Two cGMP-dependent protein kinases have opposing effects on molt-inhibiting hormone regulation of Y-organ ecdysteroidogenesis

Growth in decapod crustaceans is restricted by the presence of a chitin-based exoskeleton that must be periodically shed through the process of molting. Molting is regulated primarily by the neuropeptide molt-inhibiting hormone (MIH) produced in the X-organ/sinus gland complex in the eyestalk ganglia, and ecdysteroids produced by a pair of molting glands or Y-organs (YOs) located in the cephalothorax (reviewed in Mykles, 2024; Fehsenfeld, 2024). The YO transitions through four physiological states over the molt cycle. These are the basal state during intermolt (stage C₄), activated state during early premolt (stage D₀), committed state during mid- and late premolt (stages D₁ and D_2–3, respectively) and repressed state during postmolt (stages A, B and C_1–3; reviewed in Mykles, 2011; Mykles and Chang, 2020). Ecdysteroid synthesis in the basal YO is attenuated by the action of MIH, resulting in low levels of YO ecdysteroid secretion and low titers of ecdysteroid in the hemolymph (Mykles, 2011; Mykles and Chang, 2020). The activated YO is characterized by an increase in ecdysteroid synthesis and secretion that initiates premolt processes in peripheral tissues, but the YO itself remains sensitive to MIH (Mykles, 2024). Increased ecdysteroidogenesis by the activated YO requires mechanistic Target of Rapamycin Complex 1 (mTORC1)-dependent protein synthesis (reviewed in Mykles and Chang, 2020; Mykles, 2021). Once the YO reaches the committed state, the YO becomes insensitive to MIH; ecdysteroid synthesis increases to a peak in late premolt, which ultimately triggers ecdysis (Mykles and Chang, 2020; Mykles, 2024). During postmolt, the repressed YO has low ecdysteroidogenic activity, resulting in the lowest hemolymph ecdysteroid titers of the molt cycle stages (Mykles, 2011). Transcriptional downregulation of MIH and mTORC1 signaling genes suggests that the repressed YO is refractory to endocrine factors until the YO transitions back to the basal state (Mykles and Chang, 2020).

MIH signaling involves cyclic nucleotide second messengers organized into a transient cAMP/Ca²⁺-dependent triggering phase linked to a sustained NO/cGMP-dependent summation phase (reviewed in Webster et al., 2012; Webster, 2015; Mykles and Chang, 2020). The transient increase in cAMP and sustained increase in cGMP, together with the in vitro inhibition of ecdysteroid secretion by analogs of either cyclic nucleotide, suggest that MIH binding stimulates cyclic AMP-dependent protein kinase (PKA) activity, which, in turn, stimulates prolonged cGMP-dependent protein kinase (PKG) activity (reviewed in Covi et al., 2009; Webster et al., 2012; Mykles and Chang, 2020). Pulsatile release of MIH from the sinus gland inhibits YO ecdysteroidogenesis and maintains the YO in the basal state (Mykles and Chang, 2020). In the proposed model, binding of MIH to a G protein-coupled receptor (GPCR) activates PKA activity, which leads to activation of a Ca²⁺-dependent NO synthase, NO-dependent guanylyl cyclase and PKG (Mykles and Chang, 2020). Although the identity of the MIH receptor has not been established with any certainty, it is hypothesized that it is homologous to insect ion transport peptide GPCRs (Kozma et al., 2024).

PKG activity is the primary effector regulating mTORC1-dependent YO ecdysteroidogenesis by MIH (Mykles, 2021). As proposed, PKG mediates mTORC1 activity via the tuberous sclerosis complex (TSC; Mykles, 2021). TSC is a heterotrimeric GTPase-activating protein that inactivates the small G protein Ras homolog enriched in brain (Rheb), an activator of mTORC1, by promoting the hydrolysis of GTP to GDP (reviewed in Shi and Collins, 2023; Wang et al., 2024). In mammalian cardiomyocytes, PKG1 phosphorylation of the TSC2 subunit activates TSC, resulting in inhibition of mTORC1 (Ranek et al., 2019; Oeing et al., 2020,b).

PKGs are members of the AGC (PKA, PKG, PKC) subfamily of serine/threonine protein kinases (reviewed in Leroux et al., 2018). PKGs occur as homodimers in the native state. Each subunit has an N-terminal regulatory region consisting of a leucine/isoleucine-zipper (LZ) domain and a cGMP-binding domain with two cyclic nucleotide binding sites, and a C-terminal region containing a protein kinase domain; an autoinhibitory site is located between the LZ and cGMP-binding domains (Kim and Sharma, 2021). The LZ domain mediates dimerization and docking with G-kinase anchoring proteins (GKAPs) that facilitate sub-cellular localization (Vo et al., 1998; Casteel et al., 2010).

In mammals, PKGs are encoded by two genes, designated prkg1 and prkg2 (Kim and Sharma, 2021). Alternative splicing produces two PKG1 isoforms, designated PKG1α and PKG1β, which differ in the N-terminal region that includes the LZ domain (Hofmann, 2020; Kim and Sharma, 2021). By contrast, PKG2 occurs as a single gene product (Hofmann, 2020; Kim and Sharma, 2021). PKG2 is known to localize to the plasma membrane by myristoylation of the N-terminus, whereas mammalian PKG1 isoforms are cytosolic (Vaandrager et al., 2005; Yuasa et al., 2008; and reviewed in Hofmann, 2020). In Drosophila, PKGs are also encoded by two genes, foraging (for) or dg2 is homologous to PKG1 and encodes 21 alternatively spliced transcripts, whereas dg1 is homologous to PKG2, encoding only one protein product (Kalderon and Rubin, 1989; Allen et al., 2017). PKG1 homologs have been linked to functional changes in food-related behaviors in fruit flies, honeybees and nematodes (Fitzpatrick and Sokolowski, 2004).

