NCAM and PSA-NCAM dependent membrane spreading and F-actin reorganization in suspended adhering neural cells

Abstract

Mediation of synchronous cell-cell interactions by NCAM and PSA-NCAM is examined here in aggregates (monolayers) of C6 polysialylated embryonic neural cells, formed rapidly (within 30 s) in suspension in an ultrasound trap. These cells express all three main isoforms of neural cell adhesion molecule (NCAM). The rate of extension of perimeter contact (i.e., membrane spreading) between closely adjacent cells and the temporal reinforcement of the Filamentous (F)-actin cytoskeleton at those regions were measured. Enzymatic removal of the cell-cell repelling polysialic acid (PSA) increases the rate of NCAM-induced membrane spreading, while removal of NCAM-120 had no detectable effect. Competitive peptide inhibition of the third immunoglobulin domain of NCAM significantly reduced the rate of membrane spreading, while NCAM siRNA transfected cells lost their ability to spread. It is argued that NCAM induced contact is the initial requirement for membrane spreading and facilitates conditions for subsequent cytoskeletal reorganization in these neural cells.

• Cell adhesion
• actin
• physical cell trap
• siRNA
• peptide

Introduction

The neural cell adhesion molecule (NCAM) has a fundamental role in the development of the nervous system mediating adhesion between neural cells, such as axon-axon interactions [1] and is a regulator of neurite outgrowth and fasciculation [2], cell motility [3], synaptic plasticity, including processes such as learning and memory consolidation [4], [5] and intracellular signalling [6].

NCAM mediates both homophilic (NCAM-NCAM) and heterophilic interactions. Homophilic interactions affect cell-cell adhesion through developmental changes in the number of homophilic bonds between NCAM molecules on apposing membranes. These interactions are modulated by the polysialic acid (PSA) content of NCAM [7], [8]. Heterophilic interactions involve the interaction of NCAM with extracellular matrix components [9], L1, CD24 as well as integrins [10].

The carbohydrate polysialic acid component of NCAM (PSA-NCAM) that is highly expressed in developmental and neural tissues modulates cell-cell and cell-substrate interactions [9]. The molecular mechanism by which PSA exercises this function remains an open question [11]. There is general agreement that the main influence of PSA on cell-cell interactions is a non-specific contribution to the physical chemical repulsive barrier at the cell surface [12], [13]. This repulsion combines an electrostatic component [14] and a simple steric mechanism [15], by which expression of PSA at the cell surface can hinder membrane-membrane and NCAM receptor-receptor apposition [16]. A more general role as a negative regulator of cell interactions (not only those mediated by NCAM) has been attributed to PSA [17], [18].

A number of assays have been explored to identify the role of NCAM and PSA-NCAM as regulators of cell-cell interactions. In classical studies chick embryonic membrane vesicles [19] and intact F11 cells in suspension [8] assayed in a 20 ml rotating suspension (90 rpm) at intervals over 15 min showed increased ‘aggregation rate’ following digestion with an α-2,8 polysialic acid cleaving endoneuraminidase. The total number of vesicles and vesicle aggregates was monitored with a Coulter Counter and an aggregation rate established from the change in this number with time. Pizzey et al. [20] applied a less hydrodynamically stressful and more sensitive assay to aggregate formation by L cell NCAM mutants formed by transfection with cDNA that were suspended in 15 µl microwell. This technique was applied subsequently to the study of aggregation when it was shown that the rotation technique could not differentiate between NCAM transformants and control cells [21] and to the study of peptide inhibition of hippoccampal neuron adhesion [22]. Immunofluorescence monitoring of the temporal progression of the molecular consequences of cell-cell aggregation, that has often been applied to cells on substrata, does not appear to have been applied to cells aggregating in suspension possibly because the incubation periods involved (t ≥ 15 min) do not provide well synchronized aggregates.

Johnson et al. [18] exploited a surface force apparatus and fluorescence correlation spectroscopy to provide direct evidence that NCAM polysialylation increases membrane repulsion and abrogates adhesion. They showed that the increase in the intermembrane repulsion following NCAM polysialylation was sufficient to overwhelm both homophilic NCAM as well as homophilic cadherin attraction.

The recent development of an ultrasound standing wave trap (USWT) (see Supplementary data Figure 1i,ii – online version only) that rapidly and synchronously forms 2-D particle or cell aggregates in a low-stress small-volume suspension in the focal plane of a light microscope [23], [24] has enabled cell adhesion to be quantified by the rate of membrane contact spreading at cell perimeters [25], fractal dimension and void analysis [26]. Subsequent immunofluorescent staining of aggregates removed from the trap showed progressive F-actin and N-cadherin accumulation at sites of cell-cell contact [25] as well as gap junctional communication between cells [27].

Figure 1.  Western blot analysis of (i) the NCAM profile of control neural cells; all three isoforms of NCAM (-120, -140, -180) are expressed by neural cells; (ii) PIPLC-treated neural cells; 70% of NCAM-120 is removed. (iii) EndoN-treated cells; embryonic neural cells are highly sialylated (PSA was detected at 250 kDa), while enzymatic digestion resulted in complete removal of PSA from NCAM. (iv) PCR analysis at 48 h post transfection of NCAM siRNA transfected cells (lane 1), cells transfected with scrambled oligos (lane 2) and untreated cells (lane 3) (n=3). β-actin was used for normalization. Data are representative of 2 or 3 independent experiments.

