Photo-caged agonists of the nuclear receptors RARγ and TRβ provide unique time-dependent gene expression profiles for light-activated gene patterning

Article abstract of DOI:10.1016/j.bmc.2004.08.022

Light-activated gene expression systems hold promise as new tools for studying spatial and temporal gene patterning in multicellular systems. Photo-caged forms of nuclear receptor agonists have recently been shown to mediate photo-dependent transcription in mammalian cells, however, because intracellularly released agonists can rapidly diffuse out of cells, the photo-initiated transcription response is only transient and limited to only a few hours in reported examples. Herein we describe a photo-caged thyroid hormone receptor agonist that provides a robust 36 h transcription response to a single irradiation event. These findings are in contrast to a closely related system, which uses a caged retinoic acid receptor agonist, which provides only a short transcription response. Comparison of the two systems, show that the duration of transcription response is not controlled by the rate of diffusion of free ligand out of the cell, but perhaps by the duration of ligand-induced transcription/stability of the active transcription complex.

Full text of DOI:10.1016/j.bmc.2004.08.022

Bioorganic & Medicinal Chemistry 12 (2004) 5949–5959
Photo-caged agonists of the nuclear receptors RARc and
TRb provide unique time-dependent gene expression profiles
for light-activated gene patterning
Kristian H. Link, Federico G. Cruz, Hai-Fen Ye,à Kathryn E. OÕReilly, Sarah Dowdell
and John T. Koh*
Department of Chemistry and Biochemistry, University of Delaware, Newark, DE 19176, USA
Received 4 July 2004; revised 13 August 2004; accepted 13 August 2004
Available online 17 September 2004
Abstract—Light-activated gene expression systems hold promise as newtools for studying spatial and temporal gene patterning in
multicellular systems. Photo-caged forms of nuclear receptor agonists have recently been shown to mediate photo-dependent tran-
scription in mammalian cells, however, because intracellularly released agonists can rapidly diffuse out of cells, the photo-initiated
transcription response is only transient and limited to only a fewhours in reported examples. Herein we describe a photo-caged
thyroid hormone receptor agonist that provides a robust 36h transcription response to a single irradiation event. These findings
are in contrast to a closely related system, which uses a caged retinoic acid receptor agonist, which provides only a short transcrip-
tion response. Comparison of the two systems, show that the duration of transcription response is not controlled by the rate of dif-
fusion of free ligand out of the cell, but perhaps by the duration of ligand-induced transcription/stability of the active transcription
complex.
ꢀ 2004 Elsevier Ltd. All rights reserved.
1. Introduction
pathways.12–15 Recently, Haselton and later Tsien re-
ported a newstrategy to directly control gene expression
with light using photo-caged nucleic acids that are tran-
scriptionally (or translationally) silent until uncaged with
light.16,17 The application of such systems to multicellular
systems often are not practical due to the difficulties of
efficiently delivering of caged biopolymers into cells. A
newand promising light-activated transcription system
that uses plant phytochrome was recently reported by
Quail and co-workers.18 This system requires addition
or co-expression of the plant phytochrome and has thus
far only been demonstrated in yeast.
Many methods are currently available to remotely regu-
late the expression of specific genes and have provided
a powerful tool to elucidate the role of specific genes in-
volved differentiation, growth, and homeostasis.1–8 The
actions of many gene products critical for the develop-
ment and homeostasis are only revealed through their
unique spatial and temporal patterns of expression.9–11
Photo-regulation of signal transduction and/or gene
expression provides a powerful new tool for the study
of gene function that can provide both spatial and tempo-
ral control of gene expression. Several studies have dem-
onstrated that expression of some genes can be regulated
by light using photo-caged enzymes, enzyme substrates,
or enzyme inhibitors involved in signal transduction
Recently, we described a new approach to regulating
gene expression using photo-caged hormones, which tar-
get nuclear hormone receptors (NHRs). NHRs function
as ligand-dependent transcriptional regulators that di-
rectly bind DNA and up-regulate gene transcription in
response to binding small molecule hormones.19,20 Sev-
eral examples of chimeric receptors in which the DNA
binding domain of NHRs have been exchanged for
DNA binding domains from other NHRs or heterolo-
gous proteins allowone to target the actions of hor-
mones to virtually any transgene of interest. In
Keywords: Nuclear receptors; Light-activated gene expression; Gene
patterning; Photo-caged.
*
Corresponding author. Tel.: +1 302 831 1947; fax: +1 302 831
6335; e-mail: johnkoh@udel.edu
These authors contributed equally to this work.
à Present address: Adolor Corporation, 371 Pheonixville Pike, Mal-
vern, PA 19355, USA.
0968-0896/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2004.08.022

5950
K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
addition, ligand receptor engineering has been used to
create selective and functionally orthogonal agonist-
NHR pairs that can be used to independently regulate
gene transcription without effecting endogenous recep-
tors.21–26 Together these observations suggest that
NHRs represent a novel platform to construct selective
photo-inducible transcriptional regulators.
