|Home | About | Journals | Submit | Contact Us | Français|
Gαi proteins play major roles in the developing and mature nervous system, ranging from the control of cellular proliferation to modulating synaptic plasticity. Although best known for transducing signals from activated seven transmembrane G-protein coupled receptors (GPCRs) when bound to GTP, key cellular functions for Gαi-GDP are beginning to emerge. Here, we show that Gαi2 is expressed in motor neuron progenitors that are differentiating to form postmitotic motor neurons in the developing spinal cord. Ablation of Gαi2 causes deficits in motor neuron generation but no changes in motor neuron progenitor patterning or specification, consistent with a function for Gαi2 in regulating motor neuron differentiation. We show that Gαi2 function is mediated in part by its interaction with GDE2, a known regulator of motor neuron differentiation, and that disruption of the GDE2/Gαi2 complex in vivo causes motor neuron deficits analogous to Gαi2 ablation. Gαi2 preferentially associates with GDE2 when bound to GDP, invoking GPCR-independent functions for Gαi2 in the control of spinal motor neuron differentiation.
The generation of neuronal diversity in the central nervous system is critical for the formation of functional neural circuits. This complex process involves the integration of signals that trigger neuronal differentiation with those specifying neuronal fates to regulate the timely differentiation of distinct neuronal subtypes (Jessell, 2000; Kintner, 2002). Perturbation of the regulatory networks that control the transition from cell proliferation to differentiation can have serious consequences such as the depletion of progenitor pools, imbalances in neuronal components, tumor formation and ultimately, the disruption of neural networks (Bertram, 2000; Kintner, 2002). However, the molecular mechanisms that regulate the process of neuronal differentiation are still not well understood.
The G protein subunit, Gα plays diverse roles in the central and peripheral nervous system that include controlling key cognitive and sensory processes such as synaptic plasticity, pain, taste, and olfaction (Wettschureck et al., 2004; Malbon, 2005). These functions primarily involve their ability to mediate signals from a varied and large number of seven transmembrane G-protein coupled receptors (GPCRs) (Neer, 1995; Neves et al., 2002). Gα proteins bind quiescent GPCRs in an inactive GDP-bound form that associates in a complex with Gβ and Gγ subunits. However, upon activation of GPCRs through chemical or physical stimulation, the Gα subunit undergoes a GDP to GTP exchange that causes the Gα/β/γ heterotrimer to release from the GPCR, and dissociate into an active Gα-GTP monomer and a Gβ/Gγ dimer. The Gα-GTP subunit subsequently binds to downstream targets or effector proteins, thereby functioning as a central component in GPCR signaling pathways (Neer, 1995; Neves et al., 2002). The Gα subunit also controls the duration of the GPCR-mediated signal by its instrinsic GTPase activity. Cleavage of GTP to the GDP form inactivates Gα, and results in the reassociation of the α/β/γ heterotrimer with the GPCR. Although initially thought to facilitate Gα binding to GPCRs, Gβ/Gγ heterodimers also mediate GPCR signaling, albeit to a lesser extent than Gα (Neer, 1995; Neves et al., 2002).
Interestingly, emerging studies provide evidence that Gαi proteins have additional functions that are distinct from their roles in transducing GPCR signals. In C.elegans, Drosophila and vertebrates, Gαi proteins function as important regulators of asymmetric cell division through their ability to orientate and position the mitotic spindle (Gotta and Ahringer 2001; Shaefer et al., 2001; Hampoelz and Knoblich, 2004; Afshar et al., 2004; Du and Macara, 2004). They mediate this function by associating with Pins (LGN in vertebrates) through the Pins GoLoco domain. In contrast to its function in GPCR signaling, the active form of Gαi in spindle positioning is Gαi-GDP, which binds to Pins/LGN instead of Gαi-GTP (Shaefer et al., 2001; Hampoelz and Knoblich, 2004; Du and Macara, 2004). Pins/Gαi complexes are also implicated in vesicle trafficking and localizing NMDA receptors to the cell membrane, a process central to synaptic plasticity (Sans et al., 2005; Knoblich, 2005). In this paradigm, Gαi-GDP elevates the levels of the NMDA receptor subunit NR2B on the membrane in the presence of Pins, thereby increasing synaptic transmission. These two examples thus indicate that GDP-bound forms of Gαi have important regulatory functions in the nervous system that are distinct from their direct roles in GPCR-mediated signaling.
