Actin polymerisation at the cytoplasmic face of eukaryotic nuclei
There exists abundant molecular and ultra-structural evidence to suggest that cytoplasmic actin can physically interact with the nuclear envelope (NE) membrane system. However, this interaction has yet to be characterised in living interphase cells.
Using a fluorescent conjugate of the actin binding drug cytochalasin D (CD-BODIPY) we provide evidence that polymerising actin accumulates in vicinity to the NE. In addition, both transiently expressed fluorescent actin and cytoplasmic micro-injection of fluorescent actin resulted in accumulation of actin at the NE-membrane. Consistent with the idea that the cytoplasmic phase of NE-membranes can support this novel pool of perinuclear actin polymerisation we show that isolated, intact, differentiated primary hepatocyte nuclei support actin polymerisation in vitro. Further this phenomenon was inhibited by treatments hindering steric access to outer-nuclear-membrane proteins (e.g. wheat germ agglutinin, anti-nesprin and anti-nucleoporin antibodies).
We conclude that actin polymerisation occurs around interphase nuclei of living cells at the cytoplasmic phase of NE-membranes.
KeywordsActin Filament Nuclear Pore Complex Actin Binding Outer Nuclear Membrane Jasplakinolide
cytochalasin D BODIPY
HeLa POM121-GFP cells
inner nuclear membrane
nuclear pore complex
outer nuclear membrane
The nuclear envelope (NE) consists of two adjacent membranes, the inner (INM) and the outer nuclear membrane (ONM). The ONM is functionally connected to the endoplasmic-reticulum (ER) and contains numerous ribosomes, while the INM contains a unique set of transmembrane proteins and maintains close contact with chromatin in the nuclear matrix through the nuclear lamina network [1, 2]. The ONM/ER comprising a single continuous membrane system is connected to the INM through nuclear pore complexes (NPC), which constitute the unique gateway for macromolecular transport across the nuclear-cytoplasmic boundary [3, 4]. The cytoskeletal protein actin may play a role in NE function. This view has received support from recent evidence suggesting that the NE-membrane system is physically connected to the cytoplasmic microfilament network. For example, ONM proteins containing giant spectrin repeats or SUN (for S ad1p, UN C-84 homology) domains are involved in nuclear anchorage and migration, probably via actin filament interactions [5, 6]. In metazoan cells, the integral ONM protein nesprin (member of the Syne/ANC-1 protein family) contains an α-actinin like actin-binding domain potentially capable to link the NE-membrane to the cytoplasmic actin cytoskeleton .
Actin is a highly conserved cytoskeletal protein that is distributed in a dynamic equilibrium between the ~42 kD monomeric (G-actin) and the filamentous polymerised (F-actin) forms. Actin's endogenous, polymeric properties are preserved in vitro and have been extensively characterised [8, 9]. Polymerisation occurs at the so-called critical concentration, where non-covalent interactions overcome a rate limiting step yielding trimeric nucleation complexes to which monomers associate spontaneously and seed filament growth . Inside living cells, the non-steady-state dynamics of actin polymerisation are highly regulated by an array of actin binding proteins (for review see [11, 12, 13]). Actin binding proteins regulate changes in actin's polymerisation state within functional sub-cellular domains or compartments allowing for actin's role in a variety of cellular processes as diverse as signal transduction and cell motility. Importantly, the ONM protein nesprin enhances actin polymerisation rates, shortens filament elongation times, and increases filament bundling in vitro. Therefore, it has been hypothesised that nesprin (or other cytoskeletal microfilament binding ONM proteins) can recruit actin to the cytoplasmic ONM interface .
Previously, ultra-structural studies using electron microscopy in fixed material have demonstrated "filament-like" networks continuous with the NE-membrane and NPC [14, 15]. Moreover, light microscopic studies have reported a fine perinuclear "shell" of actin filaments around the nuclei of fixed cultured cells . In the current study we visualised actin polymerisation in proximity to the NE-membrane inside living cells. This approach has been technically challenging, in part because the compartmentalised gradient is expected to be small (compared with large backgrounds coming from actin involved in other processes). We overcame these difficulties, using a variety of fluorescent probes for actin and confocal fluorescence imaging methods. We report perinuclear actin inside intact living cells during interphase, and we provide evidence that perinuclear actin polymerisation involves binding at the ONM of the NE-membrane.
Cytochalasin D binds NE-membranes
Actin accumulates at the cytoplasmic phase of NE-membranes
Further, we observed that the distribution of micro-injected fluorescent actin in 3D rendered reconstruction from through stack image series comprises a punctuate discontinuous labelling across the NE-membrane [see Additional file 1]. A pattern strongly analogous to that observed using cells stained with CD-BODIPY and POM121-GFP [see Additional file 2].