The decapod YO expresses all the components of the MIH signaling pathway, including PKG (Das et al., 2018; Mykles, 2021). In the blackback land crab, Gecarcinus lateralis, PKG1 transcript expression in the YO is molt-stage dependent, with the highest expression occurring in the basal YO (Das et al., 2016). Here, we report a comprehensive analysis of crustacean PKGs using bioinformatic tools and biochemical assays. Taking advantage of the CrusTome database (Pérez-Moreno et al., 2023), phylogenetic analysis identified contiguous sequences encoding four PKG1 isoforms and PKG2 in decapod crustaceans. Multiple sequence alignments identified conserved motifs that distinguished PKG1 isoforms. In vitro assays of YOs from G. lateralis and the green shore crab (Carcinus maenas) determined the effects of PKG inhibitors on MIH control of ecdysteroid secretion. Surprisingly, the results indicate that PKG1 and PKG2 had opposing effects, which may explain how the basal YO maintains low ecdysteroid secretion rates in the presence of MIH in intermolt animals.

Adult male Gecarcinus lateralis (Fréminville 1835) were collected in the Dominican Republic and shipped overnight via air cargo to Denver, CO, USA. Gecarcinus lateralis were collected under Contract for Access to Genetic Resources for Research Purposes DJC-1-2019-01310 and Collection and Export Permit No. VAPS-07979 from the Ministry of Environmental and Natural Resources of the Dominican Republic. Animals were housed in group cages containing 10–15 animals each with an aspen bedding substrate moistened with 5 ppt Instant Ocean at 27°C and 80% relative humidity on a 12 h:12 h light:dark schedule as described in Covi et al. (2010). Animals were fed lettuce, carrots and raisins twice a week. Molt stage was determined by limb regenerate growth, condition of the membranous layer of the exoskeleton, and hemolymph ecdysteroid titer (Covi et al., 2010). For the YO assays, intermolt animals were eyestalk ablated (ESA) 3 days prior to the assay to activate the YOs (Lee et al., 2007a).

Adult male Carcinus maenas (Linnaeus 1758) were collected from Bodega Harbor, Bodega Bay, CA, USA, and maintained at the Bodega Marine Laboratory in a flow-through seawater system at ambient seawater temperatures (12–16°C). Crabs were fed weekly with fish.

Ecdysteroid concentrations in hemolymph and YO media samples were quantified by a competitive enzyme-linked immunosorbent assay (ELISA; Kingan, 1989) as modified by Abuhagr et al. (2014).

The full-length mature G. lateralis MIH (Gl-MIH, without the signal peptide) was synthesized using the solid phase Fmoc (FluorenylMethylOxyCarbonyl) methodology by Pepmic Co., Ltd (Suzhou, China). The specific synthesis steps followed previously established procedures (Mosco et al., 2012) and chemical synthesis took into account the three disulfide bridges through the use of semi-synthetic tethers (Thalluri et al., 2018). The Gl-MIH amino acid sequence is as follows: AVINDECPNVIGNRDIFKKVDWICEDCANIFRIDGLATLCRKNCFRNIDFLWCVYASERQAEKDELTRYVSILRAGSV with three disulfide bridges (Cys7:Cys44, Cys24:Cys40 and Cys27:Cys53; indicated by bold font). The mature G. lateralis MIH sequence shares 67% amino acid identity and 84% similarity with C. maenas.

An in vitro assay quantified the effects of MIH with or without PKG inhibitors on YO ecdysteroid secretion. YOs from adult male intermolt C. maenas or from 3-day post-ESA G. lateralis were used. The G. lateralis physiological saline consisted of 316 mmol l⁻¹ NaCl, 5.4 mmol l⁻¹ KCl, 8.8 mmol l⁻¹ CaCl₂, 6.8 mmol l⁻¹ MgSO₄ and 1 mmol l⁻¹ Na₃PO₄ and 10 mmol l⁻¹ HEPES-NaOH (pH 7.4). The C. maenas physiological saline consisted of 430 mmol l⁻¹ NaCl, 5 mmol l⁻¹ K₂SO₄, 7 mmol l⁻¹ MgCl₂, 4.5 mmol l⁻¹ CaCl₂ and 10 mmol l⁻¹ HEPES-NaOH (pH 7.2). A general PKG inhibitor, Rp-8-Br-PET-cGMPS (2-bromo-3,4-dihydro-3-[3,5-O-[(R)-mercaptophosphinylidene]-β-D-ribofuranosyl]-6-phenyl-9H-imidazo[1,2-a]purin-9-one; Cayman Chemical, Ann Arbor, MI, USA) and a PKG2-specific inhibitor, AP-C5 (4-[4-(1H-imidazol-1-yl)phenyl]-N-2-propyn-1-yl-2-pyrimidinamine; R&D Systems Inc., Minneapolis, MN, USA), were dissolved in dimethyl sulfoxide (DMSO) and diluted 100-fold in crab saline to achieve final inhibitor concentrations of 1, 10 or 100 µmol l⁻¹ for C. maenas and 100 µmol l⁻¹ for G. lateralis (a single concentration was used owing to a limited number of animals).

