ii) the segment is homo-dimeric in nature; and (iii) it contains an ⁸⁴⁴MSEPIE⁸⁴⁹ NCδ regulatory motif. Incorporation of these principles in the fold recognition model of the dimeric form of CCD disclosed that CCD consists of eight β strands and six α-helices (Venkataraman et al., 2008). Its prominent feature is that the two CCD chains are antiparallel, a feature later confirmed experimentally (Duda et al., 2012b). In general, it supported its previously proposed homology-based three-dimensional CCD 1AWL model (Liu et al., 1997), yet it incorporated a significant advancement. An additional seven-residue ⁹¹¹TFRMRHM⁹¹⁷ helix motif in CCD structure was present which represented the docking site, ⁸³⁶V-L⁸⁵⁷ (previously mis-numbered as ⁸³⁷V-L⁸⁵⁸) between ROS-GC1 and NCδ (Venkataraman et al., 2008).

As of to-date, no crystal structure of any MGCs CCD module exists. However, subsequent to (Venkataraman et al., 2008) publication, crystal structures of CCD of the two forms of guanylate cyclases have been solved, eukaryotic (Cyg12) unicellular green algae Chlamydomonas reinhardtii (Winger et al., 2008) and Cya2 cyanobacterium Synechocystis (Rauch et al., 2008). Cyg12 represents atypical soluble and Cya2 the bacterium MGC.

With the model system of the recombinant ROS-GC1 the present study decodes the precise structure of its CCD, elucidates its biochemical principles at the sub-molecular level, through experimentation validates them for its regulation by Ca²⁺ sensors GCAP1, GCAP2 and S100B and bicarbonate operative in phototransduction, and finally, proposes their application to the general MGC family.

Three-dimensional model of ROS-GC1 CCD monomer was built using structural information on eukaryotic soluble guanylate cyclase (Cyg12) CCD of the green algae Chlamydomonas reinhardtii (Winger et al., 2008) as a template, UniProt entry P55203 with A1013R as the query sequence, and Iterative Threading and ASSEmbly Refinement, I-TASSER (web server version)¹. The top-3 unique templates identified by I-TASSER were (PDB IDs), 3uvj_A, 3et6_A and 4p2f_A. Note that 3uvj_A denotes PDBID_ChainName. Based on the secondary structure predictions and I-TASSER C-score, the top-ranked model for CCD was chosen as a representative structure referred to it as the default model for CCD. Two copies of I-TASSER built CCD monomer models were structurally aligned with the experimental soluble guanylate cyclase dimer structure, PDI ID: 3et6 (Winger et al., 2008) to create a homo-dimer ROS-GC1-CCD model. FATCAT² method was used for the structural alignments and creating the dimer model of CCD. The details of the modeling are provided in the Supplemental Materials.

(1) Point mutations for the creation of the D⁸³⁴A, E⁸⁷⁴A, D⁸⁷⁸A, R⁹²⁵A, C⁹⁴⁶A, N⁹⁵³A, R⁹⁵⁷A, and E⁸⁷⁴A/C⁹⁴⁶A mutants were introduced to ROS-GC1 cDNA by polymerase chain reaction using appropriate mutagenic primers. The mutations were verified by sequencing. (2) Membraneous abridged forms of ROS-GC1: ΔExtD mutant was constructed by introducing two Hpa1 restriction sites at nucleotide positions 241 and 1446 in ROS-GC1 cDNA. The 1.2 kb Hpa1 fragment was excised, and the remaining part re-ligated; ΔExtD,ΔJMD,ΔKHD mutant was constructed from the ΔExtD mutant by introducing two BglII restriction sites at nucleotide positions 1557 and 2503 and excision of a 1 kb BglII fragment (amino acids 447–761); for the ΔSHD mutant two HpaI sites were introduced at nucleotide positions 2533 and 2663 allowing excision of a fragment amino acid residues 772–808; for the ΔCTE mutant a TGA STOP codon was introduced at position 972. (3) Soluble constructs of ROS-GC1: partKHD/SHD/CCD/CTE (aa 733–1054) fragment was amplified by PCR from the ROS-GC1 cDNA by PCR and cloned in frame into pET30a bacterial expression vector; CCD/CTE (aa 817–1054), CCD (aa fragment 817–965), and CTE (aa fragment 986–1054) fragments were amplified by PCR from ROS-GC1 cDNA and cloned in frame into pET30aLIC vector.