The USWT is an ultrasound resonator where the acoustic pathlength in the cell suspension is a single half wavelength. The resonator has a pressure node plane half way through the cell suspension and parallel to the transducer (see Supplementary data Figure 1ia). The cell trap exploits the fact that particles in general experience an axial direct acoustic radiation force when in an ultrasound standing wave field. This force drives them towards a node plane (Supplementary data Figure 1ib). They then move, within that plane, to accumulate at the centre of the field (Supplementary data Figure 1ic). This in-plane movement arises because variations in the pressure distribution within the node plane give rise to a weak lateral radiation force (DRFl) and because cells are convected by acoustic streaming [23–25].

The resulting 2-D aggregate provides good resolution capability for optical examination of cells in situ in the trap and in the immununofluorescence characterization on aggregate removal from the trap. The above developments in cell manipulation as well as in molecular techniques (for example siRNA for the targeted silencing of the gene of interest as well as the use of peptides specifically designed to inhibit the action of membrane receptors without the significant steric effects imposed by use of antibodies) make it timely to revisit cellular systems and refine some understandings of NCAM and PSA-NCAM modulation of cell-cell adhesion.

In the present study, we aim to examine the influence of NCAM and PSA-NCAM on cell-cell adhesion and perimeter membrane contact spreading in adhering neural cell monolayers in suspension in an ultrasound trap.

Materials and methods

Culture of 921202-6 (Clone 6 (C6)) neural cells

All reagents were from Sigma (UK) unless otherwise stated. The conditionally immortalized neural cell line (derived from embryonic rat mesencephalon) was maintained as a replicative culture at 33°C/5% CO₂ in neuronal supplemented medium (NSM) and supplemented with 5 ng/ml fibroblast growth factor (FGF-2), as previously described [25].

Phosphatidyl-inositol-specific phospholipase C (PIPLC) and Endoneuraminidase (EndoN) treatment

Neural cells were prepared as described above and diluted to 3×10⁶/ml. Cells were suspended in NSM containing 2 U/ml of phosphatidyl-inositol-specific phospholipase C (PIPLC) [2] or 20 U/ml of Endoneuraminidase (EndoN, AbCys S. A., Paris, France) [28], and incubated for 1 and 2 h respectively at 33°C/5% CO₂. Cells were then pelleted by centrifugation (5 min, 28 g) and suspended in NSM at the same concentration prior to introduction into the ultrasound trap.

Western blot analysis

Cells suspended in NSM or treated with NSM containing 2 U/ml PIPLC or NSM containing 20 U/ml EndoN, were lysed in boiling extraction buffer (10 mM tris (pH 7.4), 1% SDS, 1 mM sodium orthovanate) and protein concentrations determined using a BCA™ protein assay kit (Pierce, Rockford, IL, USA). Equal amounts of protein (20 µg) from three independent experiments were separated on a 7.5% SDS-PAGE gel and Western blotting performed as described previously [29] using a monoclonal anti-NCAM antibody that recognizes all three NCAM isoforms (clone 123c3, (1:100), Abcam, UK) or a monoclonal anti-PSA-NCAM antibody (clone 2-2B, (1:500), Chemicon, UK), which only recognizes PSA attached to NCAM, and not NCAM alone.

Peptide inhibition of NCAM

Neural cells were prepared as described earlier and resuspended in NSM containing 0.2, 0.5, 1, 3 or 5 mg/ml of the 21 amino acid peptide KHIFSDDSSELTIRNVDKNDE (resuspended in PBS) (Severn Biotech Ltd., UK) corresponding to the homophilic binding site of the Ig3 domain of NCAM between Lys-243 and Glu-263 [21]. A suspension of NSM with PBS, or NSM with an equivalent concentration of the peptide KYSFNYDGSELIIKKVDKSDE (Severn Biotech Ltd., UK), a chicken homologue of the inhibitory peptide described above (the sequence homology between the rat and chicken peptides is 52%) were used as negative controls to confirm the peptide-specific inhibition of NCAM.

siRNA treatment of cells

Neural cells were plated on a 24-well plate at a concentration of 25×10⁴ cells per well in 1 ml of NSM media. Cells were incubated overnight at 33°C/5% CO₂ to allow cells to reach confluence. Cells were incubated with either 25 ng of NCAM siRNA (Santa Cruz Biotechnology, Inc) specifically designed to knock out all three isoforms of NCAM, or an unrelated scrambled siRNA (5′-UAAUGUAUUGGAACGCAUATT-3′). Transfections were performed according to the manufacture's protocol (INTERFERin™ siRNA transfection reagent; Autogen Bioclear, UK). For experiments in the trap, cells were plated in a 75 cm² flask at a concentration of 10⁶/ml and treated with 175 ng of NCAM siRNA.

Confirmation of NCAM knock down: RNA extraction and RT-PCR

Cells were lysed into 1 ml Trizol® reagent (Invitrogen, UK) at 24, 48 and 72 h post-transfection. Total RNA was extracted with chloroform and precipitated overnight with isopropanol as previously described [29]. cDNA was produced using 0.5 µg/µl oligo (dT)₁₅ (Promega, UK) as previously described [29]. Abolition of NCAM mRNA was assessed, by PCR, using primers specifically designed against exons 9–11 of the rat sequence, which are present in all 3 NCAM isoform transcripts (S 5′-TAGATGGGCACATGGTGGTA-3′ and AS 5′-GGGTAGGCAAAGACCTCACA-3′, 0.27 µM each, annealing temperature 58°C, 232 bp product). 25 cycles were carried out in a Techne TC-312 thermal cycler (PCR analysis determined that 25 cycles was within the logarithmic phase of amplification for both genes). Amplified products were analysed on 2% agarose gels with a 100–1000 bp DNA ladder for size identification. Amplification of the housekeeping gene β-actin was used for normalization using previously published primer sequences [30].

