J. Neurochem. (2010) 115, 1024–1034.
Temporal carbohydrate expression patterns at cell surfaces are thought to be of crucial regulatory significance during developmental processes. Hitherto, however, data on individual membrane proteins undergoing development-associated changes in glycosylation are sparsely. Here, we show that the two natriuretic peptide receptors, guanylyl cyclase-A (GC-A) and GC-B are subject to pronounced size alterations in the rat brain between postnatal day 1 and adult. Comparable size changes were not detectable for GC-A and GC-B in peripheral tissues and for three other membrane proteins (insulin receptor, insulin-like growth factor-II/mannose-6-phoshate receptor, neutral endopeptidase) in brain, indicating remarkable specificity. As revealed by treatments with carbohydrate-digesting enzymes, both GC-A and GC-B are hyperglycosylated at N-linked glycosylation sites in the developing brain. At postnatal day 1, the vast majority of GC-B (but not GC-A) molecules contain additionally an O-linked carbohydrate modification of about 1 kDa in mass and a further modification of similar size which is resistant to enzymatic removal. The glycoforms exhibited functional activity in membrane GC assays, indicating proper folding and signaling capability. These data link recently reported roles of natriuretic peptides during brain development for the first time with specific glycosylation states of their cyclic GMP-generating receptors.
atrial natriuretic peptide
C-type natriuretic peptide
- GC-A, GC-B
natriuretic peptide receptors-A and B
insulin-like growth factor-II/mannose-6-phoshate
neutral endopeptidase (neprilysin)
natriuretic peptide receptor
postnatal day 1
sodium dodecyl sulfate–polyacrylamide gel electrophoresis
Guanylyl cyclase-A (GC-A) and GC-B are plasma membrane receptors for members of the natriuretic peptide (NP) family. GC-A (also known as NP receptor-A, NPR-A) is the common receptor for atrial NP (ANP) and B-type NP, whereas GC-B (NPR-B) specifically interacts with and is activated by C-type NP (CNP; Potter et al. 2006; Kuhn 2009). The receptors are composed of an extracellular ligand binding, a single membrane-spanning and an intracellular GC domain which generates accumulations of cyclic GMP (cGMP) upon agonist binding. In addition, they contain a kinase homology domain, where phosphorylation/dephosphorylation at several serine/threonine residues modulates receptor activity (Potter et al. 2006).
Natriuretic peptide signaling via GC-A/GC-B has broad physiological impact by regulating blood pressure, fluid volume homeostasis, and controlling a variety of organ-specific functions (Potter et al. 2006; Kuhn 2009). Reported neuronal disorders in GC-B knockout (Ko) mice (Tamura et al. 2004), errors in axonal branching observed in CNP mutant mice (Schmidt et al. 2009), and accumulating evidence for distinct and stage-specific activities of NPs and their receptors in the developing brain (Simpson et al. 2002; DiCicco-Bloom et al. 2004; Jankowski et al. 2004; Waschek 2004; Cao and Yang 2008; Müller et al. 2009) recently raised the attention to important functions in the CNS.
Guanylyl cyclase-A and GC-B are structurally closely related proteins with very similar cDNA-deduced molecular masses in the range of 115 kDa (SwissProt). Their native molecular weights are significantly higher (by about 5–15 kDa) because of glycosylation (Lowe and Fendly 1992; Fenrick et al. 1997; Miyagi et al. 2000; Müller et al. 2002). The rat proteins contain six (GC-A) and seven (GC-B) consensus N-glycosylation sites, respectively (SwissProt), and N-linked glycosylation appears to represent the major size-affecting post-translational modification (Fenrick et al. 1997; Miyagi et al. 2000; Müller et al. 2002). However, the physiological consequences of glycosylation, including potential effects on ligand binding and cell surface expression, are still poorly understood (Kuhn 2009). Moreover, the post-translational modifications of the two receptors in vivo and their potential regulation during developmental processes are largely unknown, since currently available data are based to a great extent on studies at the mRNA level and investigations with receptor-transfected cell lines.