One set of experiments determined the biological activity of MIH. YO pairs were incubated with either saline control or 50 nmol l⁻¹ MIH. A second set of experiments tested the effects of PKG inhibitors ±50 nmol l⁻¹ MIH. Hemolymph samples were taken for each animal before dissection to assess molt stage based on circulating ecdysteroid titers. One YO in the pair from each animal served as the MIH control treatment, while the other YO served as the MIH and PKG inhibitor experimental treatment. YOs were preincubated for 30 min at room temperature in 0.5 ml saline containing either 1% DMSO (control treatment) or PKG inhibitor and 1% DMSO (experimental treatment) on an orbital shaker at 100 rpm. After preincubation, YOs were transferred to 0.5 ml saline containing either 50 nmol l⁻¹ MIH and 1% DMSO (control) or 50 nmol l⁻¹ MIH, PKG inhibitor and 1% DMSO (experimental) for 4.5 h at 100 rpm at room temperature. Hemolymph (100 µl) and a subset of media (200 µl) samples were combined with three volumes of methanol and stored at −15°C until quantification with the competitive ELISA. In C. maenas, animals determined to be in premolt (>100 pg µl⁻¹) or postmolt (<20 pg µl⁻¹) based on hemolymph 20E were excluded from analysis. Data are expressed as percent of control [(experimental 20E secretion/control 20E secretion)×100]. Statistical significance between the means was determined by a two-tailed paired t-test in RStudio (v. 2024.4.1.748).

Protein reference sequences were obtained from the NCBI Reference Sequence Database (RefSeq; O'Leary et al., 2016), GenBank (Sayers et al., 2020) and UniProtKB (Consortium, 2022) and used as the query for four iterative BLAST searches against the CrusTome database (v.0.1.0) to ensure broad phylogenetic representation among crustaceans (Altschul et al., 1990; Pérez-Moreno et al., 2023). Protein sequences, including reference sequences and BLAST results, were aligned using MAFFT-DASH with the flag ‘-originalseqonly’ (v.7.508; Rozewicki et al., 2019). Multiple sequence alignments (MSAs) were trimmed using ClipKit with the smart-gap parameter to trim alignment gaps while maintaining phylogenetically informative sites (Steenwyk et al., 2020). Phylogeny was inferred using IQ-TREE (v.1.6.12; Trifinopoulos et al., 2016) to construct a maximum likelihood phylogenetic tree, and evolutionary models were estimated using ModelFinder (Kalyaanamoorthy et al., 2017) as optimized in IQ-TREE.

Initial trees were refined by removing partial sequences with 200 or fewer amino acids, those that did not contain at least two of the conserved domains found in PKGs (leucine-zipper, two cGMP binding domains and a kinase domain), sequences with ambiguous residues, those that appeared to be other related proteins (e.g. those containing kinase or cyclic-nucleotide binding domains) and those that were contamination from nematodes that often infect crustaceans. Nematode contamination was initially identified by long branch lengths and erroneous position within phylogenies, and subsequently validated with NCBI BLAST. Final trees were constructed with the refined dataset using the above parameters and rooted at the midpoint. Branch support for phylogenetic relationships was estimated with ultrafast bootstrap approximation (UFBoot; Minh et al., 2013) with 1000 iterations, as well as an approximate Bayes test. Final tree files, input files and multiple sequence alignments are deposited in Dryad at https://doi.org/10.5061/dryad.1g1jwsv7b.

CD-Search in the NCBI Conserved Domain Database (CDD, v3.20; Wang et al., 2023) and EMBL-EBI's InterProScan (IPS, v5.69-101.0; Paysan-Lafosse et al., 2023) were used to predict and annotate the position of conserved domains and essential residues in G. lateralis and C. maenas sequences. MSAs were generated using MAFFT and visualized using the plot_msa.py script (Kozma et al., 2024), available at https://github.com/invertome/scripts/tree/main/plots.

Relative transcript expression in transcripts per million reads (TPM) of intermolt YOs was determined for G. lateralis using RNAseq data from Das et al. (2018). PKG1 isoforms were not distinguished in this dataset owing to the method of transcript assembly available at that time. Therefore, PKG1 transcript expression was assumed to be pooled expression of all reads from PKG1 sequences that mapped to the PKG1β contig sequence. Relative transcript expression for C. maenas YOs was obtained from the intermolt YO RNAseq data generated by Oliphant et al. (2018), assembled in the CrusTome database and quantified into TPM using Salmon (v.1.7.0). Salmon quantifications were run with the following flags ‘--seqBias --gcBias –validateMappings’ to ensure accurate quantification by avoiding potential biases. Expression was graphed as means±s.e.m. in RStudio (v.2024.04.1.748).

Sequencing of PCR products was used to validate the identities of the G. lateralis and C. maenas PKGs. Combined cDNA from YO, heart and claw muscle was used as template for PCR using sequence-specific primers directed to the N-terminal region (Table S1). As the primers for the four PKG1 (α1–3, β) isoforms identified in these species spanned the putative alternative splicing site, the same reverse primer was used for each isoform; the forward primer was isoform-specific. PCR conditions were as follows: initial denaturation at 95°C for 2 min, followed by 35 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 45 s, and a final extension cycle at 72°C for 10 min. PCR products were purified from agarose gels using the QIAEX II Gel Extraction Kit (QIAGEN, Hilden, Germany) and sequenced by GeneWiz (Azenta Life Sciences, South Plainfield, NJ, USA).

