Cytochalasin D binds NE-membranes
Actin binding drugs alter the gross shape of cell nuclei in eukaryotic cells during interphase. In untreated HeLa cells (or those treated with DMSO vehicle) we noticed that the nuclear shape (revealed by Hoechst staining) was in general characterised by a smooth, round shape, whereas ~20% of the nuclei were irregular in shape. Irregular nuclei were characterised by asymmetric abnormalities including lobes and/or invaginations. In HeLa cells treated for thirty minutes with the actin binding drugs lantruculin A (LatA), cytochalasin D (CD) or jasplakinolide (Jaspl) a significant two-to four-fold increase in the percentage of cells displaying NE-membrane irregularities was observed (fig.1A; see also fig.1B for an example of cells after the respective treatment). Of the drugs tested, CD was of particular interest. This fungal metabolite binds the free-barbed-end of actin filaments and thereby blocks actin polymerisation. The effect of CD on nuclear shape raised the possibility that the NE-membrane is directly a target for the drug. To address this question, we profited from the availability of a fluorescent CD conjugate: CD-BODIPY TMR (tetramethylrhodamine) to detect CD binding-sites at the NE-membrane using fluorescence microscopy.
Before performing live cell in situ experiments we examined the actin binding characteristics of CD-BODIPY-TMR in vitro and in fixed cells. CD-BODIPY fluorescence was measured in vitro at various concentrations (2–1000 nM; fig.2A) in solutions containing fully polymerised rabbit skeletal muscle actin at 10 μM (a concentration far above the critical concentration for polymerisation wherein ~100% of available actin is in the polymeric form ). Under these in vitro conditions a concentration dependent, sigmoid shaped increase in CD-BODIPY fluorescence was observed, which closely resembled its behaviour when added to fixed cells at the same concentration (fig.2A). Both in vitro and in fixed cells these results suggest that the drug bound a saturable actin-binding site in the absence of dynamic actin filament turnover. In a second, in vitro assay we tested CD-BODIPY's fluorescence characteristics in the presence of polymerised actin and lantruculin A (LatA). LatA specifically binds G-actin (1:1), but not F-actin, and the resulting reduction in available monomers, shifts the dynamic steady-state equilibrium for actin filament formation towards depolymerisation . In the presence of a constant concentration of actin (~10μM) and CD-BODIPY (50 nM), the effect of increasing concentrations of LatA was a net reduction in CD-BODIPY fluorescence (fig.2B). This effect was linear at drug concentrations between 1–10μM, and half-maximal at ~5μM, reflecting the predicted stoichiometry given the concentration of actin present. Since only the concentration of LatA was modified in this assay, the three-fold change in CD-BODIPY fluorescence could be explained by fluorescence enhancement of the drug's tetramethylrhodamine (TMR) moiety, upon binding to free-barbed-end protomers. Fluorescent enhancement resulting from TMR-drug binding to actin has been previously reported for TMR-phalloidin, and attributed to the sensitivity of TMR-X moiety in proximity to the hydrophobic environment of actin protomers [19,20]. This increase in fluorescence is a favourable property for the drug's use as a probe because it enhances specific signal detection. In light of this fact, we next examined CD-BODIPY fluorescence in a pure solution of actin around the critical concentration. Accordingly, we found that CD-BODIPY (50 nM) fluorescence peaked in pure solutions of actin at the critical concentration (~1.1μM; ) as shown in fig.2C(i)). Furthermore, relative to the total amount of actin present (fig.2C(ii)), CD-BODIPY fluorescence was diminished at levels at and above the critical concentration (fig.2C(iii)). In contrast, below the critical concentration the CD-BODIPY fluorescence relative to total actin present was at its highest levels, consistent with the view that free-barbed-end protomers and tri-meric nucleation complexes were most abundant at these concentrations (fig.2C(iii)). These in vitro results strongly suggested CD-BODIPY's utility as a fluorescent probe to detect CD binding sites in situ.
