New photoactivators for multiphoton excited three-dimensional submicron cross-linking of proteins: Bovine serum albumin and type 1 collagen

Article abstract of DOI:10.1562/0031-8655(2002)076<0135:NPFMET>2.0.CO;2

We report the synthesis and optical characterization of two new photoactivators and demonstrate their use for multiphoton excited three-dimensional free-form fabrication with proteins. These reagents were developed with the goal of cross-linking Type 1 collagen. This cross-linking process produces structures on the micron and submicron size scales. A rose bengal diisopropyl amine derivative combines the classic photoactivator and co-initiator system into one molecule, reducing the reaction kinetics and increasing cross-linking efficiency. This derivative was successful at producing stable structures from collagen, whereas rose bengal alone was not effective. A benzophenone dimer connected by a flexible diamine tether was also synthesized. This activator has two photochemically reactive groups and is highly efficient in cross-linking bovine serum albumin and Type 1 collagen to form stable, robust structures. This approach is more flexible in terms of cross-linking a variety of proteins than by traditional benzophenone photochemistry. The photophysical properties vary greatly from that of benzophenone, with the appearance of a new, lower energy absorption band (λ_max～370 nm in water) and broad, visible emission band (～500 nm maximum). This absorption band is highly solvatochromic, suggesting it arises, at least in part, from a charge transfer interaction. Collagens are typically difficult to cross-link photochemically, and the results here suggest that these two new activators will be suitable for cross-linking other forms of collagen and additional proteins for biomedical applications such as the de novo assembly of biomimetic tissue scaffolds.

Full text of DOI:10.1562/0031-8655(2002)076<0135:NPFMET>2.0.CO;2

New Photoactivators for Multiphoton Excited Three-dimensional Submicron
Cross-linking of Proteins: Bovine Serum Albumin and Type 1 Collagen
Author(s): Jonathan D. Pitts, Amy R. Howell, Rosa Taboada, Ipsita Banerjee, Jun Wang, Steven L.
Goodman, and Paul J. Campagnola
Source: Photochemistry and Photobiology, 76(2):135-144.
Published By: American Society for Photobiology
https://doi.org/10.1562/0031-8655(2002)076<0135:NPFMET>2.0.CO;2
URL: http://www.bioone.org/doi/full/10.1562/0031-8655%282002%29076%3C0135%3ANPFMET
%3E2.0.CO%3B2
BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and
environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published
by nonprofit societies, associations, museums, institutions, and presses.
Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of
BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.
Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries
or rights and permissions requests should be directed to the individual publisher as copyright holder.
BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research
libraries, and research funders in the common goal of maximizing access to critical research.

Photochemistry and Photobiology, 2002, 76(2): 135–144
New Photoactivators for Multiphoton Excited Three-dimensional
Submicron Cross-linking of Proteins: Bovine Serum Albumin and Type 1
Collagen¶†
Jonathan D. Pitts‡1, Amy R. Howell2, Rosa Taboada2, Ipsita Banerjee§2, Jun Wang#2,
Steven L. Goodman**1,3 and Paul J. Campagnola*3,4
3
4
1Center for Biomaterials, Department of Physiology and Center for Biomedical Imaging Technology, University of
Connecticut Health Center, Farmington, CT and
2University of Connecticut, Department of Chemistry, Storrs, CT
Received 13 December 2001; accepted 28 April 2002
ABSTRACT
zophenone, with the appearance of a new, lower energy
absorption band ( 370 nm in water) and broad, vis-
500 nm maximum). This absorption
band is highly solvatochromic, suggesting it arises, at
least in part, from a charge transfer interaction. Colla-
gens are typically difficult to cross-link photochemically,
and the results here suggest that these two new activators
will be suitable for cross-linking other forms of collagen
and additional proteins for biomedical applications such
as the de novo assembly of biomimetic tissue scaffolds.
␭
ϳ
max
We report the synthesis and optical characterization of
two new photoactivators and demonstrate their use for
multiphoton excited three-dimensional free-form fabri-
cation with proteins. These reagents were developed with
the goal of cross-linking Type 1 collagen. This cross-link-
ing process produces structures on the micron and sub-
micron size scales. A rose bengal diisopropyl amine de-
rivative combines the classic photoactivator and co-ini-
tiator system into one molecule, reducing the reaction
kinetics and increasing cross-linking efficiency. This de-
rivative was successful at producing stable structures
from collagen, whereas rose bengal alone was not effec-
tive. A benzophenone dimer connected by a flexible di-
amine tether was also synthesized. This activator has two
photochemically reactive groups and is highly efficient in
cross-linking bovine serum albumin and Type 1 collagen
to form stable, robust structures. This approach is more
flexible in terms of cross-linking a variety of proteins
than by traditional benzophenone photochemistry. The
photophysical properties vary greatly from that of ben-
ible emission band (
ϳ
INTRODUCTION
In the 1990s there has been an explosion in the use of non-
linear optical excitation in a variety of fields, including bio-
medical imaging and material science. It has been shown
that two- and three-photon excitation coupled with a laser
scanning microscope produces fluorescence or other optical
imaging signals with intrinsic three-dimensionality (1). This
methodology has in fact revolutionized biological confocal
microscopy. Multiphoton absorption occurs via a virtual
nϪ1
state(s), where the absorption probability is given by pn/
where p is the laser pulse energy, n is the number of photons
to be absorbed and is the laser pulse width. The power can
␶
,
¶Posted on the web site on 17 May 2002.
†This paper is dedicated to the memory of Dr. Gary Epling, who
passed away in September 2001.
‡Current address: MediSpectra, Lexington, MA, USA.
§Current address:Department of Chemical Engineering, University
of Notre Dame, Notre Dame, IN, USA.
#Current address: GlaxoSmithKline Pharmaceuticals, King of Prus-
sia, PA, USA.
␶
be adjusted in a microscope such that excitation occurs ex-
clusively at the point of focus, i.e. the only location where
sufficiently high peak power exists. It is this dependence on
the peak power that in practice necessitates the use of ultra-
fast lasers for multiphoton excitation. The use of femtosec-
ond lasers has the added advantage of keeping the average
laser power low, while maintaining high peak power, there-
by avoiding photodamage to sensitive biological specimens.
The same multiphoton excitation axial confinement now
commonly used for fluorescence imaging can also be utilized
for optical fabrication. It is possible to construct or modify
three-dimensional structures in essentially any shape on size
scales ranging from the submicron optical diffraction limit
through the millimeter range. Some of the current fabrication
efforts using multiphoton absorption have taken the form of
photobleaching within a preformed matrix (2) as well as
modifying a photoresist by ablation (3). Perhaps the most
potentially powerful embodiment is in photopolymerization
**Current address: Imago Scientific Instruments Corporation, Mad-
ison, WI, USA.
*To whom correspondence should be addressed at: University of
Connecticut Health Center, Center for Biomedical Imaging Tech-
nology, MC-1507, 263 Farmington Avenue, Farmington, CT
06070, USA. Fax: 860-679-1039; e-mail: campagno@neuron.
uchc.edu
Abbreviations: BPD, benzophenone dimer; BSA, bovine serum al-
bumin; C-T, charge transfer; DPPF, 1,1-bis(diphenyphosphino)
ferrocene; EI, electrospray ionization; ISC, intersystem crossing;
MPE, multiphoton excitation; MS, mass spectra; NA, numerical
aperture; NMR, nuclear magnetic resonance; PBS, phosphate-
buffered saline; THF, tetrahydrofuran; TLC, thin layer chroma-
tography; TPEF, two-photon excited fluorescence.