Given the importance of Gα proteins in nervous system function, we considered the possibility that Gαi proteins may be required for regulating neuronal diversity. One model system where the molecular pathways that regulate neuronal differentiation and subtype specification are relatively well characterized is in developing spinal motor neurons (Jessell, 2000; Price and Briscoe, 2004). Spinal progenitors located in the ventricular zone (VZ) of the spinal cord are patterned into discrete dorsal-ventral domains through integrating sonic hedgehog (shh), fibroblast growth factor (FGF), and retinoic acid signals (RA) (Diez del Corral et al., 2003; Novitch et al., 2003). Each progenitor domain expresses a unique profile of transcription factors that ultimately regulates their capacity to generate a particular neuronal subtype (Jessell, 2000; Price and Briscoe, 2004). In the case of motor neurons, RA signals induce the bHLH protein Olig2 in ventral progenitors, which acts as a key determinant of motor neuron identity by priming cells to implement motor neuron fate specification programs (Mizuguchi et al., 2001; Novitch et al., 2001; Novitch et al., 2003; Lee et al., 2005). RA subsequently initiates the differentiation of Olig2+ progenitors into postmitotic motor neurons through upregulating the expression of GDE2, a six-transmembrane protein containing an extracellular glycerophosphodiester phosphodiesterase (GDPD) domain (Nogusa et al., 2004; Rao and Sockanathan, 2005; Yanaka et al., 2007; Yan et al., 2009). GDE2 GDPD activity triggers Olig2 downregulation, and synchronizes Ngn2-dependent neurogeneic pathways and motor neuron fate specification networks to drive the differentiation of postmitotic motor neurons (Rao and Sockanathan, 2005; Yan et al., 2009).
Here, we investigate the expression of Gαi proteins in spinal motor neurons and the function of Gαi2 in motor neuron differentiation with respect to known regulators of motor neuron differentiation. We find that different members of the Gαi family are expressed in differentiating and postmitotic motor neurons, suggesting sequential roles for Gαi proteins in motor neuron development. Using loss of function and overexpression assays in the chick spinal cord, we show that Gαi2 plays roles in regulating motor neuron differentiation, and that its function is mediated in part through its interaction with GDE2. Strikingly, Gαi2 preferentially interacts with GDE2 when bound to GDP. These findings identify a role for Gαi2 in spinal motor neuron development, and invoke GPCR-independent functions for Gαi2 in regulating motor neuron differentiation.
Embryos were prepared for immunohistochemistry and in situ hybridization as described (Sockanathan and Jessell, 1998). Tissues were embedded in Tissue-Tek O.C.T. (Sakura Finetek) and 12-μm serial sections were obtained. Primary antibodies used are as follows: K5 (rabbit anti-Isl1/2), 1:2500; guinea pig anti-Isl1/2, 1:10000 (provided by T.M. Jessell); 4H9 (anti-Isl2), 1:100; 81.5C10 (anti-HB9/MNR2), 1:100 (Developmental Studies Hybridoma Bank, [DHSB]); rabbit anti-MNR2, 1:8000 (provided by B. Novitch); goat anti-βGal, 1:3000 (Arnel); rabbit anti-GFP, 1:2000 (Molecular Probes); mouse anti-PCNA, 1:2,000 (SIGMA); rat anti-BrdU 1:100 (Abcam); rabbit anti-phospho-Histone H3(Ser10), 1:200 (Milipore); mouse anti-Nkx6.1 F55A10, 1:50 (DHSB); mouse anti-Pax6 1:250 (DHSB); rabbit anti-Nkx2.2, 1:4000 (provided by T.M. Jessell); rabbit anti-Irx3, 1:8000 (provided by T.M. Jessell ); mouse anti-Lim1/2 (1:1; DHSB). Images were captured using a Zeiss LSM 5 Pascal confocal microscope. In situ hybridization was performed as described (Schaeren-Wiemers and Gerfin-Moser, 1993). Quantitation of neuronal numbers was carried out using 5–10 sections/embryo from 5 embryos.
All cDNAs were derived from the chick and subcloned into pCAGGS or a 250bp fragment from the mouse HB9 promoter-based vector for in ovo electroporation (MNe; Lee et al., 2004). For siRNA experiments, Gαi2 siRNA duplexes were electroporated as previously described (Rao et al., 2004). Gαi2 siRNA sequences (Dharmacon) are as follows: 5′ ACAUCCAGAGCAAGUUUGAUU 3′; 3′UCAAACUUGCUCUGGAUGUUU 5′ Control DsRed siRNA sequences are as published in Rao et al., (2004).
