During morphogenesis, neuronal precursor cells migrate from the zone where they are born to their final destination, which, in some cases, is at a distance of several millimeters[1], [2]. After reaching their destination, neurons must establish appropriate synaptic connections by sending out from their soma projections called neurites. The motion of neurites is guided by growth cones located at their tips[3], [4]. Growth cones contain a variety of chemical and mechanical receptors and sophisticated biochemical machinery that couples these receptors to the cytoskeleton[5]–[7]. Extruding from the tip of the growth cone are highly motile structures called filopodia and lamellipodia that are used to explore and probe the environment[3], [6]. All these complex events, which are at the basis of neuronal development and differentiation, involve cell motility requiring a precise control of cellular and molecular motors. The motion of these structures has been analyzed and characterized to some extent by time-lapse microscopy[8]–[12].However, little is known about how neurons use these structures to sense the mechanical properties of their environment and about what range of forces these structures exert during their exploratory motion.

Analysis of the forces exerted by neurons has been limited to theoretical considerations; experimental analysis has been limited to samples of isolated filaments[13]–[17]or migrating cells[18], [19]. Measured forces range from 1 or 2 pN in isolated actin filaments and microtubules to 1 nN in migrating keratocytes. Quantitative characterization of the force exerted by lamellipodia and filopodia during neuronal differentiation could help to elucidate how neurons sense the environment and process mechanical information. Precise description of the mechanical and dynamic events that occur during neuronal differentiation and migration would provide new insights regarding the molecular events controlling these biological functions. In addition, it would offer a more precise way for evaluating the role of molecular motors in cell motility under physiological conditions and in neurodegenerative disease.

In this study, we used optical tweezers[20]–[22] to measure the force exerted by filopodia and lamellipodia during neuronal differentiation. Unlike other force measurement methods, optical tweezers are non-invasive and provide direct high temporal resolution for position detection (<10 nm) and force measurement (<1 pN), highly relevant in biological systems[21]. We found that a single filopodium exerts a force not exceeding 3 pN. In contrast, lamellipodia exert forces of 20 pN or more lasting less than 1 s to approximately 30 s. Treatment of growth cones with the selective myosin light chain kinase (MLCK) inhibitor ML-7[23] or the microtubule depolymerizing agent nocodazole[24] drastically reduced the motion and force exerted by lamellipodia, while filopodia continued to move and exert forces up to 3 pN. Growth cones treated with the actin depolymerizing agent latrunculin A[24] did not exert any detectable force. These findings suggest that no force can be produced in the absence of actin polymerization and that development of forces larger than 3 pN requires microtubule polymerization. This study shows that not only do neurons process information, but also they act on the environment, exerting forces varying 1 to 2 orders of magnitude.

During neuronal differentiation and development, the growth cone of each neurite extends its filopodia and lamellipodia to explore the chemical nature of the environment and to probe the rigidity and composition of the extracellular matrix[23]. Under these circumstances, cell motility is strictly linked to the generation of forces. Therefore, we used optical tweezers[20]–[22] to measure the force exerted by the growth cones of differentiating neurons.

A typical growth cone can be divided into two regions: the central region and the peripheral region. The latter consists of filopodia and a lamellipodia[33]–[36]; the motion of these structures is a major component of neuronal differentiation. This is the first report of a quantitative determination with a millisecond temporal resolution of the force exerted by filopodia and lamellipodia in differentiating neurons. The force developed over time, with a maximal rate of increase of 10 pN s-1. Thin filopodia, during a protrusion or lateral collision (Fig. 2), exerted a force not exceeding 3 pN. In contrast, lamellipodia exerted a force of up to 20 pN and possibly more, which could increase in discrete steps of approximately 0.2 pN (Fig. 4d). These steps had properties very similar to those observed during the assembly of isolated microtubules[16]. The measured forces were smaller than forces involved in cellular traction force or measured in migrating keratocytes[15], [18]. Measured forces here reported, may not fully represent the ability that lamellipodia have because, at least in some cases, only a fraction of the forces exerted is picked up by the beads and therefore the value of 20 pN here reported for lamellipodia is the maximal force that was measured. Indeed we expect lamellipodia to exert larger forces, possibly up to hundreds of pN, as in migrating epithelial cells[37]. The diameter of filopodia tips is approximately 100 nm, i.e. 10 times smaller than the diameter of the used beads, therefore the maximal force measured for filopodia is expected to be a reliable estimate of the force exerted by these structures.

Force measurements with optical tweezers require test beads to be in the harmonic potential well of the trapping optical force and to be displaced from its equilibrium position inside the well only by the force to be measured. When adhesion forces between the bead and the growth cone and/or between the bead and the substrate become dominant, the stiffness of the optical trap is profoundly modified and it is impossible to obtain an accurate force measurement. Therefore, it is necessary to place the bead at 1 micron or so from the substrate where neurons are growing. As exploring filopodia and lamellipodia lift up from the substrate, forces can be reliably measured under these conditions.

Filopodia are composed of bundles of actin filaments and occasional microtubules [5], [24]. We observed that the force exerted by a protruding filopodium is in the pN range, not exceeding 3 pN. Its amplitude is of the same order as that measured during polymerization of actin filaments[14], [17] and microtubules[16]. This similarity implies that the protrusion force generated by polymerization is minimally counterbalanced by the membrane surrounding actin bundles and microtubules, indicating that the membrane at the filopodia tips has a low stiffness[38], [39].