The biological activity of the synthetic Gl-MIH was tested by incubating YOs in media without (control) or with 50 nmol l⁻¹ Gl-MIH. The synthetic peptide was more effective on C. maenas YOs (see Data availability for raw data). In C. maenas, MIH significantly inhibited ecdysteroid secretion by 63% (37±8% of control, n=7, P=0.006). In G. lateralis, ecdysteroid secretion with MIH was 51% lower than the control. However, the difference in the means did not reach significant levels (49±11% of control, n=8, P=0.054).

A general PKG inhibitor, Rp-8-Br-PET-cGMPS, countered the inhibitory effect of MIH on YO ecdysteroid secretion in C. maenas (Fig. 1A, Table 1). YO secretion in media containing 10 or 100 µmol l⁻¹ Rp-8-Br-PET-cGMPS and 50 nmol l⁻¹ MIH was significantly higher than that of the paired control YOs incubated with 50 nmol l⁻¹ MIH alone (217% and 311% of control, P=0.017 and P=0.006, respectively). In G. lateralis, 100 µmol l⁻¹ Rp-8-Br-PET-cGMPS had no effect on YO ecdysteroid secretion (101% of MIH control, P=0.92; Fig. 1B, Table 1).

A PKG2-specific inhibitor, AP-C5, enhanced the inhibitory effect of MIH on ecdysteroid secretion in both C. maenas and G. lateralis YOs (Fig. 1C,D, Table 1). In C. maenas, AP-C5 showed a concentration-dependent effect on YO secretion (Fig. 1C, Table 1). In media containing 1, 10 or 100 µmol l⁻¹ AP-C5 and 50 nmol l⁻¹ MIH, YO secretion was significantly lower than that of the paired control YOs incubated with 50 nmol l⁻¹ MIH alone (60%, 24% and 10% of the control; P=0.04, P=0.0002 and P=0.0002, respectively). In G. lateralis, 100 µmol l⁻¹ AP-C5 and 50 nmol l⁻¹ MIH significantly lowered ecdysteroid secretion with respect to 50 nmol l⁻¹ MIH alone (24% of the control, P=0.0004; Fig. 1D, Table 1).

Phylogenetic analysis of pancrustacean and tardigrade sequences from CrusTome and selected vertebrate sequences from RefSeq (NCBI), GenBank (NCBI) and UniProtKB showed that PKG sequences partitioned into PKG1 and PKG2 clades (Fig. 2). PKG1 formed a monophyletic group consisting of 206 sequences among 108 species, with subclades in vertebrates and decapods distinguishing PKG1α and PKG1β isoforms (Fig. 2). PKG2 formed a monophyletic group consisting of 60 sequences among 53 species (Fig. 2). The distribution of PKG1 and PKG2 sequences in the Ecdysozoa in the CrusTome database is summarized in Table 2. In 42 decapod species, 107 sequences were identified; PKG1α sequences were more common than PKG1β and PKG2 sequences. Copepoda had 20 sequences in 8 species and Isopoda had 19 sequences in 14 species (Table 2). The number of PKG sequences ranged between one and nine in the other ten taxa.

Within decapods, PKG1 isoforms were classified as alpha or beta based upon homology with annotated reference sequences and the position in the phylogenetic tree, as well as identification of the leucine zipper as either DD_cGKI-alpha (cd12085) or DD_cGKI-beta (cd12086) by NCBI CDS software. Importantly, predicted PKG1 sequences deduced from the NCBI Eukaryotic Genome Annotation Pipeline (accession numbers beginning with XP) match transcripts found in CrusTome (Table 3). Phylogenetic analysis identified four alpha (PKG1α1, PKG1α2, PKG1α3 and PKG1α4) isoforms and one beta (PKG1β) isoform in decapod species (Fig. 3). The naming convention used here provides clear annotation of PKG1 isoforms as either alpha or beta, as well as subtypes of alpha, rather than the vague ‘isoform X1’ used in reference sequence databases.

PKG2 was identified in all decapod taxa, except penaeid shrimp (Fig. 4, Table 3). A single PKG2 transcript was identified in each of the decapod species, including identical transcripts in dactyl and lateral flagella (D, LF) and YO in the blue crab, Callinectes sapidus (Fig. 4). The Procambarus clarkii PKG2 sequence predicted from genomic data matched that of the Pc-PKG2 contig sequence assembled in CrusTome (Table 3). The Eriocheir sinensis PKG2 sequence predicted from genomic data contained a full-length open reading frame (ORF), whereas the Es-PKG2 contig sequence assembled in CrusTome is missing a portion of the N terminal (Table 3).

The domain organization of PKGs was highly conserved in metazoan species (Kim and Sharma, 2021). Fig. 5 shows the functional domains of G. lateralis PKGs as representative of the decapod sequences. The N-terminal regulatory region contained an LZ domain, an autoinhibitory (AI) site and two cyclic nucleotide binding (CNB) domains in the cGMP-binding domain. The C-terminal catalytic region contained a kinase domain with ATP and substrate binding sites. The PKG1 isoforms differed in the N-terminal sequence that included the LZ domain and AI site (Fig. 5).