᭧
2002 American Society for Photobiology 0031-8655/02
$5.00ϩ0.00
135

136 Jonathan D. Pitts et al.
of de novo structures. Several groups have employed this
approach with the goal of producing three-dimensional op-
tical storage devices, waveguides and optical limiters (4–8).
These structures have typically been produced from ure-
thanes and acrylates using well-known photochemistries for
free radical polymerization but utilizing multiphoton exci-
tation. Concurrent efforts have focused on developing new
photoactivators that have large two- and three-photon ab-
sorption cross sections with the goal of accelerating the pro-
cess, as well as reducing the required photon flux to avoid
photodamage in fragile samples (9–13). Other work has tried
to parallelize the process through simultaneous multifocal,
multiphoton excitation (14). Our work to date in this area
has focused on biological applications of this technology,
where we have utilized this fabrication process to cross-link
proteins and biocompatible polymers (e.g. polyacrylamide),
and protein–synthetic hybrids in aqueous environments
(15,16). We have examined the relative efficiencies of sev-
eral different photoactivators (xanthenes) and co-initiators
(triethanol amine, triethyl amine) in cross-linking a wide
range of proteins (15). We also measured sustained release
rates of entrapped bioactive molecules in hydrogel and pro-
tein matrices created by this process and demonstrated that
bioactivity of entrapped enzymes was maintained (16).
In this paper we report our efforts in developing and char-
acterizing new, more efficient photoactivators for cross-link-
ing minimally soluble proteins by multiphoton excitation. A
specific motivation for this work is the ability to photochem-
ically cross-link Type 1 collagen because collagens, those of
Type 1 in particular, are difficult to optically cross-link. We
have been unsuccessful in this task using several classes of
photoactivators including previously utilized xanthenes
(15,16), triarylmethanes and quiolines, and we have not been
able to locate any published work that was not invasive.
Collagen is an important protein in biomedical applications
in that it is the primary building block for many tissues in-
cluding bone, skin and muscle. The collagens are also the
primary proteins in forming topological complex extracel-
lular matrices. Historically collagen has most commonly
been chemically cross-linked by gluteraldehyde; however,
this approach has biocompatibility problems and further has
limited spatial control relative to that achievable by photo-
chemistry. Therefore, having the ability to photochemically
assemble collagens into any three-dimensional architecture
at the submicron scale would be of considerable utility in
the fabrication of synthetic biomimetic tissue, engineering
scaffolds and other devices.
Figure 1. Structures of the two new photoactivators: (a) rose bengal
diisopropyl amine conjugate; (b) modified BPD connected by a flex-
ible diamine tether.
The first molecule, shown in Fig. 1a, is a conjugate of the
photoactivator rose bengal and co-initiator diisopropyl
amine. Rose bengal and its derivatives have been widely
utilized in visible light excitation schemes (18,19). Our de-
rivative combines the classic photoactivator and co-initiator
pair of molecules into a single compound, reducing the re-
action kinetics from three-body to two-body. This type of
photoactivator typically initiates the photochemistry and is
either regenerated or degraded in the process. The second
molecule in Fig. 1b is a benzophenone dimer (BPD) linked
by a flexible diamine tether, where the potassium salt form
provides significantly increased aqueous solubility. This
photoactivator differs in that it binds directly to proteins and
thus becomes an integral part of the cross-linked protein ma-
trix. Modified proteins chemically conjugated to benzophe-
none derivatives have been used to cross-link several pro-
teins including actin, myosin and troponin (20–22). This new
approach is fundamentally different, simpler and more flex-
ible in that the BPD has two photoreactive ends that essen-
tially simultaneously cross-link proteins without relying on
an initial chemical conjugation and purification step. To the
best of our knowledge, there are no previous reports of this
type of bifunctional photoactivator. This molecule also dis-
plays photophysical properties that vary greatly from ben-
zophenone, possessing both lower energy absorption and
emission bands. These properties were investigated and
were, in part, found to resemble those of Michler’s ketone.
We will show that both the rose bengal derivative and BPD
are successful in cross-linking Type 1 collagen.
A second motivation for this work lies in the benefit of
using minimal photon flux in fragile biological systems. For
example, accidental three-photon absorption of endogenous
nucleic acids and proteins in live tissue can be quite dam-
aging (17). Increased photochemical efficiency can be
achieved by either creating new activators with large mul-
tiphoton excited cross sections, or alternatively, by synthe-
sizing new activators with better photoreactive quantum
yields. Here we take the latter tactic of improving the re-
active step. We describe the synthesis and characterize the
photophysics of two new photoactivators and measure their
relative efficiencies in cross-linking Type 1 collagen using
bovine serum albumin (BSA) as a benchmark for compari-
son.
MATERIALS AND METHODS
Reagents. Rose bengal (Sigma, St Louis, MO) and BSA (Calbi-
ochem, San Diego, CA) were used without further purification. Type
1 collagen (Becton Dickinson, Franklin Lakes, NJ) is acid extracted
from rat-tail tendon. Toluene was distilled from sodium. Petroleum
ether was purchased from JT Baker (Phillipsburg, NJ) and distilled
from CaCl2. With the exceptions noted, all starting reagents were
purchased from Aldrich (St. Louis, MO) and used without further
purification.
Analysis. Nuclear magnetic resonance (NMR) spectra were ob-
1
tained on a Bruker Avance DRX-400 (400 MHz H, 100 MHz 13C)
1
NMR spectrometer. The H and 13C chemical shifts (
␦
) are reported
in units of parts per million (ppm) from tetramethyl silane. Infrared
spectra were recorded on a JASCO FT/IR-410 spectrometer and are

Photochemistry and Photobiology, 2002, 76(2) 137
mixture was extracted with Et2O (3 50 mL), and the combined
ϫ
organic extracts were dried (MgSO4) and concentrated in vacuo.
Purification by flash chromatography on silica gel (petroleum ether–
ethyl acetate [EtOAc] 90:10) afforded (4-bromophenyl)(4-methyl-
phenyl)methanol (1.53 g, 62%) as a white solid: mp 80–81
(film): 3280 (broad), 3030, 2900, 1470, 1350, 1000, 900 cmϪ1; H
NMR (400 MHz, CDCl3): 7.44 (m, 2H), 7.23 (m, 4H), 7.13 (m,
2H), 5.76 (s, 1H), 2.33 (s, 3H), 2.13 (s, 1H); 13C NMR (100 MHz,
CDCl3): 142.9, 140.5, 137.7, 131.5, 129.3, 128.1, 126.5, 121.3,
75.5, 21.1; MS (EI); m/z: 278 (Mϩ 2), 276 (Mϩ), 263, 261, 185,
ЊC; IR
1
␦
␦
ϩ
183, 119 (100), 91, 77, 65; Anal. calcd for C14H13OBr: C, 61.09; H,
4.40. Found: C, 60.92; H, 4.74.
(4-Bromophenyl)(4-methylphenyl)methanone (4). A solution of
(4-bromophenyl)(4-methylphenyl)methanol (3) (1.04 g, 3.8 mmol),
KMnO4 (1.56 g, 9.9 mmol) and water (18 mL) was refluxed for 5
h. The reaction mixture was cooled and then acidified to pH 2 with
concentrated HCl. The mixture was filtered and the filtrate extracted
with Et2O (3
ϫ 50 mL). The combined organic extracts were dried
(MgSO4), filtered and concentrated. Purification by flash chromatog-
raphy on silica gel (petroleum ether–EtOAc 90:10) afforded (4-
bromophenyl)(4-methylphenyl)methanone (0.68 g, 67%) as a white
1
solid: mp 138–139
1H NMR (400 MHz, CDCl3):
2.44 (s, 3H); 13C NMR (100 MHz, CDCl3):
134.5, 132.5, 131.4. 130.7, 129.3, 127.2, 21.6; MS; m/z: 276 (Mϩ
2), 274 (Mϩ), 185, 183, 119 (100), 91, 76; Anal. calcd for
Њ
C; IR (film): 3100, 2900, 1680, 1550, 1300 cmϪ
;
␦
7.69–7.60 (m, 6H), 7.27 (m, 2H),
195.3, 143.5, 136.7,
␦
Figure 2. The reaction scheme for the synthesis of the BPD. The
coupling of the monomers using a palladium catalyst had the lowest
yield.
