Aizenberg Calcitic Microlenses Brittlestars Nature 2001
Aizenberg Calcitic Microlenses Brittlestars Nature 2001
Aizenberg Calcitic Microlenses Brittlestars Nature 2001
m
)
a
(
m
)
1
2
3
4
5
a
0
5
10
15
0 5 10
d
1 2 3 4 5
0
h
2
= 60
m
h
1
= 48
m
x
a
1
a
2
d
b
Figure 3 Analysis of the focusing effect of the lens layer. a, Superposition of the
micrographs shown in Fig. 2ce. Photoresist is selectively exposed under the lens areas
outlined by dashed lines. b, Schematic representation of the focusing action of lenses
located along the dotted line in a. Lenses with a focal plane above h
1
produce narrow
features in the rst lithographic experiment and no features in the second (lens 1). Lenses
with a focal point below h
2
produce narrow features in the second experiment and no
features in the rst (lens 5). When the focal plane of a lens is located signicantly above h
1
,
no pattern in either photoresist layer is formed (lens 3). Patterns in both photoresist layers
are recorded under lenses with a focal point between h
1
and h
2
(lenses 2 and 4).
c, Distances from the focal points of lenses to the rst imaging plane (x) as a function of
the lens diameters (L). d, Sizes of the spots in photoresist (a) as a function of x.
2001 Macmillan Magazines Ltd
letters to nature
NATURE | VOL 412 | 23 AUGUST 2001 | www.nature.com 821
lens to the rst imaging plane:
x a
1
h
2
2 h
1
=a
1
a
2
2
where a
1
and a
2
are the measured diameters of the spots recorded at
the distances h
1
and h
2
from the top surface of the lens array,
respectively. Figure 3c shows that the distances x depend linearly
on the lens diameters L:
x 1:14L 2 44:8 3
where r
2
is 0.93.
Equations (1) and (3) allowed us to calculate the experimental
position of the focal plane for the lenses operating in PDMS:
d h
1
x 2 t <0:25L <6:512:5 mm 4
Using basic considerations of the geometric optics for a thick lens
18
that has the characteristic shape of the microlenses, we determined
the relationship between d and the effective focal length f:
f <d1 2 1 2 n
1
=n
c
t=R
u
21
<1:5d 5
where R
u
is the curvature radius of the upper surface of the lens, and
n
1
(1.46) and n
c
(1.66) are the refractive indices of PDMS and calcite
along the c-axis, respectively. The position of the focal point in the
actual biological environment, (d
b
), can be determined using the
thick-lens formula
18
for two media with different refractive indices:
d
b
<dn
c
=n
1
2 1=n
c
=n
2
2 1 <0:14L <47 mm 6
where n
2
(1.34) is the refractive index of body uids
11
.
Figure 3d presents the sizes of the spots in the photoresist (a) as
a function of x for the entire lens array. Using a linear approxi-
mation of the data, ax 1:26x 2:9 (r
2
0:90), one can
estimate the size of the spot at the focal plane, a
0
ax 0
<2:9 mm, and the operational diameter of a typical lens,
L
0
ax f <20 mm. The intensity of the incoming light is thus
enhanced at the focal point by a factor of E L
0
=a
0
2
<50.
For a thick lens formed by two spherical surfaces, a
0
is related to
L
0
and f by the equation
18
a
0
<L
2
0
=f . For such a lens, the experi-
mental values of a
0
and f would then correspond to the maximum
operational diameter L
0
<56 mm and the light enhancement
factor E <34. The experimentally determined values of L
0
and E
(see above equations) are signicantly higher, clearly indicating that
there must exist a marked compensation for spherical
aberration
11,12,18
. To verify this conclusion, we calculated the optimal
prole for the compensated calcitic lens in a biological medium(red
line in Fig. 1g) and found it to be in a good agreement with the
actual shape of the operational part of a lens. The proportionality of
t and d to L (equations (1), (4) and (6)) suggests that all lenses are
rescaled replicas of each other, implying that the entire array is
compensated for aberration.