MSA of decapod sequences identified four PKG1α isoforms, designated PKG1α1, PKG1α2, PKG1α3 and PKG1α4 (Fig. 6). PKG1α isoforms differed in the N-terminal sequence that included the LZ domain. The isoforms were first distinguished by the length and amino acid sequence upstream from the LZ domain (Fig. 6A–D). PKG1α1 sequences in Anomura and Brachyura began with MPL[P/S] (reference positions 1–4) and ranged from 14 to 20 amino acids in length before the beginning of the LZ domain (Fig. 6A). PKG1α2 was identified in all decapod taxa. This isoform began with MPQ[L/F] (reference positions 1–4) and was 14 to 15 amino acids in length (Fig. 6B). PKG1α3 was identified in Anomura, Astacidea, Brachyura and Caridea. The protein began at the LZ domain sequence MGS[L/M] (reference positions 21–24; Fig. 6C). PKG1α4 was identified only in Achelata, Astacidea and Caridea. This isoform began with MS[E/L/K/Q] (reference positions 1–3; Fig. 6D). The LZ domain commonly began with MES (reference positions 19–21), rather than MGS found in the other alpha isoforms (Fig. 6D). Other PKG1α sequences that occurred in Astacidea and Caridea could not be assigned to the four isoforms. These sequences were designated ‘PKG1α-’, as they appeared to be an additional isoform unique to Astacidea and Caridea (Fig. 3). Sequences for Gl-PKG1α and Cm-PKG1α isoforms were validated by direct sequencing of purified PCR products.

MSA of decapod sequences identified one PKG1β isoform. Compared with PKG1α isoforms, the PKG1β sequences had an extended N-terminal sequence rich in glycine residues that ranged in length between 33 and 47 amino acids upstream of the LZ domain (Fig. 6E). The PKG1β LZ domain sequence began with LRDT (reference positions 63–65) and diverged significantly from the alpha isoforms until the 3′ end of the LZ domain, in which all the PKG1 isoforms ended with LDKxxSV[I/M/L]P (Fig. 5). Sequences for Gl-PKG1β and Cm-PKG1β isoforms were validated by direct sequencing of purified PCR products (see Data availability for raw data).

Sequence analysis of the G. lateralis and C. maenas contigs indicated that the PKG1α1, PKG1α2, PKG1α3 and PKG1β isoforms in CrusTome were alternatively spliced mRNA products of the same gene. The PKG1 isoforms in each species had identical 3′-untranslated regions (UTRs) but varied in 5′-UTR sequences. The ORFs differed in the DNA sequence of the 5′ region encoding the LZ domain and much of the region before the first CNB site (Fig. 7). A conserved serine (reference position 180; indicated by a star) marked the beginning of DNA and protein sequence identity in all four isoforms in each species (Fig. 7). Conserved amino acids predicted to be important for function were identified in the LZ, cGMP-binding and kinase domains (Fig. 7).

Analysis of the Gl-PKG2 and Cm-PKG2 sequences indicated that the contigs were products of a single gene. The sequences were validated by direct sequencing of purified PCR products. Both proteins had a long N-terminal extension consisting of a ∼190-amino acid sequence rich in glycine, proline and a consecutive stretch of histidine (reference positions 150–162; Fig. 8). An annotated LZ sequence, including conserved residues, for PKG2 was identified through IPS as CDD did not identify the LZ domain in these sequences. Conserved amino acids predicted to be important for function were identified in the LZ, cGMP-binding and kinase domains (Fig. 8).

GKAP-docking residues in the LZ domain predicted by CDD and IPS were unique to each isoform. The four predicted GKAP-docking residues were [R/K]DxExK in PKG1α1, [R/H]RxExK in PKG1α2, KQxExV in PKG1α3 and [Q/K]RxExE in PKG1β (reference positions 55, 56, 58 and 60; Fig. 7). The predicted GKAP-docking residues in PKG2 were DVxExQ (reference positions 209, 210, 212 and 221; Fig. 8). The location of putative decapod GKAP-docking residues within the LZ aligns with that of annotated reference sequences in human and mouse (alignment available on Dryad at https://doi.org/10.5061/dryad.1g1jwsv7b). Glutamic acid (E) is the third amino acid that was conserved in all human, mouse and decapod GKAP-docking sites for PKG1 and PKG2. However, the first, second and fourth GKAP-docking residues varied between mammalian and decapod sequences.

Unexpectedly, PKG1 sequences with a 14- to 17-amino acid insertion located in the kinase domain were identified in 10 decapod and one stomatopod species (Fig. 6F; Table S2). For all but two species, a matching transcript was identified for the same isoform that did not contain the insertion (Fig. 6F). This conserved insertion was identified in several different PKG1 isoforms, including PKG1α2, PKG1α3, PKG1α4, the ungrouped PKG1α- and PKG1β (Fig. 3). Sequences containing this insertion were denoted with ‘ins’ to differentiate them from the corresponding isoform sequence lacking the insertion (Fig. 6F). The insertion appeared to be unique to crustaceans and was located upstream of the activation loop in the catalytic site. It added amino acids in the region between conserved residues involved in substrate binding, ATP binding and the active site (Fig. 6F). The insertion did not contain additional functional residues annotated by NCBI CDS.

The relative expression of PKG1 and PKG2 in G. lateralis and C. maenas YOs was calculated using transcriptomic data from intermolt animals. In G. lateralis, only the Gl-PKG1β isoform was identified (Das et al., 2018), which was most likely because of the assembly methods used, which did not distinguish between isoforms. Therefore, the Gl-PKG1α isoforms were likely included with Gl-PKG1β, resulting in an overall quantification of Gl-PKG1 expression. For C. maenas, all Cm-PKG1α isoforms were identified in the CrusTome database. However, Cm-PKG1α1 and Cm-PKG1α2 were expressed in both the YO and the central nervous system, whereas PKG1α3 was only expressed in the central nervous system (Table 3). Regardless of isoform, PKG1 transcript expression in the intermolt YO far exceeded that of PKG2 in both species. Gl-PKG1 expression was two orders of magnitude greater than that of Gl-PKG2 (Fig. 9A). Cm-PKG1 expression was three orders of magnitude greater than that of Cm-PKG2 (Fig. 9B).