ϩ
C14H11OBr: C, 61.30; H, 4.05. Found: C, 61.26; H, 3.82.
4-(4-Bromobenzoyl)benzoic acid (5). (4-Bromophenyl)(4-methyl-
phenyl)methanone (4) (1.75 g, 6.4 mmol) dissolved in warm acetic
acid (6.55 mol) was kept just below the boiling point. A solution of
chromic acid (1.73 g, 0.0173 mol) in H2O (3.95 mL), acetic acid
(6.43 mL) and concentrated sulfuric acid (1.24 mL) was added slow-
ly to the solution. The reaction was monitored by TLC until the
starting material was consumed. The reaction mixture was then fil-
tered, and the solid was rinsed with water. The white residue was
purified by flash chromatography on silica gel (ethyl acetate/acetic
acid 99.5:0.5) to give 4-(4-bromobenzoyl) benzoic acid (1.36 g,
reported as wave numbers (cmϪ1). Combustion analyses were per-
formed by NuMega Resonance Labs, Inc., San Diego, CA. Melting
points were observed in open Pyrex capillary tubes and are uncor-
rected. Low-resolution mass spectra were obtained on an HP 5970
series GC-MSD system and are reported in units of mass to charge
(m/z). High-resolution mass spectra were obtained on a JEOL JMS-
AX505HA instrument, and electrospray ionization (EI) mass spectra
(MS) were obtained on a Micromass Quattro LC, both at the Uni-
versity of Notre Dame. Flash chromatography was performed on
78%) as a white solid: mp 270–271
ЊC (lit. mp (26) 274ЊC); MS (EI);
m/z: 306 (Mϩ 2), 304 (Mϩ), 185 (100), 183, 149, 76, 65.
ϩ
Silica Gel, 40 ␮m, 32–63 flash silica from Scientific Adsorbent Inc
Methyl 4-(4-bromobenzoyl)benzoate (6). 4-(4-Bromoben-
zoyl)benzoic acid (5) (0.20 g, 0.66 mmol) was added to a solution
of methanol (2 mL) and concentrated sulfuric acid (0.23 mL) and
stirred under reflux. The solution was then concentrated in vacuo.
Purification by flash chromatography on silica gel (petroleum ether–
EtOAc 90:10) afforded methyl 4-(4-bromobenzoyl) benzoate (0.16
(Atlanta, GA). Thin layer chromatography (TLC) was performed on
silica gel (EM Science Silica Gel 60 F254) glass plates, and the com-
pounds were visualized by UV with 5% phosphomolybdic acid in
ethanol or 0.5% potassium permanganate in 0.1 M aqueous NaOH.
Syntheses. The synthesis of the diisopropyl amine rose bengal
derivative has been previously described in detail (23). Briefly, the
g, 81%) as a slightly pink solid: mp 177–178
178 8.15 (m, 2H), 7.81 (m, 2H),
C); 1H NMR (400 MHz, CDCl3):
7.67 (m, 4H), 3.97 (s, 3H); 13C NMR (100 MHz, CDCl3):
194.8,
ЊC (lit. mp [26] 177–
first step is the synthesis of the intermediate 2Љ-bromoethyl ester of
Њ
␦
rose bengal by reaction with dibromoethane. The rose bengal 2
Љ-
␦
bromoethyl ester was then reacted with diisopropyl amine to form
the rose bengal amine derivative.
166.2, 140.9, 135.7, 133.5, 131.8, 131.5, 129.6, 129.6, 128.1, 52.4;
MS (EI); m/z: 320(Mϩ
ϩ
2), 318 (Mϩ), 185, 183 (100), 163, 104,
To synthesize the BPD, we planned to link the monomeric units
via palladium-catalyzed aryl amination strategies, similar to that de-
veloped by Hartwig (24) and Buchwald and coworkers (25). The
scheme for the synthesis of the dimer is shown in Fig. 2. The syn-
thesis of compound 8 began with a reaction between the Grignard
reagent derived from commercially available 4-bromotoluene (1)
and 4-bromobenzaldehyde (2). Attempts to directly oxidize the re-
sultant alcohol 3 to benzophenone monomer 5 were unsuccessful.
Oxidation of 3 with KMnO4 provided benzophenone 4, and the
methyl group was oxidized with CrO3 to obtain 5. Esterification
provided key benzophenone building block 6. Modest success in the
bifunctional coupling between pentane diamine and 6 was achieved
by using 1,1-bis(diphenyphosphino)ferrocene (DPPF) as the catalyst.
For cross-linking studies 7 was hydrolyzed to provide the dipotas-
sium salt 8.
(4-Bromophenyl)(4-methylphenyl)methanol (3). Magnesium turn-
ings (0.39 g, 0.016 mol) were dried in a flask and then cooled under
nitrogen. Anhydrous diethyl ether (Et2O) (3 mL) was added, and
then 4-bromotoluene (3.0 g, 1.8 mmol) was added. The mixture was
refluxed, and more anhydrous Et2O was added to keep the solution
volume constant. The resultant Grignard solution was added drop-
wise to a solution of 4-bromobenzaldehyde (1.66 g, 9.0 mmol) in
anhydrous Et2O (6 mL). The reaction was heated at reflux for 30
min. The cooled solution was poured into a solution of HCl (3.7 mL
concentrated HCl in 17 g ice-water) and stirred for 1 h. The resulting
76.
Methyl 4-(4-[5-(4-[4-methylcarboxylatebenzoyl]phenylamino)penty-
lamino]benzoyl)benzoate (7). A dried Schlenk tube was charged with
methyl 4-(4-bromobenzoyl)benzoate (6) (0.20 g, 0.63 mmol),
tris(dibenzylideneacetone)dipalladium Pd2(dba)3 (0.029 g, 0.032
mmol), DPPF (0.035 g, 0.063 mmol) and toluene (8 mL) under
nitrogen. The reaction was capped with a polytetrafluoroethylene
septum, and pentanediamine (0.026 g, 0.252 mmol) was added via
a syringe. The reaction mixture was then heated to 90ЊC for 12 h.
The cooled mixture was diluted with diethyl ether (30 mL), filtered
and then concentrated in vacuo. Purification by flash chromatogra-
phy on silica gel (petroleum ether–EtOAc 50:50) afforded dimer 7
(0.072 g, 21%) as a pale yellow solid: IR (CDCl3): 3350, 3050,
2900, 1750, 1680, 1680, 1300, 1150, 990, 780 cmϪ1; H NMR (400
1
MHz, CDCl3):
6H), 3.23 (m, 4H), 2.04 (s, 2H), 1.72 (m, 4H), 1.23 (m, 2H); 13C
NMR (100 MHz, CDCl3): 194.0, 166.3, 152.1, 142.9, 132.8, 129.3,
129.1, 128.8, 125.6, 111.1, 52.2, 42.9, 28.9 24.3; HRMS (FAB)
calcd for C35H35N2O6 (Mϩ
1) 579.2495. Found: 579.2476.
