BioMEMS for Cell Biology - Introduction to BioMEMS - page 345

concentrations/gradients used have to be guessed. he irst studies used rudimentary gradient
generators such as pipettes or the Boyden/Zigmond/Dunn chambers, which result in very low
throughputs and poor reproducibility.
he irst axon guidance study of neurons in a microfabricated gradient was performed by
a collaborative team led by Venkatesh Murthy and George Whitesides at Harvard University
( Figure 6.50 ). To simplify the experiment, the cells (rat hippocampal neurons) were exposed
to an immobilized laminin gradient, which had been produced before seeding the cells. It is
not clear whether such laminin gradients are present in vivo (in the hippocampus or elsewhere
in the brain). Nevertheless, the work does raise an intriguing question: what is the appropriate
mode of presentation of an axon guidance molecule (soluble or immobilized) in a cell culture
experiment? In vivo, most axon guidance factors have motifs that make them bind to the ECM,
at least partially, so they are not strictly “freely difusing.” It may be that a freely difusing mol-
ecule is too mobile for the receptor to bind to it and transduce the signal to the intracellular
signaling machinery. On the other hand, if the axon guidance factor is too bound by the surface
(or the ECM), it may not signal properly to the cell.
Does the growth cone need to receive continuous feedback from the gradient to navigate, or
can the signal be discontinuous as long as it gets integrated more oten than the reaction time
of the growth cone? In 2006, a German team led by Friedrich Bonhoefer (Max Planck Institute
in Tübingen) and Martin Bastmeyer (University of Karlsruhe, Germany) proposed to substitute
the continuous gradients with discrete gradients, which are easier to microstamp as dots of
varying spacing/dimensions ( Figure 6.51 ). To form the densest part of the gradient, stripes as
narrow as 0.3 μm separated by 0.3 μm were stamped. Growth cones of chick temporal retinal
axons (which display a “stop reaction” in vivo at a particular location of ephrin gradients), but
not nasal axons, were able to integrate discontinuous ephrin gradients and stop at a distinct zone
in the gradient while still undergoing ilopodial activity.
However, as we have seen, microstamping can be impractical (proteins must be dried and
the stamp needs to be fabricated at high resolution), so Bonhoefer and Bastmeyer repeated the
experiment with a microluidic gradient generator capable of patterning a “ladder” of proteins
from solution ( Figure 6.52 ). he biological indings were identical: temporal axons (but not
nasal ones) were guided by discrete, immobilized ephrin gradients, and they were also stopped
by certain concentration-dependent gradient values. his suggests a general validity of the

a b c

Laminin

BSA Laminin

BSA

Inlets

Serpentine
channel

PDMS (oxidized)

Poly- L -lysine

Gradient
mixer

d

50

2nd longest neurites

40

Longest neurites

30

Gradient
Outlet

20

10

0

0 90 180

0 50 100 150 200

Position [µm] Length of neurite [µm]

Channel wall

FIGURE 6.50 Axon.growth.dictated.by.absorbed.laminin.gradients..(From.Stephan.K..W..Dertinger,.
Xingyu.Jiang,.Zhiying.Li,.Venkatesh.N..Murthy,.and.George.M..Whitesides,.“Gradients.of.substrate-
bound.laminin.orient.axonal.speciication.of.neurons,”. Proc. Natl. Acad. Sci. U. S. A. .99,.12542-
12547,. 2002.. Copyright. (2002). National. Academy. of. Sciences,. U.. S.. A.. Figure. contributed. by.
George.Whitesides.)