Flag or 6XHis epitope tags were fused to the N-terminus of GDE2 or C-terminus of Gαi2 and subcloned into pCAGGS or pCS2 vectors. Transiently transfected HEK293T cells were harvested and homogenized in lysis buffer using standard procedures (Yan et al., 2009). Lysates were incubated with anti-Flag M2 (Sigma) antibody and GammaBind G Sepharose beads (GE Healthcare) or anti-Flag M2 (Sigma)-bound agarose beads overnight at 4°C under constant rotation. After centrifugation and extensive washing, the precipitated proteins were analyzed by SDS-PAGE and western blot using rabbit anti-His (Santa Cruz), anti-FLAG M2 or rabbit anti-GDE2 antibodies (Yan et al., 2009). For the GDP/AlF4− experiments, HEK293T cells were separately transfected with either Flag-GDE2 or Gαi2-6xHis plasmids. Cells were lysed, using 0.1% TritonX lysis buffer (0.1% Triton-X100; 25mM Tris-Cl, pH7.5; 150mM NaCl; 10mM MgCl2; 1/200 SIGMA Proteinase Inhibitors), spun to remove debris, and supernatants transferred to fresh tubes. Gαi2 lysates were split into aliquots, and water, GDP (10μM final), or NaF (10mM final) + AlCl3 (30μM final) were added. Gαi2 lysates were incubated for 30min. at RT, prior to mixing with GDE2 lysates. Anti-Flag antibody was added to each tube, rotated at 4°C for 3h, spun, and supernatants were transferred to fresh tubes containing gamma-bind slurry (pretreated with 2% BSA and washed in lysis buffer). Beads and lysates were rotated 1h at 4°C, and washed 4 times with lysis buffer supplemented with appropriate concentrations of GDP or NaF /AlCl3. Samples were eluted in sample buffer and run on 10% SDS/polyacrylamide gels, transferred to PVDF membranes, and bands visualized with ECL+ chemiluminiscence autoradiography (Amersham).
Large scale coimmunoprecipitation experiments and LC-MS/MS analysis were performed as described in Yan et al., 2009. Rabbit anti-mGDE2 CT antibodies were used at dilutions 1:1000.
To calculate the proliferative index, electroporated embryos were exposed to BrdU for 30min before dissection and processing as described above (Yan et al., 2009).
As a first step to define the function of Gαi proteins in neuronal development, we examined the expression of three members of the Gαi family in the ventral spinal cord of Hamburger Hamilton stage (St) 20 chick embryos: Gαi2, Gαi3, and Gαi1. At this stage of development, cells at different stages of differentiation are present; progenitor cells are located medially in the VZ, newly differentiating cells are present in the intermediate zone (IZ), and fully differentiated motor neurons are beginning to settle in the lateral marginal zone (MZ) of the ventral horn (Hollyday, 2001). In situ hybridization analysis shows that Gαi2 is expressed medially in ventral progenitors, newly differentiating and newly born motor neurons (Figure 1B). In contrast, Gαi3 is expressed in newly differentiating and postmitotic motor neurons, while Gαi1 is detected in postmitotic motor neuron populations (Figure 1D, E). The medial to lateral distribution of these Gαi family members in developing motor neurons suggests that these proteins may have sequential roles in the progression of motor neuron development, with Gαi2 regulating early steps in motor neuron differentiation. Consistent with Gαi2 function in early stages of motor neuron differentiation, Gαi2 expression initiates in laterally located cells in the ventral spinal cord just prior to postmitotic motor neuron generation (arrow, Figure 1A). Interestingly, Gai2 expression is maintained in the ventral spinal cord in medial cells and in postmitotic motor neurons after neurogenesis is complete, suggesting that Gαi2 is required for later aspects of motor neuron development or function (Figure 1C).