Previous studies using conventional approaches in fixed cells have described evidence for perinuclear actin in a variety of cell types including giant polyploid silk gland epithelial cells , insect cells , amobea  and fungi . In vertebrate tissue cells a perinuclear actin basket, revealed by phalloidin has also been reported [29, 30]. However, these studies were limited to a visualisation of actin after cell fixation, and lacked information on actin polymerisation, in situ, ie inside living cells. In the current study, we provide evidence for perinuclear actin polymerisation using spinning disk confocal microscopy. Our results show that inside intact living cells actin polymerises at NE-membranes, in a manner involving free-barbed-end protomers and through yet unidentified molecular components at the cytoplasmic face of the NE-membrane.
Actin is one of the most abundant proteins in eukaryotic cells and this exacerbates detection of specifically localised subcellular compartments or gradients, because the vast majority of measured signal is derived from actin bound in very bright structures (e.g.: stress fibers) or G-actin. Evidently, this is less problematic for studies investigating actin polymerisation in more accessible regions for imaging e.g. actin polymerisation at the leading edge of the plasma membrane [8, 31, 32, 33]. By contrast, in the current study our interest was focused on perinuclear actin. To visualise actin in the perinuclear cytoplasm we used spinning disk confocal microscopy in order to maximise three-dimensional resolution. With this approach it was striking that a nuclear actin ring was observed using several types of probes for visualisation, but was not readily observed using conventional wide-field imaging. The need for spinning disk confocal microscopy followed by deconvolution can explain why the perinuclear actin ring has until now not been described inside living cells.
Using confocal imaging our results reveal that cytoplasmic micro-injection led, within minutes, to accumulation of fluorescent actin at the NE-membrane. This rapid accumulation of micro-injected actin fits with the idea that the perinuclear pool of actin and cytoplasmic actin are in dynamic equilibrium. One mechanism for cytoplasmic G-actin to interact with the perinuclear actin pool could be that the latter concentrates actin binding sites that present free-barbed-ends to which free cytoplasmic G-actin may further associate. Certainly, our results using CD-BODIPY are highly consistent with this view. We have shown that CD-BODIPY that binds free-barbed-end actin with high affinity in vitro, yields a labelling pattern coincident with the NE-membrane in situ. Pertinently, the NE-membrane associated CD-BODIPY signal was observed to extend into the perinuclear cytoplasm, but never into the nuclear matrix, suggesting that the signal is continuous with the cytoplasmic actin cytoskeleton. However, given the sensitivity and resolution limits of optical microscopy this does not rule out the possibility that actin polymerisation could be physically located at the inner-face of the NE-membrane.
A recent study has shown that microinjected actin is localised to the NE-membrane of starfish oocytes, in a ring pattern around the nucleus  analogous to our findings presented in the current work. This study showed that a transient burst of actin polymerisation occurs upon NE-membrane fenestration marking the beginning of nuclear envelope breakdown (NEBD), and M-phase progression. Using ultra-structural electron microscopy Lenard et al. localised actin polymerisation to the inner face of the NE-membrane in accordance with their model. Our current study using high-resolution multidimensional light microscopy raised the same question: i.e. whether polymerisation occurs at the inner-(nuclear matrix) or outer-(cytoplasmic) face of the NE-membrane. By contrast to the actin polymerisation upon NEBD, our results suggest that perinuclear actin polymerisation in interphase nuclei of metazoan cells is driven from physical interactions at the cytoplasmic side of the NE-membrane. In vitro, we observed long actin filaments propagating outwards from intact nuclei isolated from differentiated rat liver cells, and incubated under conditions favouring actin polymerisation. Furthermore, nuclei were intact and apparently impermeable (in the absence of GTP) to actin because we never observed fluorescent actin inside the nuclear matrix. Additional argument supporting the view that actin binds the cytoplasmic face of isolated nuclei was derived from our results indicating that actin polymerisation was strongly diminished in the presence of antibodies (or lectin) targeting ONM proteins, but not INM. Both antibody (mAb414) and lectin targeting ONM nucleoporins inhibited actin polymerisation as strongly as anti-nesprin (a non-nucleoporin target). Therefore we hypothesise that these effects were due to molecular steric hindrance of a cytoplasmic ONM actin binding site closely juxtaposed with nucleoporins and/or non-nucleoporins. This interpretation is also consistent with our observation that no actin labelling method yielded precise co-localisation with the nucleoporin POM121. Despite the uncertainty concerning whether nucleoporins and/or non-nucleoporins provide the actin binding site, our results do demonstrate that intact interphase nuclei support actin polymerisation at the cytoplasmic face of ONM. Interestingly, previous studies on isolated patch-clamped nuclei have provided evidence that NPC ionic permeability measured by electrophysiology can be modulated by actin added to the ionic bathing solution suggesting that the effect involved cytoplasmic ONM interactions . Moreover, a role for NPC-associated cytoplasmic cytoskeleton filaments in regulation of NPC transport functions has been postulated . In either case, taken together with our current results, these hypotheses provide for a plausible functional impact linking the signalling milieu of the cytoplasmic actin cytoskeleton with NPC structure and function.