Regarding structural features of NPRs, the brain is of particular interest. Previous studies demonstrated a CNS-specific N-linked glycosylation of GC-A in adult rats, resulting in an 8 kDa lower molecular mass than in peripheral tissues (Müller et al. 2002). The same investigation, however, discovered that the size of GC-A in the early postnatal brain is similar to that in peripheral tissues, indicating marked but not yet molecularly characterized alterations during brain development. The latter issue is of interest, considering the pivotal role of alterations in the glycosylation state of cell surface receptors during development (Haltiwanger and Lowe 2004; Laughlin et al. 2008) and their impact for proliferation and differentiation (Lau et al. 2007). Specific spatial and temporal expression patterns of diverse carbohydrate structures, involving both N- and O-linked protein glycoconjugates, are a hallmark of developmental processes (Haltiwanger and Lowe 2004; Zhang et al. 2008; Montpetit et al. 2009). Thus, attempts to elucidate the molecular mechanism(s) underlying the different sizes of GC-A in the maturing versus adult brain were consequential.
While brain GC-A levels increase in the course of development, GC-B is most highly expressed in the perinatal brain (Müller et al. 2009). These findings and characterizations of the GC-B-expressing cells provided evidence that CNP/GC-B signaling specifically controls perinatal stages of brain maturation (Müller et al. 2009). This raised the obvious question, whether GC-B, like its relative GC-A, is subject to structural alterations during brain development.
After extensive efforts to establish and optimize detection of GC-A and GC-B in tissue extracts, we were able to analyze the native receptor proteins in the rat brain during postnatal development. We found that both GC-A and GC-B undergo profound size changes in brain but not in peripheral tissues during the course of development. Revealing a molecular basis and indicating a strikingly similar regulation, both NPRs are hyperglycosylated in the neonatal brain at N-linked glycosylation sites. The identification of additional post-translational modifications, one of which represents O-linked glycosylation, discriminates GC-B from GC-A. These brain- and developmental stage-specific receptor modifications are proposed to be associated with and of functional importance for reported activities of NP signaling during neurogenesis/brain maturation.
Materials and methods
Unlabeled peptides were purchased from Bachem (Weil am Rhein, Germany), 125I-labelelled ANP, insulin, and insulin-like growth factor-II (IGF-II), 2 kCi/mmol each, from Amersham (Freiburg, Germany).
Animals and tissues
Wistar rats (Charles River Laboratories, Sulzbach, Germany) of different developmental stages were dissected after decapitation, and the tissues were immediately frozen in liquid nitrogen and stored at −80°C. Total brain tissues contained olfactory bulbs but not spinal cord. Isolated olfactory bulbs (n = 8–18), hypothalami (n = 6–14), and aortae (n = 3–4 each) were pooled prior to homogenization. Animals were used according to government principles regarding the care and use of animals with permission (G8151/591–00.33) of the local regulatory authority. GC-A Ko mice and respective wild-type controls (Yurukova et al. 2007) were 7 months old.
Preparation of membrane protein fractions
Frozen tissues were pulverized in a mortar, suspended in 10 mL/g tissue of ice-cold homogenization buffer (Müller et al. 2002) and homogenized by 10 strokes in a Potter–Elvehjem homogenizer (Wheaton, Millville, NJ, USA). After centrifugation at 3000 g for 8 min at 4°C to remove cell debris and nuclei, the supernatant fractions were centrifuged for 30 min at 100 000 g at 4°C. After withdrawal of the supernatants, the crude membrane pellets were re-suspended in 50 mm Tris-HCl buffer, pH 7.5, and stored at −80°C. Protein concentrations were determined by using a kit from Bio-Rad (Munich, Germany) with bovine serum albumin (fraction V; Sigma, St Louis, MO, USA) as standard.
Receptor affinity labeling
Protocols for photoaffinity labeling of GC-A by 125I-ANP (Müller et al. 2002) have been reported before. In brief, membranes were incubated with radiolabeled ANP (0.5 nm), ligand/receptor cross-links were induced by UV light irradiation, and reaction products were analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) under reducing conditions in 7.0% acrylamide separation gels. After staining with Coomassie blue R-250 to visualize co-migrated size marker proteins (SDS-6H; Sigma), gels were dried and subjected to autoradiography at −70°C using Kodak XAR-5 films (Kodak, Rochester, NY, USA) and intensifying screens. Insulin and IGF-II/mannose-6-phoshate (Man-6-P) receptors were photoaffinity labeled by UV light irradiation (see before; 125I-insulin: 1 nm; 125I-IGF-II: 2 nm).