COS-7 cells were induced to express ROS-GC1 or its membrane-bound mutants using a calcium-phosphate coprecipitation technique (Sambrook et al., 1989). Sixty hours after transfection, the cells were harvested and their membranes prepared (Duda et al., 2016). The mutations did not affect membrane targeting of the proteins and their half-lives as verified by immunostaining. Some of the harvested cells were seeded on coverslips, fixed in 4% paraformaldehyde, stained with ROS-GC1 antibody and the immunoreaction was visualized after incubation with secondary antibody conjugated with DyLight488. The membraneous expression of the mutants was comparable.

The soluble ROS-GC1 constructs were individually expressed in BL21 bacterial cells as a His-tag fusion proteins and purified by Ni affinity chromatography. Purity of the protein was analyzed by SDS-PAGE. Concentration of the protein was determined by Bradford method at A₆₀₀.

Membrane samples were incubated individually without or with varying concentrations of recombinant bovine GCAP1 or GCAP2 (purified as described in Duda et al., 2011), recombinant mouse S100B (purified as in Pozdnyakov et al., 1997) or NaHCO₃. The assay mixture (25 μl) consisted of (mM): 10 theophylline, 15 phosphocreatine, and 50 Tris-HCl; pH 7.5, and 20 μg creatine kinase (Sigma). In experiments with GCAP1 and GCAP2, 1 mM EGTA was added to the reaction mixture; with S100B, 1 μM Ca²⁺ was present, and when the bicarbonate effect was tested, neither EGTA nor Ca²⁺ were added. The reaction was initiated by addition of the substrate solution (4 mM MgCl₂ and 1 mM GTP, final concentrations) and maintained by incubation at 37°C for 10 min. The reaction was terminated by the addition of 225 μl of 50 mM sodium acetate buffer, pH 6.2, followed by heating on a boiling water bath for 3 min. The amount of cyclic GMP formed was determined by radioimmunoassay (Nambi et al., 1982). All assays were done in triplicate and except where stated otherwise, were performed three times.

The catalytic activities of the soluble ROS-GC1 deletion mutants were assayed identically, except that purified protein instead of cell membranes was present in the reaction mixture and the cyclic GMP formed was measured by radioimmunoassay.

The guanylate cyclase activity is presented as average ± SD of three experiments done in triplicate.

To correlate the catalytic changes brought about by the mutations, the activities of the mutants were compared with wild type recombinant ROS-GC1 through Michaelis plots for the ligand used, fitting the data to the Hill equation, v = V_max (S)ⁿ/K_M + Sⁿ. V_max is the activity, S is the concentration of the ligand, K_M is the substrate concentration at which half-maximal velocity is achieved, and n is the Hill coefficient.

Photoreceptor ROS-GC1 is linked with phototransduction through its four limbs by distinct pathways: two modulated by Ca²⁺-sensors, GCAP1 and GCAP2; one by Ca²⁺-sensor, S100B and, the fourth by CO₂ (bicarbonate) via a Ca²⁺-independent mechanism (Figure 1C) (reviewed in Sharma et al., 2016). The first two pathways are specific for rod photoreceptors (Makino et al., 2008, 2012; Koch and Dell’orco, 2013); GCAP1- and S100B-modulated for the cone photoreceptors (Wen et al., 2012; Sharma et al., 2014) and the GCAP1 and CO₂/bicarbonate-modulated pathways for the red cone photoreceptors (Duda et al., 2015). Notably, the migratory patterns of these pathways are very different, yet they all are translated to generate cyclic GMP at a common CCD center. The origin and flow of the GCAP1 signal pathway is exceptional, it originates in an intracellular, JMD site and is then successively processed at the KHD and SHD sites before being transmitted to the CCD for final processing operation (Figure 1C). In contrast, GCAP2 and S100B signals originate on the CTE and then are transmitted to the CCD. A ⁶⁵⁷WTAPELL⁶⁶³ motif is critical for the signaling of both GCAPs, however, it has no role in controlling the basal catalytic activity of the cyclase or in the binding of the GCAPs (Duda et al., 2011). Bicarbonate signal, in a unique mode to itself, originates and gets translated at the CCD (Duda et al., 2015, 2016). The CENTRAL POINT is that all the transmitted signals, once they arrive at the CCD, are translated by the identical transduction steps. The functional (regulatory) specificity of the signals resides only in their migratory pathways.