Potassium 4-(4-[5-(4-[4-potassiumcarboxylatebenzoyl]phenylamino)
pentylamino]benzoyl)benzoate (8). A solution of methyl 4-(4-[5-(4-[4-
methylcarboxylatebenzoyl]phenylamino)pentylamino]benzoyl)benzoate
(7) (0.030 g, 0.052 mmol) and potassium hydroxide (0.005 g, 0.10
mmol) in ethanol (3 mL)–water (0.16 mL) was stirred at room tem-
␦ 8.11 (m, 4H), 7.73 (m, 8H), 6.57 (m, 4H), 3.95 (s,
␦
ϩ

138 Jonathan D. Pitts et al.
perature for 2.5 h. The solvents were removed by freeze-drying, and
the residue was triturated with Et2O. The product 8 (0.02 g, 63%)
was isolated as a pale yellow solid: IR (KBr): 1634, 1594, 1261,
1
1099, 1024 cmϪ1; H NMR (400 MHz, CD3OD):
␦
8.00 (d, J
8.8 Hz, 4H), 3.24 (m, 4H), 1.91
(m, 4H), 1.16 (m, 2H); 13C NMR (100 MHz, CD3OD):
195.8;
173.0, 153.8, 140.6, 140.5, 128.5, 128.3, 123.9, 110.6, 48.4, 28.5,
27.0; MS; m/z (ESI-positive ion mode): 665 (Mϩ Kϩ); MS; m/z
(ESI-negative ion mode): 549 (MϩϪ 2Kϩ Hϩ).
ϭ 8.2
Hz, 4H), 7.62 (m, 8H), 6.60 (d, J
ϭ
␦
ϩ
ϩ
Fabrication instrument. The fabrication apparatus was described
previously (15) and consists of a laser scanning confocal microscope
(MRC600, Biorad, London, UK; Axioscope, Zeiss, Jena, Germany)
modified for near-infrared excitation and a femtosecond titanium
sapphire oscillator (900-F, Coherent, Santa Clara, CA). The laser
was operated between 780 and 850 nm with a repetition rate of 76
MHz at 100 fs pulse duration with an average power at the sample
of approximately 50 mW, or about 1 nJ per pulse. With a 0.75
numerical aperture (NA) objective lens (Zeiss) this corresponds to a
peak power of roughly 1010 W/cm2. This NA corresponds to lateral
and axial resolutions of approximately 700 nm and 2
tively.
␮m, respec-
The laser scanner is configured to perform line and rectangular
raster scans, where the latter scan consists of 768 512 pixels,
which for a 20 lens yields a full field of view of 670 by 447
ϫ
ϫ
␮m
and is scanned in about 1 s. Rectangles were also created at higher
zoom. Optical diagnostics are measured via either two-photon ex-
cited fluorescence of the photoactivator (confocal fluorescence de-
tection) or transmitted light. Line scans were repetitively performed
until polymerization was observed via the appearance of a change
in the refractive index. Although not rigorously quantitative in that
we are not measuring final yields, this method is suitable for com-
parison of the relative cross-linking efficiency of different photoac-
tivators by maintaining constant laser power and protein concentra-
tion. The results are found to be self-consistent between sets of ex-
periments (15).
The absorption and emission spectra were measured using a Cary
(Varian, Palo Alto, CA) spectrometer and a Spex (Edison, NJ) spec-
trometer, respectively. The fluorescence quantum yield of the BPD
was measured relative to Rhodamine B. Power dependence mea-
surements to determine if the excitation of the BPD occurred via
two- or three-photon absorption were performed by monitoring the
relative fluorescence intensities. These determinations were per-
formed at 780, 810 and 847 nm by an epi-illumination fluorescence
Figure 3. (a) Jablonski diagram for rose bengal showing the rele-
vant states involved in multiphoton excitation. The work here ex-
cited the higher energy band. (b) Reaction efficiency of rose bengal
4
and the rose bengal diisopropyl amine in BSA (1.5
ϫ
10Ϫ M, pH
7.2 PBS) as a function of photoactivator concentration. The excita-
tion wavelength was 800 nm and average power at the sample was
50 mW. The photon flux is integrated until the threshold for cross-
linking is optically observed.
setup as described previously (15). Briefly, a 0.25 NA 5
(Zeiss) was used for the excitation, and the fluorescence was col-
lected and detected at 90 to the excitation. The laser power was
attenuated with a /2 plate and a Glan Laser polarizer (CVI Laser,
ϫ objective
Њ
␭
Albuquerque, NM). The fluorescence was separated with a long
wave–pass dichroic (550 nm, CVI Laser), and the residual laser was
blocked by BG-39 color glass filters. The filtered signal was detected
by a photomultiplier (R4632, Hammamatsu, Bridgewater, NJ),
preamplified (SR350, Stanford Research Systems, Sunnyvale, CA)
and boxcar-averaged (SR250 Stanford Research Systems).
The fluorescence lifetime of the BPD was measured using time
correlated single photon counting. A Picoquant (Berlin, Germany)
Time Harp PC card was used for the acquisition, and the detector
was a photon counting photomultiplier module (7421, Hammamat-
su). After deconvolution, the instrument response function is ap-
proximately 80 ps.
electronic state, which subsequently decays nonradiatively
with near unit efficiency to the S1 state (15). Rose bengal in
water has a fluorescence quantum yield of about 0.01, and
the singlet oxygen quantum yield is between 0.75 and 0.85
(28). Thus, at least this fraction of population is converted
from the excited singlet to the triplet state. In xanthenes the
lifetime of the first triplet state is known to be about 2 ms
(28).
We are interested in comparing the cross-linking perfor-
mance of the rose bengal derivative with rose bengal, as well
as with the BPD. Rather than measure final yields, we de-
termine the relative rates (or efficiencies) of the photoacti-
vators by running repetitive line scans until the onset of re-
action is determined by a visible change in refractive index
and measuring the integrated photon flux. By maintaining
constant protein concentration and laser power and varying
the photoactivator concentration, this method provides a
self-consistent manner in which to compare different pho-
toactivators. The resulting concentration dependence of the
two-photon excited cross-linking efficiency of both rose ben-
gal and the amine derivative on BSA is shown in Fig. 3,
where in both cases the two-photon excitation was ␭ ϭ 800
RESULTS
Rose Bengal diisopropyl amine
Multiphoton excitation proceeding through a virtual or non-
stationary state was predicted in the early part of the 20th
century by Go¨ppert-Mayer (27). Figure 3a shows a Jablonski
diagram that describes the two-photon excitation for the pho-
toactivator rose bengal. The S1 state may be reached via a
single photon excitation of 550 nm (␭max) or in the corre-
sponding simultaneous two-photon approach via two 1100
nm photons proceeding through a virtual state. Alternatively,
two-photon excitation can occur at 800 nm to access the S2

Photochemistry and Photobiology, 2002, 76(2) 139
Ϫ
nm and the BSA concentration was 1.5
ϫ
10 4 in phosphate-
buffered saline (PBS), pH 7.2 The relative rate is measured
in terms of photon flux required to observe the onset of
reaction in units of J/cm2. These units are not rigorous for
integrating the flux in a multiphoton process because of the
peak power dependence. For one-photon absorption, the ab-
sorption probability is independent of the mode of excitation,
i.e. continuous wave or pulsed, and there is no dependence
on the pulse width for the latter. By contrast, multiphoton
n
Ϫ
absorption scales as pn/a
␶
1, where p is the pulse energy in
is the laser pulse
Joules, a is the spot size in cm2, and
␶
Figure 4. Fluorescence image of cross-linked collagen rods using
width. There are two different operative powers, the average
power and peak power, where the latter is the operative pa-
rameter in multiphoton absorption. It must be stressed that
the values herein are thus not comparable with the one-pho-
ton excited polymerization and cross-linking literature
(29,30). Thus, a lower photon flux at higher photoactivator
concentration corresponds to faster cross-linking. This same
type of rapid drop-off, i.e. increase in rate, as a function of
photoactivator concentration has been seen by others in one-
photon excitation in terms of polymer yields (29,30). The
steepness of the slope is in part a consequence of the ex-
perimental apparatus, as this corresponds to one scan. Al-
though in our previous work we observed that the presence
of a triethanol amine co-initiator was required to observe
polymerization of synthetic polymers with rose bengal, it
had a large inhibitory effect on the cross-linking of proteins,
by at least one or two orders of magnitude (16). Here we
observe that both rose bengal and the amine derivative con-
jugate are indeed effective in fabricating BSA, although rose
bengal alone is a few-fold faster. However, the advantages
of this new activator will be shown subsequently.