SDS–PAGE and receptor size determinations
Proteins were separated by SDS–PAGE under reducing conditions in 7% acrylamide (acrylamide to bisacrylamide ratio = 37.5 : 1; in g each) gels. To verify small size differences and/or to optimize protein band resolution, aliquots were also analyzed in running gels containing 6% or 8% acrylamide and different (e.g., 75 : 1) acrylamide to bisacrylamide ratios.
Apparent molecular masses of native and deglycosylated proteins, detectable after immunoblotting, were calculated from graphs of the mobility of protein standards (SDS-6H; Sigma) as a function of their molecular masses (45, 66, 97, 116, 205 kDa).
After SDS–PAGE, proteins were transferred to nitrocellulose membranes as specified before (Müller et al. 2004). Blots were stained with Ponceau S (Sigma), and the positions of co-migrated reference proteins (SDS-6H; Sigma) were marked. Records of the protein images were derived by scanning prior to destaining the blots in H2O and served to control protein loading. Blots were pre-treated with blocking solution (#1096176; Roche Molecular Biochemicals, Indianapolis, IN, USA) as described (Müller et al. 2004). For GC-A detection, either affinity-purified rabbit polyclonal antibodies generated in our laboratory (against the peptide sequence CKGKVRTYWLLGERGSSTRG, which is located near the C-terminus of GC-A) or commercial ones (PGCA-101AP; FabGennix, Frisco, TX, USA) were used with comparable results. Affinity-purified polyclonal (PGCB-201AP; FabGennix) or monoclonal (NCL-CD10–270; Menarini Diagnostics, Berlin, Germany) antibodies were used for detection of GC-B and neutral endopeptidase (NEP), respectively. Goat anti-rabbit or anti-mouse IgG (Pierce, Rockford, IL, USA), linked to peroxidase, served as secondary antibodies, and signals were detected via enhanced chemiluminescence (Müller et al. 2004). For GC-A and GC-B detection, blots were routinely incubated with blocking solution, then stripped (Müller et al. 2010) and re-incubated with blocking solution prior to antibody exposure. This treatment was found to further reduce background staining of the blots, leading to enhanced resolution of specific immunoreactivity.
Membrane samples (40 up to 120 μg of protein) were incubated in 100 μL of 30 mm sodium phosphate buffer, pH 7.5, containing 10 mm EDTA, 1% (v/v) Triton X-100, 0.1% (w/v) SDS, 2.5 mm dithiothreitol, 0.2 mm phenylmethyl sulphonylfluoride, and 4 U recombinant N-glycosidase F (# 11 365 169; Roche) for 15 h (unless stated otherwise) at 37°C. For analyses of O-linked carbohydrate modifications, O-glycosidase (# 11 347 101; Roche) and neuraminidase (# 10 269 611) were present at 2.5 and 50 mU/100 μL, respectively. Analogous incubations in the absence of the enzymes served to prove specificity of reactions. Samples were stored at −80°C until reaction product analysis via immunoblotting.
Membrane GC assays and measurement of cGMP
Basal and NP-stimulated GC activity in membrane preparations was characterized as described (Müller et al. 2004). In brief, membranes (10 μg protein) were incubated for 12 min at 37°C in either the absence or presence (1 μm) of NPs prior to measurements of cGMP formation by a commercial ELISA (Müller et al. 2004).
Generation of cGMP and homologous desensitization of GC-A in intact aortae
The experimental details are provided as Appendix S1.
Experimental reproducibility and data analysis
The figures shown were representative of at least five independent experiments performed. Densitometric quantifications were carried out using ImageJ (NIH, open source). Data were graphed and analyzed using Prism 3.02 (GraphPad Software Inc., San Diego, CA, USA). The significance of effects was assessed using Student’s t-test.