In order to validate the critical role of the model-predicted CE residues in ROS-GC1 function in photoreceptor outer segments, we analyzed the effect of their mutations on the basal and regulatory catalytic modes of the guanylate cyclase. Each residue was individually mutated to alanine and the resulting mutant was assessed for its basal and ligand (Ca²⁺ sensor GCAPs and S100B or Ca²⁺- independent bicarbonate)-dependent activities. Because the site-directed mutation/expression results are very similar for each CE residue, only their combined essentials are provided.

The Mg²⁺ binding residues: D⁸³⁴, CE motif 1; D⁸⁷⁸, CE motif 3. D⁸³⁴ mutation results in the inhibition of 71.3% and D⁸⁷⁸ of 73.7% of basal ROS-GC1 catalytic activities (Figure 4A). Thus, each residue controls more than 70% of the basal saturation activity of ROS-GC1.

Connected with these losses, the two mutations also disable all of the ROS-GC1’s four-limbed modulatory activities. GCAP1: D⁸³⁴ mutation lowers the ROS-GC1 V_max from 440 to 71 pmol cyclic GMP min^-1 (mg protein)^-1 and D⁸⁷⁸ mutation to 81 pmol cyclic GMP min^-1 (mg protein)^-1 (Figure 4B). The EC₅₀ values (∼0.8 μM) and Hill coefficients (∼2) for both mutants remain unchanged. GCAP2: D⁸³⁴ mutation lowers ROS-GC1 V_max from 504 to 53 pmol cyclic GMP min^-1 (mg protein)^-1 and D⁸⁷⁸ mutation, to 45 pmol cyclic GMP min^-1 (mg protein)^-1 (Figure 4C). The EC₅₀ values (∼4 μM) remain the same, yet Hill coefficient values are lowered, to 1.35. S100B: Recall, In contrast to the GCAP sensors, S100B senses and stimulates ROS-GC1 catalytic activity in a Ca²⁺-dependent manner with a K_1/2 of 0.3 to 0.8 μM (reviewed in Sharma et al., 2016). D⁸³⁴ mutation lowers the ROS-GC1 V_max from 452 to 72 pmol cyclic GMP min^-1 (mg protein)^-1 and D⁸⁷⁸ to 91 pmol cyclic GMP min^-1(mg prot)^-1 (Figure 4D). The S100B EC₅₀ value of 0.8 μM and stimulatory Hill’s coefficient of 2 remain unchanged. CO₂: This phototransduction-linked limb of the ROS-GC1 is a recent discovery (Duda et al., 2016; reviewed in Sharma et al., 2016). It is a bicarbonate-modulated Ca²⁺-independent signal transduction pathway. Our scattered ongoing studies have begun to show that this pathway is signaled by CO₂ through carbonic anhydrase (CAII) enzyme, which converts CO₂ to bicarbonate and, bicarbonate, in turn, serves the second messenger of CO₂; importantly, in support of this hypothesis, electric recording studies demonstrate that the carbonic anhydrase inhibitor hinders the bicarbonate-dependent generated electric impulses in the red cones of the salamander (Makino et al., 2017). Like the three Ca²⁺-modulated pathways, the D⁸³⁴A and D⁸⁷⁸A mutations partially disable the bicarbonate modular operation (Figure 4E). D⁸³⁴A mutation lowers the V_max from 465 to 57 pmol cyclic GMP min^-1(mg prot)^-1 and D⁸⁷⁸A to 60 pmol cyclic GMP min^-1(mg prot)^-1. The EC₅₀ values of ∼30 mM bicarbonate remain similar to the 25 mM for the wild type ROS-GC1. The Hill’s coefficients are 1.93 and 1.64 for the D⁸³⁴A and D⁸⁷⁸A mutants, respectively, slightly lower than the 2.3 for ROS-GC1.