In our initial study a photo-caged form of estradiol was
used to mediate light-dependent expression of estrogen
responsive genes.27 Caged hormones (or NHR agonists)
readily diffuse across cell membranes and therefore have
many advantages over methods, which employ caged
biopolymers, which can be difficult to introduce into
multicellular systems. Caged hormone receptor agonists
have been used to spatially resolved patterns of reporter
gene expression,28 however, when hormones are locally
uncaged within multicellular systems, the duration of
transcription response is only transient, presumably be-
cause the released hormone ultimately diffuses out of the
cell. In the case of our previously reported nitroveratryl
estradiol, intracellular uncaging provides a relatively
weak (40% of max) and short duration reporter gene re-
sponse that is limited to a fewhours. Lawrence and co-
workers using a caged ecdysone and cells expressing the
ecdysone receptor obtained transient expression profiles
that were slightly longer and more intense but were still
of relatively short duration (P50% max expression for
2h).28 These short duration transcription responses are
often too short to provide a practical method to study
the effects of gene patterning without having to con-
stantly re-expose the expressed pattern.
Figure 1. Light-stable agonists and photo-caged analogs.
and calcitriol contain light-sensitive polyenes and trii-
odothyronine contains aryl iodides sensitive to photo-
isomerization or degradation. Therefore, we synthesized
3 and 4 as potential photo-caged analogs of the known
synthetic agonists of RAR and TR, 1 and 2 (GC-1)
(Fig. 1).33,34
2.2. Photo-caged agonists 3 and 4 efficiently release
photo-stable agonists in vitro
The stilbene 1 was first reported by Charpentier et al. as
a potent RARc agonist.33 Under our assay conditions
using transiently transfected HeLa cells, 1 has an EC50
of 467nM with our DR4-luc reporter. This compound
is relatively stable to irradiation with a fluorescent labo-
ratory UV lamp showing less than 1% decomposition
after 180s of irradiation. Similarly, the potent halogen-
free thyromimetic GC1,34 2 (EC50 = 48nM), shows less
than 11% decomposition after 180s of irradiation,
demonstrating that 1 and 2 are stable to conditions of
photo-deprotection. The caged analogs 3 and 4 were
easily synthesized from the previously reported com-
pounds 1 and 7 (Scheme 1),33,34 and were confirmed to
contain less than 0.2% of the parent agonists by HPLC.
Although it is difficult to directly compare the duration
of gene response between these two studies, which in-
volve different cell types and different reporter gene
constructs, it is intriguing to consider that other hor-
mone-receptor pairs might be able to provide a longer
and more robust transcription response to transient
photo-patterning. Toward this goal we have studied
the properties of caged retinoic acid receptor (RAR)
and thyroid hormone receptor (TR) agonists in closely
related experimental systems using the same cell line
and exploiting the well-known phenomenon that RAR
and TR can both regulate reporter gene expression from
the identical DR4 hormone response element.29–32
The caged compounds 3 and 4 were photo-deprotected
under cell free conditions to determine the rate and effi-
ciency with which caged compounds were converted to
agonist products upon irradiation. Methanolic solutions
of compounds 3 and 4 (100lM) were placed in polystyr-
ene cell culture plates and exposed in an analogous
manner used in our cell culture experiments (see Exper-
imental). HPLC analysis confirms that 3 and 4 efficiently
liberate the agonists 1 and 2 with greater than 90% con-
version eꢀffi2ciency. However, the rate of deprotection of 4
2. Results
2.1. Photo-stable agonists for TR and RAR
In previous work we demonstrated that a photo-caged
form of estradiol could be used to control transactivation
function from the steroid hormone receptor, estrogen
receptor (ER) in an exposure dependent manner.27 In
this study we sought to examine the characteristic prop-
erties of light-activated gene expression mediated by two
members of the nuclear receptor family, which activate
gene transcription in a related mechanism that is distinct
from that of steroid hormone receptors. Coincidentally,
the natural ligands for the nuclear receptors, TR,
RAR, RXR, and VDR, are light sensitive. Retinoic acid
(1.1 · 10
s
ꢀ1) is three times slower than that of 3
(3.1 · 10ꢀ2 sꢀ1) such that only 20% of 4 is converted to
2 during a 90s exposure (Fig. 2).
2.3. Photo-caged compounds 3 and 4 are stable under
cellular conditions
The successful application of 3 and 4 to mediate light-
activated gene transcription requires that our photo-

K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
5951
Scheme 1. Synthesis of caged agonists 3 and 4. Reagents and conditions: (a) H2SO4, EtOH, reflux (86%), (b) K2CO3, nitrovarearyl bromide, DMF
(91%), (c) LiOH, THF/H2O (75%), (d) NaOH, rt (95%).
graded nor prematurely uncaged under intracellular
conditions (Fig. 3).
