Modulation of calcium-induced cell death in human neural stem cells by the novel peptidylarginine deiminase-AIF pathway

Abstract

PADs (peptidylarginine deiminases) are calcium-dependent enzymes that change protein-bound arginine to citrulline (citrullination/deimination) affecting protein conformation and function. PAD up-regulation following chick spinal cord injury has been linked to extensive tissue damage and loss of regenerative capability. Having found that human neural stem cells (hNSCs) expressed PAD2 and PAD3, we studied PAD function in these cells and investigated PAD3 as a potential target for neuroprotection by mimicking calcium-induced secondary injury responses. We show that PAD3, rather than PAD2 is a modulator of cell growth/death and that PAD activity is not associated with caspase-3-dependent cell death, but is required for AIF (apoptosis inducing factor)-mediated apoptosis. PAD inhibition prevents association of PAD3 with AIF and AIF cleavage required for its translocation to the nucleus. Finally, PAD inhibition also hinders calcium-induced cytoskeleton disassembly and association of PAD3 with vimentin, that we show to be associated also with AIF; together this suggests that PAD-dependent cytoskeleton disassembly may play a role in AIF translocation to the nucleus. This is the first study highlighting a role of PAD activity in balancing hNSC survival/death, identifying PAD3 as an important upstream regulator of calcium-induced apoptosis, which could be targeted to reduce neural loss, and shedding light on the mechanisms involved.

Keywords: Apoptosis inducing factor (AIF); Cell death; Citrullination–deimination; Human neural stem cell; Peptidylarginine deiminase (PAD, PADI); Vimentin.

Graphical abstract

Fig. 1

PAD expression in the developing human central nervous system (brain and spinal cord) and in hNSCs. A) Real time RT-PCR analysis of PAD3 and PAD2 in fetal brains (left panel) and spinal cords (right panel) from human embryos at 42, 63 and 70 days of gestation. PAD2 expression increases while PAD3 expression decreases with development. Human liver was used as a positive control for all PADs. Asterisk indicates statistically significant differences (p < 0.05). B) PAD2 and PAD3 transcript detected by in situ hybridization in human spinal cord at 46 days of gestation (dg). Scale bars are 200 μm. C) PAD2 and PAD3 protein detected by Western blot in developing human brain (Br) and spinal cord (SC); no dramatic change in PAD protein expression is observed.

Fig. 2

PADs are expressed in human neural stem cells (hNSCs) and PAD3 inhibition increases hNSC proliferation. A) PAD2 and PAD3 transcript detected by RT-qPCR in hNSC derived from embryonic brain and spinal cord (SC). B) Detection of PAD2 and PAD3 by immunocytochemistry (red) in hNSCs: both proteins are detected in cytoplasm and nucleus (counterstained with Hoechst dye). Scale bars: 25 μm. All pictures are at the same magnification. C) Analysis of cell growth determined by the methylene blue assay after treatment with 100 μM Cl-amidine for 24, 48 or 96 h. Cl-amidine significantly increases hNSC proliferation as compared to controls at 48 and 96 h. * = p < 0.05, ** = p < 0.01 by ANOVA and Student's t-test. D) Analysis of cell growth determined by the methylene blue assay after transfection with siRNA against PAD2 (siPAD2) and PAD3 (siPAD3) or scrambled siRNA. A significant increase in cell growth as compared to controls is observed at 48 h only in cells transfected with siPAD3 (p < 0.05; two-way ANOVA). Error bars indicate SDM; n ≥ 3.

Fig. 3

Cl-amidine treatment reduces dose-dependent cell death induced by thapsigargin in hNSCs. A) Quantification of cell death induced by 5 μM thapsigargin (Thaps) detected by propidium iodide (PI) staining. The number of PI-positive cells increases upon thapsigargin treatment and is reduced both by pre-treatment or post-treatment with 100 μM Cl-amidine (Cl-am); n = 8; error bars = s.d.; ** = p < 0.01 by ANOVA and Student's t-test. B) Cell survival determined by methylene blue assay after 24 hour treatment with different concentrations of thapsigargin either alone or following 100 μM Cl-amidine treatment (n ≥ 3; error bars = SDM). Cl-amidine treatment significantly increases cell survival (p < 0.05; ANOVA).

Fig. 4

Effect of thapsigargin on PAD expression and citrullination in hNSCs. A) RT-qPCR analysis of PAD2 and PAD3 transcripts after thapsigargin treatment: note up-regulation of PAD3, but not PAD2 transcript, in treated cells; * = p < 0.05 by ANOVA and Student's t-test (n ≥ 3; error bars indicate SDM). B) Western blot analysis of citrullinated proteins detected by F95 monoclonal antibody and of citrullinated histone H3 (Cit-H3) following treatment with either Cl-amidine (100 μM) or thapsigargin (5 μM) alone, or both compounds for 24 h. Cl-amidine was added to the culture medium 15 min before thapsigargin treatment. Actin was used as a loading control. Note that PAD activation by thapsigargin results in protein citrullination and this is reduced by the PAD inhibitor, Cl-amidine.

