In this study we show that DS addition to dissociated neuronal cultures promotes cellular clustering and the formation of fiber fascicles which serve as substrates for migrating neurons. We use this new cell culture model to identify kinetic distinctions between slow and fast moving cortical neurons.
At present, cortical migration studies rely on in vivo
analysis and organotypic slices preparations. Improvements in acute cortical slice preparations have permitted detailed analysis of migration movements [2
], but because the migrating cells are entirely contained within the slice of tissue, the slice preparation may not allow rapid application of some reagents (recombinant proteins, function blocking antibodies and pharmacological reagents) at defined concentration, to the migrating cells. Moreover, the slice preparation does not permit phase contrast and DIC microscopy that are extensively used in other studies of cellular motility.
analysis of neocortical migration, where cells are completely accessible, has primarily relied on one model called the imprint assay [20
]. This assay involves cutting 200 μm thick slices of cortex, digesting them briefly with a protease, and then imprinting the slice onto a sticky substrate (Cel-Tak). After washing away the non-adherent cells, a layer of cells is left on the substrate. The layer contains many cells including radial glial cells with attached, migrating neurons. The assay is a useful tool in the analysis of radial glial guided migration and has provided important insights into critical signaling pathways [20
]. Lattice cultures complement existing models and have a number of strengths including ease of preparation, physiological migration speed and suitability for imaging and pharmacology.
Lattice culture formation is induced by DS addition to the media. Although the mechanism of DS induced conversion of dissociated cortical cultures is not known in detail, the negatively charged DS may neutralize the positively charged PDL substrate thereby reducing cellular adhesion [23
]. This reduction in substrate adhesion appears to change the adhesive preference of cells, promoting cell-cell adhesion and allowing cell bodies to aggregate while their processes fasciculate. Similar formations were observed subsequent to lectin treatment of granule cell cultures and were termed 'cable cultures' [24
]. Once these fiber networks are established DS does not appear to be required to sustain migration as removal of DS by media exchange does not change migration rate.
Migrating neurons display a polarity, with a distinct leading process, trailing processes and asymmetric localized organelles [5
]. The migrating cell's leading process appears to extend and retract in a fashion that does not directly anticipate the forward movement of the nucleus [7
]. Instead a dilation of the basal portion of the migrating neuron's leading process precedes the forward movement of the nucleus. This dilation is enriched in membrane vesicles, microtubules, and the centrosome, a microtubule-organizing center [7
]. Centrosomal movements appear to precede nuclear translocation [7
] and recent work reveals that centrosome advancement is essentially constant in cortical neurons but the nucleus moves in a saltatory fashion into the forming dilation [27
The average migration rate in lattice cultures is ~1.5-fold higher than the reported migration rate of locomoting (saltatory) neurons in slice explants [2
] and 4–5 fold higher than diI (1,1'-dioctadecyl-3,3,3',3'- tetramethylindocarbocyanine perchlorate)-labeled neurons in slice culture [19
] or neurons migrating in imprint cultures [20
]. However the average migration rate in lattice cultures is the same as that reported for cortical neurons exiting slice explants and migrating into matrigel matrix (49.6 ± 6.6 μm/hr; p = 0.64) [27
] indicating that cortical neurons can travel at ~50 μm/hr in the absence of surrounding tissue. The difference in average speed between explants and in vitro
approaches may reflect simple steric hindrance or a more complicated interplay of adhesion and chemotropic factors.
In our study we find that overall migration speed is correlated with the frequency of saltatory events in slow cells and the amplitude of saltatory events (distance moved) in faster cells. This relationship emerges from the observation that mean step speed (i.e. nucleokinetic speed), in isolation, cannot account for the 5-fold difference in average cellular speed between the slowest and fastest quartiles of cells. While pathological conditions may alter the speed of nuclear movement, the relative constancy of step speed in our study suggests that the molecular mechanisms that move the nucleus are not a primary determinant of average speed in healthy neurons. Average step distance accounted for most (R = 0.96) of the variance between slow and fast cells (Fig. ), despite the different classes of migrating neurons (Fig. ), and the mixed neuronal and glial character of the fascicles (Fig. ). Thus the relationship between overall speed and average step distance may be a fundamental feature of fiber-guided migration. This finding draws attention to the mechanisms that dictate average step distance.
