Insulin receptor signaling regulates synapse number, dendritic plasticity, and circuit function in vivo

Abstract

Insulin receptor signaling has been postulated to play a role in synaptic plasticity; however, the function of the insulin receptor in CNS is not clear. To test whether insulin receptor signaling affects visual system function, we recorded light-evoked responses in optic tectal neurons in living Xenopus tadpoles. Tectal neurons transfected with dominant-negative insulin receptor (dnIR), which reduces insulin receptor phosphorylation, or morpholino against insulin receptor, which reduces total insulin receptor protein level, have significantly smaller light-evoked responses than controls. dnIR-expressing neurons have reduced synapse density as assessed by EM, decreased AMPA mEPSC frequency, and altered experience-dependent dendritic arbor structural plasticity, although synaptic vesicle release probability, assessed by paired-pulse responses, synapse maturation, assessed by AMPA/NMDA ratio and ultrastructural criteria, are unaffected by dnIR expression. These data indicate that insulin receptor signaling regulates circuit function and plasticity by controlling synapse density.

Figure 1. Insulin receptor localization in the visual system of X. laevis

(A) Diagram of the Xenopus visual circuit. Optic tectal neurons receive direct visual input from retina ganglion cells of the eye. (B-D) Insulin receptor immunostaining in the eye (B) and brain (C, D). Insulin receptor is widely distributed in the Xenopus visual circuit. Area in box in C is enlarged in D. Insulin receptor is present in the cell body (CB) region and dendrites and is concentrated in the neuropil (NP) of the tectum and retina. R: rostral, C: caudal, arrow: primary dendrites of tectal neurons. Scale bars: 50μm in B & C, 20μm in D.

Figure 2. Manipulation of insulin receptor signaling

(A) Model of insulin receptor signaling strength in cells with only endogenous insulin receptor (Endo. IR, left panel) and cells expressing exogenous wtIR (Exo. wtIR, middle panel) or dnIR (Exo. dnIR, right panel). Ectopic expression of wtIR increases total insulin receptor and therefore can increase insulin receptor signaling upon ligand stimulation. The insulin receptor monomer is composed of α, β subunits bridged by intrinsic disulfide bond (thin short bar). Two α/β subunits dimerize by an extrinsic disulfide bond (thick short bar) to generate a functional receptor. Note that, exogenous insulin receptor can dimerize with both endogenous and exogenous insulin receptor. Therefore, ectopic expression of dnIR could decrease insulin receptor signaling by blocking the transphosphorylation of insulin receptor heterodimers and possibly by sequestering ligand from endogenous insulin receptor homodimers. (B) dnIR decreases insulin receptor signaling. COS1 cells were transfected with GFP, wtIR:CFP:HA or dnIR:CFP:HA. Cells were stimulated with insulin for increasing periods before immunoprecipitation with the anti-HA antibody. Insulin receptor was detected with anti-IR antibody (top panel). Endogenous COS1 cell insulin receptor is detected in lanes where ectopically expressed Xenopus wtIR and dnIR were immunoprecipitated with HA antibody, indicating that ectopic wtIR and dnIR can interact with endogenous COS1 cell insulin receptor. Note that ectopic insulin receptor is shifted upward on the gel by virtue of the CFP. Insulin receptor activation detected by anti-phosphotyrosine antibody (α-pY, bottom panel) shows that wtIR and endogenous co-IP’ed COS1 insulin receptor can be phosphorylated upon insulin stimulation. However, in cells transfected with dnIR, phosphorylation of neither Xenopus mutant insulin receptor nor endogenous COS1 insulin receptor can be detected. (C) moIR decreases insulin receptor protein. Western analysis of insulin receptor in HEK293 cells transfected with Xenopus wtIR:CFP:HA followed by moCTRL or moIR. Equal amounts of protein were loaded as indicated by anti-tubulin antibody (bottom panel). Insulin receptor detected by anti-HA antibody (top panel) shows that ectopically expressed Xenopus insulin receptor was decreased with the moIR transfection compared to moCTRL.

Figure 3. Insulin receptor signaling is required for normal tectal cell responses to visual input

Whole-cell recordings from tectal neurons in intact tadpoles demonstrate visual stimulation evoked compound synaptic currents to light OFF responses over a wide range of stimulus intensities. (A) Representative recordings from neurons expressing GFP, wtIR, dnIR and moIR at different LED intensities. Superimposition of 20 consecutive responses (gray) and the averaged trace (black) are shown. (B, C) Integrated total charge transfer over the 1.5s window starting from the decrement of the light. dnIR-expressing tectal neurons (B) and moIR-transfected neurons (C) have significantly smaller visual stimulation evoked responses than controls and wtIR-expressing cells. (D, E) Visual stimulation evoked excitatory responses, defined by integrated initial charge transfer over the 50ms window starting from the onset the evoked responses, are significantly smaller in dnIR expressing neurons (D) and moIR-transfected neurons (E) compared to controls, indicating insulin receptor is required for normal excitatory synaptic function in response to visual inputs. Statistical differences are comparisons between test groups and controls.