We focused our analysis on Gαi2 as its high expression in IZ cells suggests a potential role in controlling spinal motor neuron differentiation. In order to determine the function of Gαi2 in developing motor neurons, we designed 21bp double-stranded RNA oligonucleotides designed against the Gαi2 coding sequence (Gαi2 siRNA) and electroporated them into embryonic chick spinal cords at St 11–13, prior to the onset of motor neuron differentiation (Rao et al., 2004). The efficacy of the Gαi2 siRNAs to ablate Gαi2 expression was tested in vitro and in electroporated chick spinal cords and shown to effectively ablate Gαi2 protein and mRNA levels when compared with control siRNA oligonucleotides (Figure 2A, F, Figure S1). Furthermore, Gαi2 siRNAs were specific to Gαi2, as they did not reduce Gαi1 and Gαi3 protein expression (Figure S1). In our hands, the siRNAs did not cause stable knockdown of Gai2 expression; we thus focused our analyses at St 20, when motor neuron differentiation in the spinal cord is at its peak.
To investigate the consequences of Gαi2 loss on motor neuron development, we first stained sections of chick spinal cords electroporated with either Gαi2 or control siRNAs with antibodies against markers specific for ventral progenitors required for dorsal-ventral patterning, and for the motor neuron progenitor determinant, Olig2 (Jessell, 2000; Price and Briscoe, 2004). Patterning of ventral progenitors and the number of Olig2+ motor neuron progenitors were equivalent in control embryos and embryos silenced for Gαi2 expression, suggesting that Gαi2 is not required for the specification of motor neuron progenitors (Figure 2B, C, G, H, M, N). However, embryos electroporated with Gαi2 siRNAs showed an approximately 25–30% reduction of postmitotic motor neurons marked by Islet1, Islet2 and HB9 expression (Figure 2D, E, I, J, M, N). Interestingly, a number of electroporated embryos also showed a broadening of the zone of cells expressing NeuroM, a marker for newly differentiating neurons, suggesting that loss of Gαi2 may lead to a lengthening of the transition from dividing progenitors to differentiated postmitotic neurons (Figure 2K; Roztocil et al., 1997). In support of this, all of the NeuroM+ cells in the ventral domain of the spinal cord in many cases continue to express Olig2 when Gαi2 is ablated, in contrast to the contralateral non-electroporated side where laterally located NeuroM expressing cells downregulate Olig2 (arrows, Figure 2L). Gαi2 is expressed in the dorsal spinal cord, leading us to ask if Gαi2 is required for dorsal spinal interneuron differentiation. Embryos electroporated with Gαi2 siRNAs showed a reduction in Lim1/2 expression, suggesting a requirement for Gαi2 in the differentiation of dI2, dI4, dI6 and V1 spinal interneuron classes (Figure S1; Helms and Johnson, 2003). Taken together, these observations are consistent with the model that Gαi2 can control neuronal differentiation in the spinal cord in multiple cellular contexts, but that it is not required for the specification of progenitor identities or for regulating progenitor number.
One alternative explanation for the loss of motor neurons when levels of Gαi2 are reduced is that the rate of progenitor proliferation is decreased. To test this possibility, we analyzed embryos electroporated with control and Gαi2 siRNAs for the ability of cells to progress through S-phase and M-phase of the mitotic cell-cycle. The frequency of cells to exit the cell-cycle could not be tested as nucleotide analogs have extremely long half lives in the egg due to the closed environment and the lack of clearing by maternal metabolic pathways (Bannigan, 1981).To analyze the number of cells in S-phase, the nucleotide analogue bromodeoxyuridine (BrdU) was added to electroporated chick embryos in ovo for 30 minutes prior to dissection in order to label S-phase cells. Sectioned embryos were then stained for PCNA, which marks all cycling cells, and the labeling index, which equals the proportion of all PCNA+ cells that had incorporated BrdU, was calculated (Chenn and Walsh, 2002). The mitotic index was computed by staining adjacent sections with the mitotic marker PH3, and obtaining the ratio of PH3+ cells to the total number of proliferating cells (Chenn and Walsh, 2002). Using these measurements of cell-cycle progression, we found that embryos electroporated with Gαi2 siRNAs showed similar S-phase and M phase profiles as control embryos, (Figure 3A-D). These observations suggest that Gαi2 does not regulate progenitor cell proliferation, supporting our model that Gαi2 regulates the differentiation of spinal motor neurons.