GC-A and GC-B size alterations during brain development
Immunoblot analyses with membrane proteins from either rat postnatal day 1 (P1) or adult brain revealed that the vast majority of P1 GC-A immunoreactivity appears as a band at 127 kDa, whereas GC-A of the mature brain migrates at 119 kDa (Fig. 1a). The calculated apparent molecular mass of the latter was fully consistent with the value of 122 kDa obtained previously by receptor affinity-labeling experiments (Müller et al. 2002), taking into account the mass (3.2 kDa) of the covalently linked GC-A ligand, ANP. To serve as a size reference, we co-analyzed GC-A expression in a peripheral tissue (penis, 129 kDa; Fig. 1a), shown to produce GC-A molecules of higher molecular weight than in brain (Müller et al. 2002). Importantly, these findings corroborated by immunological detection previous evidence (Müller et al. 2002) for size alterations of GC-A during brain development.
Comparative analyses of GC-B uncovered a striking similarity to GC-A. Like GC-A, GC-B has a lower apparent molecular mass in the adult brain than in penis and P1 brain (Fig. 1a). However, unlike that of GC-A, GC-B expression in brain decreases between P1 and adult (Müller et al. 2009) and shows a certain size heterogeneity in adult brain tissue. Molecular mass determinations revealed mean values of 132 (penis), 130 (P1 brain), and predominantly 122 (adult brain) kDa (Fig. 1a). These findings demonstrated that GC-B (like GC-A) is subject to pronounced molecular weight changes during brain development.
Decreases in the molecular weights of GC-A and GC-B between P1 and adult were likewise evident when different brain areas (cerebral cortex, cerebellum, brain stem) were examined separately (Fig. S1). RT-PCR analyses with primers flanking reported sites of alternative splicing in GC-A (Hartmann et al. 2008) and GC-B (Tamura and Garbers 2003) did not show development-dependent structural changes at the mRNA level (Fig. S2). Moreover, the neonatal sizes of GC-A and GC-B were indiscriminative between males and females or when brain tissues were homogenized in either the absence or presence of phosphatase inhibitors (Fig. S3).
We next addressed at which stage during development the NPR expression in brain switches from the neonatal (peripheral-like) to the adult forms. Immunoblot analyses revealed (Fig. 1b) that the size changes is a gradual process, involving a detectable molecular weight reduction already between P1 and P5, and further proceeds to the adult forms during later stages. These findings were confirmed by affinity-labeling experiments, demonstrating the size reduction of GC-A between P10 and P20 (Fig. 1c).
The structural difference between adult peripheral (penis) and central (brain) GC-A and GC-B was similarly evident in mice (Fig. 1d). Co-analyses of GC-A Ko mice served to prove specific immunoblot detection of the two NPRs.
The size changes of GC-A and GC-B during development are brain- and NPR-specific
Developmental size alterations of GC-A were not observed in previous studies with rat testis and lung (examination from P10 to P90) tissue (Müller et al. 2004). Such effects remained also undetectable for GC-A and GC-B (Fig. 2a shows representative immunoblots) between P1 and adulthood (P90) in lung. Further evidence for the absence of size changes in peripheral tissues (kidney, liver) is provided in Fig. S4.
We then asked whether other glycosylated membrane proteins undergo comparable size alterations in brain during development. For this investigation, we first selected the ANP- and CNP-degrading NEP (also known as neprilysin, EC 220.127.116.11) considering its functional relationship to NP signaling (Potter et al. 2006). The transmembrane glycoprotein NEP degrades several peptides besides ANP and CNP and has attracted particular attention in brain physiology by its role in amyloid β protein proteolysis with impact for cerebral amyloid β protein plaque formation (Hemming et al. 2007). We found low levels of NEP in brain (as compared with lung) by immunoblotting (Fig. 2b) and recognized that the protein has a smaller size (97 kDa) in brain than in lung (∼ 105 kDa) and other peripheral tissues (kidney, adrenal; data not shown) examined. The protein concentrations in brain were similar at P1 and adult, and we did not observe development-dependent size changes (Fig. 2b). Thus, the developmental regulation of NEP in brain is dissimilar to that of GC-A and GC-B, and NEP does not undergo the size shift observed in the case of the two NPRs.