Ultrasound trap

The ultrasound trap employed in the present work had four layers; a piezoceramic ultrasound transducer (Ferroperm, Kvistgard, Denmark) in a housing of radial symmetry, a steel layer coupling the ultrasound to a one half wavelength (λ/2 or 0.5 mm depth, where λ is the wavelength of sound in water), an aqueous layer and a reflector that provided optical access from above (Supplementary data Figure 1ii) (we note that the acoustic frequency (f) is given by the formula f = c/λ, where c is the speed of sound in water). The constant transducer drive voltage came from a function generator (HM 8138, HAMEG, Germany). The ultrasound trap has been previously described in detail [25], [31], [32]. The sound pressure amplitude (P₀) was 0.54 MPa (corresponding to 20 V_p-p) for the first 30 s of ultrasound initiation and subsequently reduced to 0.27 MPa (i.e., 10 V_p-p) for 30 s and 0.06 MPa (i.e., 2.2 V_p-p, the level at which the acoustic radiation force just balanced the gravitational force on the aggregate and thus maintained the aggregate in suspension) for the remaining 29 min of aggregate levitation in the trap [27].

Optical system

A fast, high-resolution CCD F-view II camera (12-bit depth) (Soft Imaging System, SIS, GmbH) mounted on an Olympus BX41M reflection epi-microscope allowed observation in the direction of sound propagation (negative z-axis) (Supplementary data Figure 1ii). Images were captured by a standard PC equipped with the Cell-P image acquisition and processing software (Soft Imaging System, SIS, GmbH).

Immunofluorescence

Cells subjected to EndoN, NCAM blocking peptide or NCAM siRNA treatments as well as control, untreated cells were levitated in the trap for 1, 3, 7, 10 or 30 min. Cells subjected to PIPLC treatment were levitated in the trap for 1 or 30 min, whereas those pre-treated with 0.5, 1, 3 or 5 mg/ml of peptide were levitated in the trap for only 30 min. Fixation and immunostaining of the aggregates was performed as previously described [25]. Cells were dual-labeled for F-actin (2 U/ml, Phalloidin-Alexa 488 conjugate; Invitrogen, UK) and NCAM (1:1000, AbCAM, UK) or PSA-NCAM (1:200, Chemicon, UK). The secondary antibody used was Alexa 594 conjugated anti-mouse IgG (5 µg/ml, Invitrogen, UK). Immunolabelling of non-sonicated neural cells subjected to all of the above treatments was performed on cells (10⁶ cells/ml) retained for 1, 3, 7, 10 or 30 min in suspension. A 100 µl cell suspension aliquot was placed on a Histobond slide, fixed and double stained for F-actin and NCAM or PSA-NCAM as described above.

Measurements of membrane spreading

Estimates of the angle θ subtended by the F-actin staining region at the centres of a pair of interacting cells were measured as previously described [33]. A detailed description and a Figure (Supplementary data Figure 1iii) illustrating how the angle θ is determined is given in the Supplementary data (online version only).

Statistical analysis

The data presented here are shown as mean±standard error, whereas those of the angle θ are presented as mean±standard deviation, unless otherwise stated. Each experiment was repeated three times and representative data are presented. Analysis of means was performed with a two-tailed two sample t-test (Minitab 13.0). Differences were considered significant at p values less than 0.05.

Results

NCAM profile of neural cells

Control (NSM-suspended) neural cells express all three isoforms of NCAM (NCAM-120, -140, -180) (Figure 1i); their relative ratios are shown in Table I. NCAM-140 is predominantly present (50%), while NCAM-180 and NCAM-120 are less abundant (at percentages of 30% and 20% respectively) (Table I). PIPLC treatment on the other hand resulted in a 70% removal of NCAM-120 as quantified from densitometric analysis of the Western blot. As expected, the relative ratios of the NCAM-140 and NCAM-180 isoforms were essentially unchanged (Figure 1ii; Table I). Treatment of cells with 2 U/ml of EndoN resulted in complete removal of PSA from NCAM (Figure 1iii). Finally, transfected siRNA NCAM cells were analysed 24, 48 and 72 h post-transfection at the mRNA level using PCR. At 24 h there was no gene silencing, while the maximum silencing (44%) was observed at 48 h post transfection (Figure 1iv). Data are representative of 2 or 3 independent experiments. NCAM mRNA levels were normalized to the housekeeping gene β-actin. NCAM expression 72 h post transfection increased reaching levels similar to the scrambled siRNA and untreated cells. The transfection efficiency reported here for neural cells lies within the range of transfection efficiencies in neural-like cells reported elsewhere [34], [35]. A summary of the NCAM related treatments applied in the current work is given in Table II.

Table I.  Summary of the percentages of the different NCAM isoforms measured by densitometric analysis for control and PIPLC-treated cells.

