5
Defined Networks of Neuronal Cells in Vitro

Andreas Offenhäusser and Angela K. Vogt

5.1

Introduction

The growth of neurons into networks of controlled geometry is of major interest in the
field of cell-based biosensors, neuroelectronic circuits, neurological implants, and phar-
maceutical testing, as well as in fundamental biological questions about neuronal interac-
tions. The precise control of the network architecture can be achieved by defined engineer-
ing of the surface material properties – this process is called neuronal cell patterning.
Within the literature, it is possible to find a wide range of methods for the formation
of well-defined structures on surfaces for cell patterning, and these will be reviewed
with regard to the formation of neuronal networks. These methods can be mainly divided
into two types. On the one hand, cell patterning can be induced by shapes and textures
formed in the substrate; this is called topographical patterning. On the other hand, the
term “chemical patterning” is used when differences in (bio)chemical adhesion properties
are produced on the surface of the substrate. The pattern-inducing effect is not always evi-
dent however, as on the one side the interaction of ECM proteins with topographical struc-
tures can result in (bio)chemical differences on the substrate, whilst on the other side the
(bio)chemical modification of a surface always includes physical effects.

In the second section of the chapter some background information will be provided

about the signaling within a biological neuronal network, together with an overview of
the history of in-vitro neuronal cell patterning. In the third section, some recent results
will be presented from current methods used in neuronal cell patterning, and this will
be followed by a discussion of the application of these methods in the field of bioelectronic
devices.

66

Nanobiotechnology. Edited by Christof Niemeyer, Chad Mirkin
Copyright

c 2004 WILEY-VCH Verlag GmbH & Co. K aA, Weinheim

ISBN 3-527-30658-7

G

5.2

Overview: Background and History

5.2.1

Physiology of Information Processing within Neuronal Networks

Neurons use differences in electrochemical potential to encode information, and mes-
sages can be passed on to other neurons through either chemical or electrical connections.
Chemical synapses comprise the conversion of an electrical signal – the action potential
arriving at an axon terminal – into a chemical signal; that is, the release of a neurotrans-
mitter into the synaptic cleft. The binding of a neurotransmitter to specific postsynaptic
receptors triggers the opening of an intrinsic ion channel in the membrane of the post-
synaptic cell. The resulting ion flux alters the transmembrane potential and facilitates
or suppresses the generation of a new action potential in this cell [1]. The excitatory neu-
rotransmitters acetycholine, glutamate, and serotonin open cation channels (Na

+

, K

+

, Ca

2+

)

which depolarize the postsynaptic cell, thus facilitating the generation of an action poten-
tial [2]. Inhibitory neurotransmitters, such as glycine and g-aminobutyric acid (GABA),
activate anion channels (Cl

–

, HCO

3

–

) that lead to hyperpolarization, thereby suppressing

neuronal firing. At electrical synapses, ion channels connect the cytoplasm of the pre-
and postsynaptic cells, and some current from the presynaptic cells also flows through
these low-resistance, high-conductance channels. This current can depolarize the post-
synaptic cell and, as a consequence, can induce an action potential [3]. The gap junction
channels of an electrical synapse thus mediate electrical signal transmission. Rectifying
and nonrectifying electrical synapses do not appear to differ in ultrastructure. At both
type of synapses, markers such as fluorescent dyes flow readily between the pre- and post-
synaptic cells through the junction. The major difference between the two classes of elec-
trical synapses may reside in the extent to which channel gating is sensitive to voltage.
However, chemical synapses – in contrast to electrical ones – exhibit plasticity and thus
are thought to be responsible for processes such as learning and memory. Therefore,
their presence in the system is absolutely crucial for the study of network behavior, plas-
ticity, and activity-dependent changes.