2.4. Photo-caged analogs 3 and 4 do not act as agonists or
antagonists with receptors below 1lM
The successful control of gene expression using caged
agonists requires that the caged compounds do not act
as agonists prior to irradiation. HeLa cells transiently
transfected with reporter vector and either pSG5hRARc
or pSG5hTRb were treated with increasing concentra-
tions of caged agonists 3 and 4. Only a small increase
in reporter gene expression is observed at concentrations
at or below1 lM demonstrating that introduction of the
nitroveratryl group onto the phenol hydroxyl of 1 or 2
Figure 2. Exposure dependence of photo-decomposition and product
formation. Compounds 3 (j) and 4 (d) are converted with greater
effectively blocks (or reduces) the potent agonist proper-
ties of the parent agonists (Fig. 4). At high concentra-
tions, greater than 2500nM, of 3 and 4, a more
significant increase in reporter gene activity is observed
suggesting that the caged analogs are possibly very weak
agonists or perhaps more likely contain trace quantities
of active agonists 1 or 2. High concentrations of 3 (3000
and 4000nM) were observed to cause visible signs of
toxicity. Therefore, caged agonists were used at or below
concentrations of 1000nM.
than 90% efficiency to their parent agonist forms 1 (h) and 2 ( ) but at
ꢁ
different rates.
caged ligands be stable under cellular conditions, only
releasing active ligand upon irradiation. To test the cel-
lular stability of our photo-caged compounds, we simu-
lated intracellular conditions by placing our compounds
in HeLa cell lysates for 36h at 37ꢁC. Because even trace
quantities of the active ligand could activate gene tran-
scription, solutions of the lysate-treated compounds
were evaluated for their ability to activate reporter gene
expression in cell based transactivation assays with cells
expressing their corresponding receptors. Cells grown in
the presence of lysate-treated compounds 3 and 4 did
not showa significant increase in gene expression com-
pared to cells grown in the presence of untreated com-
pounds at the same concentration. In addition, cells
grown in the presence of either lysate-treated or un-
treated caged compounds showessentially equivalent re-
porter gene expression after irradiation with light,
indicating that the caged compounds were neither de-
Compounds 3 and 4 were also evaluated for their ability
to act as potential antagonists in competition with the
parent agonists 1 and 2. HeLa cells transiently trans-
fected with either pSG5hRARc or pSG5hTRb were
grown in the presence of either 800nM of agonist 1 or
100nM agonist 2 were treated with increasing concen-
trations of caged analogs 3 and 4. No change in reporter
gene expression was observed below 1000nM indicating
that the caged compounds do not act as effective anta-
gonists below1 lM (Fig. 5). In vitro radio-ligand com-
petition binding assays confirmed that compound 3
has an apparent Kd greater than 8mM.

5952
K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
Figure 3. Treatment of caged analogs 3 and 4 with cell lysates does not effect transactivation function in unexposed cells (white bars) or irradiated
cells (black). (A) HeLa cells transiently transfected with pSG5hRAR and 3. (B) Cells transiently transfected with pSG5hTRb and treated with 4.
Figure 4. Compounds 3 and 4 showalmost no transcriptional activity below1000nM in cells expressing (A) RAR or (B) TR.
Figure 5. Caged compounds 3 and 4 do not antagonize RAR or TR dependent transactivation. (A) Cells expressing RAR treated with 800nM 1 and
increasing concentrations of 3. (B) Cells expressing TR treated with 100nM 2 and increasing concentrations of 4.

K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
5953
2.5. Caged ligands 3 and 4 can mediate exposure
dependent RAR and TR dependent transactivation
response
tially attaining the maximum possible gene expression
with a 90s exposure.
2.6. Intracellular uncaging provides a transient transac-
tivation response
A fluorescent UV lamp (kmax = 365nm), described previ-
ously, was used to irradiate cells directly through the
polystyrene lid of a standard cell culture plate, which
acts as a filter for potentially damaging short wavelength
UV (<345nm).27 Cells transiently transfected with re-
porter and either pSG5hRARc or pSG5hTRb grown
in the absence of ligand did not display any exposure
dependent reporter gene expression. Exposure times
greater than 5min were accompanied by some visible
changes in cell morphology or viability. Therefore, sub-
stantially shorter exposure times of 90s or less were used
to avoid undesirable effects of prolonged irradiation.
The application of photo-caged hormone receptor agon-
ists to form spatially discrete patterns of gene expression
may be limited by the diffusion of ligand out of cells
after being intracellularly uncaged. To simulate the con-
ditions expected when compounds are uncaged intracell-
ularly in only a local subpopulation of cells, HeLa cells
were grown in six-well plates with media containing
either the caged compounds 3 or 4 for 12h. Just prior
to irradiation, the media was removed; the cells were
washed twice with PBS buffer and the media replaced
with fresh media without caged agonist such that the
only caged compound present during irradiation is
intracellular. The fluorescent property of the parent
agonist 1 allows us to directly visualize the likely cellular
localization of structurally related caged analogs 3 and
4. Cells grown in the presence of 1, showintense intra-
cellular fluorescence demonstrating that such com-
pounds are likely localized in the cytoplasm and not
strongly associated with the membrane or cell surface
after media exchange (Fig. 7).