	NCAM-120	NCAM-140	NCAM-180
Control	20%	50%	30%
PIPLC-treated	6%	61%	33%

Table II.  Summary of the NCAM-related treatments together with their effect on NCAM and PSA-NCAM.

	Polysialylation	NCAM-related membrane properties
Control	Yes	NCAM-120, NCAM-140, NCAM-180
PIPLC	Yes	70% less NCAM-120
Endo-N	No	No PSA
NCAM siRNA	No	Reduced expression of all three NCAM isoforms

Aggregate growth and morphology

Neural cells suspended in NSM formed a small aggregate, of an approximate diameter of 500 µm, within 20 s of ultrasound initiation at the central region of the nodal plane (Figure 2a), while a more closely packed aggregate was formed within 3 min of initiation of cell-cell contact (Figure 3b) [25]; the first signs of extended perimeter cell-cell contacts (i.e., membrane spreading) were already noticeable at that time (Figure 2b, arrows). An aggregate was fully formed within 5 min of ultrasound exposure, as characterized by the absence of free single cells from the field of view. No detectable difference could be observed in the aggregate formation pattern between control (i.e., neural cells suspended in NSM), enzyme treated (PIPLC and EndoN) and NCAM siRNA transfected cells (unpublished data). Continuous microscopic observation showed that the aggregates were formed as follows: incoming single cells and dendritic small clusters formed strong contact instantly with the peripheral of the growing aggregate and did not roll along the periphery. The short dendritic clusters rotated, without internal reorganization, about their first contact point with the main aggregate so that it developed a more closely packed form over an extended area [25]. We have previously shown [36] that at the operational acoustic pressures (0.06–0.54 MPa) employed here, the stress imposed on cells at the pressure nodal plane by acoustic streaming flow is much less than that experienced by cells during standard centrifugation procedures, i.e.,<28 g.

Figure 2.  (a) Neural cell aggregate levitated in the trap for 20 s. Cell aggregate morphology was open with a number of voids. Scale bar is 60 µm. (b) Zoom-in image of part of an aggregate levitated in the trap for 3 min; the initial signs of an extended perimeter contact are shown (arrows). Scale bar is 20 µm.

Figure 3.  (i) Representative micrographs captured from different fields of view of the distribution of F-actin (a, b, c) and NCAM (d, e, f) in neural cells levitated in the trap for 1 (a, d), 7 (b, e) and 30 min (c, f). At 1 min (a) cells showed a small and thin interfacial actin (arrows), while at 7 min (b) the length and thickness of the interface has increased (arrows); at 30min (c) strong actin accumulation at the cell- cell interface is shown (arrows). NCAM was present at the cell interface from the initial minutes of cell-cell contact (1 min, d, arrows) with progressive accumulation at the interface over the ensuing 29 min (e-f, arrows).Scale bar is 50 µm. (ii) Close-up representative micrographs showing the distribution of PSA-NCAM in cells levitated in the trap for 1 (a) and 30 min (b). PSA is initially (1 min, a) located pericellularly in a spot-like pattern (arrow), while after 29 min (b) a continuous PSA-NCAM interface is observed (arrow). Scale bar is 15 µm. This Figure is reproduced in colour in Molecular Membrane Biology online.

F-actin reorganization and NCAM distribution in cell aggregates subjected to different treatments

Neural cell aggregates removed from the trap after 1 min of ultrasound exposure dissociated into small 2-D clumps (typically 10–50 cells) upon deposition on a glass slide. In contrast, 30 min aggregates remained intact when placed on a slide (indicating that the mechanical strength of the aggregate had increased), but subsequently broke up into large fragments of 100 or more cells during the fixation process [27]. Images were captured from all the fragments of the aggregate (approximately 100 fragments), providing guidance on the typical recurring patterns characteristic of a sample. It is noted that, while samples were removed from the trap after 1, 3, 7, 10 and 30 min of levitation, the immunofluorescent images correspond to the ‘real’ times of 2.5, 4.5, 8.5, 11.5 and 31.5 min due to the time gap (typically 1.5 min) from termination of ultrasound to specimen fixation.

Control (NSM-suspended cells), PIPLC- and EndoN-treated cells

F-actin in control cells, following 1 min of ultrasound exposure was distributed in a cortical arrangement at the cell surface (Figure 3ia). A thin and short region of high staining at the cell-cell contact region was also observed (Figure 3ia, arrows). After 7 min, actin label progressively increased at the interfaces of the cells (Figure 3ib, arrows), while in the ensuing 23 min, F-actin strongly accumulated at the cell-cell contact interface (Figure 3ic, arrows). There was no detectable difference in F-actin organization, as observed by immunofluorescence, between control, PIPLC-treated (unpublished data) or EndoN-treated cells (Figure 4ia).