If the calcitic microlenses are involved in photoreception as light
guides and concentrators, one should expect the presence of
receptors positioned at their focal points. Indeed, a transmission
electron microscopy study
6
of thin sections of decalcied DAPs
revealed bundles of nerve bre located at the predicted distance d
b
beneath the lens layer (Fig. 4). The diameters of the neural bundles
(b <24 mm) correspond well to the estimated size of the focused
spot a
0
. Furthermore, what is seen as a diurnal colour change
of O. wendtii (Fig. 1b) is, in fact, a ltering and diaphragm action
of chromatophores. These chromatophores regulate the intensity
of light reaching the lenses by extending their pigment-lled
processes to cover the lens during the day and retracting them
to a lateral position between the lenses during the night
6
(Fig. 4).
Although the characteristics of echinoderm dermal photoreceptors
and their precise locations are poorly understood and
controversial
3,6,19,20
, the likelihood of sub-lens photoreceptors in
O. wendtii also corresponds with neurophysiological
21,22
and
biochemical
20
data.
On the basis of our results, we suggest that the array of calcitic
microlenses with their unique focusing effect and underlying neural
receptors may form a specialized photoreceptor system with a
conceivable compound-eye capability. For a compound eye to
operate, each unit (lens plus photoreceptor) must respond only to
light coming from a single direction
23
. The angular selectivity
is determined by three principal parameters: rst, the angular
resolution of the lens, f a
0
=f . Our experiments yield a f-value
of ,108. The second parameter is the ratio of the detector diameter
to the focal length, m. Although we do not know the effective size of
the receptors in O. wendtii, we can estimate the maximum value of
m that is limited by the diameter of the sub-lensar nerve bundles:
m
max
b=f <f. The last parameter that determines the angular
selectivity is the alignment of the detector with respect to the
optical axis. From our observation that the lenses are compensated
for aberration, it follows that when light deviates from the
optimal direction by w, the spot size increases linearly with
jwj : aw 2 a
0
<jwjL
0
. This implies that the focal intensity
decays as jwj w*
22
, that is, it is sharply peaked in the vicinity
of the optimal incident angle, with the characteristic width of the
peak (w* <a
0
=L
0
) being of the order of the angular resolution f.
This effect suggests a plausible mechanism for targeting specic
receptors aligned with the lens axis, on the basis of signicantly
stronger receptor response to the incoming focused light.
The selected nerve bundle will then efciently detect signals that
come only from one direction with the angular selectivity of ,108,
making the calcitic microlenses suitable for a compound-eye unit. It
is doubtful, however, that one DAP could function as a single eye
even though it has a slightly curved shape, the individual lenses are
all oriented along the crystallographic c-axis, minimizing the effect
of birefringence. This suggests that the entire array is a highly
redundant optical element free of aberration that detects the light
from a particular direction. As a brittlestar has a large number of
DAPs and additional arrays of lenses on the lateral surfaces, all of
which are oriented differently and are capable of changing position
as the armmoves, it could potentially extract a considerable amount
of visual information about its environment. Althoughwe have only
limited evidence that the lens apparatus operates at a distance
8
,
which would conform to the denition of an eye, our results suggest
that the presence of such structures may be sufcient to elicit the
rapid, coordinated behaviours, such as detection of predators and
retreat toward crevices, that imply the occurrence of vision. These
are the very abilities that are central to the survival of individuals of
O. wendtii in their natural habitat
7
.
The demonstrated use of calcite by brittlestars, both as an optical
element and as a mechanical support, illustrates the remarkable
ability of organisms, through the process of evolution, to optimize
one material for several functions, and provides new ideas for the
fabrication of `smart' materials
24,25
. M
L
N
PP P
1 m
Figure 4 Transmission electron micrograph of the decalcied section of the DAP of
O. wendtii. Sample preparation is described in ref. 6. L, lens area; N, nerve bundle; P,
pigment.
2001 Macmillan Magazines Ltd
letters to nature
822 NATURE | VOL 412 | 23 AUGUST 2001 | www.nature.com
Received 29 March; accepted 25 June 2001.
1. Hyman, L. H. The Invertebrates Vol. 4 (McGraw-Hill, New York, 1955).
2. Millott, N. The photosensitivity of echinoids. Adv. Mar. Biol. 13, 152 (1975).
3. Yoshida, M., Takasu, N. & Tamotsu, S. in Photoreception and Vision in Invertebrates; (ed. Ali, M. A.)
743771 (Plenum, New York, 1984).
4. Lowenstam, H. A. & Weiner, S. On Biomineralization 123134 (Oxford Univ. Press, Oxford, 1989).
5. Wainwright, S. A., Biggs, W. D., Currey, J. D. & Gosline, J. M. Mechanical Design in Organisms (John
Wiley, New York, 1976).