Cells transiently transfected with pSG5hRARc or
pSG5hTRb were grown in the presence of 800nM 3 or
500nM 4, such that a 90s exposure would be expected
to liberate 1 or 2 at a concentration approximately twice
the EC50 of these ligands for their respective receptors.
Both ligands showan exposure dependent production
of the reporter gene in cells expressing their corresponding
receptors (Fig. 6). Compound 3 mediates RAR-depend-
ent reporter gene expression, approaching a 4-fold induc-
tion of reporter gene expression or 90% of the maximum
inducible expression level with a 90s exposure. Although
maximal level of gene expression mediated by 3 is large,
the induction of gene expression relative to unexposed
cells is relatively modest, 4-fold induction, and is largely
limited by the significant background activity observed
in the absence of exposure. Cells treated with 500nM 4
showexposure dependent TR-mediated reporter gene
expression corresponding to a 24-fold induction, essen-
HeLa cells pretreated with 1000nM 3 and ÔwashedÕ prior
to irradiation for 90s, showa unique temporal depend-
ence of gene expression. After exposure, luciferase activ-
ity increases over the initial 4h at a rate similar to cells
treated with 3 but not washed prior to exposure. Gene
product increases to a maximum at 6h (54%), and then
decreases presumably as the result of the loss of intracel-
lular ligand due ultimately to diffusion (Fig. 8A). The
Figure 6. Reporter gene activity determined 24h after irradiation is exposure and ligand dependent. (A) Cells expressing hRARc with 800nM 3 (Ç)
or with no added ligand (h). (B) Cells expressing hTRb with 500nM 4 (Ç) or with no added ligand (h).

5954
K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
reaches almost 80% the maximum expression level in
12h and then decreases. Transcription is sustained over
50% of the maximum level for over 36h, substantially
longer than has been attained for any caged hormone
studied thus far (e.g., ER, RAR, or EcR) (Fig.
8B).27,28 These studies showthat dramatically different
time-dependent transcription profiles can be attained
by different nuclear receptors even when using the same
cell line and reporter gene. Although 100% of the max-
imal expression level is not attained, it is important to
recognize that genes are generally not expressed at their
maximal levels and the similar profiles of time-depend-
ent gene expression are observed in many cellular proc-
esses such as cell division or differentiation.35 Therefore,
the light-activated transcription system composed of 3
and TR provides a physiologically significant expression
response to a single, exposure event offering substantial
practical advantage over existing systems. Furthermore,
caged nuclear receptor agonists can provide unique
time-dependent transcription profiles that are distinct
from those expected from direct photo-activation of
caged DNA or caged proteins involved in signal trans-
duction pathways.
Figure 7. Fluorescence micrograph of HeLa cells incubated with 1 and
observed under upright microscope with dipping lens.
duration of transcription response, measured as the
length of time gene product is 50% of the maximum level
or greater, is limited to approximately 2h. This is com-
parable with the duration of transcription response pre-
viously reported for ER and EcR mediate gene
patterning,27,28 therefore, similar to ER and EcR, this
light-activated RAR system likely does not provide a
practical method to pattern expression in multiple tis-
sues without having to constantly rephoto-pattern the
system.
3. Discussion
3.1. The magnitude and duration of transactivation
response may be dependent upon the duration of active
transcription
2.7. Cells expressing TR provide a long duration response
HeLa cells expressing TR and pretreated 500nM 4 dis-
play a unique expression profile that differs substantially
from that observed with RAR and 3 in both intensity
and duration of gene response. In this case reporter gene
The caged agonists 3 and 4 provide exposure dependent
RAR and TR mediated reporter gene expression cul-
tured cells. This exposure dependence, suggests that
light-activated gene expression using hormone receptors
Figure 8. Time dependence of luciferase formation after 90s exposure. (A) Cells expressing RARc pretreated with 800nM 3. (B) Cells expressing
TRb pretreated with 4; (j) ligand uncaged in media and cells; (d) ligand uncaged in cells only after prewashing; ( ) not irradiated.
ꢁ

K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
5955
may be used to create both spatially resolved binary pat-
terns and potentially gradients of inducible gene prod-
ucts in multicellular systems.
rate of free ligand diffusion out of the cell, which would
be expected to be very fast, on the order of 2 · 105 sꢀ1
for diffusion across a lipid bilayer.36 Furthermore, agon-
ists 1 and 2, though not identical, are structurally very
similar, and would be expected to diffuse across mem-
branes at comparable rates, yet they exhibit markedly
different time dependent transcription profiles.