Figure 4.  Representative micrographs of the distribution of F-actin (i–iv a), PSA-NCAM (i b) and NCAM (ii–iv b) in cells pre-treated with (i) 20 U/ml EndoN, (ii, iii) 175 ng of NCAM siRNA and (iv) 0.2 mg/ml of peptide. (i) EndoN pre-treated cells showed an F-actin distribution at 7 min (a) comparable to control cells at 30 min (Figure 4ic), while no PSA-NCAM (b) was detected. (ii) Little F-actin could be detected at the cell-cell interface (a) in NCAM siRNA transfected cells levitated in the trap for 30 min; (b) NCAM -/- staining in these cells; (c) superimposed image of F-actin with NCAM. Attention is drawn to a set of four NCAM -/- cells present in this field of view embedded in a red rectangle and highlighted in (iii). (iii) F-actin (a) in NCAM knockdown cells was diffusely localized in the cells; a short and thin interface is also shown. Very little or no NCAM (b) could be detected; (c) superimposed image of F-actin with NCAM. (iv) In peptide pre-treated cells, 2.76-fold less F-actin (a) is concentrated at the cell-cell interface over 30 min; NCAM (b) was to a great extent localized pericellularly (block arrow), whereas interfacial NCAM staining appeared quite punctate (dotted arrow); (c) superimposed image. Scale bar in (i, ii, iv) is 100 µm and in (iii) 10 µm. This Figure is reproduced in colour in Molecular Membrane Biology online.

NCAM on the other hand, in control cells was detected at the cell-cell interface from the first minutes of ultrasound exposure (i.e., at 1 min of ultrasound exposure) (Figure 3id); however, the interfacial staining was occasionally non-continuous (Figure 3id, arrows). NCAM staining at the interface was more continuous 6 min later (Figure 3ie, arrows), while after 30 min NCAM was distributed evenly at the cell-cell contact area (Figure 3if, arrows). As was the case for F-actin, no detectable difference in the temporal distribution of NCAM could be observed between control, PIPLC- or EndoN-treated cells (unpublished data).

NCAM siRNA knockdown cells

As shown in Figure 1iv, there was an overall 44% reduction in the amount of NCAM mRNA expressed by siRNA transfected cells at 48 h. Selection of a targeted cell population (as in the present case of NCAM -/- cells) by flow cytometry (FACS) has been achieved elsewhere [34]; however, no rat NCAM antibody, including the NCAM antibody we have used in this study for Western blotting and immunofluorescence, is currently available for FACS analysis, to our knowledge. Consequently, the unselected neural cell population was introduced into the trap. Removal of aggregates from the trap at different time points, fixation of aggregates and immunofluorescence proceeded as usual; however, measurements of the F-actin integral intensity and the angle θ of the cell-cell interface ‘chord’ were made only from pairs of cells that were NCAM -/-; a total of 20 randomly pairs of NCAM -/- cells were analysed (n=20). Figure 4iia shows the F-actin distribution at sites of cell-cell contact of a 30 min aggregate consisting of cells expressing different levels of NCAM (Figure 4iib). A range of NCAM intensities is observed; pairs of cells having a high NCAM staining intensity are found adjacent to cells of low NCAM staining intensity respectively. Attention is drawn to a set of four NCAM -/- cells present in this field of view (Figure 4iib and highlighted in iiib). F-actin in NCAM knockdown cells was localized in a very short and thin interface (Figure 4iia and iiia), while no cortical organization could be observed (Figure 4iia and iiia). No difference could be observed between knockdown cells in contact for 1 or 30 min, i.e., no F-actin reorganisation was detectable. Figure 4iic and iiic show almost no co-localization of F-actin with NCAM in the NCAM siRNA transfected cells.

The above qualitative observations were validated by quantifying the temporal accumulation of F-actin at the cell-cell interface (Figure 5). Intensity measurements of the phalloidin staining of 20 randomly selected paired cells (with the exception of NCAM siRNA cells where measurements were performed on a selected NCAM -/- subpopulation of 20 paired cells) showed that the amount of F-actin at the cell-cell contact region progressively increased over 30 min of cell-cell contact duration (Figure 5). A similar increase was quantified for the PIPLC pre-treated cells (Figure 5). In contrast, the amount of F-actin recruited to the cell-cell interface of EndoN-treated cells was lower than that of control cells; however it increased similarly to control cells to reach the same final integral intensity at 30 min (Figure 5). There was no significant difference (p>0.05) in the F-actin integral intensity of control, PIPLC and EndoN pre-treated cells in contact for 1 and 30 min. Finally, NCAM siRNA cells showed no significant F-actin reinforcement at sites of cell-cell contact from 1 to 30 min (p<0.05, i.e., significantly different to control cells) (Figure 5).

Figure 5.  Measurements of the integral intensity of AlexaFluor 488-Phalloidin staining at the cell-cell interface of 20 randomly selected pairs of cells (n=20) pre-subjected to different treatments. Measurements were performed in aggregates of cells being in contact in the trap for 1, 3, 7, 10 and 30 min. Error bars represent one standard error of the mean.

PSA-NCAM distribution in control and EndoN-treated cells

PSA-NCAM in control (NSM-suspended) cells in contact for 1 min was localized at the cell-cell contact area in a punctate arrangement (Figure 3iia, arrow). Concentration of PSA at the cell-cell interface increased over the following 29 min of cell-cell contact duration and exhibited a continuous, clearly defined staining pattern at the cell-cell contact area (Figure 3iib, arrow); no PSA-NCAM was immunofluorescently found in cell-cell contact free regions. In contrast, no PSA-NCAM could be detected by immunolabelling aggregates of cells pre-treated with EndoN in contact for 1 or 30 min in the trap. Figure 4ib shows that negative PSA-NCAM immunofluorescence was obtained in EndoN pre-treated cells in contact for 7 min; as mentioned earlier no re-expression of PSA-NCAM at the cell surface was observed over the 30 min experimental period.