3
the rose bengal diisopropyl amine of concentration 2.5
The Type 1 collagen concentration was 1.5
ϫ
10Ϫ M.
6
ϫ
10Ϫ M, and the pH
was approximately 4. The fluorescence is because of entrapped rose
bengal. These rods were adhered to the glass substrate and were
stable over the observation period.
bleaching and polymer yields in air (relative to nitrogen)
coupled with the strong photoactivator concentration depen-
dence on the reaction efficiency indicates that not all the
cross-linking occurs via a singlet oxygen mechanism.
Although the rose bengal amine derivative was slower
than rose bengal alone in cross-linking BSA, it is still a
potentially powerful new photoactivator. BSA is a serum
protein, highly soluble (Ͼ100 mg/mL) and readily cross-
linked by a variety of methods. Type 1 collagen, however,
is difficult to solubilize and thus to cross-link efficiently.
Figure 4 shows a fluorescence image of several parallel rods
of Type 1 collagen (rat-tail tendon) successfully cross-linked
by the rose bengal amine derivative, where the concentra-
tions of the rose bengal diisopropyl amine and collagen were
Ϫ
3
Ϫ6
2.5
ϫ 10 and 6 ϫ 10 M, respectively. The contrast here
There are two generally accepted modes in which proteins
can be cross-linked by photoactivation: hydrogen atom ab-
straction and formation of singlet oxygen to photooxidize
protein residues (31). Given singlet oxygen yields of 75–
85% for rose bengal, we cannot rule out the involvement of
singlet oxygen in the protein cross-linking because these
measurements were not made in an oxygen-free environ-
ment. However, Banerjee (23) showed that the rose bengal
amine derivative bleached approximately twice as fast under
nitrogen as under air. This is indicative of bleaching occur-
ring at least in part by the formation of radicals via a charge
transfer (C-T) process, rather than the singlet oxygen mech-
anism. Further, for the case of acrylamide, the polymeriza-
tion was roughly twice as fast under nitrogen as in air. In
the current work (Fig. 3b), the dependence of the photoac-
tivator concentration on the relative rate (at constant BSA
concentration) indicates that it is being consumed, which is
suggestive of, at least in part, a hydrogen abstraction mech-
anism. This is because the rose bengal would be regenerated
in a mechanism involving the formation of singlet oxygen
because simple energy transfer occurs from the triplet state,
allowing the rose bengal to return to the ground state. By
contrast, in the case of hydrogen abstraction, the ketone on
rose bengal would become reduced to the alcohol, and the
resulting molecule is no longer photoactive. Note that we
have previously determined (16) that the rose bengal chro-
mophore is not significantly bleached during the course of
these measurements, and photobleaching thus has little or no
effect on these measurements. The suppression of photo-
arises from both the change in refractive index that occurs
after fabrication, as well as from the fluorescence from en-
trapped rose bengal amine inside the collagen. Note we have
previously shown the dye is readily washed away in water
postfabrication, whereas the structural integrity of the cross-
linked protein is maintained (16). Although these collagen
structures are not as well formed as those from BSA (15,16)
(data not shown), they were well-adhered to the glass and
had long-term stability (at least over the examination period
of 1 h). By contrast, our best efforts at cross-linking collagen
using rose bengal alone produced only diffuse structures that
disintegrated within seconds of formation. Other dyes we
tried, including xanthenes, the triarylmethanes crystal violet
and FD&C Green #3 and the quinoline dye FD&C Yellow
#10, did not produce even transient evidence of collagen
cross-linking.
A practical limiting factor in forming more tightly cross-
linked structures is the miscibility between the rose bengal
amine and collagen. When rose bengal or the amine deriv-
ative was added to acid-extracted Type 1 collagen, some
collagen fibers (micron-size) fell out of solution. These ap-
peared to have adsorbed rose bengal because they were fluo-
rescent. We have observed this behavior with other dyes
including the triarylmethanes crystal violet and FD&C Green
#3, and the quinoline dye FD&C Yellow #10.
We do expect this rose bengal amine photoactivator to be
successful with a variety of forms of collagen and other min-
imally soluble structural proteins, including laminin, myosin
and actin because most proteins possess a number of ex-

140 Jonathan D. Pitts et al.
posed appropriate reactive sites. These include residues con-
taining amines and ketones for the H atom abstraction mech-
anism and residues containing aromatics, heterocycles and
olefins such as tryptophan, tyrosine and histidine for the sin-
glet oxygen mechanism. (31,32) We have now in fact been
successful in cross-linking every protein we have attempted.
Optical characterization of BPD.
Benzophenones, upon excitation of the ␲ → ␲* transition
(␭
max
ϳ 260 nm) are known well to form ketyl biradicals
that are inert to water and can extract a hydrogen from a
nearby carbon–hydrogen bond and form a carbon–carbon
bond (33). Each end of the BPD shown in Fig. 1b can react
in this way and becomes an integral part of the cross-linked
matrix. Because this BPD is a new photoactivator, its basic
photophysical and photochemical properties needed to be
characterized. The UV–visible absorption spectrum in sev-
eral solvents is shown in Fig. 5a. The normal ␲ → ␲* tran-
sition associated with benzophenone (
␭
ϳ
260 nm) is pre-
max
sent, in addition to a comparably strong new band with
␭
max
ϳ
370 nm. The weak n → ␲* band (
␭
ϳ 325 nm in
max
benzophenone) is presumably obscured in the overlap of the
two stronger bands. The appearance of the second band
could be caused by, in part, an inter- or intramolecular C-T
band. To investigate the first possibility, the relative absorp-
tion intensities of the two bands (260 and 370 nm) were
measured as functions of concentration and were found to
be unchanging over an order of magnitude of concentration
(data not shown). This concentration independence rules out
an intermolecular C-T process. This new band is highly sol-
vatochromic, with
␭
max
(H2O) Ͼ ␭
(methanol) Ͼ ␭
max
max
(CH3CN) ഠ ␭
(tetrahydrafuran), where the shift in ␭
max
max
was approximately 20 nm between these solvents. This spec-
tral ordering is consistent with the stabilization of a C-T
band expected with these solvents, on the basis of their re-
spective polarities and solvation energies. Furthermore, the
lower absorption bands (ϳ290 nm) of aniline and aniline
derivatives are known to possess some intramolecular C-T
character (34). Each monomeric moiety of the BPD is in
Figure 5 (a) The absorption spectra of the BPD in PBS, methanol,
tetrahydrofuran (THF) and acetone. The lower energy bands display
solvatochromism, with an approximately 20 nm shift in the absorp-
tion maximum between PBS and THF. (b) The absorption spectra
of the BPD, Michler’s ketone, and benzophenone in methanol
(Michler’s ketone was insoluble in water). (c) The emission spec-
trum of the BPD in PBS excited at 375 nm. The quantum efficiency
was approximately 0.005. Excitation at 260 nm produced the same
emission spectrum.
