We studied the attachment of astroglial cells on smooth silicon and arrays of silicon pillars and wells with various widths and separations. Standard semiconductor industry photolithographic techniques were used to fabricate pillar arrays and wells in single-crystal silicon. The resulting pillars varied in width from 0. 5 to 2.0 micrometer, had interpillar gaps of 1.0-5.0 micrometer, and were 1.0 micrometer in height. Arrays also contained 1.0-micromter-deep wells that were 0.5 micrometer in diameter and separated by 0.5-2.0 micrometer. Fluorescence, reflectance, and confocal light microscopies as well as scanning electron microscopy were used to quantify cell attachment, describe cell morphologies, and study the distribution of cytoskeletal proteins actin and vinculin on surfaces with pillars, wells, and smooth silicon. Seventy percent of LRM55 astroglial cells displayed a preference for pillars over smooth silicon, whereas only 40% preferred the wells to the smooth surfaces. Analysis of variance statistics performed on the data sets yielded values of p > approximately.5 for the comparison between pillar data sets and < approximately.0003 in the comparison between pillar and well data sets. Actin and vinculin distributions were highly polarized in cells found on pillar arrays. Scanning electron microscopy clearly demonstrated that cells made contact with the tops of the pillars and did not reach down into the spaces between pillars even when the interpillar gap was 5.0 microm. These experiments support the use of surface topography to direct the attachment, growth, and morphology of cells. These surfaces can be used to study fundamental cell properties such as cell attachment, proliferation, and gene expression. Such topography might also be used to modify implantable medical devices such as neural implants and lead to future developments in tissue engineering.

According to the treadmill hypothesis, the rate of growth cone advance depends upon the difference between the rates of protrusion (powered by actin polymerization at the leading edge) and retrograde F-actin flow, powered by activated myosin. Myosin II, a strong candidate for powering the retrograde flow, is activated by myosin light chain (MLC) phosphorylation. Earlier results showing that pharmacological inhibition of myosin light chain kinase (MLCK) causes growth cone collapse with loss of F-actin-based structures are seemingly inconsistent with the treadmill hypothesis, which predicts faster growth cone advance. These experiments re-examine this issue using an inhibitory pseudosubstrate peptide taken from the MLCK sequence and coupled to the fatty acid stearate to allow it to cross the membrane. At 5-25 microM, the peptide completely collapsed growth cones from goldfish retina with a progressive loss of lamellipodia and then filopodia, as seen with pharmacological inhibitors, but fully reversible. Lower concentrations (2.5 microM) both simplified the growth cone (fewer filopodia) and caused faster advance, doubling growth rates for many axons (51-102 microm/h; p <.025). Rhodamine-phalloidin staining showed reduced F-actin content in the faster growing growth cones, and marked reductions in collapsed ones. At higher concentrations, there was a transient advance of individual filopodia before collapse (also seen with the general myosin inhibitor, butanedione monoxime, which did not accelerate growth). The rho/rho kinase pathway modulates MLC dephosphorylation by myosin-bound protein phosphatase 1 (MPP1), and manipulations of MPP1 also altered motility. Lysophosphatidic acid (10 microM), which causes inhibition of MPP1 to accumulate activated myosin II, caused a contracted collapse (vs. that due to loss of F-actin) but was ineffective after treatment with low doses of peptide, demonstrating that the peptide acts via MLC phosphorylation. Inhibiting rho kinase with Y27632 (100 microM) to disinhibit the phosphatase increased the growth rate like the MLCK peptide, as expected. These results suggest that: varying the level of MLCK activity inversely affects the rate of growth cone advance, consistent with the treadmill hypothesis and myosin II powering of retrograde F-actin flow; MLCK activity in growth cones, as in fibroblasts, contributes strongly to controlling the amount of F-actin; and the phosphatase is already highly active in these cultures, because rho kinase inhibition produces much smaller effects on growth than does MLCK inhibition.

No abstract

Although scientists for the past 100 years have been studying the effects of surface contact on the growth and development of mammalian cells, it has only been during the last two decades that various researchers around the world have been employing the tools developed by the semiconductor industry to fabricate complex surface features for use in cell culture studies [1]. The work presented here concerns the study of the effects of structural cues on the attachment, spreading and outgrowth of mammalian central nervous system cells. This topic of research and others like it stand at the common boundaries between physics, engineering, and biology in the relatively new field of "Nanobiotechnology".Our motivation for studying the interactions between central nervous system cells and topographically modified surfaces lies in the hypothesis that structural cues on the cellular size-scale might significantly impact the formation of detrimental glial scars around implanted neural devices. The formation of such an electrically insulating glial sheath renders a probe useless for long-term in vivo applications [2]. Thus, early work focused on the responses of astroglial cells to various topographically patterned substrates [3]. Focus was then redirected to study the effects of these same structural cues on the attachment and process outgrowth of hippocampal neurons as a function of the geometries presented to the cells.Standard semiconductor fabrication techniques were used to fabricate micrometer-sized columnar structures in single-crystal silicon wafers (see Figure 1). Photolithography and reactive-ion etching were the cornerstone methods employed in the fabrication process. Methods of hot-embossing were also used to fabricate topographically patterned polymer substrates for use in time-lapse imaging. Substrates were patterned with a range of feature sizes from 0.5 µm up to 2.0 µm, separated by a range of gaps from 0.5 to 5.0 µm.Prior to cell culture, sterilized patterned substrates were incubated in 1 mg/ml solutions of poly-llysine in borate buffer to promote neuron cell adhesion. Cells were then plated on surfaces and allowed to adhere and spread for periods from 24 hours to 21 days. At pre-determined time points, cells were fixed and stained immunochemically to visualize cell membranes, processes, and the expression of specific proteins via fluorescence microscopy (see Figure 2A). Fixed samples were also critical-point dried and gold-coated for scanning electron microscopy (SEM) analysis (see Figure 2B). Data contained in the fluorescence and SEM images were used to quantify the responses of hippocampal neurons to the columnar structures. The rates of process (axonal and dendritic) outgrowth and the degree of process alignment to the patterns on the substrate were measured as a function of the inter-pillar spacing, or geometric constraint, presented to the cell. Additionally, these results and the images obtained in time-lapse studies were used to assist in the development of