A dynamic pool of actin filaments is linked to the cytoplasmic face of nuclei in metazoan cells during interphase. Further studies will now aim at the identification of the factor (or factors) responsible for targeting actin polymerisation to the NE-membrane, and precisely defining the functional implications therein.
Cell lines and cell culture
HeLa (ATCC) were cultured in complete DMEM (DMEM Glutamax (Gibco BRL), 10% FCS, 100 U ml-1 streptomycin, 100 μg ml-1 penicillin) at 37°C; 5% CO2. H-PomGFP cells: nuclear pore labelling in HeLa cells was achieved by transfecting cells (Fugene6, Boehringer-Mannheim) with a plasmid containing POM121-GFP (kind gift of E. Hallberg, University College, Huddinge). Citrin-actin (kind gift of A. Hoppe, University of Michigan) was expressed where stated in HeLa cells using the same transfection method. Citrin-actin is a fusion construct of actin with an improved YFP variant ; the plasmid used in the current study has been described elsewhere .
Constructing HeLa cells stably expressing LaminA-dsRed
Prelamin A cDNA cloned into pSVK3 plasmid was amplified by PCR, with an Eco RI restriction site engineered at the 5' end of the sense primer (5'-GCG AAT TCT ATG GAG ACC CCG TCC CAG CGG-3') and a Kpn I restriction site engineered at the 5' end of the antisense primer (5'-GC GGT ACC TTA CAT GAT GCT GCA GTT CTG-3') (as described in ). The reaction product was digested with Eco RI and Kpn I and ligated in frame into purified, linearised pDsRed-C1 (Clontech, Inc, Palo Alto, CA) after digestion with Eco RI and Kpn I. HeLa cells cultured to 80% confluency were transfected in chamber slides with Lipofectamine PLUS (Life Technologies, Inc., Gaithersburg, USA.), following the manufacturer's instructions. Stable transfected cells were selected in the presence of antibiotic (1 mg ml-1 G418; Gibco Life Technologies Ltd). Selection of dsRed-positive cells was performed with FACS (fluorescence-activated cell sorter) analysis.
Actin-binding drug treatment of HeLa cells
HeLa cells were grown on cover-slides and treated for 30 minutes with cytochalasin D 1μM (Sigma), latrunculin A 1μM (Molecular Probes) or jasplakinolide 200 nM (Molecular Probes) at 37°C. Then cells were washed with PBS, fixed for 5 minutes with 3%PFA at RT, stained with Hoechst 33342 (Sigma-Aldrich) (10 min, 0.001 mg ml-1) and mounted on slides. The nuclei were imaged with a widefield microscope (see below) and nuclear shape was evaluated and counted in two independent experiments. For each experimental condition at least 150 nuclei from at least ten microscopic fields were counted.
In vitro actin polymerisation assay on isolated rat liver nuclei
Nuclei were purified from rat liver : Rats were asphyxiated with CO2, and were then decapitated. The livers were isolated, placed into ice-cold low sucrose buffer (0.25 M sucrose, 50 mM Tris-HCl pH 7.5, 25 mM KCl, 5 mM MgCl2, 1 mM DTT, 0.5 M PMSF and 1 μg ml-1 leupeptin/aprotinin/pepstatin) and minced. Then, 1 volume of minced liver and 2 volumes of the same buffer were homogenised in a Potter-Elvehjem homogeniser (0.025 cm clearance) for 15 full strokes at 1,700 rpm. The homogenate was filtered through 4 layers of cheese cloth and centrifuged at 800 g for 15 minutes at 4°C. 1 volume of loose pellet was mixed with 2 volumes of high sucrose buffer (2.3 M sucrose, 50 mM Tris-HCl pH 7.5, 25 mM KCl, 5 mM MgCl2, 1 mM DTT, 0.5 M PMSF and 1 μg ml-1 leupeptin/aprotinin/pepstatin) before overlaying 1 part of high sucrose buffer with 5 parts of the initial mixture and ultra-centrifuged at 28000 rpm for 1 hour at 4°C in a SW-28 rotor (Beckman, USA). The nuclei containing pellet was re-suspended in the low sucrose buffer and was again homogenised with 2 full strokes as described above. This homogenate was centrifuged at 800 g for 15 minutes at 4°C and the pellet contained the highly purified rat liver nuclei. These nuclei were re-suspended in the low sucrose buffer, and aliquots were stored at-80 °C. The number of nuclei was determined by measuring the OD260 (OD260 = 1 equals 3 × 106 nuclei). The quality and entireness of isolated nuclei was assured by light microscopy.