fact similar to Michler’s ketone, (4,4Ј-bis(dimethylamino)-
benzophenone. The absorption spectra of benzophenone,
Michler’s ketone, and the BPD in methanol are shown in
Fig. 5b. The lower energy bands of the BPD and Michler’s
ketone nearly overlap in both energy and intensity. This red-
der absorption band in Michler’s ketone is known to arise
from C-T interaction (35), and in that it is similar to the
BPD. However, the higher energy band of the BPD more
closely resembles that of benzophenone itself.
The emission spectrum of the BPD, shown in Fig. 5c, also
departs greatly from that of benzophenone and Michler’s
ketone. Despite the similarity in the absorption spectra be-
tween the BPD and Michler’s ketone, the latter is entirely
nonfluorescent in methanol. The BPD has a broad emission
band extending through the visible region, with a maximum
of 500 nm, whereas this region of the emission spectrum is
essentially missing in the benzophenone, which only has a
weak blue band. The emission spectrum in Fig. 5c is iden-
tical for both 260 and 375 nm excitation, suggesting the final
state is the same for either excitation band. Both bands are
indeed photoactive in terms of cross-linking proteins.
The fluorescence quantum yield of the BPD was measured
relative to Rhodamine B and found to be
ϳ0.005 in water.
The fluorescence lifetime in water was determined to be 170
ps, indicating the singlet excited state is heavily quenched,
which is consistent with the small fluorescence quantum
yield.
Because our work uses multiphoton excitation, power de-
pendencies were measured at three wavelengths to determine
if the route of excitation for the BPD was a two- or three-
photon process or some combination thereof. This is impor-
tant because the mode of excitation can affect the minimum
feature sizes that are obtainable: at the same wavelength,
three-photon excitation provides finer resolution than two-
photon absorption (36). This is because the optical resolution
in a multiphoton process is not rigorously governed by the
diffraction limit on account of the cooperative nature of mul-
tiphoton excitation within the focal volume (36). These de-

Photochemistry and Photobiology, 2002, 76(2) 141
4
Figure 7. (a) Reaction efficiency of the BPD on BSA (1.5
ϫ
10Ϫ
M, pH 7.2 PBS) as a function of BPD concentration. The excitation
wavelength was 800 nm. (b) Fluorescence image of a two-tiered
collagen structure cross-linked by the BPD at 850 nm. The BPD
3
6
and collagen concentrations were 2.5
respectively.
ϫ ϫ
10Ϫ M and 6 10Ϫ M,
Figure 6. (a) Log–log plot of the power dependencies for the BPD
for multiphoton excitation at 780, 810, and 847 nm, as determined
by integrated fluorescence intensity measurements. In PBS pH 7.2
(b) A proposed Jablonski diagram showing the states for the BPD,
showing the band associated with benzophenone as well as the new,
lower energy, C-T band that resembled Michler’s ketone. Excitation
of both bands was successful in cross-linking both BSA and collagen
and is consistent with excitation of either band having the same
emission spectrum, indicating the final state is the same in both
cases.
rules for one- and multiphoton absorption can be quite dif-
ferent. For example, consider the symmetric chromophore of
fluorescein. The S0
→ S2 transition is symmetry-forbidden
for one-photon excitation but is in fact the strongest transi-
tion for two-photon excitation. Further, because of pertur-
bations, the multiphoton absorption spectra do not always
simply coincide with the corresponding energy of the one-
photon absorption spectrum. A proposed Jablonksi diagram
showing the relevant states of the BPD and the photoreactive
route is shown in Fig. 6b, where we suggest that under our
conditions (780 to 850 nm excitation) the ␲ → ␲* transition
is primarily a three-photon absorption and the lower energy
C-T band is excited by two photons.
The effect of the BPD concentration on BSA cross-linking
efficiency is shown in Fig. 7a. It is observed that the BPD
displays a strong concentration dependence on the reaction
efficiency, indicating that it is either being degraded in the
photochemistry (like rose bengal) or becoming incorporated
into the matrix. We suggest that the latter case predominates
on the basis of analogy with the well-known benzophenone
photochemistry (33).
terminations were made by measuring the fluorescence in-
tensity of the broad visible emission band of the BPD as a
function of laser power at several excitation wavelengths.
This approach works because the emission spectrum is iden-
tical for excitation of either the 260 or 375 nm band via one-
photon excitation. These results for 780, 810, and 847 nm
multiphoton excitation are shown in Fig. 6a as log–log plots,
where the slopes represent the absorption order. It is seen
that the 780 nm multiphoton excitation has a slope of 2.84,
corresponding to a primarily three-photon absorption, as ex-
pected by simple analogy from the monomeric benzophe-
none chromophore. By contrast, 847 nm excitation has a
slope of 2.18, implying it is mostly caused by two-photon
absorption, with some three-photon overlap.
The relative and absolute intensities of these bands could
not be readily determined on account of the overlap of two-
and three-photon absorption because no consistent normali-
zation factor could be used. ‘‘Pure’’ two- and three-photon
absorption wavelengths would have been necessary, and this
was not experimentally achievable. Note that the selection
The required integrated flux to cross-link BSA (same con-
centration) was approximately one order of magnitude high-
er for the BPD dimer than for rose bengal. This can be be-
cause of either differences in the multiphoton absorption
cross section or relative reaction efficiencies after excitation

142 Jonathan D. Pitts et al.
(intersystem crossing [ISC] yield and triplet state quenching
rates). Rose bengal and the BPD have similar fluorescence
quantum yields (0.01 and 0.005) and comparable fluores-
cence lifetimes (37) (92 and 170 ps, respectively). These
data would suggest comparably high ISC efficiencies for
these molecules. We can roughly compare the triplet quench-
ing rates of analogs of each with the reaction quantum ef-
ficiencies after ISC. The triplet state quenching rate by ox-
ygen for rose bengal is approximately one-tenth diffusion
to occur by an intramolecular reaction. This increases the
cross-linking efficiency by two means: (1) the reaction ki-
netics are reduced in terms of the number of diffusive spe-
cies; i.e. three-body to two-body; (2) the C-T conversion rate
effectively increases because of the constrained proximity of
the reactive pair during the finite triplet state lifetime. Fur-
ther, it allows lower concentration of the co-initiator amine
to be used—in the normal scheme co-initiator concentrations
in the 100 mM range are used, whereas submillimolar con-
centrations of the derivative were sufficient to observe cross-
linking.
Ϫ
Ϫ
Ϫ1
limited (28), or 10 8 M 1 s . The quenching rate by oxygen
Ϫ
8
for benzophenone (33) is nearly diffusion limited at 4
ϫ
10
Ϫ
1
Ϫ1
M
s . This suggests the decreased relative rate for cross-
The BPD was designed to be a reagent to be used to
fabricate a wide range of proteins. Previous work on cross-
linking structural proteins with benzophenone had begun
with the chemical conjugation of benzophenone to a protein
(e.g. via succinimydyl ester or isothiocyanate chemistries),
followed by the photochemical step to bind a second protein.