Simple mechanical considerations show that the force exerted by a wandering filopodium during a lateral collision (Fig. 2a–c) can be accounted for by the elastic force expected from its flexural rigidity[19], [40] and its bending or buckling. No additional contribution from other force-generating mechanisms is required. The exact mechanisms causing filopodia to bend and/or buckle are not understood. Thermal fluctuations certainly provide a constant driving force, but a variety of other motor proteins[19], [29,29] present in the growth cone could intervene, although their relative contribution is still unknown. Indeed, inhibition of myosin II and microtubule polymerization blocked lamellipodia motion and drastically reduced the force produced by growth cones (Fig. 6), while filopodia continued to move and were able to exert forces in the pN range. In contrast, with blockade of actin polymerization, filopodia and lamellipodia produced no measurable forces. Thus, in the absence of actin polymerization, growth cones cannot exert any force, and microtubule polymerization is necessary for development of forces exceeding 3 pN. Therefore, actin filaments and microtubules cooperate and interact in a complex way so as to generate a wide range of forces.

The motion of filopodia and lamellipodia seems to follow stereotyped patterns wherein the stiffness of an obstacle is first probed. Often, an isolated filopodium changed its direction of growth after colliding with a trapped bead. In contrast, lamellipodia could remove an obstacle, often by growing underneath it and lifting it. Exploring filopodia exerted forces in the pN range, whereas migrating cells exert forces in the nN range[18]. A migrating neuron must be able to displace large obstacles; hence, it uses large forces. Filopodia gently explore their environment using only weak forces, and lamellipodia can exert a larger force opening the way for the growth cone. Thus, not only do neurons process information but they are also able to mechanically modify their environment by selecting forces varying from less than 1 pN to 1 nN[18]. Indeed, differentiating neurons sense the mechanical and chemical properties of barriers in front of their neurites and appear to have smart molecular motor planning, which guides and modifies the ultimate direction taken by neurites in the developing nervous system. Notably, these capabilities are in sharp contrast with metal and/or silicon components used for commercial information processing, which lack motility and motor planning.

Rats (P10−12) were anesthetized with CO2 and sacrificed by decapitation in accordance with the Italian Animal Welfare Act. DRGs were incubated with trypsin (0.5 mg/ml), collagenase (1 mg/ml), and DNase (0.1 mg/ml) in 5 ml Neurobasal medium in a shaking bath (37°C, 35–40 min). They were mechanically dissociated, centrifuged at 300 rpm, resuspended in culture medium, and plated on poly-L-lysine-coated (0.5 ug/ml) coverslips. Cells were incubated for 24 to 48 h, and nerve growth factor (50 ng/ml; Alomone, Israel) was added before measurements were obtained.

The optical tweezers setup was built as previously described.[41] The dish containing the differentiating neurons and the beads (PSI-1.0NH2; G.Kisker GbR, Steinfurt, Germany) was placed on a microscope stage, which could be moved by a 3-axis piezoelectric nanocube (17 MAX 301; Melles Griot Inc., USA). The temperature of the dish was maintained at 37°C using a Peltier device. Bead position was determined in the x,y plane with an accuracy of 10 nm, using back focal plane (BFP) detection which relies on the interference between forward scattered light from the bead and unscattered light[22], [42]. The BFP of the condenser was imaged onto a QPD, and the light was converted to differential outputs digitized at 4 kHz and low-pass filtered at 2 kHz. The bead displacement d = (dx,dy) from the equilibrium position inside the optical trap was also determined by video tracking using correlation methods with sub-pixel resolution. The lateral trap stiffness κx,y = (kx,ky) and the detector sensitivity were calibrated using the power spectrum method[22], with voltage signals filtered and digitized at 5 and 20 kHz, respectively. For multiple trapping experiments, computer-generated diffractive optical elements were projected onto the liquid crystal display of the phase-programmable modulator (PPM X8267-11; Hamamatsu Photonics, Japan)[41], [43] in order to generate multiple spots in the specimen with a Gaussian intensity profile. For experiments where a single Gaussian beam was required, the PPM was switched off. In multiple trapping experiments, only the direction of the force was determined but not its amplitude.

For statistical analysis, QPD traces were low-pass filtered at 50 Hz. Collisions selected for statistical analysis had to satisfy three criteria: 1) maximal amplitude larger than 5σ, 2) duration longer than 100 ms, and 3) presence of a colliding filopodium or lamellipodium in contact with a bead verified by visual inspection of the movie. The collision duration was calculated as the interval between two consecutive crossings of 5σ. The force exerted by the neurite Fneu was calculated as -Ftrap. When the displacement of the bead from its equilibrium position inside the trap was less than 400 nm, Ftrap = (Fx, Fy) was calculated as Fx = dxkx and Fy = dyky[22]. When the bead was also moved along the vertical axis, the lateral displacement measured with the QPD was compared with data obtained from video tracking; the data were discarded if lateral displacements measured with the two methods differed by more than 50%. The axial force along the z axis was not measured.