Intensity measurements of AlexaFluor-594 staining revealed that the amount of PSA present at the cell-cell interface increased by 2-fold, from 503 at 1 min to 1011 µm² at 30 min of cell-cell contact duration. The mean intensity of PSA was derived from measurements of 20 randomly selected pairs of cells from each of two replicate experiments. The variances and means of the intensity distributions of the sets of measurements were not significantly different (p>0.05). AlexaFluor-594 intensity measurements for EndoN-treated cells were not performed as no staining was detected (Figure 4ib).

Peptide inhibition of NCAM-NCAM interactions and their effect on cytoskeletal reorganization and membrane spreading

A 0.2 mg/ml concentration of the NCAM blocking peptide showed a 41% decrease in the F-actin integral intensity (Figure 6i) and a 42% decrease in the angle θ of the ‘cell-cell interface chord’ (Figure 6ii), following 30 min contact compared with the control value for that time. These indices fell further to limiting values at 0.5 mg/peptide (Figure 6i, ii). There was no significant difference between the mean F-actin integral intensity and angle θ values of cells suspended in 0.5–5 mg/ml of peptide; in contrast, control cells and cells suspended in 0.2 mg/ml of peptide were significantly different than cells suspended in 0.5–5 mg/ml of peptide (p<0.05). Cells suspended in NSM containing PBS or a scrambled peptide showed no significant difference (p>0.05) in the amount of F-actin recruited to the cell-cell interface or the angle θ of the cell-cell interface ‘chord’ in relation to control cells.

Figure 6.  Measurements of (i) the integral intensity of AlexaFluor 488-Phalloidin staining at the cell-cell interface of 20 randomly selected pairs of cells (n=20), and (ii) of the angle θ formed by paired cells pre-subjected to 0–5 mg/ml of NCAM-blocking peptide. Error bars represent one standard error of the mean.

F-actin temporal reorganization in the presence of 0.2 mg/ml of peptide

A 2.8-fold increase of F-actin integral intensity over 30 min in cells of aggregates formed in suspension in NSM containing 0.2 mg/ml of peptide was significantly different (p<0.05) from the 4.4-fold increase in the respective value of control cells (Figure 5).

Immunofluorescence observations detected little F-actin at the interface of cells in contact for 1 min (unpublished data). Some peripheral (cortical) F-actin staining was also observed (unpublished data) as was the case for control, NSM-suspended cells (Figure 3ia). In the following 29 min there was some accumulation of F-actin at the interface (Figure 4iva, arrow) but the levels of accumulation were lower than those observed for control cells (Figure 3ia). NCAM, on the other hand, was to a greater extent localized pericellularly (Figure 4ivb, block arrow), whereas interfacial NCAM staining appeared quite punctate (Figure 4ivb, dashed arrow). The limited extent of co-localization of F-actin and NCAM in the peptide-treated cells is shown in Figure 4ivc.

Assessing the rate of membrane spreading under different conditions

The angle θ of the cell-cell interface ‘chord’ of paired cells, as an indication of the spreading ability of cells, was measured for 20 different pairs of cells subjected to all of the above treatments (PIPLC, EndoN, NCAM blocking peptide, NCAM siRNA) as well as for control, untreated cells (Table III). The rates and the asymptotic extents of spreading of control, EndoN and peptide pre-treated cells (but not for PIPLC-treated cells as only two time points were investigated, nor for siRNA NCAM transfected cells as no detectable difference was observed between the angle θ value of cells in contact for 1 or 30 min) were determined from an Origin 7 best fit of the nonlinear curve (Figure 7) (Table III). The following phenomenological equation (Equation 1) was employed:1 where α is the asymptotic value of the angle θ reached after 30 min of cell-cell contact, b is the rate of membrane spreading and t is the cell-cell contact duration time (min).

Figure 7.  Measurements of the angle θ of the cell-cell interface ‘chord’ of cells subjected to different treatments over time (min). Measurements were conducted in 20 randomly selected pairs of cells (n=20); error bars represent the standard deviation.

Table III.  Summary of the different rates of spreading ‘b’ extracted from Origin 7 for the different NCAM-related treatments applied in this study, together with the values of the asymptotic extent of spreading ‘α’.±represents the estimated standard error of ‘α’ and ‘b’ after 10 iterations as generated in Origin 7.

Treatment	Asymptotic extent of spreading ‘α’ (°)	Rate of spreading ‘b’ (min^−1)
Control	104.1°±4.75	0.16±0.02
EndoN	112.2°±3.11	0.42±0.05
Peptide	62.4°±3.26	0.37±0.08
PIPLC^a, ^b	105°±2.4	N/A
SiRNA NCAM^a, ^b	30.7°±3.8	N/A

^aIt is noted that the values of ‘α’ for the PIPLC- and siRNA NCAM-treated cells do not represent the asymptotic extent of spreading and thus were not derived from Origin 7, but are the values obtained directly from the measurements performed in Cell P software. ^bThe rate of spreading was not calculated for the PIPLC-treated cells as only two time points were investigated (1 and 30 min); finally, as no difference was observed between the angle θ value of the NCAM siRNA transfected cells at 1 or 30 min no rate of spreading could be obtained.

Control and Endo-N treated cells

α in cells suspended in NSM alone (control) progressively increased over 30 min to reach 104.1°±1.6 after 30 min of cell-cell contact in the trap (Figures 7 and 8a), while in EndoN-treated cells α is 112.2°±1.6 (Figures 7 and 8b).