For these studies, the identical reporter gene construct
and cell line was used to measure both RAR and TR
mediated gene expression. These systems provides a un-
ique opportunity to explore the factors that may lead to
different time-dependent transcription profiles that are
independent of cell type or the nature of the reporter
gene. Presumably, gene product formation triggered by
intracellularly released hormone receptor agonist stops
as the result of the loss of intracellular ligand. However
the duration of reporter gene response obtained with
GC1 and TR is too long to be attributed simply to the
In an attempt to better understand the factors leading to
the different time-dependent transcription responses
from RAR and TR, a mathematical model for transcrip-
tion/translation developed by Hargrove and Schmidt
was fit to our two systems.37 The kinetic parameters
governing the RAR and TR systems are significantly
underdetermined. However, relative and qualitative
Scheme 2. The formation of protein product is represented as an intermediate between transcription/translation and protein degradation. The
duration of transactivation response is limited by the decomposition of the receptor associated transcription complex (kd0).
Figure 9. Transcription/translation models fit to observed time-dependent transcription profiles using the equation of Hargrove and Schmidt (j,
solid lines): (A) RAR with 8, ks1 = 0.34pmol/g/d, ks2 = 10,000pmol/d/pmol RNA, kd1 = 2.77dꢀ1, kd2 = 5.54dꢀ1, P0 = 114pmol/g, R0 = 0.08pmol/g,
and R2 = 0.93. (B) TR with 5, ks1 = 0.51pmol/g/d, ks2 = 10,000pmol/d/pmol RNA, kd1 = 2.77dꢀ1, kd2 = 5.54dꢀ1, P0 = 13.9pmol/g, R0 = 0.18pmol/g,
and R2 = 0.99; Intracellular uncaging fit as first order exponential decay of the transcription complex (d, dashed line): (A) RAR with 8,
kd0 = 0.18dꢀ1 (half-life = 3.14dꢀ1), a = ꢀ0.66pmol/g, ks1 = 0.34pmol/g/d, ks2 = 10,000pmol/d/pmol RNA, kd1 = 2.77dꢀ1
, ,
kd2 = 5.54dꢀ1
P0 = 92.7pmol/g, R0 = 0.08pmol/g, and R2 = 0.71. (B) TR with 5, kd0 = 0.025dꢀ1 (half-life = 0.27dꢀ1), a = ꢀ0.36pmol/g, ks1 = 0.51pmol/g/d,
ks2 = 10,000pmol/d/pmol RNA, kd1 = 2.77dꢀ1, kd2 = 5.54dꢀ1, P0 = 13.9pmol/g, R0 = 0.18pmol/g, and R2 = 0.99.

5956
K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
insight into the functions controlling the duration of re-
porter gene expression in response to a transiently intro-
duced stimulus can be approximated, using literature
values of mRNA and protein half-lives because the
mRNA and protein products are identical for both
systems.
the cell.40–49 As noted above, the diffusion of ligand
out of the cell is clearly not controlling the rate of
decay of active transcription because the rate of diffu-
sion of free ligand out of the cell is expected to be
approximately 10-orders of magnitude faster than the
modeled rate of decomposition. This suggests that that
the rate of loss of active transcription complex, perhaps
via proteolysis or chaperone mediated disassembly of
the transcription complex and/or ligand off-rate likely
control the rate of transcription loss and ultimately
the duration of reporter gene response to intracellularly
released hormone.
It has been previously demonstrated, using a two com-
partment mathematical model, that the amount of ac-
tive protein P(t), is represented as an intermediate
between mRNA and protein products (Scheme 2). This
model shows that the maximal amount of gene product
formed, the Ôfold inductionÕ, and the half-life for forma-
tion of protein are critically dependent on the mRNA
and protein stabilities, in addition to the rates of tran-
scription and translation.
4. Conclusion
Newmethods to create and control spatial and temporal
patterns of expressed gene products may serve as a pow-
erful tool to elucidate the function of the many genes,
which elicit their function only through their unique spa-
tial and temporal patterns of expression. Light-activated
gene expression using caged agonists of NHRs provides
several unique advantages over current methods to
photo-pattern gene expression in multicellular systems,
however, the application of this method to some systems
is potentially limited by the relatively short duration of
expression response reported thus far. It is nowdemon-
strated under similar experimental conditions that the
duration of transcription response is receptor depend-
ent. Under identical conditions RAR shows a relatively
short duration response whereas thyroid hormone
receptor agonist that provides a robust 36h reporter
gene response to a single irradiation event. This is far
longer (16 times) than any other system NHR-depend-
ent system thus far. Comparison of the RAR and TR
dependent systems showthat the duration of transcrip-
tion response is not controlled by the rate of diffusion of
free ligand out of the cell, but rather by the rate of tran-
scription loss by the inactivation of the active transcrip-
tion complex.
Since the same DR4-luciferase reporter was used in the
TR and RAR studies, ks2, kd1, kd2 should be identical.
Thus the rate of transcription, ks1, could be responsible
for the differences in the temporal responses for TR and
RAR. The rates of transcription were determined by fit-
ting the data for the intra and extracellular deprotection
experiments of RAR and TR to the mathematical model
of Hargrove and Schmidt, using literature values for the
half-life of luciferase mRNA and protein, and a generic
rate of protein synthesis.37–39 Our experimental data fits
reasonably well to this model. The results obtained from
the regression analysis shows that the rate of transcrip-
tion of RAR is approximately 66% of TR based on
the maximum absolute luciferase activity of RAR is
¼
ks1ks2
approximately 60% that of TR, and Pmax
9, solid lines).