Depending on the cell class (e.g. cerebellar granule vs. cortical neuron), step distance may be determined by the distance between the nucleus and the dilation [7
], or by the distance between the nucleus and the centrosome immediately prior to nucleokinesis [16
]. If centrosome advancement was continuous rather than saltatory [27
], step distance would be limited by the position of dilation formation in the basal leading process. Little is known about the mechanism that triggers the formation of the dilation, although the possibility that the dilation represents a site of cellular adhesion has been proposed [18
]. This suggestion outlines a model in which reducing the number of cellular adhesion sites would increase step distance and potentially increase average migration rate. In situations of low cellular adhesion like those thought to occur at the end of migration, cellular movement would be relatively continuous and rapid, as has been observed in terminally translocating cells [2
Neurons migrating in the radial direction establish a specialized junction with the radial glial fiber termed the "interstitial density." This density is filled with filamentous material and is characterized by a 20 nm dilation between apposed plasma membranes [20
]. Although the specific proteins that form this junction are not known, normal radial migration appears to require connexin 43, connexin 26 [28
], and IgCAM family member CHL1 [29
]. Although neuronal subtype specific migration defects are found in mice lacking CHL1 [29
], no single adhesion molecule has been identified which is absolutely required for radial migration per se
. For example, suppression of connexin 43 [30
] impairs, but does not prevent migration through the cortical plate. These findings imply that migrating neurons possess multiple, partially redundant adhesion systems. Similarly, tangential migration of interneurons occurs along axons that express the cell adhesion molecule Tag-1 during early cortical development [32
]. Tag-1 has been functionally implicated in tangential migration [6
] however, mice deficient in Tag-1 do not show pronounced cortical interneuron migration deficits [33
]. Observations of rapid switching between migration modes (e.g. tangential to radial) in organotypic slice cultures [19
] also suggest the coincident expression of multiple adhesion systems. Given the mixture of axonal and radial glial fibers found within the fiber fascicles and the multiple redundant adhesion systems deployed in vivo
, it seems likely that migrating neurons in lattice cultures will deploy multiple, functionally redundant classes of adhesion molecules.
The mechanism governing step frequency is unknown. In our study, cells traveling above 26 μm/hr exhibit a step frequency of ~8 steps per hour. However these steps lack rhythm, suggesting that there is not an intrinsic pattern generator underlying step frequency. In migrating cerebellar granule cells, the advancement of the soma is coincident with transient elevations of intracellular Ca2+
and, imposing or blocking these transients alters the rate of migration [35
]. If migrating cortical neurons exhibit similar calcium transients during migration, such transients might regulate the saltatory component of migration, namely nucleokinesis, rather than the apparently smooth centrosomal advancement. In this emerging model, step frequency would be related to the frequency of dilation formation in the leading process as well as the frequency of saltatory event initiation.
Recent studies have begun to shed light on the motors underlying nucleokinesis. Insufficiency in the lissencephaly gene Lis1 reduces migration rate [36
] by either decoupling the centrosome from the nucleus [26
] or preventing the forward advancement of the centrosome and the nucleus [27
]. Lis1 appears to regulate the activity of cytoplasmic dynein, a minus-end directed microtubule motor required for centrosomal advance and nucleokinesis [27
]. Dynein is found around the nucleus and in the dilation of the basal leading process, suggesting it may exert a "pulling" force on the nucleus. However, dynein is not the only motor involved in nucleokinesis; multiple studies, including our own, have shown that blockade of myosin-II prevents somal advance [7
]. As myosin-II immunoreactivity is found primarily behind the nucleus [7
], it may exert a "pushing" or protrusive force in concert with dynein's pulling force. The relationship between these two force-generating systems is not completely clear but there is evidence that microtubule destabilization may activate myosin-II to initiate nucleokinesis [18
]. Sorting out the mechanisms underlying saltatory movement will require further pharmacological and genetic studies in which fiber-guided migration can be explored quantitatively.