The ribose positioning residue, N⁹⁵³, CE motif 6, controls 55% of basal ROS-GC1’s catalytic activity (Figure 5A) and also disables most of the modulatory activities of its four-limbed pathways (Figures 5B–E). GCAP1: N⁹⁵³ mutation lowers the ROS-GC1 V_max from 440 to 95 pmol cyclic GMP min^-1 (mg protein)^-1. The EC₅₀ values (∼0.8 μM) and Hill coefficients (∼2) for both mutants remain the same as for wild type ROS-GC1 (Figure 5B). GCAP2: N⁹⁵³ mutation lowers the ROS-GC1 V_max from 508 to 92 pmol cyclic GMP min^-1 (mg protein)^-1 (Figure 5C). The mutation does not affect the EC₅₀ (∼4 μM) and Hill coefficient of 2. S100B: N⁹⁵³ mutation lowers the ROS-GC1 V_max from 452 to 97 pmol cyclic GMP min^-1 (mg protein)^-1 (Figure 5D). The EC₅₀ value of 0.8 μM and stimulatory Hill’s coefficient of 2 remains unchanged. CO₂: The N⁹⁵³ mutation results in lowering the ROS-GC1 V_max from 465 to 94 pmol cyclic GMP min^-1 (mg protein)^-1 (Figure 5E) but has no effect on EC₅₀ (∼25 mM bicarbonate) and Hill coefficient (2.2) values.

The triphosphate angling residues R⁹²⁵, motif CE4, and R⁹⁵⁷, motif CE7. They control, respectively, 66 and 53% of basal ROS-GC1 catalytic activity, and also disable most of the modulatory activities of its four-limbed pathways (Figure 6). GCAP1: R⁹²⁵ mutation lowers the ROS-GC1 V_max from 440 to 50 pmol cyclic GMP min^-1 (mg protein)^-1 and R⁹⁵⁷ mutation to 70 pmol cyclic GMP min^-1 (mg protein)^-1 (Figure 6B). The EC₅₀ values (∼0.8 μM) and Hill coefficients (∼2) for both mutants remain unchanged, however. GCAP2: R⁹²⁵ mutation lowers the ROS-GC1 V_max from 508 to 63 pmol cyclic GMP min^-1 (mg protein)^-1 and R⁹⁵⁷ mutation to 57 pmol cyclic GMP min^-1 (mg protein)^-1 (Figure 6C). The EC₅₀ values (∼4 μM) remain unchanged and also the Hill coefficient value of 2.0. S100B: R⁹²⁵ mutation lowers the ROS-GC1 V_max from 452 to 70 pmol cyclic GMP min^-1 (mg protein)^-1 and R⁹⁵⁷ to 90 pmol cyclic GMP min^-1 (mg prot)^-1 (Figure 6D). The S100B EC₅₀ value of 0.8 μM and stimulatory Hill’s coefficient of 2 remains unchanged. CO₂: Like the three Ca²⁺-modulated pathways, the R⁹²⁵ and R⁹⁵⁷ mutations disable the bicarbonate operation. R⁹²⁵ mutation lowers the ROS-GC1 V_max from 465 to 63 pmol cyclic GMP min^-1 (mg protein)^-1 and R⁹⁵⁷ to 145 pmol cyclic GMP min^-1 (mg prot)^-1 (Figure 6E). The EC₅₀ for bicarbonate is ∼25 mM for both mutants. The Hill coefficient of 1.1 for the R⁹²⁵A mutant is significantly lower than 2.3 for wild type ROS-GC1 but for the R⁹⁵⁷A mutant, 1.7, is close.