Although certainly possible, this procedure would be rather
time consuming for cross-linking every protein of interest.
The availability of two photoreactive moieties on the same
molecule adds a great deal of flexibility to the fabrication
approach, in that several different proteins can be simulta-
neously photochemically cross-linked in situ without having
an initial chemical step. Further, unlike rose bengal, at the
millimolar concentrations used the BPD had no miscibility
problems with the acid-extracted Type 1 collagen. We fur-
ther point out that the cross-linking ability of the BPD is
maintained in a neutral or acidic environment. By contrast,
the photoproperties of rose bengal and other xanthenes de-
grade in acidic solution, and it is much more effective in a
neutral or mildly basic solution.
To the best of our knowledge, we have published the only
data on multiphoton excited cross-linking of proteins; thus,
it is not possible to compare our results with other literature.
Other work in the multiphoton fabrication area has focused
on the use of synthetic polymers (2–8). We have measured
the relative rates for BSA cross-linking using several other
dyes including the triarylmethanes crystal violet and FD&C
Green #3, and the quinoline dye FD&C Yellow #10. These
dyes were all approximately 10- to 50-fold less efficient than
rose bengal. None of these chemistries was successful with
Type 1 collagen.
linking BSA for the BPD versus rose bengal in part arises
from the former having a smaller multiphoton absorption
cross section. In all likelihood, the multiphoton absorption
cross section of rose bengal is significantly larger than that
of the BPD because of its substantially increased conjugation
(38). The extinction coefficent of the BPD (⑀ ϳ 16 000) is
five-fold smaller than that of rose bengal, and to a first ap-
proximation the two-photon cross section should then scale
as the square of this ratio or about 25-fold. Because of the
two- and three-photon overlap described previously, it is not
possible to rigorously extract absolute multiphoton absorp-
tion cross sections.
The power of this new BPD photoactivator is shown in
the fluorescence image of a two-tiered rectangle in Fig. 7b.
The concentration of the BPD was approximately 2.5 mM
in PBS, pH 7. The pH of the acid-extracted collagen was
approximately 4 and the collagen concentration was 6
ϫ
10 6 M. The bottom tier of the rectangle adhered to the glass
substrate is created by performing 50 consecutive raster
scans of duration 1 s each, where the scan consists of 768
Ϫ
ϫ
512 pixels at Zoom 4 (158
then lowered by approximately 2
increased to 5 (127 84 m), and then exposed by 50
consecutive raster scans. Each layer was approximately 4
ϫ
106
␮m). The stage was
␮m, and the zoom was
ϫ
␮
␮s
high. The image contrast arises from the change in refractive
index as well as from the residual fluorescence from the
BPD. Although collagen possesses both intra- and intermo-
lecular cross-links, the solution is transparent to transmitted
light and has uniform fluorescence intensity before it is fur-
ther cross-linked by the BPD (or rose bengal derivative).
This three-dimensional structure is well-formed (as judged
by sharpness of the edges and uniform contrast), well-ad-
hered to the glass substrate, and had long term stability (over
a several hour period of observation after fabrication).
CONCLUSIONS
We demonstrate the use of two new photoactivators for mul-
tiphoton-excited three-dimensional free-form fabrication
with proteins. The rose bengal diisopropyl amine maintains
the same photophysical and photochemical properties as rose
bengal and combines the photoactivator and co-initiator sys-
tem into one molecule, thereby reducing the overall kinetics.
This derivative produced stable structures from Type 1 col-
lagen monomers, whereas this was not achievable with rose
bengal alone. We also synthesized and characterized the
photophysics of a new BPD connected by a flexible diamine
tether. This activator has two photochemically reactive
groups and is highly efficient in cross-linking collagen to
form stable, robust structures. It is also more flexible than
traditional benzophenone schemes in that no chemical con-
jugations are required. The photophysical properties are
quite different from that of benzophenone, with the appear-
DISCUSSION
The rose bengal diisopropyl amine shown in Fig. 1a was
designed to have the favorable photochemical reactivity of
rose bengal without changing the chromophore because the
absorption spectrum is essentially identical to that of rose
bengal itself (data not shown) (23). Rose bengal also has a
Ϫ
respectable two-photon (15) (
ϳ
10
ϫ
10 50 cm4 s) absorption
cross section. Most highly conjugated dyes have two-photon
absorption cross sections of this order (39), although reports
of larger cross sections have appeared in the literature
(10,12). The addition of the covalently bound diisopropyl
amine allows the intermolecular C-T reaction between the
photoactivator and co-initiator forming radical ions (30,40)

Photochemistry and Photobiology, 2002, 76(2) 143
multifocal array for multiphoton microscopy and micromachin-
ing. Optics Lett. 25, 1213–1215.
15. Campagnola, P. J., A. R. Howell, D. Delguidas, G. A. Epling,
J. D. Pitts and S. L. Goodman (2000) 3-Dimensional sub-micron
polymerization of acrylamide by multi-photon excitation of xan-
thene dyes. Macromolecules 33, 1511–1513.
16. Pitts, J. D., P. J. Campagnola, G. A. Epling and S. L. Goodman
(2000) Reaction efficiencies for sub-micron multi-photon free-
form fabrications of proteins and polymers with applications in
sustained release. Macromolecules 33, 1514–1523.
17. Konig, K., P. T. C. So, W. W. Mantulin and E. Gratton (1997)
Cellular response to near-infrared femtosecond laser pulses in
two-photon microscopes Optics Lett. 22, 135–136.
18. Valdes-Aguilera, O., C. P. Pathak, J. Shi, D. Watson and D. C.
Neckers (1992) Photopolymerization studies using visible light
photoinitiators. Macromolecules 25, 541–547.
19. Bi, Y. and D. C. Neckers (1994) A visible light initiating system
for free radical promoted cationic polymerization. Macromole-
cules 27, 3683–3693.
ance of a new, lower energy C-T absorption band (
␭
ϳ
max
370 nm) and a broad, visible emission band (max
ϳ 5000
nm). These two new activators may prove effective in fab-
ricating with other forms of collagen and other structural
proteins (e.g. actin, myosin) for biomedical applications such
as de novo directed assembly of submicron featured tissue
engineering scaffolds and the creation of submicron featured
sustained release devices. This overall approach has better
three-dimensional capabilities and is more flexible and bio-
compatible than either photolithography or microcontact
printing.
Acknowledgments We gratefully acknowledge support by the
University of Connecticut Health Center, NSF Academic Re-
search Infrastructure, NIH R01 GM60703, and the Donaghue
Foundation.
20. Leszyk, J., J. H. Collins, P. C. Leavis and T. Tao (1987) Cross-
linking of rabbit skeletal muscle troponin with the photo-reac-
tive reagent 4-maleimodobenzphenone: identification of resi-
dues in troponin 1 that are close to cysteine-98 of troponin C.
Biochemistry 26, 7042–7047.
21. Tao, T., M. Lamkin and C. J. Scheiner (1985) The conformation
of the C-terminal region of actin: a site-specific photocrosslink-
ing study using benzophenone-4-maleimide. Arch. Biochem.
Biophys. 240, 627–634.
REFERENCES
1. Denk, W., J. H. Strickler and W. W. Webb (1990) Two-
photon laser scanning fluorescence microscopy. Science 248,
73–76.