Phylogenetic analysis established that ecdysozoan species have two PKG genes, designated PKG1 and PKG2. These were homologous to the prkg1/PKG I and prkg2/PKG II genes in chordates (Fig. 2), which indicates a remarkable conservation of PKG genes across metazoan taxa. It appears that there has been no gene duplication since the protostome and deuterostome lineages split over 500 million years ago. The crustacean and vertebrate PKGs have the same domain organization (Fig. 5) (Leroux et al., 2018; Kim and Sharma, 2021). Moreover, the similarity extends to the major PKG1/PKG I isoforms produced by alternative mRNA splicing. Two PKG1 and PKG I isoforms, designated as alpha and beta, have identical cGMP-binding and C-terminal regions, but differ in the N-terminal region (Figs 2, 5) (Kim and Sharma, 2021). An exception is Drosophila, in which four alternative promoters and alternative mRNA splicing of the foraging (PKG1) gene produce 21 isoforms (Allen et al., 2017). These isoforms share the same kinase domain but differ in the N-terminal region and cGMP-binding domain (Allen et al., 2017). By contrast, PKG2/PKG II is expressed as a single sequence, indicating that alternative mRNA splicing does not occur (Fig. 5) (Kim and Sharma, 2021). It is notable that the representation of annotated PKG2 sequences is scarce even in repositories, such as RefSeq (NCBI) and UniProtKB, as much of the research has focused on PKG1 and its role in cardiac function (reviewed in Hofmann, 2018; Leroux et al., 2018; Adler et al., 2020; Cuello and Nikolaev, 2020).

Decapod PKGs are likely encoded by two genes, PKG1 and PKG2, that are orthologous to the two corresponding single-copy genes found in humans, mice, Drosophila and other model organisms (Fitzpatrick and Sokolowski, 2004; Kim and Sharma, 2021). Although genomic sequences for G. lateralis or C. maenas are not available, analyses of the mRNA and amino acid sequences are consistent with transcripts that were generated from two distinct PKG genes. First, the G. lateralis or C. maenas PKG1 mRNAs shared a common 3′-UTR but differed in the 5′-UTRs. Human and mouse prkg1/PKG I transcripts also share a common 3′-UTR but differ in the 5′-UTR sequences between the α and β isoforms (Sellak et al., 2013). Second, the splice site that joins either the α or β N-terminal sequence to the C-terminal human and mouse prkg1/PKG I sequences (Sandberg et al., 1989) aligns at the same position in both nucleotide and amino acid sequences with all G. lateralis and C. maenas PKG1 isoforms. Third, PKG2 is likely a single and separate gene, as only one sequence was identified for G. lateralis and C. maenas and in 10 other decapod species (Fig. 4) and model organisms (Hofmann, 2020). Moreover, PKG1 and PKG2 transcripts differed in both the 3′-UTR and 5′-UTR sequences. The absence of PKG2 in penaeid shrimps was likely due to low transcript expression, as opposed to loss of the gene.

Crustacean species express numerous PKG1 isoforms. Identification of multiple isoforms was made possible by taking advantage of the CrusTome RNA sequence database, which consists of reassembled transcriptomes representing different tissue types and 189 species from the major crustacean taxonomic groups (Pérez-Moreno et al., 2023). Five PKG1α isoforms, designated α-, α1, α2, α3 and α4, and one PKG1β isoform were identified in decapod species (Fig. 3). Sequencing of PCR products from the sequences that spanned the LZ domain and putative splice site in the Gl-PKG1 and Cm-PKG1 isoforms confirmed that the isoforms were alternatively spliced products of the same gene. In addition, a novel alternatively spliced variant was identified. Designated ‘ins’, this PKG1 isoform had a 14- to 17-amino acid insertion in the kinase domain (Fig. 6F). In most cases, corresponding PKG1α and PKG1β isoforms without the insertion were identified (Figs 3, 6F). This suggests that alternative splicing occurs outside of the amino-terminal region in crustaceans. Although located in a highly conserved region between residues involved in substrate binding, ATP binding and kinase activity (Fig. 6F), the functional significance of this insertion is unknown.

PKG1 isoforms and PKG2 sequences differed in the N-terminal region, which has structural and regulatory functions. The LZ domain mediates isoform-specific homodimer formation and subcellular localization (Francis and Corbin, 2013; Kim and Sharma, 2021; Sharma et al., 2022; Hofmann, 2020). The length and amino acid sequences of the LZ domains differed among the PKGs, suggesting that dimerization only occurs between the same proteins (Fig. 6). The LZ domain also had four amino acid residues responsible for binding GKAPs, which are isoform-specific and mediate subcellular localization (Casteel et al., 2010). Variation among the predicted GKAP-docking residues for decapod isoforms suggests that each isoform selectively interacts with different GKAPs and may therefore have a unique subcellular localization, which mirrors PKG isoform-specific differences in mammalian cells (Corradini et al., 2015). The specificity of GKAP interactions with individual PKG isoforms is mainly mediated by the physiochemical properties of the amino acid residues in each isoform. For example, negatively charged residues in human PKG1β interact with positively charged residues on associated GKAPs (Casteel et al., 2005; Corradini et al., 2015). A positively charged lysine in PKG1α mediates the formation of the PKG1α-GKAP complex (Sharma et al., 2008). In contrast, the LZ of PKG2 interacts with a characterized GKAP through van der Waals forces (Corradini et al., 2015). The physiochemical properties of decapod GKAP-docking sites are unique to each isoform, such as charged amino acids in PKG1α1, PKG1α2, PKG1α4 and PKG1β, and hydrophobic or uncharged polar amino acids in PKG1α3 and PKG2 (Figs 6–,8).