Figure 4. Insulin receptor signaling regulates frequency of AMPA receptor-mediated mEPSC

(A) Representative traces of spontaneous AMPA mEPSCs, superimposed from 30 consecutive traces. (B) Examples of individual mEPSC. (C) mEPSC frequency for GFP-, wtIR- and dnIR-expressing neurons. dnIR-expressing neurons have significantly lower mEPSC frequency compared to GFP controls and wtIR-expressing neurons. (D) Cumulative distributions of the first 30 inter-mEPSC event intervals. dnIR-expressing neurons have longer inter-event intervals than GFP- or wtIR-expressing neurons (p<0.0001, Kolmogorov-Smirnov test). Only event intervals <20s were plotted and three data points from dnIR cells with intervals of 22.2, 24.3 and 51.2s were omitted in this plot. (E& F) Averaged mEPSC amplitude (E) and cumulative distributions of mEPSC amplitudes from the first 30 mEPSCs (F) for cells expressing GFP, wtIR and dnIR are comparable.

Figure 5. Insulin receptor signaling does not alter presynaptic vesicle release probability or electrophysiological synapse maturation

(A) Representative traces of evoked monosynaptic EPSCs in response to pairs of stimuli. (B) Paired-pulse ratio, defined as the ratio of peak AMPA amplitudes (EPSC2/EPCS1), is not different between GFP-, wtIR- and dnIR-expressing neurons. (C) Overlays of 30 consecutive sample traces recorded at -70mV and +45mV in response to RGC axon stimulation. Arrows and bars show where AMPA and NMDA currents were measured. (D) AMPA/NMDA ratio is comparable between GFP-, wtIR- and dnIR-expressing cells.

Figure 6. Insulin receptor regulates synapse numbers

(A) Electron micrographs show ultrastructural morphology of synaptic terminals that contact GFP-, wtIR- and dnIR-expressing dendrites. Postsynaptic areas, presynaptic area and the clustered synaptic vesicle were highlighted in light blue, green and pink, respectively. (B) dnIR-expressing dendrites receive significantly fewer synapse contacts compared to GFP- and wtIR-transfected dendrites. (C) Synapses that contact GFP-, wtIR-and dnIR-expressing dendrites show comparable ultrastructural synaptic maturity, determined by the area occupied by clustered synaptic vesicles relative to the area of the presynaptic terminal. (D) GFP-, wtIR- and dnIR-expressing neurons have comparable ranges of postsynaptic profile sizes, represented by the short diameter of labeled postsynaptic area.

Figure 7. Insulin receptor signaling modulates experience-dependent dendritic structural plasticity

(A) Top: Schematic of the visual stimulation protocol. In vivo time lapse images were collected before and after animals were exposed to 4h of dark and 4h of enhanced visual stimulation. Images and 3D reconstructions of representative neurons are shown for each group. Growth rates for individual neurons (gray) with the mean ± SEM (black) are shown on the right panel. GFP- and wtIR-expressing neurons significantly increase arbor growth rates with visual stimulation, whereas dnIR-expressing neurons do not increase growth rates in response to visual stimulation. (B) Dendritic arbor growth rates seen with visual stimulation normalized to the growth rate in the dark. GFP- and wtIR-expressing cells double their growth rate during the visual stimulation period whereas dnIR-expressing neurons do not respond to visual stimulation and have significantly reduced relative growth rates compared to GFP- and wtIR-expressing neurons. (C) Branch length retraction and extension of existing branches during the dark and visual stimulation periods. GFP- and wtIR-expressing neurons retract branch length at the same rate in the dark and with visual stimulation, but extend significantly more branch length during the visual stimulation period. dnIR-expressing neurons retract significantly more branch length and fail to increase branch extensions with the visual stimulation. (D) Classification of individual branch tips into stable, lost and added dynamic branch categories during the dark and visual stimulation periods. (E) Number of stable, lost and added branch tips during the dark and visual stimulation periods. All groups of neurons have more stable branches with visual stimulation than in the dark and they have comparable numbers of added branches during the visual stimulation period compared to the dark period. However, dnIR-expressing neurons lost significantly more branches during the visual stimulation treatment while other groups have the same rates of branch tip loss in the dark or with visual stimulation. TDBL: total dendritic branch length.