Figure 8.  Zoom-in typical micrographs of sets of cells stained with AlexaFluor 488-Phalloidin after 30 min in contact in the trap; (a) control, (b) EndoN, (c) peptide and (d) siRNA NCAM pre-treated cells. The angle θ is measured from immunofluorescence F-actin images, as F-actin reorganization is a marker of the extent of perimeter contact increase. Scale bar is 10 µm. This Figure is reproduced in colour in Molecular Membrane Biology online.

Control and PIPLC-treated cells

PIPLC-treated cells showed no significant difference (p>0.05) in their ability to spread in relation to control cells; α was 105°±2.4 after a 30 min period of cell-cell contact duration.

Control and peptide-treated cells

α for cells treated with 0.2 mg/ml of NCAM-blocking peptide was significantly reduced by 1.75-fold (Figures 7 and 8c) in relation to control cells (p<0.05) (Figures 7 and 8a).

Control and NCAM siRNA cells

Membrane spreading in NCAM siRNA transfected cells was significantly impaired (p<0.05) in relation to control cells. The value of α was 31.5°±2.5 after 1 min and remained essentially unchanged over the 30 min period of cell-cell contact duration (Figures 7 and 8d), i.e., 3.6 times lower than that reported above for control, NSM-suspended cells (Figures 7 and 8a).

The experimental data gave values of b of 0.16±0.02 min⁻¹ and 0.42±0.05 min⁻¹ for control and EndoN-treated cells respectively (Table III), whereas for the peptide pre-treated cells b was 0.37±0.08 min⁻¹ (Table III). The ratio of b of EndoN and control cells and its standard error were 2.63±0.076, whereas the ratio of α was 1.07±0.003. In contrast, the ratio of α and b of control and peptide (0.2 mg/ml) pre-treated cells were 2.31±0.14 and 0.6±0.003 respectively. b of EndoN pre-treated cells was significantly different (p<0.05) than that for controls as was that for peptide pre-treated cells (p<0.05).

Discussion

NCAM has a fundamental role in nervous system development including the mediation of adhesion between neural cells and control of cell motility [3]. Therefore we undertook the present study to examine the specific influence of NCAM and PSA-NCAM on cell-cell adhesion and perimeter membrane contact spreading in adhering neural cell monolayers in suspension using a novel ultrasound trap.

One of the key findings was the observation that the asymptotic extent of spreading α and the rate of membrane spreading b were affected when NCAM was modulated using NCAM siRNA, NCAM blocking peptide, PIPLC or EndoN treatments (Table III). In the NCAM siRNA transfected cells, in which NCAM expression was to a great extent abolished, α was significantly impaired (Table III). Similarly, addition of the competitive NCAM peptide also reduced α in a concentration dependent manner indicating the necessity of NCAM for membrane spreading (Table III). At the highest peptide concentration, α was comparable to the limiting value obtained in the NCAM siRNA transfected cells. The early (i.e., 10 min) occurrence of the asymptotic value α in the temporal study of change in θ in the presence of 0.2 mg/ml peptide (Figure 7), suggests that the decreased value of α observed in the peptide inhibition experiments at 30 min (Figures 7 and 8c) is not controlled by change in the spreading rate b. However, the rate of spreading of peptide pre-treated cells is higher than that of control cells (Table III). Although this result was confirmed in three independent experiments, we can offer no hypothesis at this stage to explain this surprising result; further work is currently in progress to elucidate this point.

PIPLC treatment, although removing 70% of the GPI-linked NCAM-120 isoform, had no detectable effect on the 1 and 30 min values of θ compared to those for control cells (Figure 7 and Table III). While it is reasonable that the partial removal by PIPLC treatment of the lowest incidence NCAM isoform might not modify adhesion the result also suggests that the lipid rafts that are uniquely associated with GPI-linked proteins (not exclusive to NCAM-120) [37] also had no clear influence on adhesion.

Enzymatic digestion, using EndoN, of PSA bound on NCAM significantly increased b (Figure 7; Table III). The increase in b quantified for the EndoN-treated cells falls within the 2.5-fold increase in a basic cell aggregation probability assay [8] and the 5-fold increase in the rate of NCAM-dependent aggregation of membrane vesicles in vitro [38]. As a result, the general role of a negative regulator of cell interactions (not only those mediated by NCAM) has been attributed to PSA [17], [18]. This suggestion well supports the experimental data presented here. The negative electrostatic and steric contributions imposed on cell adhesion by PSA have been removed thus allowing dynamic changes in membrane contacts [10]; consequently cells are able to spread around each other at a faster rate. The asymptotic value α of the angle θ however reaches the same level as control cells after 30 min of cell-cell contact (Figure 7, Table III). The molecular mechanism by which PSA exercises this function remains however an open question [11].

Clearly, in the absence of NCAM receptor interaction there is no subsequent spreading of perimeter contact or increase in length of membrane below which the F-actin cytoskeleton is reorganised. We have also demonstrated, in suspended neural cells, that while the spreading process was complete within 7 min of initiation of cell-cell contact in EndoN pre-treated cells (Figure 7), F-actin integral intensity continued to increase at later times through recruitment of F-actin to the cell-cell interface (Figure 5). More specifically, EndoN pre-treated cells had already reached 70% of its asymptotic value after 1 min of contact concomitant with an F-actin integral intensity of 23% of its final value. Therefore, F-actin is laid down only up to the limiting points that membrane perimeter contact has been achieved. It is well established that cadherins influence F-actin reorganization [39], however little is known about the interactions of NCAM with cytoskeletal proteins. NCAM-140 and -180 can interact directly with isolated β1-subunits of spectrin [40], α-actinin [41] and α- and β-tubulin [41]. There is as yet no specific evidence of which we are aware, that NCAM alone can trigger F-actin cytoskeletal reorganization.