The AI site is a six amino acid motif located between the LZ and CNB-A domains (Fig. 5). AI sites are pseudo-substrate motifs that block substrate binding within the catalytic core (Francis and Corbin, 2013; Sharma et al., 2022). The AI sites characterized in human PKG sequences are largely conserved in the decapod sequences (Francis and Corbin, 2013; Sharma et al., 2022). The AI motif sequence (RAQGIS) in human PKG1α is nearly identical in all the decapod PKG1α isoforms (Fig. 7). For PKG1β, the human and selected decapod AI sequences differ by one residue (KRQAIS and KRTAIS, respectively; Fig. 7). For PKG2, the human and selected decapod AI sequences differ at two resides (AKAGVS and KKQGVS, respectively; Fig. 8; Francis and Corbin, 2013; Sharma et al., 2022). These differences in the LZ domain and AI site determine isozyme-specific activation constants and affinities for cGMP (Sharma et al., 2022). Given the strong similarities between decapod and mammalian PKGs, it is likely that the biological properties of these proteins are retained in decapods. These data suggest that expressing multiple PKG1 isoforms broadens its function by expanding the subcellular localization of the protein and/or allowing for activity across a range of cGMP concentrations.

PKG1 and PKG2 had opposite effects on MIH-dependent YO ecdysteroid synthesis and secretion. Experiments using PKG inhibitors±Gl-MIH showed that PKG1 inhibits ecdysteroidogenesis, whereas PKG2 stimulates ecdysteroidogenesis (Fig. 1). Inhibition of both PKG1 and PKG2 with a cGMP analog reversed the effects of MIH in the C. maenas YO, but not in the G. lateralis YO (Fig. 1A,B). In addition, Gl-MIH alone was more effective in inhibiting YO ecdysteroid secretion in C. maenas than in G. lateralis (see Results). Although PKG1 had a dominant role in both species, these data suggest that the two species differed in the relative contributions of PKG1 and PKG2 activity to MIH signaling. RNAseq data showed that PKG1 was expressed at levels approximately two and three orders of magnitude greater than PKG2 in G. lateralis and C. maenas, respectively (Fig. 10). Assuming that protein levels are correlated with transcript levels, the PKG1:PKG2 ratio is estimated to be approximately 10-fold greater in C. maenas than in G. lateralis. As MIH activates both PKGs, a lower PKG1:PKG2 ratio would lessen the effect of the Gl-MIH±cGMP analog on the G. lateralis YO ecdysteroid secretion. In other words, PKG2 counters the inhibitory effect of PKG1, and the greater the PKG2 activity, the greater the countering effect would be. Consequently, Gl-MIH would have a greater inhibitory effect on C. maenas YO than on G. lateralis YO. Moreover, the cGMP analog inhibitor would be more effective in reversing the effect of Gl-MIH on the C. maenas YO than on the G. lateralis YO, which is supported by the results (Fig. 1, Table 1). These data are supported by previous studies in which ecdysteroid secretion by activated YOs was inhibited by the cGMP analog 8-Br-cGMP to 50% in G. lateralis and 34% in C. maenas (reviewed in Covi et al., 2009). Notably, a difference in which PKG1 isoform predominates in the YO may also contribute to species-specific differences in responsiveness to MIH, although at this time we do not know which isoforms are dominant or how differences in the alternatively spliced regions may affect cellular location and/or affinity for cGMP binding.

This study has revealed a fascinating example of paralogs that have opposing, rather than redundant, effects. It also raises the possibility that PKG1 and PKG2 have opposing roles in vertebrates. Three separate paralogous gene pairs associated with Huntington's disease have opposing effects and interact with a similar network of proteins (Vagonia et al., 2022). One such pair is inhibitor of nuclear factor kappa B kinase subunit beta (IKBKB) and inhibitor of nuclear factor kappa-B subunit alpha (IKKA). DNA damage has opposite effects on the paralogs: increased IKBKB activity and decreased IKKA activity promotes cleavage of the huntingtin protein (HTT; Khoshnan et al., 2009; Vagonia et al., 2022). Conversely, increased IKKA activity and decreased IKBKB activity blocks HTT cleavage, which protects against the neurodegenerative effects of Huntington's disease (Khoshnan et al., 2009; Vagonia et al., 2022).

The discovery that the PKG2 inhibitor AP-C5 enhanced the inhibitory effect of Gl-MIH on YO ecdysteroidogenesis answers a longstanding question in decapod endocrinology. The magnitudes of the synergistic effect of Gl-MIH and AP-C5 on C. maenas and G. lateralis YO secretion (90% and 76% inhibition compared with Gl-MIH controls; Fig. 1, Table 1) are larger than those reported for MIH alone (Mattson and Spaziani, 1985; Webster, 1986; Saidi et al., 1994; and reviewed in Lachaise et al., 1993; Mykles, 2021). Numerous studies have shown that YOs incubated with high concentrations of MIH or sinus gland extract reach a maximum of 60–70% inhibition, depending upon the species (Lachaise et al., 1993; Covi et al., 2012). The opposing effects of PKG1 and PKG2 explain how MIH prevents molting, while allowing for basal ecdysteroidogenesis in the intermolt YO. As PKG1 is expressed at much higher levels than PKG2, PKG1 has the dominant role in MIH suppression of ecdysteroid synthesis, whereas PKG2 counteracts that suppression and allows for basal ecdysteroid secretion during intermolt.