5.2.2

Topographical Patterning

The employment of topographical patterning techniques started in the early 1960s [4] and
1970s. At this time, typically planar substrates with etched or scribed grooves or glass
fibers were used as the means to study cellular patterning. With increasing evidence that
curvature was the main effect in cell guidance on surfaces, more groups began to examine
the effects of varying groove depth, width, and spacing. In the 1980s, lithography was used
to microfabricate grooved surfaces, in particular by utilizing anisotropic etching of silicon
wafers. For further details about concepts, materials, surface structures, and possible cel-
lular and biomolecular mechanisms for topographically patterning that were presented
over the past, the reader is referred to reviews by Curtis and Wilkinson [5, 6] and by
Jung et al. [7]. During the late 1980s and the early 1990s, the Glasgow group was starting

67

5.2 Overview: Background and History

to study the relative effects of groove depth and pitch, and cell guidance by ultrafine struc-
tured quartz and silicon surfaces produced by electron beam lithography [8]. Later, a com-
bination of adhesive stripes and topographic features was studied in regard of cell pattern-
ing [9]. These authors showed that for strong adhesive stripes and shallow grooves, the
cells aligned along the adhesive stripes. As groove depth increased, the degree of pattern-
ing increased along the groove direction. An example of topographical patterning is
shown in Figure 5.1. Further discussion of topographical methods for neuronal cell
patterning can be found later in this chapter.

5.2.3

Chemical Patterning

The first results of chemical cell patterning were introduced during the mid-1960s when
Carter et al. discovered that fibroblasts adhered preferentially to palladium islands evapo-
rated onto a polyacetate surface [10]. In 1975, Letourneau was using this method to study
the alignment of chick dorsal root ganglion neurons on palladium regions on polymeric
substrates [11]. He could demonstrate that the cells adhered well on the metal regions
when evaporated onto tissue culture plastic, but showed only weak adhesion when the pal-
ladium was surrounded by polyornithine. These studies showed that differences
(’contrast’) between adjacent regions are necessary in order to obtain cell patterning.
Later, the role and function of extracellular matrix (ECM) proteins were used to improve
cell adhesion and growth and promote neurite extension [12–18]. These proteins were ad-
sorbed onto solid surfaces in order to study cell adhesion in vitro. A major advance was
the identification of the Arginine-Glycine-Aspartate (RGD) cell adhesive domain present
in some ECM proteins [19, 20]. The RGD domain binds specifically to integrin receptors
on the outer membrane of the cell. The authors have also used a covalent tethering
method of the active peptide sequence to the surface. Later, other recognition subunits
in proteins responsible for cell adhesion were identified, including the B1 chain of the
laminin [17, 21]. By using only the protein recognition sequences in combination with
a spacer molecule to achieve cell adhesion, any issues of protein conformation can be
neglected.

68

5 Defined Networks of Neuronal Cells in Vitro

Figure 5.1

Left: Two nerve cells (dorsal

root ganglion, rat) growing in adjacent
12 mm-wide, 3 mm-deep grooves with
superimposed parallel adhesive tracks
(laminin) (From S. T. Britland et al.,

Exp.

Biol. Online 1996, 1, 2). Right: Nerve
cells on microelectrode array with to-
pographical guidance structures (From
M. Denyer et al.,

Cell. Eng. 1997, 2,

122–131).

In 1988, Kleinfeld, Kahler and Hockberger used photolithographic techniques for the

patterning of silanes on surfaces for the controlled adhesion and growth of neurons
[15]. This report probably had the highest impact on the field of neuronal cell patterning.
In 1992, the Curtis group photolithographically patterned laminin in lines of 6 mm width
for the outgrowth of neurites [22]. At about the same time, the Fromherz group demon-
strated that photolithography could be used to pattern ECM proteins for the guided out-
growth of leech neurons in culture [23, 24], while in the mid-1990s the Aebischer group
covalently modified polymeric substrates for neuronal cell patterning [25, 26].

Similar to the photolithographic methods, organic thin films can be patterned by using

photochemical reactions. The Wheeler group used selective laser ablation to grow rat hip-
pocampal neurons on grids of polylysine with varying line width, intersection distance,
and nodal diameter resulting in a very high compliance (Figure 5.2) [27]. Photoablation
was also used to pattern ultrathin polymer layers in order to control the adsorption of pro-
teins and the adhesion and spatial orientation of neuronal cells onto surfaces (for exam-
ple, see Ref. [28]). Further examples of photochemical patterning of neuronal cells will be
discussed later.