In order to investigate the mechanism by which Gαi2 might regulate motor neuron differentiation, we took a candidate approach to identify proteins known to regulate motor neuron differentiation that may interact with Gαi2. While GPCRs are likely candidates for this purpose, no GPCRs have been implicated in the regulation of motor neuron differentiation to date. Moreover, the large number and diversity of GPCRs decreases the feasibility of screens using the candidate approach. Known mediators of motor neuron differentiation include transcription factors such as Neurogenin 2; but, these factors are unlikely to be relevant contenders, as Gαi2 localizes to the cell membrane and its presence in the nucleus has not been reported (Scardigli et al., 2001; Lee and Pfaff, 2003; Lee and Pfaff, 2005). However, one plausible candidate is the six transmembrane protein GDE2, which controls spinal neuronal differentiation through extracellular GDPD activity (Nogusa et al., 2004; Rao and Sockanathan, 2005; Yanaka et al., 2007; Yan et al., 2009). In addition to being membrane bound, loss of GDE2 results in a similar phenotype to Gαi2 ablation in that progenitor specification is not perturbed, but there is a major loss of motor neurons, and specific classes of ventral and dorsal interneurons (Rao and Sockanathan, 2005; Yan et al., 2009).
To explore if Gαi2 might exert its effects on neuronal differentiation through GDE2, we examined if Gαi2 and GDE2 are able to form a complex using coimmunoprecipitation (IP) assays. Plasmids expressing FLAG-tagged GDE2 (FLAG-GDE2) and His-tagged Gαi2 (chGαi2-His) were transfected into HEK293T cells, and GDE2 containing complexes were IP-ed using anti-FLAG antibodies. Western blots of the IP-ed complexes using anti-His antibodies detected a band of the appropriate molecular weight for chGαi2-His, only when GDE2 and Gαi2 were coexpressed, consistent with the ability of GDE2 and Gαi2 to interact (Figure 4A). In addition, we performed unbiased proteomic screens where we carried out large scale IPs of GDE2 containing complexes in transfected HEK293T cells followed by Liquid Chromatography-Mass Spectrometry/Mass Spectrometry (LC-MS/MS). Analysis of a protein band of approximately 40kD that corresponds to the molecular weight of Gα proteins and that was detected in GDE2 containing complexes but not controls, identified three separate tryptic peptides corresponding to endogenous human Gαi2 (Figure 4B). Taken together, these data are consistent with the model that Gαi2 and GDE2 are capable of interacting to form a complex in vivo.
GDE2 contains a 43 amino acid intracellular N-terminal region, six transmembrane domains, an intracellular C-terminal domain of 82 residues, and two 13 amino acid intracellular connecting loops between the transmembrane domains. As a first step to determine which of these regions binds to Gαi2, we generated versions of GDE2 that lack the first 38 amino acids of the N-terminal domain (GDE2ΔN38). IP analysis using extracts from HEK293T cells transfected with chGαi2-His and FLAG-GDE2ΔN38 show that Gαi2 is capable of interacting with GDE2ΔN38, suggesting that the N-terminal region of GDE2 is not required for Gαi2 binding (Figure 4A). We next constructed variants of GDE2 that lack the C-terminal intracellular domain (GDE2ΔC); however, deletion of the C-terminal domain rendered GDE2 unstable and prone to degradation (data not shown). To circumvent this problem, we expressed three different forms of the C-terminal (CT) region of GDE2 in HEK293T cells. Western blots of extracts containing the GDE2 CT domain alone (CT-cyt), a myristoylated/palmitoylated form of the GDE2 CT region that targets it to the membrane (CT-lyn) and the CT region attached to the last transmembrane domain of GDE2 (CT-tm6), showed that all three forms of GDE2 were expressed in HEK293T cells. This observation is consistent with the increased stability of the CT domain compared with GDE2ΔC, although CT-cyt was consistently expressed at lower levels compared with CT-lyn and CT-tm6. To determine if the CT of GDE2 was sufficient to interact with Gαi2, we performed IP assays using extracts from HEK293T cells transfected with chGαi2-His and either CT-cyt, CT-lyn or CT-tm6. All three forms of the GDE2 CT domain were found to complex with Gαi2, suggesting that sequences located within this 82 amino acid domain are sufficient for Gαi2 and GDE2 to interact (Figure 4C).
To obtain further evidence that the CT domain is the site of GDE2/Gαi2 interaction, we tested the ability of the GDE2 CT domain to interfere with GDE2/Gαi2 complex formation in the context of the full-length GDE2 protein. We repeated the IP assay using extracts from HEK293T cells transfected with FLAG-GDE2, chGαi2-His, and plasmids expressing HA-tagged CT-cyt. Western blots of the complexes IP-ed with anti-FLAG antibodies showed that GDE2 and Gαi2 could interact in the absence of CT-cyt, but that GDE2/Gαi2 complex formation was abolished in the presence of the GDE2 CT domain (Figure 4D). Notably, addition of the CT-domain did not interfere with interactions between GDE2 and its known activator Prdx1 (Figure S2). Taken together, these data suggest that Gαi2 interacts with the intracellular C-terminal domain of GDE2 and that the CT region of GDE2 can effectively interfere with GDE2/Gαi2 complex formation.