(Fig.
kd1kd2
In this study, we have selectively released the agonists 1
and 2 intracellularly by removing the extracellular caged
compound 3 and 4 prior to irradiation. Under identical
conditions, it was determined that the duration and
magnitude of the reporter gene response was much
greater with TR + 2 then RAR + 1. In an attempt to
gain further insights into the molecular events responsi-
ble for controlling these differences in reporter gene re-
sponse, the intracellular only deprotection was fit to a
modified form of the Hargrove and Schmidt equation,
where active transcription was to a first approximation
modeled as a first order exponential decay, kd0, while
keeping the kinetic parameters derived above constant.
The results of the regression analysis of the intracellular
only deprotection of TR + 2 and RAR + 1, showthat
the rate of decay of active transcription by RAR is
approximately seven times faster than that of TR, which
is 3 · 10ꢀ7 sꢀ1. Using this model, it can further be shown
that the different rates of transcription, ks1 observed for
TR and RAR is insufficient to account for the very large
difference in the duration of transcription response (data
not shown).
5. Experimental
5.1. General
All compounds were purchased from Aldrich chemical
company unless otherwise noted. NMR spectra were re-
corded on a Bruker DRX-400 spectrometer. Chroma-
˚
tography was performed using ICN SiliTech (60A)
Flash Silica. HeLa cells were obtained from ATCC
(American Type Culture Collection) maintained at the
University of Delaware Biology Core Facility. Lucif-
erase reporter gene activity of transiently transfected
HeLa cells was determined using the Dual-Luciferase
Reporter Assay System (Promega #E1960) following
the manufacturerÕs protocol. DR4-Luc+ is derived from
pGL3 (Promega), an enhanced luciferase promotor by
inserting three nestled copies of the DR4 hormone
response element (TGACCTAAAATGACCTAAAAT-
There are several potential molecular events that may
account for the differences in the apparent rate at
which transcription decreases for the RAR and TR
based systems including disassembly/decomposition of
the transcription complex, the off-rate of the ligand
from the receptor and diffusion of the ligand out of
GACCTAAAATGACCT) upstream of
thymidine kinase promotor (tk) corresponding to bases
a minimal
ꢀ113 to +31.

K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
5957
5.2. Transient transfixion assays (general)
tRA for 16h at 4ꢁC. The protein was then absorbed
onto hydroxyapatite (0.5mL) for 30min on ice, washed
four times with 0.8mL of wash buffer (10mM Tris
(pH7.5) ꢀ 0.5% tritron · 100). The hydroxyapatite pel-
let was then extracted two times with 0.8mL of 2:1
methanol/chloroform and dissolved in 15mL of scintil-
lation fluid. The data was fit to the following equation
by non-linear regression analysis using GraphPad InPlot
(Graphpad software).
Twenty-four hours before transfection, HeLa cells were
seeded at a density of 40,000 cells per well in six-well cul-
ture plates (Costar# 3516) and grown in DulbeccoÕs
Modified Eagle Medium, DMEM (without phenol red)
supplemented with 10% fetal bovine serum (FBS; Celle-
gro, lot# SBF30-1126) and gentamycin. Three hours
prior to transfection, the media was changed to DMEM
containing 10% charcoal-resin stripped FBS. FBS was
depleted of endogenous hormones according to the pro-
cedure of Samuels et al.50 Transfections were performed
by calcium phosphate method using pSG5hRARc or
pSG5hTRb as receptor DNA, DR4-Luc+ (reporter),
and pRLCMV (control). Twenty hours after the trans-
fection, the media was removed and replaced with media
containing an appropriate concentration of ligand. All
manipulations involving compounds 2 and 4 were per-
formed in the dark with the use of a photographic dark-
room safelight. The cells were allowed to incubate in
media containing caged ligand for 12h prior to exposure
(see Exposure Protocol). After exposure the cells were
allowed to incubate for 24h before harvesting by passive
lysis. Cell extracts were immediately assayed using the
Dual Luciferase Assay (Promega) using a Dynex lumi-
nometer. Bioluminescence activity is reported in relative
light units (RLU) determined as the ratio of the firefly
luminescence divided by the luminescence of the renilla
luciferase control.
5.5. Intracellular deprotection by prewashing
In order to determine if ligand was being deprotected in-
side cells, HeLa cells were transiently transfected with
respective receptor, reporter, and control. The cells were
allowed to incubate for 12h in the presence of active
agonists or their photo-caged analogs. The cells were
then rapidly washed twice with PBS buffer, which was
replaced with ligand-free media. Immediately after the
media was changed, the cells were irradiated with 30
and 180s to liberate any photo-caged agonists. The cells
were allowed to incubate for 0–48h before harvesting by
passive lysis.