The guanine recognition residues, E⁸⁷⁴, motif CE2 and C⁹⁴⁶, motif CE5. Individually, residues E⁸⁷⁴ and C⁹⁴⁶ control, respectively, 72 and 65% of the basal catalytic activities (Figure 7A) and both mutations together disable all the basal activity of ROS-GC1 (Figure 7A: mutant E⁸⁷⁴A/C⁹⁴⁶A). GCAP1: E⁸⁷⁴A mutation lowers the ROS-GC1 V_max from 440 to 115 pmol cyclic GMP min^-1 (mg protein)^-1 and the C⁹⁴⁶A mutation to 87 pmol cyclic GMP min^-1 (mg protein)^-1 (Figure 7B). The EC₅₀ values of ∼0.8 μM are the same as for the wild type ROS-GC1 and the Hill’s coefficients of 1.75 and 1.71 for the mutants are close to the 2 value for the wild type cyclase. The double mutant is unresponsive to any concentration of GCAP1 tested (Figure 7B). GCAP2: Mutation of E⁸⁷⁴ lowers the ROS-GC1 V_max from 508 to 60 pmol cyclic GMP min^-1 (mg protein)^-1 and mutation of C⁹⁴⁶ to 80 pmol cyclic GMP min^-1 (mg protein)^-1 (Figure 7C). The EC₅₀ values (∼4 μM) remain unchanged; the Hill’s coefficients are above 1, but slightly lower than that for wild type ROS-GC1, being 1.57 for the E⁸⁷⁴A mutant and 1.50 for C⁹⁴⁶A. The double mutant does not respond to GCAP2 (Figure 7C). S100B: E⁸⁷⁴A mutation lowers the ROS-GC1 V_max from 452 to 93 pmol cyclic GMP min^-1 (mg protein)^-1 and C⁹⁴⁶A, to 122 pmol cyclic GMP min^-1 (mg prot)^-1 (Figure 7D). The values of EC₅₀ of ∼0.7 μM S100B and the Hill’s coefficients of 1.7 for both mutants are the same as for the wild type ROS-GC1. There is no measurable activity of the double mutant either with or without S100B (Figure 7D). CO₂: Like the three Ca²⁺-modulated pathways, E⁸⁷⁴A and C⁹⁴⁶A mutations affect bicarbonate operation. Mutation of E⁸⁷⁴ lowers the ROS-GC1 V_max from 465 to 54 pmol cyclic GMP min^-1 (mg protein)^-1 and C⁹⁴⁶ to 61 pmol cyclic GMP min^-1(mg prot)^-1 (Figure 7E). The mutations do not affect significantly the EC₅₀ and Hill coefficient values. They remain comparable to the wild type ROS-GC1 values being EC₅₀ ∼30 mM for both mutants and Hill coefficient 1.91 for E⁸⁷⁴A and 1.53 for C⁹⁴⁶A. Importantly, there is no detectable catalytic activity of the double mutant (Figure 7E).

An earlier study (Tucker et al., 1998) concluded that both residues, E⁸⁷⁴ and C⁹⁴⁶ (E⁹²⁵ and C⁹⁹⁷, corresponding human ROS-GC1 residues), individually control the total catalytic activity of the ROS-GC1. In this study the authors used the HEK 293 cell system for the expression of guanylate cyclases; and they did not evaluate the mutant with double mutations (E⁹²⁵/C⁹⁹⁷) and the residues were mutated, respectively, to K and D instead of A. Our results revise this conclusion; only two mutations together, E⁸⁷⁴/C⁹⁴⁶, totally disable the catalytic activity of ROS-GC1.

Alanine mutation of CE residues does not affect cooperativity of CCD active sites. Homodimeric antiparallel structure of the guanylate cyclase catalytic domain results in the existence of two equivalent catalytically active sites (Liu et al., 1997; Venkataraman et al., 2008). It was now sought to determine how mutation of an individual CE would affect the communication between the active sites in the mutant-cyclases. Activity assays were carried out in the presence of constant concentration of Mg²⁺ and varying concentrations of GTP (0–3 mM) and K_M as a measure of a mutant’s affinity for GTP and Hill coefficients indicative of interaction between active sites, were determined. The results are summarized in Table 1. For the wild type ROS-GC1 the K_M value for GTP was 0.45 mM. All ROS-GC1 CE mutants exhibited K_M values around 0.5 mM, ranging from 0.48 for D⁸⁷⁸A to 0.64 for C⁹⁴⁶A. As expected, ROS-GC1 exhibited positive cooperativity, with a Hill coefficient of ∼2 (2.01 ± 0.32), conforming the presence of two, interacting with each other, active sites. All the mutants exhibited also positive cooperativity, with Hill coefficients significantly above 1, varying from 1.68 ± 0.22 for C⁹⁴⁶A to 2.11 ± 0.23 for D⁸⁷⁸A. These results demonstrate that although the CE mutations considerably hinder the catalytic processes of the active sites, leading to drastically lowered V_max values, the sites remain cooperative in their catalytic functions.