5.3

Methods

Based on the overview presented, we will now focus in more detail on the methods
currently used to produce defined networks of neuronal cells in vitro.

5.3.1

Topographical Patterning

Topographical cell patterning methods have been continuously developed over the past de-
cade. These methods are based on lithography and structuring techniques developed in
microelectronics industry, and are used either to pattern resist on the surface of a silicon

69

5.3 Methods

Figure 5.2

Networks of neurons on

a pattern of polylysine on glass. The
polylysine was patterned through a
quartz mask with a pulse from a 193 nm
wavelength excimer laser. (From Ref.
[27].)

wafer and selectively etch away the material of interest, or selectively to deposit a layer of
material to yield topographical patterns on the substrate [29].

Craighead et al. used micron-size topographical features to influence the pattern forma-

tion of neuronal cells [30]. They observed the preferred attachment of astrocytes and neu-
rons to arrays of silicon pillars, although this mechanism is not completely understood.
For smaller feature sizes electron beam lithography is used to create feature sizes down
to tens of nanometers for applications in neuronal cell patterning as shown by the Cornell
and the Glasgow groups [31, 32].

Topographical patterning was also used in a different way: cells and neurites are immo-

bilized in deep structures rather than modulating the cytoskeleton by imposing mechan-
ical restrictions on the plasma membrane. This approach is based on observations that the
mechanical forces generated by the cells will move or rearrange the neurites during the
culture period. Structures with a high-aspect ratio can be realized by lithography and
structuring methods developed for the production of Micro-Electro-Mechanical Systems
(MEMS). Maher et al. have used this approach to grow neuronal cells in deep pits on a
silicon substrate, and recorded data from them using metal microelectrodes [33]. Griscom
et al. explored a three-dimensional (3D) microfluidic array to influence cell placement and
neural guidance. The complex 3D high-aspect ratio structures of poly-dimethylsiloxane
(PDMS) were made directly on structured silicon wafers and using EPON SU-8 negative
photoresist [34].

Recently, Merz and Fromherz have built on the results of such experiments, and have

obtained well-defined networks of cultured neurons from the pond snail Lymnaea stagnalis
by growing them in a microstructured polyester photoresist (SU-8) on a silicon substrate
[35]. By applying electrophysiological techniques, they studied pairs of nerve cells that had
formed connections, which were identified to be electrical synapses.

5.3.2

Photolithographic Patterning

Photolithographic techniques are well established for mass production of silicon chips
with a resolution and alignment precision in the sub-mm range. The pattern in the photo-
resist, which is generated by light exposure through a mask followed by chemical devel-
opment, can be transferred into thin films of molecules immobilized on a surface [15].
However, organic solvents and alkaline solutions used in the process may influence the
stability of functional molecules. Clark et al. [36] showed that photolithography could
be used to pattern laminin, which guided neurite outgrowth similarly well as other meth-
ods [14]. Standard photoresist techniques have been adapted to generate micropatterns of
proteins on glass by using lift-off and plasma-etching techniques [37, 38].

5.3.3

Photochemical Patterning

Photochemical patterning can be used to pattern self-assembled monolayers (SAM) or
thin films of organic molecules by exposing the surface to UV light through either a
photomask or a metal mask. Usually, illumination with UV light causes oxidation of