5.6. Evaluation of compound stability to cell lysates
In order to determine if the ligands were being depro-
tected in the cells in the absence of light exposure, com-
pounds were preincubated in cell lysate and assayed for
activity. Solutions of 3 and 4 (0.2lM), in media contain-
ing cell lysates from 1.5 · 107 HeLa cells, was allowed to
incubate for 36h at 37ꢁC. These lysated-treated ligand
solutions were diluted to appropriate concentrations in
media and added to transfected cells. These cells were
compared with transfected cells grown in the presence
of 3 and 4 that were not pretreated with cell lysates.
5.3. Cell exposure procedure
Cells growing in six-well culture plates were placed be-
lowa UV lamp (Spectroline model XX-15A; long wave
UV 365nm). The samples were irradiated for 10–300s
through a 4mm thick glass plate and the polystyrene
culture plate lid, which acted as a filter for short wave
length UV. The intensity was determined to be 400mW.
5.7. Synthesis
5.7.1. Ethyl 4-[(E)-2-[3-(1-adamantyl)-4-hydroxyphen-
yl]ethenyl]-benzoate (5). To a solution of 1 (0.45g,
1.2mmol) in 5mL of ethanol was slowly added
0.064mL of concentrated sulfuric acid (98%). The solu-
tion was refluxed for 12h, neutralized with sodium
bicarbonate, and extracted 2 · 10mL ether. The com-
bined organic extracts were dried over MgSO4, and
evaporated in vacuo to afford 0.416g (85.6%) of a yellow
solid.
5.4. Binding assay
The ligand binding domain of RARc, residues 178–423,
was overexpressed in E. coli as a N-terminal his6 tag in
pER15b following a modification of the procedure of
Moras and co-workers.51 BL21(DE3) cells grown at
37ꢁC to an OD600 of 0.6–0.7, were induced with 1mM
IPTG and incubated at 25ꢁC for 3.5h. Collected cell pel-
lets were resuspended in lysis buffer (Tris–HCl (pH8.0),
5mM imidazole, 500mM NaCl, 0.5mM PMSF, 2lg/
mL Leupeptin) french pressed twice, sonicated 3 · 10s
(setting 3, Branson sonicator with microprobe), and
incubated 20min at 0ꢁC. Centrifugation (10,000rpm,
SS-34 rotor) produced crude soluble cell lysates, which
were stored in 10% glycerol at ꢀ20ꢁC. Ligand binding
assays were carried out following the procedure of
Wecksler and Norman.52 Briefly, RARc (LBD) was di-
luted to a concentration of approximately 2nM (active
protein) in binding buffer (10mM Tris (pH8.0), contain-
ing EDTA (15mM), DTT (2mM), Leupeptin (2lg/mL),
and 10% glycerol) and incubated with [20-methyl-3H]
tRA with various concentrations of unlabeled competi-
tor, in the absence or presence of 100-fold excess cold
1H NMR (400MHz, CDCl3): d 1.38 (t, 3H), 1.78 (s, 6H),
2.12 (S, 9H), 4.39 (q, 2H), 6.80 (d, J = 8Hz, 1H), 6.93 (d,
J = 16Hz, 1H), 7.16 (d, J = 16Hz, 1H), 7.22 (dd, J = 8,
2Hz, 1H), 7.34 (d, J = 2Hz, 1H), 7.51 (d, J = 8Hz, 2H),
8.01 (d, J = 8Hz, 2H); 13C NMR (100MHz, CDCl3): d
166.50, 164.95, 159.15, 142.34, 138.75, 131.51, 129.94,
128.65, 125.91, 125.41, 125.37, 124.99, 111.87, 55.10,
40.53, 37.86, 29.07, 14.37. MS (FAB) calcd for
C27H30O3 (MH+) 402.2165, found: 402.1147.
5.7.2. Ethyl 4-[(E)-2-[3-(1-Adamantyl)-4-[(nitroveratr-
yl)oxy]phenyl]ethenyl]-benzoate (6). To a suspension of
5 (0.100g, 0.25mmol) and K2CO3 (0.069g, 0.50mmol)
in 1mL of dry DMF was added dropwise a solution of

5958
K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
(0.069g, 0.25mmol) nitroveratryl bromide in 1mL of dry
DMF. The resulting solution was allowed to stir for 12h
and extracted with 2 · 10mL ether. The combined or-
ganic extracts were dried over MgSO4, evaporated in va-
cuo, and purified by flash chromatography hexanes/ethyl
acetate (50:50) to afford 0.125g (83.1%) of a yellowsolid.
5.7.5.
benzyl)phenoxyl]acetic acid (4). To
[3,5-Dimethyl-4-(30isopropyl-40-nitroveratryl-
solution of
a
0.0628g (2.07mmol) of 8 in 5mL of methanol was added
1mL of 1N NaOH. The reaction mixture was allowed to
stir for 1h at ambient temperature and then acidified
with 1N HCl (10mL) and extracted 2 · 10mL ether.
The combined organic extracts were dried over MgSO4
and concentrated in vacuo. The resulting residues was
purified by flash chromatography hexane/ethyl acetate
(90:10) to yield 4 (95.2%) as a white crystalline.