For in situ visualisation inside living cells we used CD-BODIPY at very low concentrations (5–50 nM) relative to the drug's expected affinity for actin. This provided sufficiently strong staining allowing for signal detection, without interfering with actin polymerisation . Within a few minutes after adding the fluorescent drug to the incubation medium we observed CD-BODIPY accumulation into plasma membrane microspikes that were co-labelled with fluorescent actin expressed in the same cell (fig.3A). This observation was consistent with the expected binding properties of CD-BODIPY, because these tiny sub-cellular compartments are characteristically concentrated with polymerising actin  and are highly sensitive to disruption by cytochalasin D. In a separate series of experiments, HeLa cells stably expressing the transmembrane nuclear pore complex (NPC) protein POM121 conjugated to the green fluorescent protein (GFP)  (H-PomGFP cells) were co-stained with CD-BODIPY. In these experiments, CD-BODIPY was observed to accumulate strongly in the deep perinuclear cytoplasm and across NE-membranes in almost all cells analysed. Based upon the distribution of the POM121 NPC signal, CD-BODIPY was localised to the NE-membrane however we did not detect a strict signal co-localisation pattern (fig.3B). Indeed across small stretches of the NE-membrane we discerned both positive and negative signal correlation among adjacent, near-sub-resolution regions of interest between CD-BODIPY and POM121-GFP fluorescence (fig.3B, lower panel, white arrow). Thus, CD-BODIPY labelling formed a discontinuous pattern comprising punctuate accumulation along the nuclear membrane. This staining pattern suggests strongly that CD-BODIPY binds directly at the NE-membrane, but is not strictly co-localised with NPCs.
Actin accumulates at the cytoplasmic phase of NE-membranes
In H-PomGFP cells we observed ~70% of NE-membrane POM121-GFP signal was spatially coincident with CD-BODIPY signal, a value significantly greater (P n = 5) than that measured from control experiments (dextran injection; see methods; fig.4A). This suggests the presence of free-barbed-end actin-filaments in close proximity to the NE-membrane. To rule out the possibility that this accumulation of NE-membrane associated actin was due to treatment with CD-BODIPY, we repeated the same imaging protocol and quantification method using H-PomGFP cells labelled by cytoplasmic microinjection of fluorescent actin (Alexa568). Within minutes of microinjection into cell cytoplasm, the fluorescent (Alexa568) actin was detected in a ring pattern around cell nuclei similar to that observed for CD-BODIPY. A series of experiments revealed ~37% of green POM121-GFP signal was coincident with cytoplasmically micro-injected actin (fig.4A) a value still significantly (P n = 5) higher than that measured in systematic control experiments. We next investigated the sub-cellular distribution of citrin-actin  expressed in a HeLa cell line stably expressing red fluorescent nuclear lamin A (LA-dsRed). Citrin-actin accumulated at the cell cortex (fig.4B and 4C; red arrows), and in a faint, but distinct ring-like pattern close to the NE-membranes (fig.4B and 4C; green arrows). Delineated by the nuclear lamina labelling, the perinuclear actin signal was observed in both the reconstructed XY (fig.4B) and orthogonal YZ (fig.4C; red arrow) section views. In these reconstructions the fluorescence intensity of perinuclear actin appeared brighter than the actin fluorescence detected in the nearby cytoplasm (fig.4B right panel). The three-dimensional analysis of signal coincidence revealed that about 65% of LA-dsRed nuclear lamina signal was significantly (P n = 5) co-incident with citrin-actin (fig.4A).
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].
The live cell imaging experiments suggested that cytoplasmically injected actin rapidly accumulated at the NE-membrane. Therefore, we presume that freely diffusing actin in the cytoplasm had direct access to the perinuclear actin pool. We therefore used an in vitro actin polymerisation assay to examine the gross formation of actin polymers around highly purified nuclei, freshly isolated from rat liver hepatocytes. Intact hepatocyte nuclei were incubated in the presence of actin under conditions favouring actin polymerisation. Under these conditions we found that fluorescent actin polymerised in a halo around isolated intact nuclei. Testing several actin concentrations we observed this phenomenon at concentrations as low as 100 nM actin (~10% of the expected critical concentration; data not shown). In the presence of only 500 nM fluorescent actin (just half the critical concentration expected for in vitro conditions) we observed polymerised, fluorescent actin microfilament bundles in a halo around approximately 80% of individual nuclei (fig.5A(a)). Pre-incubation of intact nuclei with antibody against FXFG-repeat containing nucleoporins (mAb414); and non-nucleoporin (anti-nuance, antibody targeting ONM non-nucleoporin nesprin/nuance) both resulted in a decrease in the number of nuclei displaying actin filaments (fig.5B). Furthermore, the non-specific nuclear transport inhibitor wheat germ-agglutinin (WGA) also significantly reduced the percentage of nuclear actin halos (Mann-Whitney test: P P > 0.05). These experiments strongly suggest that actin polymerisation is localised to the cytoplasmic phase of intact isolated NE-membranes in vitro.