70

5 Defined Networks of Neuronal Cells in Vitro

the molecules in the exposed areas (for example alkanethiolate oxidize to alkanesul-
fonate), and this alters the properties of the organic molecules – that is, their solubility.
By immersing the patterned substrate in a solution with another organic molecule, the
illuminated region can be modified with a second monolayer [39]. This method has
been used successfully for the adhesion and growth of rat hippocampal neurons on cir-
cuit-like patterns employing mm features [40]. SAMs of silanes on glass have been used
in combination with deep-UV photopatterning: trimethoxysilylpropyldiethynenetriamine
(DETA) supports cell adhesion and outgrowth; and (tridecafluoro-1,1,2,2-tetrahydrooc-
tyl)-1-1dimethylchlorosilane (13F) is cell repellent [41]. It was shown that networks of neu-
rons grown on DETA patterns against a background of 13F develop normal chemical sy-
napses in culture [42]. Photochemical patterning in combination with versatile crosslink-
ing chemistry and tailored peptides has proved to be very effective in patterning the
growth of hippocampal neurons [43]. The patterning was achieved by applying UV-photo-
masking technique and the chemically attachment of a synthetic peptide derived from a
neurite-outgrowth-promoting domain of the B2 chain of laminin. The attachment was car-
ried out by coupling the peptide to an amine-derived glass surface using a heterobifunc-
tional crosslinker.

5.3.4

Microcontact Printing

Microcontact printing (mCP) uses an elastomeric stamp to create patterns of organic mo-
lecules on surfaces, and was initially developed by Whitesides group to print patterns of
monolayers of alkanethiols onto gold substrates [44]. The method allows the patterning of
surfaces with biomolecules, and has been studied extensively in the context of biosensors
and high-throughput bioassays (see Chapter 3). Application of this technique to the inves-
tigation of cell–substrate interactions [45] has mainly focused on endothelial cell adhesion
and the control of neuronal process outgrowth for the creation of defined neuronal
networks [46–51], as shown in Figure 5.3.

The procedure starts with a photolithography step to produce the mold (master stamp).

The patterning of neuronal networks requires high-relief stamps which can be realized,
for example, by photoresists with a high-aspect ratio and a thickness of more than
5 mm. The elastomeric stamp is prepared by casting PDMS against the patterned photo-
resist. The PDMS stamp is inked with a solution of organic molecules, dried, and placed
in contact with a surface. The organic molecules are transferred only at those regions
where the stamp contacts the surface. The patterned surface can be in the range of several
cm

2

in size, and the features can have an edge resolution in the sub-mm range. When a

pattern of cell-attracting components is stamped onto a background material that repels
cell adhesion, the attachment and outgrowth of cells – for example neurons – is restricted
to the regions where transfer took place, confining the geometry of the forming networks.
The cells in such networks have been shown to be interconnected by chemical synapses
allowing signal transduction along the predefined pathways (Figure 5.4) [51a].

71

5.3 Methods

72

5 Defined Networks of Neuronal Cells in Vitro

Figure 5.3

Network of rat embryonic

cortical neurons grown on a pattern of
ECM proteins created by microcontact
printing onto polystyrene. (A) Phase-
contrast microscopy. (B) The connec-
tivity of the cells is visualized by mi-
croinjection of three different fluores-
cent dyes into three cells in the course
of a patch–clamp measurement. (Fig-
ures provided courtesy of A. K. Vogt.)

Figure 5.4

Signal transduction in a

geometrically confined network of rat
embryonic neurons. The cells were
cultivated on a pattern of ECM proteins
against a background of polystyrene.
Synaptic connectivity was observed
through patch–clamp measurements. A
signal evoked experimentally in cell 1
(C1) traveled through two synapses
via cell 2 to cell 3, indicating a simple
functional network. (Figures provided
courtesy of A. K. Vogt.)

5.4

Outlook

Neuro from the nerves, the silver paths. Romancer. Neuromancer. (...) “I met Neuro-
mancer. He talked about your mother. I think he’s something like a giant ROM
construct, for recording personality, only it’s full RAM. The constructs think
they’re there, like it’s real, but it just goes on forever.” (...)
Case chewed his lower lip and grazed out across the plateaus of the Eastern Seaboard
Fission Authority, into the infinite neuroelectronic void of the matrix.

William Gibson Neuromancer

Bioelectronic interfacing is a topic that inspires and fascinates not only science-fiction wri-
ters and movie makers. Indeed, the possibilities arising from an interweaving of neuronal
networks with microelectronics – a “marriage of biological systems with technology” – are
probably many more than we can envision to date. Cell-based hybrids as biosensors, neu-
ronal prostheses, neuroelectronic circuits and artificial intelligence are only the first issues
we are aiming at.