MIH signaling controls mTORC1-dependent ecdysteroid synthesis in the decapod YO. Mykles (2021) proposed a model in which PKG activates the TSC complex by phosphorylating the TSC2 subunit; TSC inactivates Rheb by promoting the hydrolysis of GTP to GDP, resulting in mTORC1 inhibition. TSC2 is regulated by various protein kinases, such as PKG, AKT and AMP kinase (Ranek et al., 2019; Oeing et al., 2020a,b; reviewed in Shi and Collins, 2023). The effects of protein kinases are determined by which target sites in TSC2 are phosphorylated. In mammals, PKG1 and AMP kinase activate TSC2, whereas AKT inhibits TSC2 (reviewed in Shi and Collins, 2023; Da Silva et al., 2022; Yin et al., 2021). As decapod PKG1 and PKG2 have opposite effects on MIH-dependent ecdysteroidogenesis, a revised model is proposed, in which PKG1 and PKG2 activate and inhibit TSC, respectively, by phosphorylating TSC2 at different sites (Fig. 10). Although MIH activates both PKG1 and PKG2, PKG1 has the dominant role, as it is expressed at two to three orders of magnitude greater than PKG2 (Fig. 9). This leads to inhibition of mTORC1-dependent ecdysteroid synthesis. However, PKG2 activity prevents total inhibition of ecdysteroid synthesis in the basal YO, resulting in low hemolymph ecdysteroid titer in intermolt animals (Mykles, 2011). Further, oxidation of PKG1α at Cys42 tempers PKG1 inhibition of mTORC1 by reducing phosphorylation and activation of TSC2 (Oeing et al., 2020a,b). Oxidation of Cys117 and Cys195 in the cGMP-binding domains of PKG1α modulates subcellular localization and protein activity, thereby conferring a redox-responsive function in mammalian cells (Cuello and Nikolaev, 2020). This may also have implications for PKG1 in the YO during premolt when increased cytochrome p450 activity, which supports increased ecdysteroid synthesis, may increase production of radical oxygen species (ROS) (Mykles and Chang, 2020). This is consistent with increases in anti-ROS proteins during premolt (Head et al., 2019).

Gecarcinus lateralis and C. maenas have been used for decades as models for the study of molting physiology (reviewed in Mykles and Chang, 2020; Mykles, 2024; Webster et al., 2012, 2013; Fehsenfeld, 2024). Although decapods share the same endocrine system that controls molting, the two species show notable differences in the effects of experimental treatments on the MIH signaling pathway. Acute withdrawal of MIH by ESA results in YO activation in G. lateralis; this method was used here and in previous studies to determine the effects of MIH and reagents on ecdysteroid secretion (Mykles and Chang, 2020; Mykles, 2024). By contrast, adult C. maenas are refractory to ESA, which was attributed to extra-eyestalk sources of MIH in the central nervous system (McDonald et al., 2011; Abuhagr et al., 2014; Pitts et al., 2017). However, when removed from the animal, the YOs are activated and begin secreting ecdysteroids immediately (Fig. 1). MIH and compounds that inhibit ecdysteroid secretion, such as PKG agonist (8-Br-cGMP), phosphodiesterase inhibitor (IBMX), NO donors (SNAP and SE175) and NO-dependent guanylyl cyclase agonist (YC-1), have a greater effect on the C. maenas YO than the G. lateralis YO in vitro (present study; Covi et al., 2009; Mykles et al., 2010). These data suggest that the cGMP turnover rate is higher in the C. maenas YO than in the G. lateralis YO. Consequently, the C. maenas YO would be more responsive to changes in MIH level in the hemolymph, which would explain the immediate activation of C. maenas YO in vitro. The difference in cGMP turnover would also contribute to the differential effect of Rp8-Br-PET-cGMPs on MIH-dependent inhibition between the two species (Fig. 1).

Future studies should be directed toward identifying the substrates of PKG1 and PKG2 in the decapod YO, to determine how MIH and mTORC1 signaling pathways are linked. In human and mouse cardiomyocytes, PKG1 activates the TSC complex by phosphorylating TCS2 at Ser1365 and Ser1366 (Ranek et al., 2019; Oeing et al., 2020a,b), resulting in Rheb inactivation and mTORC1 inhibition (Shi and Collins, 2023). Furthermore, quantifying PKG mRNA and protein levels over the molt cycle may elucidate the roles of PKG1 and PKG2 as the YO transitions through the basal, activated, committed and repressed states.

We thank Hector C. Horta and Rafael Polanco for the collection of Gecarcinus lateralis in the Dominican Republic, and Karl L. Menard and Samuel S. Briggs of the University of California Davis Bodega Marine Laboratory for the collection and husbandry of Carcinus maenas. We thank Dr Mihika T. Kozma for initial characterization of Gl-MIH biological activity. Finally, we acknowledge the use of the Supercomputing Center for Education & Research (OSCER) at the University of Oklahoma for providing high-performance computing resources for the bioinformatic work.