Fig. 5

PAD activity is increased in HEK293T cells expressing PAD3-EGFP A) Live images of cells transfected either with EGFP alone or PAD3-EGFP 40 h after transfection with corresponding nuclear staining and detection of endogenous PAD3 (ePAD3) and PAD3-EGFP (P3-G) by Western blot in untransfected cells (1), cell transfected with the EGFP plasmid (2) and cells transfected with PAD3-EGFP. All panels are at the same magnification. Scale bar is 50 μm. B) PAD activity in HEK293T whole cell lysate assessed by the BAEE assay 24 h after transfection. A significant increase (p < 0.05; two-way ANOVA) in PAD activity is observed in HEK293T cell expressing PAD3-EGFP, but not in HEK293T cells expressing the mutated PAD3-EGFP lacking enzymatic activity (ΔPAD3-EGFP) where activity is as in untransfected (WT) cells. C) Cl-amidine (10 μM) significantly (p < 0.05; two-way ANOVA) reduces PAD activity in PAD3-EGFP HEK293T cell lysate in the BAEE assay. D) Cell death determined by propidium iodide (PI) staining 40 h after transfection with PAD3-EGFP (P3); * = p < 0.05, ** = p < 0.01 by Anova and Student's t-test. E) Cell survival determined by methylene blue assay after 24 hour treatment with thapsigargin. Significantly (p < 0.05; two-way ANOVA) higher cell death is observed in PAD3-EGFP cells than in HEK293T cells transfected with the EGFP, ΔPAD3-EGFP or WT (labels are as in (B)). Error bars indicate SDM; n ≥ 3.

Fig. 6

PAD3 is the main PAD involved in hNSC cell death. Death/survival of hNSCs treated for 24 h with thapsigargin was determined by the methylene blue assay and staining for the apoptosis marker annexin V. A) Cell survival is significantly (p < 0.05; two-way ANOVA) reduced in hNSCs carrying PAD3-EGFP as compared to cells transfected with PAD3 lacking the active site (ΔPAD3-EGFP), EGFP alone, or no transfection (WT). B) Example of cells expressing PAD3-EGFP (PAD, green) and Annexin-V (AnV, red) counted for quantification (arrows); note the significantly higher percentage of PAD3-EGFP/Annexin-V-positive cells; * = p < 0.05 by ANOVA and Student's t-test. C) Cell survival is significantly (p < 0.05; two-way ANOVA) increased in hNSC transfected with PAD3 but not PAD2 siRNA. Scale bars are 50 μmin B. Error bars indicate SDM; n ≥ 3.

Fig. 7

PAD3 expression in hNSCs modulates thapsigargin-induced cell death via AIF translocation. A) Effect of staurosporine (Staur) and thapsigargin (Thaps) treatments for 4 h on caspase 3 activation (cl-Casp3) and AIF translocation to the nucleus assessed by immunocytochemistry and confocal microscopy. Thapsigargin unlike staurosporine does not activate caspase 3, but both induce AIF translocation. B) Confocal images showing translocation of AIF to the nucleus in thapsigargin-treated cells. PAD inhibition by Cl-amidine (Cl-am) blocks AIF translocation. Scale bars in A and B are 25 μm. C) Immunoprecipitation of hNSC protein extract with the PAD3 antibody (Ly: full lysate; Ft: flow through; w1: wash 1; w3: wash 3; El: eluate). Note that AIF, vimentin (Vim) and actin are co-immunopreciptated only in thapsigargin-treated cells. D) PAD3 is immunoprecipitated under all conditions tested; Con: control. E) Western blot of protein extracted from control, thapsigargin- or Cl-amidine and thapsigargin-treated hNSC. The cleaved form of AIF (tAIF) is detected in thapsigargin-treated extracts but not in extract from cells pre-treated with Cl-amidine. F) Immunoprecipitation of control and thapsigargin-treated hNSC protein extracts with the anti-citrullinated protein antibofdy, F95: AIF and vimentin are co-immunoprecipitated only in thapsigargin-treated cells.

Fig. 8

Thapsigargin induces cytoskeletal disorganization in hNSCs that is reduced by PAD inhibition and affects AIF–vimentin association. Confocal images of hNSC stained for actin (detected by phalloidin, green) and vimentin (green) and AIF (red) 3 h after treatment; nuclei are counterstained with Hoechst dye (blue). Double-labeling for actin or vimentin and AIF. Scale bars = 25 μm; all images are at the same magnification. B) High magnification images showing possible association of vimentin and AIF and disorganization of the cytoskeleton upon thapsigargin treatment. C) Immunoprecipitation of hNSC protein extract with the AIF antibody (Ly: full lysate; Ft: flow through; w1: wash 1; w3: wash 3; El: eluate). Vimentin but not actin is co-immunoprecipitated by AIF.