If Gαi2 mediates its effects on spinal motor neuron differentiation through its interactions with GDE2, then disruption of endogenous GDE2/Gαi2 complexes should result in motor neuron phenotypes that are comparable to when Gαi2 is ablated. To test this hypothesis, we wanted to express GDE2 CT-cyt in cells that express GDE2 and Gαi2 in the spinal cord, namely, within cells that are located in the IZ that directly abut cycling VZ progenitors (Figure 1B). We specifically did not want to perturb Gαi2 function in VZ cells given the established roles of GPCRs in regulating progenitor proliferation in the spinal cord (Megason and McMahon, 2002). Two fragments derived from the mouse HB9 promoter have been consistently used to express heterologous genes in developing spinal motor neurons; a 250bp element (MNe) in combination with a minimal promoter, and a larger 3.0kb fragment (3.0HB9) (Lee and Pfaff, 2003; Lee et al., 2004; Lee and Pfaff, 2005). To compare the onset of gene expression driven by these promoters we electroporated MNe-GFP and HB9-3.0GFP into chick spinal cords to monitor when GFP expression initiates in relation to VZ progenitors. The lateral margin of the VZ is marked by cells that are in S-phase, while cells in the IZ lie immediately lateral to this boundary (Hollyday, 2001). Examination of electroporated spinal cords pulsed with BrdU showed that GFP expression driven by HB9-3.0 GFP initiates in cells in the final cell-cycle, while the MNe element drives GFP expression in IZ cells immediately adjacent to the VZ (Figure 5A, B). Thus, in order to express GDE2 CT-cyt in IZ cells where endogenous GDE2 is normally expressed, we subcloned GDE2 CT-cyt under the control of the MNe promoter element (MNe-GDE2 CT).
We electroporated MNe-GDE2 CT into chick spinal cords at St 11–13 prior to motor neuron differentiation and assayed the effects of GDE2 CT expression on motor neuron development at St 20. Expression of GDE2 CT in IZ cells resulted in the loss of postmitotic motor neurons assessed by the expression of Islet1, Islet2 and HB9, with no changes in dorsal-ventral patterning or the number of Olig2+ progenitors (Figure 5C, D, F, -H). In addition, we observed a similar expansion of cells expressing NeuroM in some cases, consistent with cells exhibiting delays in their transition to a differentiated state (Figure 5E). This phenotype is remarkably similar to the changes observed when Gαi2 expression is reduced using siRNAs (Figure 2) and suggests that the ability of Gαi2 to regulate spinal motor neuron differentiation occurs in part, through its interaction with GDE2.
A major function of Gαi2 is to transduce signals from activated GPCRs to downstream effector molecules. GDP bound Gαi2 interacts with Gβ and Gγ to form an inactive heterotrimer that binds to seven transmembrane GPCRs. Upon stimulation of relevant GPCRs, GDP to GTP exchange occurs on Gαi2, and G-protein heterotrimers disassociate to GTP bound Gαi2 and Gβ/Gγ dimers, both of which can regulate downstream effector proteins (Neer, 1995; Neves et al., 2002). Our finding that Gαi2 is involved in regulating spinal motor neuron differentiation through its interaction with GDE2 raises the possibility that GDE2 is a target of GPCR signaling pathways. If so, then one prediction is that GTP and not GDP bound forms of Gαi2 will preferentially bind GDE2. To investigate the GDP/GTP status of Gαi2 when bound to GDE2, we took advantage of two point mutants of Gαi2 that are known to alter the equilibrium of GDP/GTP exchange on Gαi2. Alteration of the glutamine (Q) residue at amino acid position 205 of the Gαi2 protein to leucine (L), limits the intrinsic GTPase activity of the protein and results in a constitutively active form of Gαi2 (Lowndes et al., 1991). Thus, Q205L Gαi2 mutants maintain a GTP bound state for significantly longer periods than WT forms of the Gαi2 protein. In contrast, Q205L Gαi2 mutants that contain a second mutation at amino acid position 273, where aspartic acid (D) is changed to asparagine (N), are unable to bind GTP or GDP and essentially exist in an “empty” state that is more analogous to the inactive GDP-bound form of Gαi2 (Goel et al., 2004). We performed IP assays using extracts from HEK293T cells transfected with FLAG-GDE2 and His-tagged WT Gαi2, Q205L Gαi2 or Q205L/D273N Gαi2 mutants. Surprisingly, the Q205L Gαi2 mutant showed decreased binding to GDE2 when compared with equivalent amounts of WT Gαi2 or the Q205L/D273N Gαi2 mutant protein (Figure 6A). This result suggests that GDP bound forms of Gαi2 are more likely to interact with GDE2 than GTP-Gαi2.
