Long-term maintenance of patterns of hippocampal pyramidal cells on substrates of polyethylene glycol and microstamped polylysine

Abstract

For neurons to attach and remain in precise micropatterns for weeks in culture, background molecules that remain nonpermissive for extended culture durations need to be identified. Nonpermissive background molecules of either polyethylene glycol (PEG) or the amino acid serine (C3H7NO3) were evaluated. The foreground regions were microstamped with 3-, 5-, or 10-micron lines of poly-D-lysine (PDL), which promotes neural attachment and growth. After 29 days in culture the foreground compliance, or the fraction of all live somata which rested on the desired PDL surface, averaged 86% for serine and 90% for PEG, with only a small decline. The background compliance, or the fraction of square areas in the pattern background which were free of neurite extension, declined from highs of 40% and 55% (midculture) to 5.5% and 12% (29 days) for serine and PEG, respectively. Images of the cultures suggest that PEG is significantly more effective as a nonpermissive substrate. We conclude that these materials, especially PEG, are adequate for the maintenance of long-term patterned cultures of neurons. We believe that this is the first report of high-quality long-term patterning of cultured neurons.

Micropatterning neuronal networks

“…22,34 Investigation of polylysine-rejecting coatings Plasma stencilling for patterning biomaterials by backlling relies on the surface coating preventing the adsorption of the biomaterial. PEGylated surfaces are well known for their protein rejecting qualities.…”

“…Alternatively reactive surfaces can be used to tether PEG species. 34,41 However, in both cases coatings were not of uniform quality resulting in limited pattern compliance. Reactive surfaces are difficult to maintain in a pristine condition prior to backlling.…”

Clustering neural spike trains with transient responses

“…A cluster corresponds to those spike trains that led to the same winning node in F 2 . Using these steady state approximations it can be shown analytically that the PART neural network defined by (11)-(15) exhibits a regular computational performance characterized by [26] 47th IEEE CDC, Cancun, Mexico, Dec. [9][10][11]2008 WeA05.2…”

“…Briefly, h ij is computed using (12)-(13) and the winning F 2 is selected. If this node passes the vigilance criterion, (11), then the top-down and bottom-up weights, respectively, z ji , z ij , are updated accorrding to (14)- (15), h ij re-computed, and so on. Typically the single committed node in F 2 is determined within just a few iterations of this procedure.…”

“…This technology has translated into novel therapies which utilize neural stimulation [7] and, for example, the neural control of robotic limbs [8], [9]. Indeed the possibility of constructing direct neural-analog interfaces is becoming a real possibility [10], [11]. However, as the number of recording electrodes grows, the limitations of our current abilities to extract useful information from the resulting large data sets become increasingly apparent.…”