These results provide experimental proof that indeed the D⁸³⁴, E⁸⁷⁴, D⁸⁷⁸, R⁹²⁵, C⁹⁴⁶, N⁹⁵³, and R⁹⁵⁷ residues of the ROS-GC1 CCD are critical for the cyclase’s catalytic activity. Although no one mutation individually leads to complete inactivation, they collectively do so. Therefore, each of these residues represents one of the seven structural units of the CE that control full catalytic activity of ROS-GC1.

Is the same true for all MGCs?

The CCD structure along with seven CE residues is conserved in the mammalian MGC family. To determine whether the CE residues are conserved in the MGC family first, the sequence of ROS-GC1 145-residue CCD was compared with the corresponding sequences of the other six members of the family: ROS-GC2, V^829-P⁹⁷⁵; ONE-GC, V^818-P⁹⁶²; ANF-RGC, Q^836-L⁹⁸⁰; CNP-RGC, Q^830-A⁹⁷⁴; STa-RGC, K^794-P⁹³⁵; GC-G, V^849-P⁹⁹³ (Figure 8). The percentile of their sequence identities with ROS-GC1 were: 93, 85, 68, 70, 75, and 69, respectively.

To assess and then to formulate a unified signal transduction concept by which the CEs operate, the question was asked whether in all members of the MGC family they exist in conserved structural motifs.

The answer was in affirmative. In ROS-GC1 the seven CEs are present in motifs: S⁸³³-T⁸³⁹; D⁸⁶⁹-G⁸⁷⁷; D⁸⁷⁸-V⁸⁸²; R⁹²⁵; M⁹⁴²-V⁹⁵²; N⁹⁵³; and A⁹⁵⁵-S⁹⁶⁰. These motifs are conserved in all seven members of the MGC family (Figure 9) and thus, they are termed CEs motifs, CEMs. It is logical to envision that, as in ROS-GC1, these CEMs are, respectively, involved in Mg²⁺ binding, ribose positioning, guanine recognition, and triphosphate angling. We therefore propose that for all MGCs the CEMs control their basal and ligand-dependent catalytic activities.

Prior to the remarkable discovery that ROS-GC1 CCD is embedded with the residence of neurocalcin δ (NCδ) recognition site (Venkataraman et al., 2008), CCD was believed to be only the translational center of the ligand-dependent signals into the production of cyclic GMP. Linkage with NCδ changed this paradigm, CCD also became a regulatory center for the NCδ-modulated Ca²⁺signals. Strikingly, in contrast to all other ligand-dependent signals, NCδ-modulated Ca²⁺ signal originates and gets transduced in CCD.

To determine if presence of the NCδ-sensing motif, V⁸³⁶-L⁸⁵⁷, is a common feature of the MGC family, sequence of this motif was compared in all members of the MGC family (Figure 10). With 100% conservation, ROS-GC1 and ROS-GC2 preserved this motif and there was about 87% conservation in ONE-GC. With the remainder four MGCs the respective percentile conservation was: ANF-RGC 68; CNP-RGC, 68; Star-GC, 68; and GC-G, 59. In accordance with this pattern, it has been experimentally validated that like ROS-GC1, ONE-GC is Ca²⁺-modulated via its sensor myr-NCδ (Duda et al., 2001; Krishnan et al., 2004; Duda and Sharma, 2008; Sharma and Duda, 2010). Importantly, with the evidence that ANF-RGC also is Ca²⁺- modulated via its sensor myr-NCδ (Duda et al., 2012a) and its NCδ sensing motif has total sequence conservation with the corresponding region in CNP-RGC (Figure 10), we conclude that this motif plays an important physiological role in the regulatory property of the MGC family. Notably, total conservation of the N-terminal VGFT and C-terminal- -LND flanking regions of the NCδ sensing motif is a family trait (Figure 10).