To address this possibility more directly, we assessed the binding of Gαi2 with GDE2 under conditions where we increased the availability of either GDP, or a stable analogue of GTP formed by the reaction of GDP and AlF4− (Chen et al., 1997). Specifically, extracts from HEK293T cells transfected with chGαi2-His and FLAG-GDE2 were preincubated with GDP, or GDP and AlF4− in the binding reaction prior to the - addition of anti-FLAG antibodies. Consistent with our previous results, addition of AlF4− resulted in decreased binding of Gαi2 and GDE2, whereas inclusion of GDP alone did not perturb Gαi2/GDE2 interactions (Figure 6B). Taken together, our results indicate that Gαi2 interacts preferentially with GDE2 when bound to GDP, suggesting that GDE2 is not a downstream effector of Gαi2 mediated GPCR signaling. Instead, our observations raise the possibility that Gαi2/GDE2 interactions may utilize non-canonical functions of Gαi2 in the control of motor neuron differentiation.
To assess the consequences of Gαi2 on motor neuron differentiation, we overexpressed Gαi2 under the MNe promoter (MNe-Gαi2), thus targeting Gαi2 expression to cells in the IZ. Chick embryos were electroporated at St 11–13 prior to motor neuron differentiation and analyzed at St 19/20, when motor neuron differentiation is at its peak. Expression of Gαi2 caused a small decrease in the number of Islet1/2 motor neurons without affecting Olig2+ motor neuron progenitor generation (Figure 6C, D). One possible explanation for this subtle phenotype is that higher cellular levels of GTP compared with GDP, combined with endogenous opposing GEF and GAP protein activities, result in a modest increase in GTP bound Gαi2, thus decreasing GDE2-dependent motor neuron differentiation (Neer, 1995; Malbon, 2005). To test this possibility, we electroporated plasmids expressing the Q205L Gαi2 mutant that prolongs the GTP bound state of Gαi2 into chick spinal cords and evaluated the production of postmitotic motor neurons (MNe-Q205L Gαi2). Expression of Q205L Gαi2 consistently caused a marked decrease in Islet1/2+ motor neurons, compared with the contralateral non-electroporated side of the spinal cord (Figure 6E, F). Taken together, these observations suggest that elevated levels of GTP bound forms of Gαi2 interfere with motor neuron generation, and are thus consistent with the model that GDP-bound Gαi2 regulates neuronal differentiation in the spinal cord.
Our study describes the expression of Gαi2 in the spinal cord, and identifies a function for Gαi2 in the control of spinal motor neuron differentiation. We show that GDP-bound forms of Gαi2 preferentially interact with GDE2, a known regulator of motor neuron differentiation and that this interaction is necessary for generating the normal complement of spinal motor neurons. Our data suggest that Gαi2 interactions are unlikely to be mediated by upstream GPCR signals and invoke instead, a non-canonical function for Gαi2 in regulating motor neuron differentiation.