Fluorescent actin-Alexa488 (Molecular Probes) in G-buffer (0.2 mM ATP, 0.2 mM CaCl2, 1 mM DTT, 10 mM Tris-Hcl, pH 8) was polymerised at 0.5 μM by addition of a 1:9 dilution of 10x KMEI (500 mM KCl, 10 mM MgCl2, 10 mM EGTA, 100 mM imidazole, pH 7) in the presence of 1 unit of isolated rat liver nuclei (~3 × 106 nuclei). Isolated nuclei were pre-incubated with antibodies overnight at 4°C with continuous, gentle agitation. Antibodies used in this assay were mAb414 (Eurogentec, 1/1000), mAb anti-Nuance (kind gift of A. Noegel, University of Cologne; 1/10), anti-Emerin (kind gift of H. Hermann, DKFZ Heidelberg; 1/200) and mAb anti c-Myc (9E10 Santa-Cruz Biotechnologies, 1/1000). Additionally, we incorporated the non-specific nuclear transport inhibitor wheatgerm agglutinin (Invitrogen) in this assay. An aliquot of the reaction mix was sealed under a glass cover slip (coated with 2% cold water fish skin gelatin) on a microscope slide and then incubated for 1 hour at 37°C to allow time for actin polymerisation to occur. Finally, the number of nuclei displaying actin microfilaments was counted using a widefield microscope.
Single cells were cytoplasmically microinjected with either 10 μM actin-Alexa568 (Molecular Probes) or 2.5 mg ml-1 dextran-rhodamine 70 kDa (Molecular Probes) using a Femtojet/Injectman micromanipulator (Eppendorf, Hamburg, Germany).
CD-BODIPY in vitro measurements
Purified, native and fluorescent (FITC) conjugated rabbit skeletal muscle actin (Cytoskeleton, Denver, U.S.A.) in 2 μl stock aliquots (10 mg ml-1; 222 μM) containing 2 mM Tris-HCl, 0.2 mM CaCl2, 0.2 mM ATP, 0.5 mM DTT, 2% sucrose, and 0.005% NaN3 (pH 8) were prepared at working dilutions (0.5–0.75 mg ml-1; 11–17 μM) by addition of low salt buffer (5 mM Tris-HCl, 0.2 mM Na-ATP, 0.5 mM DTT, 0.2 mM CaCl2, pH 8). For each experiment a single fresh actin stock was prepared. Drugs were added, yielding the final concentrations stated. Samples were vigorously mixed by pipetting, vortexed and then left on ice for 1 hour before centrifugation (100,000x g, 4°C, 30 minutes). Supernatants were added to 5x concentrated high salt polymerisation buffer giving final concentrations of 100 mM KCl and 10 mM Mg-ATP. After mixing, a 5 μl aliquot was sealed under a glass cover slip on a microscope slide and incubated at 37°C for at least 3 hours to allow actin to fully polymerise to a steady state  before semi-quantitative observation using fluorescence microscopy. Each experiment (triplicate samples) was performed at least twice yielding comparable results.
Image acquisition, processing, and analysis
For live cell imaging, cells cultured on 35 mm glass bottom dishes (MatTek Corporation, USA) were exchanged into OptiMEM (Gibco BRL), and transfered to the microscope stage that was maintained at 37°C (Carl Zeiss "Tempcontrol 37–2 digital" incubator support). For experiments using CD-BODIPY, experimental medium was freshly prepared by serial dilution of a 1 mM stock solution (DMSO) to the stated final concentrations (2–50 nM). Facile and sufficient staining of cells was achieved by incubation of cells in this solution for 20 minutes before recordings began. In the presence of such very low concentrations, we found that cells could be maintained for several hours, or even days, without any deleterious effects. Using bright-field observation, we routinely ensured that observed cells were flat, adherent and healthy. Fluorescent imaging was performed using a high-speed spinning disk confocal system (UltraView RS Perkin-Elmer, USA) equipped for dual wavelength excitation (Kr-Ar laser; 488 nm and 568 nm) and axial z-stack sampling (PI objective piezo-drive). The system was built around a Carl Zeiss (Germany) Axiovert 200 microscope. Cells were visualised using a Hamamatsu (Japan) ERII camera (exposure time 200 ms) using a 100x NeoFluor NA 1.3 oil-immersion objective (Carl Zeiss). Fast axial (through-stack) sampling used a piezo-objective drive stepping at 200 nm throughout the cell volume.