The cultivation of neurons on field effect transistors for extracellular stimulation and

signal recording – one of the requirements for such applications – is already possible,
as shown in Figure 5.5. Being able to precisely pattern neuronal networks is another
scientific advance that takes us a step towards these goals. In order to communicate reli-
ably with a silicon chip, cell adhesion must be confined precisely to defined areas on the
chip, as must be the pathways of connectivity [52–54]. However, control over polarity and
synapse formation is required for the controlled design of networks of neuronal cells. The
potential applications for such neuron–chip systems are ambitious – for example, artificial
photoreceptors which can be implanted into an irreversibly damaged eye and are able to
communicate directly with the optical nerve. Other ideas for neuronal protheses include
the bridging of damaged sections of the spinal cord by neurosilicon chips, or the targeting

73

5.4 Outlook

Figure 5.5

Brainstem neurons

cultured for 7 days on a laminin-coated
triple gate structure for extracellular
recordings. (Figure provided courtesy
of S. Ingebrandt.)

of neurite outgrowth after injury, specifically to re-innervate damaged tissue. Apart from
medical applications, neuronal networks on silicon are thought to be implementable into
neurocomputers to support data processing directly, without the need to unravel the prin-
ciples underlying neuronal signal transduction.

On the other hand, exactly this unraveling of neuronal information processing is per-

haps the most fascinating problem addressed by researchers working with patterned neu-
rons. As simplified systems with defined and manipulatable geometry, these networks
may teach us much about the impact of connectivity on the input and output of a network,
about the way neurons integrate incoming signals, and how different parameters influ-
ence the transmission, routing, and processing of such signals. The rules underlying
these actions, which lead to the selective amplification of some signals and concurrent
attenuation of others, may be regarded as the neuronal language or code. Deciphering
this code is of fundamental interest in diverse scientific disciplines, since it seems to
be the basis for many abilities of the brain, such as learning, associative recognition,
and memory.

Attempts to model the function of the human brain with computers have been made

already. The term “neuronal network” also stands for algorithms that are supposed to en-
able a computer to learn for example to recognize and associate patterns in the way a per-
son would when they are “fed” with the proper teaching material. The abilities of such a
system of course are limited to very particular types of tasks, and they are also rather sen-
sitive with respect to the correct training experiments [55]. Implementing new findings on
principles of neuronal signal integration and processing into such algorithms may help us
to refine them, ultimately creating computers with abilities that were so far restricted to
humans, such as associative memory or creativity.

References

74

5 Defined Networks of Neuronal Cells in Vitro

[1]

Nicholls, J. C., Martin, A. R., Wallace, B. G.,
From Neuron to Brain, 3rd edition, Sinauer
Associates, Sunderland, Mass., 1992.

[2]

Hucho, F., Weise, C., Angew. Chem. – Int.
Edit. 2001, 40, 3101–3116.

[3]

Kandel, E., Schwartz, J., Jessell, T., Princi-
ples of Neural Science; 3rd edition, Prentice-
Hall International, Inc., London, 1991.

[4]

Rosenberg, M. D., Science 1963, 139,
411–412.

[5]

Curtis, A., Wilkinson, C., Biomaterials
1997, 18, 1573–1583.

[6]

Curtis, A. S. G., Wilkinson, C. D., J. Bio-
mater. Sci.-Polymer Ed. 1998, 9, 1313–1329.

[7]

Jung, D. R., Kapur, R., Adams, T., Giuliano,
K. A., Mrksich, M., Craighead, H. G.,
Taylor, D. L., Crit. Rev. Biotechnol. 2001,
21, 111–154.

[8]

Clark, P., Connolly, P., Curtis, A. S. G.,
Dow, J. A. T., Wilkinson, C. D. W., Develop-
ment 1990, 108, 635–644.

[9]

Britland, S., Morgan, H., Wojiak-Stodart,
B., Riehle, M., Curtis, A., Wilkinson, C.,
Exp. Cell Res. 1996, 228, 313–325.

[10]

Carter, S. B., Nature 1965, 208, 1183–1187.