6. Hendler, G. & Byrne, M. Fine structure of the dorsal arm plate of Ophiocoma wendti (Echinodermata,
Ophiuroidea). Zoomorphology 107, 261272 (1987).
7. Hendler, G. Brittlestar color-change and phototaxis (Echinodermata: Ophiuroidea: Ophiocomidae).
Mar. Ecol. 5, 379401 (1984).
8. Cowles, R. P. Stimuli produced by light and by contact with solid walls as factors in the behavior of
ophiuroids. J. Exp. Zool. 9, 387416 (1910).
9. Donnay, G. & Pawson, D. L. X-ray diffraction studies of echinoderm plates. Science 166, 11471150
(1969).
10. Ameye, L., Hermann, R., Wilt, F. & Dubois, P. Ultrastructural localization of proteins involved in sea
urchin biomineralization. J. Histochem. Cytochem. 47, 11891200 (1999).
11. Clarkson, E. N. K. &Levi-Setti, R. Trilobite eyes and the optics of Des Cartes and Huygens. Nature 254,
663667 (1975).
12. Gal, J., Horvath, G., Clarkson, E. N. K. & Haiman, O. Image formation by bifocal lenses in a trilobite
eye? Vision Res. 40, 843853 (2000).
13. Towe, K. M. Trilobite eyes: calcied lenses in vivo. Science 179, 10071010 (1973).
14. Doderlein, L. Ueber `Krystallkorper' bei Seesternen. Denkschr. Med. Nat. Ges. Jena 8, 491494 (1898).
15. Smith, A. B. in Special Papers in Palaeontology 25, 181 (Palaeontology Association, London, 1980).
16. Xia, Y. N. & Whitesides, G. M. Soft lithography. Annu. Rev. Mater. Sci. 28, 153184 (1998).
17. Aizenberg, J., Rogers, J. A., Paul, K. E. & Whitesides, G. M. Imaging the irradiance distribution in the
optical near eld. Appl. Phys. Lett. 71, 37733775 (1997).
18. Flint, H. T. Geometrical Optics (Methuen, London, 1936).
19. Cobb, J. L. S. & Moore, A. Comparative studies on receptor structure in the brittlestar Ophiura
ophiura. J. Neurocytol. 15, 97108 (1986).
20. Johnsen, S. Identication and localization of a possible rhodopsin in the echinoderms Asterias forbesi
(Asteroidea) and Ophioderma brevispinum (Ophiuroidea). Biol. Bull. 193, 97105 (1997).
21. Cobb, J. L. S. & Hendler, G. Neurophysiological characterization of the photoreceptor system in
brittlestars. Comp. Biochem. Physiol. 97A, 329333 (1990).
22. Stubbs, T. R. in Echinoderms (ed. Lawrence, J. M.) 403408 (Proc. Int. Echinoderms Conf., Tampa
Bay, 1982).
23. Land, M. F. in Comparative Physiology and Evolution in Invertebrates B: Invertebrate Visual Centers and
Behavior I (ed. Autrum, H.) 471592 (Springer, Berlin, 1981).
24. Mann, S. & Ozin, G. A. Synthesis of inorganic materials with complex form. Nature 382, 313318
(1996).
25. Belcher, A. M., Hansma, P. K., Stucky, G. D. & Morse, D. E. First steps in harnessing the potential of
biomineralization. Acta Mater. 46, 733736 (1998).
Acknowledgements
We thank P. Wiltzius and M. Megens for helpful discussions.
Correspondence and requests for materials should be addressed to J.A.
(e-mail: jaizenberg@lucent.com).
.................................................................
Delineation of prognostic biomarkers
in prostate cancer
Saravana M. Dhanasekaran*, Terrence R. Barrette*, Debashis Ghosh,
Rajal Shah*, Sooryanarayana Varambally*, Kotoku Kurachi,
Kenneth J. Pientak, Mark A. Rubin*# & Arul M. Chinnaiyan*#
Departments of * Pathology, Biostatistics, Human Genetics, Urology,
k Internal Medicine, and Comprehensive Cancer Center, University of Michigan
Medical School, Ann Arbor, Michigan 48109, USA
# These authors share senior authorship
..............................................................................................................................................