To investigate whether other membrane receptors undergo size alterations during development, we used affinity-labeling approaches to examine the brain expression of insulin and IGF-II/Man-6-P receptors. The insulin receptor is more abundant in the developing than the mature brain (Chiu and Cline 2010) and exhibits, like GC-A (Müller et al. 2002), a lower molecular weight in brain than in other organs (Adamo et al. 1989). Cross-linking with 125I-labeled insulin demonstrated the size difference of the insulin receptor α subunit between peripheral (liver) and brain (olfactory bulb) tissues as well as a preferential expression in the neonatal (P1) brain (Fig. 2c). However, unlike that of GC-A and GC-B, the apparent molecular mass of the insulin receptor was indiscriminative between P1 and adult.
The multifunctional IGF-II/Man-6-P receptor is unrelated to the insulin receptor and binds, in addition to IGF-II, a large number of proteins that are tagged by Man-6-P residues (Kim et al. 2009). Cross-linking with 125I-labeled IGF-II revealed the expression of this receptor in brain and provided evidence for the absence of size (and concentration) changes during postnatal development (Fig. 2d; the whole autoradiogram and specificity controls are shown in Fig. S5). Together, these findings showed a remarkable specificity of the structural alterations of GC-A and GC-B during brain development.
Developmental alterations in N-glycosylation contribute to the size changes of GC-A and GC-B in brain
Previous studies have shown that differences in the degree of N-linked glycosylation explain the size difference of GC-A between CNS and peripheral tissues, since treatments with N-glycosidase F reduced the molecular weights of both receptor variants to the same value of 116.000 (Müller et al. 2002). Considering that these experiments were performed with receptor molecules cross-linked to 125I-ANP (3.2 kDa), analogous size calculations by SDS–PAGE of deglycosylated proteins without coupled ligand should reveal a lower value. To examine whether the peripheral-like size of P1 brain GC-A is also explained by higher amounts of N-linked oligosaccharides, membranes from neonatal brain and adult aorta were treated with N-glycosidase F, and the reaction products were analyzed by immunoblotting. These studies revealed a GC-A size reduction to the same value of 113 kDa (Fig. 3a). When analogous assays were carried out with membranes from adult brain (data not shown), the resulting apparent molecular mass of GC-A was likewise 113 kDa. Thus, specific modes of N-linked glycosylation are not only responsible for the different molecular weights of GC-A between peripheral and adult CNS tissues, but also contribute essentially to the receptor size alterations in brain during development. However, the constant appearance of a second immunoreactive band at ∼ 115 kDa (Fig. 3a, marked by an asterisk) after intensive digestion of P1 brain membrane proteins with N-glycosidase F suggested the existence of a (minor) GC-A subpopulation distinguished by an additional modification unrelated to N-linked glycosylation.
When N-glycosidase F-treated P1 brain membranes were analyzed by immunoblotting for GC-B immunoreactivity, two bands at 113 and 115 kDa became evident (Fig. 3b). This result resembled that obtained for GC-A with the conspicuous exception that the 115 kDa band (marked by an asterisk) in the case of GC-B represents the major receptor variant. Analogous N-glycosidase F treatments of adult brain and penis membranes reduced GC-B molecular masses to 113 kDa (Fig. 3c). Thus, the size difference between native GC-B in adult brain (mainly 122 and 126 kDa) and penis (132 kDa) is explained by different degrees of N-linked glycosylation. GC-B expression in the neonatal brain, however, is exceptional by the existence of an additional (and predominating) receptor population that cannot be converted to the 113 kDa band by N-glycosidase F activity.
To address whether alterations in phosphorylation could induce size-relevant effects in our assays, we examined the migration of GC-A after pre-treatment with ANP, known to elicit receptor desensitization (Potter and Garbers 1992; Müller et al. 2006), which is correlated with dephosphorylation at several sites in the receptor kinase homology domain (Potter et al. 2006; Schröter et al. 2010). Such treatments, performed with freshly isolated aortas, evoked significant desensitization of GC-A (Fig. 3d), but size alterations remained undetectable in the native protein as well as after deglycosylation (Fig. 3e). Likewise, analogous treatments carried out with CNP did not generate detectable size effects in the case of GC-B (Fig. 3f).