2. Cheng, P. C., J. D. Bhawalker, S. J. Pan, J. Wiatakiewicz, J. K.
Samarabandu, W. S. Liou, G. S. He, G. E. Ruland, N. D. Kumar
and P. N. Prasad (1996) Two-photon generated three-dimen-
sional photo-bleached patterns in a polymer matrix. Scanning
18, 129–131.
3. Witzgall, G. R. Vrijen, E. Yablonovitch, V. Doan and B. J.
Schwartz (1998) Single-shot two-photon exposure of commer-
cial photoresist for the production of three-dimensional struc-
tures. Optics Lett. 23, 1745–1747.
4. Strickler, J. H. and W. W Webb (1991) Three-dimensional op-
tical data storage in refractive media by two-photon point ex-
citation. Optics Lett. 16, 1780–1782.
5. Maruo, S., O. Nakamura and S. Kawata (1997) Three-dimen-
sional microfabrication with two-photon-absorbed photopoly-
merization. Optics Lett. 22, 132–134.
6. Kawata, S., H.-B. Sun, T. Tanaka and K. Takada (2001) Mi-
crofabrication: finer features for functional microdevices. Nature
412, 697.
7. Sun, H.-B., T. Tanaka, K. Takada and S. Kawata (2001) Two-
photon photopolymerization and diagnosis of three-dimensional
microstructures containing fluorescent dyes. Appl. Phys. Lett.
79, 1411–1413.
22. Agarwal, R., K. N. Rajasekharan and M. J. Burke (1991) Iden-
tification of the site of phot-crosslinking formed in the absence
of magnesium nucleotide from SH2 (Cys-697) in myosin su-
bfragment 1 labeled with 4Ј-maleimidylbenzophenone. J. Biol.
Chem. 266, 2272–2275.
23. Banerjee, I. (2001) Photochemistry and applications of protein-
bound dyes and dye-amine conjugates in the presence of visible
light. Ph.D. thesis, University of Connecticut.
24. Hartwig, J. F. (1997) Palladium-catalyzed amination of aryl ha-
lides: mechanism and rational catalyst design. Synlett 1997,
329–340.
25. Wolfe, J. P., S. Wagaw, J.-F. Marcoux and S. L. Buchwald
(1998) Rational development of practical catalysts for aro-
matic carbon-nitrogen bond formation. Acc. Chem. Res. 31,
805–818.
26. Parham, W. E. and Y. A. Sayed (1974) Synthesis of isomeric
methyl benzoylbenzoates and substituted o-, m-, and p-ben-
zoylbenzoic acids. J. Org. Chem. 39, 2053–2056.
27. Go¨ppert-Mayer, M. (1931) Ober Elementarakte mit zwei Quan-
tensprungen. Ann. Phys. 9, 273–294.
28. Neckers, D. C. (1989) Rose bengal. J. Photochem. Photobiol.
A 47, 1–29.
8. Day, D., M. Gu and A. Smallridge (1999) Use of two-photon
excitation for erasable rewritable three-dimensional bit optical
data storage in a photorefractive polymer. Optics Lett. 24, 948–
950.
29. Neckers, D. C. and O. M. Valdes-Aguilera (1993) Photochem-
istry of the xanthene dyes. In Advances in Photochemistry, Vol.
18. (Edited by D. Volman, G. S. Hammond and D. C. Neckers),
pp. 315–393. John Wiley & Sons, New York, NY.
30. Eaton, D. F. (1986) Dye sensitized photopolymerization. Adv.
Photochem. 13, 427–487.
31. Balasubramanian, D., X. Du and J. S. Zigler Jr. (1990) The
reaction of singlet oxygen with proteins, with special reference
to crystallins. Photochem. Photobiol. 52, 761–768.
32. Webster, A., D. Britton, A. Apap-Bologna and G. Kemp (1989)
A dye-photosensitized reaction that generates stable protein-pro-
tein crosslinks. Anal. Biochem. 179, 154–157.
33. Ledger, M. B. and G. Porter (1972) Primary photochemical pro-
cesses in aromatic molecules. Part 15. The photochemistry of
aromatic carbonyl compounds in aqueous solution. J. Chem.
Soc. Faraday. Trans. 1 68, 539–553.
34. Klessinger, M. and J. Michl (1995) Excited States and Pho-
tochemistry of Organic Molecules. VCH Publishers, New
York.
9. Belfield, K. D., X. Ren, E. W. Van Stryland, D. J. Hagan, V.
Dubikovsky and E. J. Miesak (2000) Near-IR two-photon pho-
toinitiated polymerization using a fluorone/amine initiating sys-
tem. J. Am. Chem. Soc. 122, 1217–1218.
10. Albota, M., D. BeLjonne, J.-L. Bredas, J. E. Ehrlich, J.-Y. Fu,
A. A. Heikal, S. E. Hess, T. Kogej, M. D. Levin, S. R. Marder,
D. McCord-Maughon, J. W. Perry, H. Rockel, M. Rumi, G.
Subramaniam, W. W. Webb, X.-L. Wu and C. Xu (1998) De-
sign of organic molecules with large two-photon absorption
cross sections. Science 281, 1653–1656.
11. Thayumanavan, S., J. Mendez and S. R. Marder (1999) Synthe-
sis of functionalized organic second-order nonlinear optical
chromophores for electrooptic applications. J. Org. Chem. 64,
4289–4297.
12. Cumpston, B. H., S. P. Ananthavel, S. Barlow, D. L. Dyer, J.
E. Ehrlich, L. L. Erskine, A. A. Heikal, S. M. Kuebler, I.-Y. S.
Lee, D. McCord-Maughon, J. Qin, S. R. Marder and J. W. Perry
(1999) Two-photon polymerization initiators for three-dimen-
sional optical data storage and microfabrication. Nature 398,
51–54.
35. Werts, M. H. V., M. A. Duin, J. W. Hofstraat and J. W. Ver-
hoeven (1999) Bathochromicity of Michler’s ketone upon co-
13. Pan, H., X. Gao, Y. Zhang, P. N. Prasad, B. A. Reinhardt and
R. Kannan (1995) A new class of heterocyclic compounds for
nonlinear optics. Chem. Mat. 7, 816–821.
ordination with lanthanide(III)
␤-diketonates enables efficient
ϩ
sensitization of Eu3 for luminescence under visible light exci-
14. Fittinghoff, D. N. and J. A. Squier (2000) Time-decorrelated
tation. Chem. Commun. 799–800.

144 Jonathan D. Pitts et al.
36. Gu, M. and C. J. R. Sheppard (1995) Comparison of three-
dimensional imaging properties between two-photon and single-
photon fluorescence microscopy. J. Microsc. 177, 128–137.
37. Ware, W. R., M. Pratinidhi and R. Bauer (1983) Performance
characteristics of a small side-window photomultiplier in laser
single-photon fluorescence decay measurements. Rev. Sci. In-
strum. 54, 1148–1156.
Bosshard and P. Gunter (1998) Structure-property relationships
in third-order nonlinear optical chromophores. J. Phys. Chem.
B 102, 4451–4465.
39. Xu, C., R. M. Williams, W. Zipfel and W. W. Webb (1996)
Multiphon excitation cross-sections of molecular fluorophore.
Bioimaging 4, 198–207.
40. Oster, G. (1954) Dye-sensitized photopolymerization. Nature
173, 300–301.
38. Tykwinski, R. R., U. Gubler, R. E. Martin, F. Diederich, C.