Organization of the intercellular space is often governed by more than one type of adhesion molecule [18]. We have previously shown that C6 neural cells express NCAM and N-cadherin and that NCAM appears at the interface of interacting cells within 1 min of initiation of cell-cell contact [25], while N-cadherin is located intracellularly (i.e., it had not yet redistributed to the cell surface). Redistribution of N-cadherin to the interfacial region occurred after 8 min from initiation of cell-cell contact [25]. The implication is that NCAM concentration at the interface precedes cadherin arrival. In support of this conclusion, Sandig et al. reported that NCAM accumulated first at the tips of filopodia-like processes followed by the arrival of cadherins [42]. This suggests that NCAM is responsible for the initial spreading of the cell perimeter contact and cadherins stabilize these contact areas by anchoring them to the cytoskeleton. Likewise, Keller et al. [43] showed that Walker carcinosarcoma cells with minimal cortical actin and no morphologically recognisable protrusions (due to colchicine and latrunculin A treatments), migrated faster on substrata than control cells with lamellipodia and an intact cortical actin layer [43]. They concluded that the cortical actin layer or actin-powered force generation at the front is redundant for locomotion [43]. Our results, demonstrating that NCAM controls membrane spreading and that cadherins control the subsequent rate of F-actin integral intensity increase, in C6 cells in suspension, are consistent with reports of cells on substrata. We recognize that the NCAM dependence of membrane spreading shown here for the C6 neural cell line cannot be a general model for cell adhesion since cells that do not express NCAM (as for example the prostate epithelial cell line PZ-HPV-7) have exhibited a spreading capacity similar to that reported here for neural cells [33].

Our data also indicate that the Ig3 domain of NCAM is capable of inducing NCAM cell-cell adhesion. This correlates well with a previous study that showed inhibition of E8 chicken embryo neuronal cell aggregation as well as aggregation of NCAM coated microspheres when an antibody against Ig3 was included in the suspending phase [44]. Domain deletion studies with NCAM-transfected cells also support a model in which Ig3 domain inhibition is sufficient to abrogate adhesion [21], [45]. It has been suggested that Ig1 [46] or Ig1 and Ig2 [47] or all five Ig domains [48] are responsible for NCAM homophilic binding. It was beyond the scope of this study to identify which other Ig domains can contribute to the homophilic binding of NCAM.

In conclusion, the results presented here provide direct evidence that the limiting extent of membrane-membrane perimeter spreading of neural cells is attributable to the functionality of NCAM at the cell surface, whilst the subsequent F-actin reorganization is attributable to cadherins localized at the cell surface. In this instance, in the absence of NCAM, no other adhesion molecule was capable of inducing membrane spreading in neural cells. Modulation of PSA content also had a significant effect on the rate of perimeter adhesion development. The cytomechanical consequences of the NCAM modulation approaches described above can also be measured (e.g., by using the dual micropipette assay recently described by Chu et al. [49]) in order to further understand the biological consequences of in vivo changes in PSA-NCAM properties.

Supplementary text

Measurements of membrane spreading

Estimates of the angle θ subtended by the F-actin staining region at the centres of a pair of interacting cells were measured as previously described [33]. Briefly, the contact region was treated as a shared chord of ‘overlapping’ circles (Supplementary Figure 1iii). It is appreciated that the cell perimeters are longer than those for perfect circles as cells have surface area in excess of that required to contain a spherical cell volume (a property necessary so that when a cell divides the two resulting smaller volumes are each continuously covered by membrane) [33]. No geometrical allowance can therefore be made for the consequent departure from circular profiles as membrane spreading continues [33]. Within that limitation it will be taken, for a ‘cell-cell contact chord’ L and cell radius R (Supplementary Figure 1iii), that θ? is the angle subtended by the chord at the centre of the cell. While the membrane dynamics and micromechanics at the end of contact regions are not known it is reasonable to assume that, for cells, the ability to continue to increase the length of a chord depends, among other things, on both the surface density of the homophilic receptors and the local resistance of the membrane to deformation [33]. Measurements of the angle θ of a cell-cell interface ‘chord’ were performed using Cell-P software.Figure 9

Supplementary Figure 1.  (i) Schematic of the temporal progression of cells in a ?/2 ultrasound trap, showing (a) particles homogeneously distributed at time zero, (b) cells already in the pressure node plane within 1 s and (c) subsequent concentration of the cells in a single aggregate. (ii) Experimental assembly: The resonator outer diameter is 25 mm. Its main components were a 1.5 MHz disc transducer that was glue-attached to a steel acoustic coupling layer, a sample volume and glass acoustic reflector. The assembly is placed on the top of an epi-fluorescence microscope's stage. (iii) Schematic diagram of an interacting cell doublet; L is the length of the F-actin staining region or ‘chord’ (drawn here as a straight line), R is the radius of the cells and ? is the angle subtended by the ‘chord’ at the centre of the cell [33].

Acknowledgements

We would like to thank Dr. G. A. Foster for providing the C6 neural cell line.