The majority of GC-B is O-glycosylated in the neonatal brain
To examine whether other post-translational modifications could explain the appearance of two distinct GC-B bands (113, 115 kDa) after digestion with N-glycosidase F, P1 brain membranes were treated with N-glycosidase F in either the absence or presence of O-glycosidase plus neuraminidase (to facilitate removal of O-linked carbohydrates). In the presence of O-glycosidase and neuraminidase, the 115 kDa band (referred to as R2, Fig. 4a) was changed to a band at 114 kDa. Thus, the GC-B population giving rise to a 115 kDa band after N-glycosidase F treatment is distinguished by O-glycosylation, and this modification is about 1 kDa in mass. In addition, the result indicates the concomitance of a further modification (also of about 1 kDa in mass), which is resistant to digestion by all deglycosylating enzymes applied. These findings are summarized in Fig. 4(b), where explicable receptor modifications in the two GC-B populations (designated as R1 and R2) are marked by N (removable by N-glycosidase F), O (removable by O-glycosidase plus neuraminidase), and X (apparently not removable by these enzymes).
To investigate the developmental regulation of the latter two modifications, analogous assays were performed with brain membranes from later ontogenetic stages. At P5, enzymatic digestions produced the same (113/115 as well as 114 kDa) bands as observed with P1 membranes (Fig. 4a). However, the intensities of the 115 (and 114) kDa bands were selectively reduced, indicating a strong and specific decrease in the R2 GC-B population between P1 and P5 (Fig. 4c). Comparative examinations of later developmental stages demonstrated a progressive disappearance of the 115 kDa band, which was replaced by a diffuse immunoreactivity between 115 and 113 kDa (Fig. 4a, lanes 5 and 7). Bands re-appeared after treatment with O-glycosidase (lanes 6 and 8), indicating that the distinct and relatively homogenous mode of O-linked glycosylation at P1 and P5 is replaced at later stages by a less pronounced and more heterogeneous kind of O-linked modification. Furthermore, the resulting bands after O-glycosidase treatment converged closer to the 113 kDa band (compare lanes 2, 4, 6, 8), providing evidence that the molecular mass amount of the digestion-resistant modification (marked as X in Fig. 4b) also decreases with development.
Since GC-A is poorly expressed in the perinatal brain (Müller et al. 2009), and only a minor portion of this receptor is converted to the brain-specific 115 kDa immunoreactive band after N-glycosidase F treatment (Fig. 3a), characterizations in analogy to those for GC-B were difficult. However, we consistently noted a specific decrease in the intensity of the 115 kDa GC-A band when P1 brain membranes were treated with O-glycosidase/neuraminidase in addition to N-glycosidase F (Fig. S6a and c). Like GC-A, NEP is poorly expressed in P1 brain and abundant in adult lung tissue. To compare the enzymatic deglycosylation between these two proteins, we co-analyzed the effect of O-glycosidase/neuraminidase treatment on P1 brain and the time course of N-glycosidase F-induced deglycosylation in lung membranes. Deglycosylation of native NEP in P1 brain (97 kDa) and lung (105 kDa) resulted in the common appearance of an immunoreactive band at 84 kDa and a second band each of slightly higher molecular mass (Fig. S6b and d). Unlike the 115 kDa GC-A band, none of these bands were affected by O-glycosidase/neuraminidase treatment. This study also showed similar reaction kinetics for N-glycosidase F-induced deglycosylation between GC-A and NEP.
Comparable agonist-dependent GC activities of the neonatal and adult glycoforms, and evidence for an outstanding expression of GC-A and GC-B in the hypothalamus
To examine whether the differential glycosylation states may affect receptor activities, we measured comparatively the ligand-induced cGMP production by GC-A and GC-B in membranes from either neonatal or adult brain. These studies revealed agonist-induced functional activity with membranes from both developmental stages and showed a correlation between receptor protein levels assessed by immunoblotting and receptor GC activities determined by cGMP-ELISA (Fig. 5a–d). These findings provided evidence that the development-specific differences in protein glycosylation do not affect, at least substantially, receptor ligand binding and activation efficiency.
Exceptionally high concentrations of both ANP and CNP in the adult hypothalamus (see Jankowski et al. 2004; and references therein) suggest particular physiological roles for NP signaling within this central structure (e.g., Inuzuka et al. 2010). To disclose the sizes (i.e., extent of glycosylation), functional activities, and relative abundance of the peptide receptors, we characterized the hypothalamic expression of GC-A and GC-B, using brain stem membranes, distinguished by relatively high NPR concentrations (Müller et al. 2009), as reference. While tissue-specific size differences remained undetectable (Fig. 5e), immunoblot analyses and GC assays (Fig. 5e and f) consistently revealed that the hypothalamus represents a paramount expression site of GC-A and particularly GC-B within the adult brain.