Adding further complexity to the modular role of the CCD, recent studies show that ROS-GC1’s CCD also contains a 108-residue, Y⁸⁵⁸–Y⁹⁶⁵, structural element (Duda et al., 2015, 2016). Except for one residue, Y⁹⁶⁵, it resides within the CCD; and it represents a Ca²⁺-independent CO₂-modulated region of the GC (Figure 1) (Makino et al., 2017). This region shows 70 to 95% identity in the MGC family, is also CO₂-modulated in ONE-GC and possibly in GC-G. Thus, MGC CCD besides containing CEs, contains also two, NCδ and CO₂, modulated regions.

Right from the time of MGC activity detection in the mammalian tissues the mystery surrounded on one of its features: why, in contrast to the adenylate cyclase which uses Mg²⁺, the preferred cofactor of MGC is Mn²⁺ (Hardman and Sutherland, 1969; Ishikawa et al., 1969; Schultz et al., 1969; White and Aurbach, 1969) (reviewed in Sharma, 2010)?

This mystery has been sustained even with the crystalized forms of the eukaryotic Cyg12 soluble (Winger et al., 2008) and the bacterium Cya2 (Rauch et al., 2008) membrane forms. These CCDs show no significant catalytic activities when the Mg²⁺-GTP substrate is used; they only show these with the use of Mn²⁺-GTP.

To address this issue, we performed domain-by-domain deletion/expression analysis of the recombinant ROS-GC1 (Figure 11). The catalytic activity [pmol cyclic GMP min^-1 (mg prot)^-1] of each truncated construct was assessed using the natural Mg²⁺-GTP as the substrate. The isolated soluble CCD construct contained only a minimal basal specific catalytic activity of 7 pmol. In contrast, the r-ROS-GC1’s basal catalytic activity and of all its truncated constructs ranged between 110 and 130 pmol, a value almost 22-fold higher than its isolated form. Importantly, whenever the constructs lacked the TMD, illustrated by the CCD–CTE and CCD, the CCD’s catalytic activity dramatically dropped from 130 to 6 pmol. To rule out the unlikely possibility that the drop in catalytic activity might be restored by anchoring CCD with any domain located N-terminally to it, the CCD’s catalytic activity was assessed in the partially truncated KHD^--SHD-CCD-CTE soluble fragment (Figure 11). No restoration occurred, however. The catalytic activity remained 6 pmol.

Our interpretation of these results is that: (1) the natural substrate of the MGC’s CCD for catalysis is Mg²⁺-GTP; (2) none of the modular domains- -ExtD, JMD, KHD, SHD and CTE- -have any role in its basic catalytic operation; and (3) TMD is the major contributor in boosting CCD’s basal catalytic activity.

To understand how TMD might contribute to the CCD’s catalytic activity, the possibility was considered that this might be due to a unique consensus motif of the TMDs. Structural comparison, however, demonstrated that this was not the case (Figure 12). Compared to the ROS-GC1, the percentile identity between the MGCs TMDs was only marginal, ROS-GC2, 24; ONE-GC, 24; Star-GC, 22; GC-G, 32; ANF-RGC, 24; CNP-RGC, 25. It is noteworthy that not even a single residue was fully conserved among the family members. Thus, we propose that TMD contributes to the CCD’s catalytic activity by making it possible that all the successive modular domains- -JMD, KHD, SHD, and CCD- are in properly fixed positions (Top Panel: Figure 1), resulting in the optimal conditions for catalysis to occur.

Encoded by seven genes, MGC is the generator of cyclic GMP which serves as an intracellular second messenger for the countless physiological processes. It is a single transmembrane spanning protein existing as a homodimer upheld by two contact regions between the monomers (Figure 1). Based on the crystal structure of the ANF-RGC ExtD (Ogawa et al., 2004), the first prototype guanylate cyclase member (reviewed in Sharma et al., 2016), there is head-to-head contact of these two domains. The second contact is formed at the CCD (Venkataraman et al., 2008) where two monomers assume an antiparallel conformation (Duda et al., 2012b) and (Figure 1). The encoded seven MGCs- -ANF-RGC, CNP-RGC, STa-RGC, ROS-GC1, ROS-GC2, ONE-GC and GC-G- -initiate cellular signaling process from different sites, the first three from the ExtD, next two from the ICD, ONE-GC from both ExtD and ICD and GC-GC, most probably from ICD (Sharma et al., 2016). Yet, regardless of the origin, all signals are transduced and translated into the generation of cyclic GMP at the CCD (Sharma et al., 2016).