Prostate cancer is the most frequently diagnosed cancer in Ameri-
can men
1,2
. Screening for prostate-specic antigen (PSA) has led
to earlier detection of prostate cancer
3
, but elevated serum PSA
levels may be present in non-malignant conditions such as benign
prostatic hyperlasia (BPH). Characterization of gene-expression
proles that molecularly distinguish prostatic neoplasms may
identify genes involved in prostate carcinogenesis, elucidate
clinical biomarkers, and lead to an improved classication of
prostate cancer
46
. Using microarrays of complementary DNA, we
examined gene-expression proles of more than 50 normal and
neoplastic prostate specimens and three common prostate-cancer
cell lines. Signature expression proles of normal adjacent pros-
tate (NAP), BPH, localized prostate cancer, and metastatic, hor-
mone-refractory prostate cancer were determined. Here we
establish many associations between genes and prostate cancer.
We assessed two of these geneshepsin, a transmembrane serine
protease, and pim-1, a serine/threonine kinaseat the protein
level using tissue microarrays consisting of over 700 clinically
stratied prostate-cancer specimens. Expression of hepsin and
pim-1 proteins was signicantly correlated with measures of
clinical outcome. Thus, the integration of cDNA microarray,
high-density tissue microarray, and linked clinical and pathology
data is a powerful approach to molecular proling of human
cancer.
We developed a 9,984-element (10K) human cDNA microarray
to analyse gene expression proles in benign and malignant prostate
tissue. As with previous cancer proling studies
710
, molecular
classication of prostate cancer was one of the goals of this analysis.
We used two distinct reference samples for comparative microarray
analysis: NAP tissue from patients with prostate cancer, and
prostate tissue from men without documented prostate pathology.
By making direct comparisons against normal tissue counterparts,
we took advantage of a `subtractive' effect, which emphasized genes
that consistently distinguished normal and neoplastic tissues.
Prostate tissues used in microarray analysis included 4 BPH
samples, 8 NAP samples, 1 commercial pool of normal prostate
tissue (from 19 individuals), 1 prostatitis sample, and 11 localized
and 7 metastatic prostate-cancer samples. Three cell lines from
metastatic prostate cancer (DU-145, LnCAP and PC3) were also
proled for gene expression. Twenty-eight additional prostate tissue
specimens were proled and the data included in the Supplemen-
tary Information (samples of 9 BPH, 1 NAP, 13 metastatic and 5
localized prostate cancers). Fluorescently labelled (Cy5) cDNA was
prepared from total RNA from each experimental sample. A second
distinguishable uorescent dye (Cy3) was used to label the two
reference samples used in this study: a pool of NAP from four
independent patients with prostate cancer and a commercial pool of
normal prostate tissues. A direct comparison between the NAP and
commercial pools was also made and notable differences in gene
expression were readily apparent (Fig. 1 and Supplementary
Information).
In all, more than 80 cDNA microarrays were used to assess gene
expression in four clinical states of prostate-derived tissues and two
distinct reference pools of normal specimens. Figure 1 provides an
overview of the variation in gene expression across the different
tissue specimens analysed (the full data set and 28 further samples
can be seen in the Supplementary Information). A hierarchical
clustering algorithm was used to group genes and experimental
samples on the basis of similarities of gene expression over all that
were tested. Relationships between the experimental samples are
summarized as dendrograms (Fig. 1a), in which the pattern and
length of the branches reect the relatedness of the samples. Benign
conditions of the prostate, such as BPH and NAP, cluster separately
frommalignant prostate-cancer cell lines or tissues, regardless of the
reference pool used. Within the prostate-cancer cluster, metastatic
and clinically localized prostate cancer formed distinct subgroups.
Eisen matrix formats
11
of the variation in gene expression show
clusters of coordinately expressed genes, highlighting relationships
between specimens (black bars in Fig. 1b, c). For example, clusters
B3 and C1 represent genes downregulated in both localized and
metastatic prostate cancer (Fig. 1b, c). By contrast, clusters B6 and
B4 highlight genes that are specically up- or downregulated in
metastatic prostate cancer, respectively (Fig. 1b). IGFBP-5, DAN1,
FAT tumour suppressor and RAB5A are examples of genes that are
downregulated specically in metastatic prostate cancer and also
have a proposed role in oncogenesis (magnied regions, Fig. 1b).
2001 Macmillan Magazines Ltd