Deglycosylation by N-glycosidase F reduced the apparent molecular masses of native GC-A and GC-B from adult brain (119 and 122 kDa, respectively) and peripheral tissues (129, 132 kDa) to the same value of 113 kDa. This data corroborates previous findings (Müller et al. 2002) that lower amounts of N-linked glycosylation discriminate CNS-resident from peripherally expressed GC-A, and shows that this holds also for GC-B. The size reduction to 113 kDa was indicative of a virtually complete removal of size-relevant post-translational modifications. These findings support at the level of endogenous receptor proteins previous investigations, where no protein modifications other than N-glycosylation and disulfide bonds were found by mass spectrometry analyses of the extracellular GC-A domain after transfection into COS cells (Miyagi et al. 2000).
Like GC-A and GC-B, various other membrane proteins, including the insulin receptor and NEP, undergo a special (brain-type) mode of oligosaccharide processing in the adult CNS, resulting in markedly lower molecular masses as compared with peripheral tissues (Adamo et al. 1989; Chen et al. 1998). In contrast, the N-linked hyperglycosylation of GC-A and GC-B in the developing brain was found to be unique, since comparable alterations in post-translational modification were not observed in the case of three other membrane proteins examined. Neither a representative of the receptor tyrosine kinase family (insulin receptor), the peptide-degrading transmembrane protease NEP nor a lectin-related protein (IGF-II/Man-6-P receptor) with diverse cellular functions ranging from lysosome biogenesis to regulation of extracellular IGF-II and cytokine levels (Kim et al. 2009) undergo such developmental changes in glycosylation. Moreover, the developmental regulation of N-linked glycosylation in GC-A and GC-B was shown to be brain-specific. These findings highlight the two cGMP-generating NPRs as exquisite targets for enzymes implicated in protein glycosylation and acting in specific temporal patterns during brain development.
The common hyperglycosylation at P1 and apparently synchronous size reductions during subsequent brain development revealed a strikingly equal regulation between GC-A and GC-B. Curiously, this implies a spatially and temporally very similar accessibility to (and efficiency of) N-glycan modification, despite a directly opposed developmental regulation of GC-A and GC-B protein levels. In contrast, and unveiling a marked difference between the two NPRs, the majority of GC-B (but not GC-A) in the neonatal brain contains an additional carbohydrate modification, which is removed by exposure to O-glycosidase and neuraminidase, an established treatment to release O-glycans (Fahrenkrug et al. 2009; Marion et al. 2009). While the structure and attachment site(s) of the O-linked carbohydrates in perinatal GC-B could not be resolved in the present investigation, our data provide at least some notable information. At P1, and until P5, the mass of this modification is rather homogenous and amounts about 1 kDa. During subsequent development, the size decreases gradually, suggesting either decrements in carbohydrate chain lengths or a progressive removal of residues at multiple attachment sites. In addition, we recognized a further post-translational modification of GC-B in the neonatal brain that was resistant to release by enzymatic digestion. The apparent mass of this modification amounts also about 1 kDa at P1 and decreases during following development. Whether this modification represents unusual glycan structures or is based on different molecular mechanisms remains to be addressed. Concerning the latter, we provided experimental evidence that different degrees of phosphorylation, however, might not be involved.
Potential functional implications
Basically, differences in glycosylation of a membrane receptor can have impact on folding (Helenius and Aebi 2004), trafficking/cell surface localization (Katada et al. 2004), receptor/ligand interactions (Lowe and Fendly 1992), and agonist-induced signal transduction (Ghanekar et al. 2004).