[11]

Letourneau, P. C., Dev. Biol. 1975, 44,
92–101.

[12]

Manthrope, M., Engvall, E., Ruoslahti, E.,
Longo, F. M., Davis, G. E., Varon, S., J. Cell
Biol. 1983, 97, 1882–1890.

[13]

Ruegg, U. T., Hefti, F., Neurosci. Lett. 1984,
49, 319–324.

[14]

Hammarback, J. A., Palm, S. L., Furcht,
L. T., Letourneau, P. C., J. Neurosci. Res.
1985, 13, 213–220.

[15]

Kleinfeld, D., Kahler, K. H., Hockberger,
P. E., J. Neurosci. 1988, 8, 4098–4120.

75

References

[16]

Kleinman, H. K., Ogle, R. C., Cannon,
F. B., Little, C. D., Sweeney, T. M.,
Proc. Natl. Acad. Sci. USA 1988, 85,
1282–1286.

[17]

Kleinman, H. K., Sephel, G. C., Tashiro,
K.-I., Weeks, B. S., Burrous, B. A., Adler,
S. H., Yamada, Y., Martin, G. R., Ann. N. Y.
Acad. Sci. 1990, 580, 302–310.

[18]

Lewandowska, K., Balachander, N., Suke-
nik, C. N., Culp, L. A., J. Cell. Physiol. 1989,
141, 334–345.

[19]

Ruoslahti, E., Pierschbacher, M. D., Cell
1986, 44, 517–518.

[20]

Ruoslahti, E., Pierschbacher, M. D., Science
1987, 238, 491–497.

[21]

Graf, J., Iwamoto, Y., Sasaki, M., Martin,
G. R., Kleinman, H. K., Robey, F. A.,
Yamada, Y., Cell 1987, 48, 989–996.

[22]

Curtis, A. S. G., Breckenridge, L., Connolly,
P., Dow, J. A. T., Wildinson, C. D. W.,
Wilson, R., Med. Biol. Eng. Comput. 1992,
30, CE33–CE36.

[23]

Fromherz, P., Schaden, H., Vetter, T.,
Neurosci. Lett. 1991, 129, 77–80.

[24]

Fromherz, P., Schaden, H., Eur. J. Neurosci.
1994, 6, 1500–1504.

[25]

Valentini, R. F., Vargo, T. G., Gardella, J. A.,
Aebischer, P., J. Biomater. Sci.-Polymer Ed.
1993, 5, 13–36.

[26]

Ranieri, J. P., Bellamkonda, R., Bekos, E. J.,
Gardella, J. A., Mathieu, H. J., Ruiz, L.,
Aebischer, P., Int. J. Dev. Neurosci. 1994, 12,
725–735.

[27]

Corey, J. M., Wheeler, B. C., Brewer, G. J.,
J. Neurosci. Res. 1991, 30, 300–307.

[28]

Bohanon, T., Elender, G., Knoll, W., Ko-
berle, P., Lee, J. S., Offenhäusser, A.,
Ringsdorf, H., Sackmann, E., Simon, J.,
Tovar, G., Winnik, F. M., J. Biomater. Sci.-
Polymer Ed. 1996, 8, 19–39.

[29]

Sze, S. M., Semiconductor devices: Physics
and Technology, John Wiley Sons, Inc.,
New York, 1985.

[30]

Craighead, H. G., Turner, S. W., Davis,
R. C., James, C., Perez, A. M., John,
P. M. S., Isaacson, M. S., Kam, L., Shain,
W., Turner, J. N., Banker, G., Biomed.
Microdev. 1998, 1, 49–64.

[31]

Turner, S., Kam, L., Isaacson, M.,
Craighead, H. G., Shain, W., Turner, J.,
J. Vac. Sci. Technol. B 1997, 15,
2848–2854.

[32]

Wilkinson, C. D. W., Curtis, A. S. G.,
Crossan, J., J. Vac. Sci. Technol. B 1998, 16,
3132–3136.

[33]

Maher, M. P., Dvorak-Carbone, H., Pine, J.,
Wright, J. A., Tai, Y. C., Med. Biol. Eng.
Comput. 1999, 37, 110–118.