For analysis, all axial image datasets were treated identically. First using geometric light reconstruction (deconvolution) based on a calculated point-spread-function, followed by a "maximum likelihood estimate" (MLE) iterative algorithm (Huygens software; Scientific Volume Imaging, Netherlands), in "Nipkow-disk" mode with parameters for the Yokagawa CSU-22 scan head. The deconvolved 3D through-stack results were then analysed for signal coincidence using the "Coloc" plug-in of Imaris (Bitplane, Switzerland). For all results we set a threshold based on the averaged background signal detected outside the region of interest in the channel used for NE-membrane detection (POM121-GFP, or dsRed-Lamin), such that the threshold identified a binary mask region delineating the NE-membrane signal. The second channel (Alexa568-actin, citrin-actin, or CD-BODIPY-TMR) was then similarly thresholded and masked. Quantification of signal co-incidence between two channels was then measured as a function of the percentage of pixels detected in the NE-membrane signal channel that were also positive for binary masked pixels derived from the actin dependent signal. This quantification process was applied to each image in any given stack, yielding an estimate of three dimensional signal co-incidence in a given cell. As a systematic control for this 3D signal coincidence we used exactly the same acquisition and analysis protocol except H-PomGFP cells were microinjected cytoplasmically with red fluorescent Alexa568-dextran. This provided us a relative measure of the microscope's ability to distinguish the NE-membrane (green) from diffusible non-specific cytoplasmic signal (red). On average (n = 5), throughout the imaged 3D volume we found that just ~12% of pixels containing specifically localised NE-membrane signal were spatially coincident with pixels also positive for red dextran signal. Given that the dimensions of the NE-membrane were below the optical resolution limits of the microscope, and that all two-color, three-dimensional analysis used the same imaging hardware/analysis this value served as a systematic threshold limit below which the NE-membrane and cytoplasmic signals could be considered "non-coincident".
Images of isolated intact nuclei for the in vitro actin polymerisation assay used a standard Zeiss Axiovert 200 M microscope equipped with a 40x objective, NA 1.3. The microscope was controlled by Till Vision (Till photonics, Germany) designed for fast live cell imaging and equipped with a xenon lamp, polychromator (PolychromeIV) for wavelength selection and a Till Photonic Imago QE CCD camera.
L. Blanchoin (CEA Grenoble, France) for helpful discussions. E. Hallberg (University College, Huddinge, Sweden) for transfection plasmid coding POM121-GFP. A. Hoppe (University of Michigan Medical School, Ann Arbor, USA) for the citrin-actin construct and helpful discussions. A. Noegel (Institute for Biochemistry, Medical Faculty, University of Cologne, Germany) for anti Nuance mAb and H. Hermann (DKFZ, Heidelberg, Germany) for anti-Emerin mAb. TD Allen (Paterson Institute for Cancer Research, Manchester, UK), G. Danuser (Scripps Research Institute, USA) and JC Courvalin (Institut Jacques Monod, Paris, France) for helpful discussion. F. Frischknecht for critical reading of the manuscript. SM is recipient of a DAAD (Doktorandenstipendium im Rahmen des gemeinsamen Hochschulprogrammes III von Bund und Ländern) and FRM grant. JE is supported by FRM and EMBO. Special thanks to Mathieu Marchand, Emmanuelle Perret, and Pascal Roux for excellent technical support. PFID is supported by Institut Pasteur, Fondation pour la Recherche Medicale, La ministre deleguee a la recherche et aux nouvelles technologies, La Région Ile-de-France (programme SESAME), and the European Union (EAMNET-FP5; AUTOMATION NEST-FP6).
- 5.Reinsch S, Gonczy P: Mechanisms of nuclear positioning. J Cell Sci. 1998, 111 ( Pt 16): 2283-2295.Google Scholar
- 10.Kuhn JR, Pollard TD: Real time measurements of actin filament polymerization by total internal reflection fluorescence microscopy. Biophys J. 2004Google Scholar
- 23.Soderqvist H, Imreh G, Kihlmark M, Linnman C, Ringertz N, Hallberg E: Intracellular distribution of an integral nuclear pore membrane protein fused to green fluorescent protein--localization of a targeting domain. Eur J Biochem. 1997, 250: 808-813. 10.1111/j.1432-1033.1997.00808.x.CrossRefPubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.