As can be concluded from the apparent molecular masses of GC-A before (P1 : 127 kDa, adult: 119 kDa) and after deglycosylation (commonly 113 kDa), the carbohydrate content of GC-A is more than twofold higher (14 vs. 6 kDa) in the neonatal than adult brain. Comparable proportions apply for GC-B. Since N-linked glycosylation resides in the extracellular (ligand binding) receptor domains, such massive differences in carbohydrate covering could severely affect peptide binding affinities. However, the present together with a recent investigation (Müller et al. 2009) demonstrated a rough correlation between immunologically detected (by western blots) protein levels and those determined by affinity-labeling (receptor/ligand cross-linking) experiments. Thus, the differential carbohydrate content of GC-A and GC-B in the neonatal versus adult brain does not seem to have critical influence on NP binding. These findings are important with respect to earlier and apparently conflicting reports that indicated either a marked role of the degree of GC-A glycosylation for ANP binding (Lowe and Fendly 1992; Marquis et al. 1999) or that GC-A-linked glycans are not at all involved in ANP binding (Miyagi et al. 2000).
Membrane GC assays revealed a correlation between receptor expression levels (at P1 vs. adult) and ligand-induced receptor enzyme activities. Together, these findings suggest the absence of aberrant folding and trafficking phenomena, and ruled out crucial effects on agonist-evoked cGMP production.
The hyperglycosylation of GC-A and GC-B in the developing brain is indicative of an expanded proportion of highly branched N-glycans (Chen et al. 1998; Ye and Marth 2004). Interestingly, recent studies identified a mechanism by which the number of N-glycans in membrane receptors and the degree of N-glycan branching act cooperatively to promote the transition from cell proliferation to arrest/differentiation (Lau et al. 2007). As an essential component of this mechanism, enhanced N-glycan branching reduces receptor endocytosis and thereby increases the cell surface expression and activity of these membrane proteins. Assuming that the enhanced degree of N-linked glycosylation of GC-A and GC-B during development also stabilizes the cell surface expression of these receptors, this could represent a meaningful mechanism to facilitate and enforce differentiation-associated NP activities during neurogenesis (Simpson et al. 2002; Müller et al. 2009). In apparent support, resolution by confocal laser-scanning microscopy revealed a punctual appearance of intracellular GC-B in the neonatal brain, consistent with a receptor regulation by endocytosis (Müller et al. 2009).
In contrast to N-glycosylation, O-glycans use different protein carbohydrate linkages and represent extremely diverse structures (Wopereis et al. 2006). The latter range from complex mucin-type O-glycans structures to post-translational modifications, in which single N-acetylglucosamine molecules are linked, in a reversible manner, to serine and threonine residues, in analogy to phosphorylation (Golks and Guerini 2008). There is convincing evidence that even subtle changes in O-glycosylation can have impact in development (Aoki et al. 2008; Ten Hagen et al. 2009). We show that the vast majority of GC-B in the neonatal brain contains O-linked glycan(s). The proportion of O-modified GC-B molecules decreases rapidly and strongly between P1 and P5, indicating a highly dynamic regulation. Recent studies revealed a pronounced perinatal expression peak of GC-B in brain and provided evidence for a specific and dynamic role in the transition process from neural stem cells to fully differentiated neurons (Müller et al. 2009). It seems likely that O-linked glycosylation is associated with and reflects this particular activity of GC-B. Reported functional implications of O-glycans, potentially attributable to GC-B, include the dynamic regulation of cell adhesion events (Zhang et al. 2008) and axon branching (Francisco et al. 2009) during neural development.
Moreover, the above findings raise the attention to GC-A and GC-B as candidate signaling molecules undergoing altered post-translational modification (and activity) in response to gene defects in so-called congenital disorders of glycosylation. Congenital disorders of glycosylation denotes a large and still growing number of inherited diseases distinguished by partially impaired glycan biosynthetic pathways and hence protein glycosylation (Freeze 2006). Remarkably, delayed brain development, leading to mental retardation, is one of the most consistent findings in these diseases (Freeze 2006).
In conclusion, advanced protein detection allowed to characterize the glycosylation of GC-A and GC-B in vivo and to identify distinctive changes during brain development. Comparable changes were not found in other membrane proteins examined. The different NPR glycoforms showed functional activity in membrane GC assays, indicating proper folding, membrane targeting, and signaling capability. Thus, the temporal carbohydrate modifications elucidated may contribute significantly to the manifold events controlling normal brain development and will further direct the focus on specific roles of NP/NPR signaling in the course of neurogenesis/brain maturation.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB 547 C13; KFO 181/1) and Bundesministerium für Forschung und Technologie (01 KY 9103/0). The authors declare no conflict of interest.