[34]

Griscom, L., Degenaar, P., LePioufle, B.,
Tamiya, E., Fujita, H., Jpn. J. Appl. Phys.
2001, 40, 5485–5490.

[35]

Merz, M., Fromherz, P., Adv. Mater. 2002,
14, 141–144.

[36]

Clark, P., Britland, S., Connolly, P., J. Cell
Sci. 1993, 105, 203–212.

[37]

Tai, H. C., Buettner, H. M., Biotechnol. Prog.
1998, 14, 364–370.

[38]

Sorribas, H., Padeste, C., Tiefenauer, L.,
Biomaterials 2002, 23, 893–900.

[39]

Dulcey, C. S., Georger, J. H., Krauthamer,
V., Stenger, D. A., Fare, T. L., Calvert, J. M.,
Science 1991, 252, 551–554.

[40]

Ravenscroft, M. S., Bateman, K. E.,
Shaffer, K. M., Schessler, H. M., Jung,
D. R., Schneider, T. W., Montgomery, C. B.,
Custer, T. L., Schaffner, A. E., Liu, Q. Y., Li,
Y. X., Barker, J. L., Hickman, J. J., J. Am.
Chem. Soc. 1998, 120, 12169–12177.

[41]

Stenger, D. A., Pike, C. J., Hickman, J. J.,
Cotman, C. W., Brain Res. 1993, 630,
136–147.

[42]

Ma, W., Liu, Q. Y., Jung, D., Manos, P.,
Pancrazio, J. J., Schaffner, A. E., Barker,
J. L., Stenger, D. A., Dev. Brain Res. 1998,
111, 231–243.

[43]

Matsuzawa, M., Liesi, P., Knoll, W., J.
Neurosci. Methods 1996, 69, 189–196.

[44]

Kumar, A., Whitesides, G. M., Appl. Phys.
Lett. 1993, 63, 2002–2004.

[45]

Singhvi, R., Kumar, A., Lopez, G. P.,
Stephanopoulos, G. N., Wang, D. I. C.,
Whitesides, G. M., Ingber, D. E., Science
1994, 264, 696–698.

[46]

Branch, D. W., Corey, J. M., Weyhenmeyer,
J. A., Brewer, G. J., Wheeler, B. C., Med.
Biol. Eng. Comput. 1998, 36, 135–141.

[47]

Branch, D. W., Wheeler, B. C., Brewer, G. J.,
Leckband, D. E., IEEE Trans. Biomed. Eng.
2000, 47, 290–300.

[48]

James, C. D., Davis, R. C., Kam, L., Craig-
head, H. G., Isaacson, M., Turner, J. N.,
Shain, W., Langmuir 1998, 14, 741–744.

[49]

Scholl, M., Sprossler, C., Denyer, M.,
Krause, M., Nakajima, K., Maelicke, A.,
Knoll, W., Offenhausser, A., J. Neurosci.
Methods 2000, 104, 65–75.

76

5 Defined Networks of Neuronal Cells in Vitro

[50]

Yeung, C. K., Lauer, L., Offenhausser, A.,
Knoll, W., Neurosci. Lett. 2001, 301,
147–150.

[51]

Kam, L., Shain, W., Turner, J. N., Bizios, R.,
Biomaterials 2001, 22, 1049–1054.

[51a]

Vogt, A. K., Lauer, L., Knoll, W. Often-
häusser, A., Biotechnol. Prog. 2003, 19,
1562–1568.

[52]

Zeck, G., Fromherz, P., Proc. Natl. Acad.
Sci. USA 2001, 98, 10457–10462.

[53]

Chang, J. C., Brewer, G. J., Wheeler, B. C.,
Biosens. Bioelectron. 2001, 16, 527–533.

[54]

Lauer, L., Ingebrandt, S., Scholl, M.,
Offenhausser, A., IEEE Trans. Biomed.
Eng. 2001, 48, 838–842.

[55]

Gasteiger, J., Zupan, J., Angew. Chem. –
Int. Edit. 1993, 32, 503–527.