In the photoreceptor ROS-GC1, a mammalian MGC, CCD is a 145-residue structural unit stretching from V⁸²⁰ to P⁹⁶⁴ (Figure 1). In its natural form, it is homodimeric. In its simulated 3D-model, each monomer is composed of 8-stranded β-sheets and six α-helices (Figure 3A). In its homodimer form two monomers assume a two-fold symmetry axis with a central gap of a circlet-like shape (Figure 3B). The circlet between the monomers contains two symmetric active sites, each formed by the conserved residues (CEs) from the two monomers (Figure 3C). The CEs are: two Mg²⁺-binding, D⁸³⁴ and D⁸⁷⁸; one, ribose positioning, N⁹⁵³; two guanine recognition, E⁸⁷⁴ and C⁹⁴⁶; and two, triphosphate angling, R⁹²⁵ and R⁹⁵⁷. The projection is that in an inactive state the dimer is in an open conformation but in closed upon binding GTP for catalysis (Figure 3D).

The CCD structure is conserved in all seven mammalian MGCs, showing a sequence identity between 92 and 65% (Figure 8). Characteristically, it is closer between the Ca²⁺-modulated guanylate cyclases – -compare ROS-GC1 vs ROS-GC2, 93% and ROS-GC1 vs ONE-GC, 85%. It is predicted that this closeness in identity is reflective of the regulatory feature of the CCD, in contrast to its basic catalytic mode, explained below.

In a uniform general mode, all seven CEs of all MGCs are arranged in structural motifs (Figure 9). These motifs, respectively, represent the common features of all MGCs. This structured theme of the motifs points out that they in an identical fashion control the folding patterns of all MGCs and, thus, their basic guanylate cyclase catalytic activities.

The presented study also discloses two important features of the mammalian MGC CCD.

In accordance with the predictions, (i) MGC CCD is homo-dimeric. (ii) Seven CEs embedded in their seven CE motifs, all are critical in controlling the basal catalytic activity of the MGCs. (iii) Mg²⁺-GTP is the natural substrate for catalysis, this basic operation is controlled by the CEM motifs, 2 and 3. (iv) The TM domain by anchoring MGC to the plasma membrane controls almost all (more than 90%) the basal CCD activity. (v) Besides its core seven-CE elemental structure, the catalytic domain is embedded with two regulatory domains, NCδ and CO₂; thus, CCD is both a core and regulatory catalytic center. (vi) CCD controls all four phototransduction pathways (discussed below).

Similar to its critical role in controlling the basic catalytic operation of the MGC family, CCD also plays a vital regulatory role in modulating all four ROS-GC1 interlocked phototransduction pathways. The mutation in any one of its seven CE elements severely impedes it’s all Ca²⁺-modulated rod and cone photoreceptor along with the Ca²⁺-modulated and CO₂/bicarbonate-modulated catalytic activity. The cyclic GMP output in the cells is compromised, and it is predicted that finally the photoreceptors will die. Notably, the ligand-modulated EC₅₀ and Hill coefficient values of the pathways are not affected, demonstrating that CEs core element has no control over the ligand binding activities of the ROS-GC.

The MGC signal transduction is a two-step process. Step one, it is contributed by the seven CE elements of the CCD and occurs in all members of the MGC family. The CCD captures Mg²⁺-GTP in its pocket and turns itself from the inactive basal to the active basal state (Figure 3D). Step two, it is a regulatory process. Upon ligand (GCAP1, GCAP2, S100B, or bicarbonate) binding, the rotation of each CCD monomer occurs around their two-fold axes (Figure 3C). This brings the CE residues into the closed maximally active position, and collectively their manifestation of the ligand dependent catalytic saturation activity.