The progression of motor neuron differentiation can be monitored by the cell-body position of prospective motor neurons within the spinal cord. Cycling progenitors are located medially within the VZ, while cells in transition to a differentiated state are situated in the IZ and terminally differentiated motor neurons are located laterally in the MZ (Jessell, 2000; Hollyday, 2001; Price and Briscoe, 2004). We show that different members of the Gαi family of proteins are expressed in overlapping patterns along the medial-lateral axis of the spinal cord. This expression pattern suggests that different Gαi proteins might function at different stages of motor neuron development, where medially expressed Gαi proteins such as Gαi2, play roles in motor neuron differentiation while more laterally expressed subunits might function in specifying postmitotic motor neuron fate or in regulating their function. Consistent with this prediction, knockdown of Gαi2 by siRNAs results in a reduction of postmitotic motor neurons, but no changes in the patterning of motor neuron progenitors, the number of Olig2+ progenitors or the rate of progenitor proliferation. We find that Gαi3 expression overlaps with Gαi2 in IZ cells of the spinal cord, suggesting potential redundant roles for Gαi3 in regulating motor neuron differentiation. Indeed, reports indicate that there is functional redundancy between Gαi2 and Gαi3, for instance, single Gαi2 and Gαi3 null mutants are viable but Gαi2/Gαi3 double knockouts are lethal (Wettschureck et al., 2004). In support of this possibility, we find that Gαi3 is capable of binding GDE2. Moreover, Gαi3 interacts with GDE2 when bound to GDP rather than GTP, suggestive of shared mechanisms between Gαi2 and Gαi3 in regulating GDE2-dependent motor neuron differentiation (G.P. and S.S., unpublished observations).
Our results suggest that Gαi2 mediates motor neuron differentiation in part through its interaction with GDE2. Interestingly, Gαi2 binds preferentially to GDE2 when bound to GDP rather than GTP. We infer from this observation that GDE2 is not an effector of Gαi2 mediated signaling from upstream GPCRs. If so, how might Gαi2 interactions with GDE2 operate to regulate motor neuron differentiation? One possibility is that GDE2 regulates downstream pathways through mechanisms utilizing Gαi2 in a manner similar to the mode of GPCR function. Some studies have described GDE2 as adopting a seven transmembrane structure, but these conclusions were based on predictions generated from computer based structural algorithms (Zheng et al., 2000; Nogusa et al., 2004; Yanaka, 2007). In contrast to these predictions, epitope tagging experiments indicate that the N and C-termini of GDE2 are intracellular while the GDPD domain is extracellular, thus providing strong biochemical evidence that GDE2 is a six transmembrane protein (Rao and Sockanathan, 2005). Since GDE2 does not conform to the typical seven transmembrane GPCR structure, it is highly unlikely that it functions as a classical GPCR. However, it is possible that the GDE family of proteins utilizes similar components of the GPCR signal transduction machinery to perform its functions. In support of this idea, the related two pass transmembrane GDPD protein GDE1 is known to bind RGS16, a protein that regulates G-protein signaling through its GTPase activity (Zheng et al., 2000).
Alternative possibilities for how Gαi2-GDP might function in concert with GDE2 to regulate motor neuron differentiation arises from known GPCR-independent roles for Gαi2-GDP in asymmetric cell division and in receptor trafficking (Hampoelz and Knoblich, 2004; Sans et al., 2005). While the possibility that Gαi2-GDP interactions with GDE2 are involved in asymmetric cell division, which is a known prerequisite for triggering neuronal differentiation, is exciting, this function is unlikely, as GDE2 and Gαi2 expression overlap in non-dividing cells of the ventral spinal cord. In terms of receptor trafficking, it remains possible that Gαi2-GDP is required for optimal transport of GDE2 to or from, the cell surface. By analogy with the NR2B subunit of the NMDA receptor, Gαi2-GDP would increase levels of GDE2 to the membrane, thereby acting as a positive regulator of GDE2 signaling and promoting motor neuron differentiation (Sans et al., 2005). This model is consistent with our results where ablation of Gαi2 causes reductions in postmitotic motor neuron numbers and a broadening of the domain of NeuroM expression in the ventral spinal cord. The sustained expression of NeuroM may result from slowing of the rate of GDE2 trafficking to the cell surface in the absence of Gαi2, and is suggestive of a delay rather than an absence of differentiation. Nevertheless, the mechanisms by which Gαi2-GDP interactions mediate GDE2-dependent motor neuron differentiation remain open and await further investigation.
In conclusion, our study identifies a role for GDP bound forms of Gαi2 in regulating neuronal differentiation that is likely to be distinct from its roles in mediating GPCR signaling. These observations raise the possibility that GDP forms of Gαi proteins may have new cellular functions in regulating multiple aspects of neuronal development and function.
We thank T.M. Jessell, B. Novitch and S. Pfaff for antibodies and reagents; Priyanka Sabharwal, Marianeli Rodriguez and Changhee Lee for comments on the manuscript, and members of the Sockanathan laboratory for scientific discussions. Funding for this work was provided by grants from NINDS (NIH) and the Muscular Dystrophy Association.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.