1H NMR (400MHz, CDCl3): d 1.40 (t, 3H), 1.78 (s, 6H),
2.11 (s, 9H), 3.87 (s, 3H), 3.99 (s, 3H), 4.39 (q, 2H), 5.63
(s, 2H), 6.81 (d, J = 8.5Hz, 1H), 6.99 (d, J = 16.3Hz,
1H), 7.18 (d, J = 16.3Hz, 1H), 7.26 (s, 1H), 7.28 (dd,
J = 8, 2Hz, 1H), 7.46 (d, J = 2Hz, 1H), 7.54 (d,
J = 8.4Hz, 2H), 7.80 (s, 1H), 8.01 (d, J = 8.3Hz, 2H);
1H NMR (400MHz, CDCl3): d 1.55 (J = 8.10Hz, 6H),
2.21 (s, 6H), 3.40 (qt, J = 8.21Hz, 1H), 3.92 (s, 2H),
3.98 (d, 6H), 4.50 (s, 2H), 5.44 (s, 2H), 6.62 (s, 1H),
6.69 (d, J = 8.24Hz, 1H), 6.80 (d, J = 8.35Hz, 1H),
6.98 (s, 1H), 7.48 (s, 1H), 7.78 (s, 1H); 13C NMR
(100MHz, CDCl3): d 169.10, 168.52, 156.04, 152.70,
141.16, 138.61, 137.60, 133.37, 129.48, 128.64, 126.10,
125.60, 114.20, 112.5, 109.4, 107.9, 34.01, 29.29, 28.28,
27.18, 20.73.
13C NMR (100MHz, CDCl3):
d 169.10, 166.50,
164.95, 162.3, 159.15, 147.7, 142.34, 141.1, 138.75,
131.51, 129.94, 128.65, 125.91, 125.41, 125.37, 124.99,
111.87, 109.40, 107.9, 81.9, 55.10, 40.53, 37.86, 29.07,
14.37. MS (FAB) calcd for C36H40O8N (MH+)
614.2729, found: 614.0856.
5.7.3. [(E)-2-[3-(1-Adamantyl)-4-[(nitroveratryl)oxy]phe-
nyl]ethenyl]-benzoic acid (3). A solution of 6 (0.100g,
0.25mmol) and 0.069g (0.12mmol) of LiOH in 2mL
THF/H2O (4:1) was allowed to react for 6h at ambient
temperature. The reaction mixture was extracted
3 · 10mL ether. The combined organic extracts were
dried over MgSO4, evaporated in vacuo, and purified
by flash chromatography hexanes/ethyl acetate (40:60)
to afford 0.075g (75.1%) of a yellowsolid.
6. Non-linear regression analysis
Time-dependent reporter gene expression was fit to the
Hargrove and Schmidt model for transcription/transla-
tion. Protein expression based on the model shown in
Scheme 2, is given by
ꢀ
ꢁ
ks1ks2
kd1kd2
ks1ks2
kd1kd2
d2t
PðtÞ ¼
ꢀ
ꢀ P0 eꢀk
1H NMR (400MHz, CDCl3): d 1.78 (s, 6H), 2.11 (s,
9H), 3.87 (s, 3H), 3.99 (s, 3H), 5.63 (s, 2H), 6.81 (d,
J = 8.5Hz, 2H), 6.99 (d, J = 16.3Hz, 1H), 7.18 (d,
J = 16.3Hz, 1H), 7.26 (s, 1H), 7.28 (dd, J = 8, 2Hz,
1H), 7.46 (d, J = 2Hz, 1H), 7.54 (d, J = 8.4Hz, 2H),
7.80 (s, 1H), 8.01 (d, J = 8.3Hz, 2H); 13C NMR
(100MHz, CDCl3): d 169.10, 166.50, 164.95, 162.3,
159.15, 147.7, 142.34, 141.1, 138.75, 131.51, 129.94,
128.65, 125.91, 125.41, 125.37, 124.99, 111.87, 109.40,
107.9, 81.9, 40.53, 37.86, 29.07. MS (FAB) calcd for
C34H35O8N (MH+) 585.2453.
ꢀ
ꢁꢀ
ꢁ
d2t
d1t
ks1
eꢀk ꢀ eꢀk
ꢀ ks2
ꢀ R0
kd1
kd1 ꢀ kd2
Identical values for ks2, kd1, kd2 were used in multiple
iterations to attain a self-consistent set of starting
mRNA values for both TR and RAR data using non-
linear regression analysis (Kaleidagraph, Synergy Soft-
ware). Values for ks1 were derived for RAR and TR
(0.32 and 0.51pmol/g/d, respectively).
The decomposition of the transcription complex that re-
sults from intracellular—only deprotection was to a first
approximation modeled as a first order exponential de-
cay using a modified form of the Hargrove and Schmidt
equation.
5.7.4.
[3,5-Dimethyl-4-(30isopropyl-40-nitroveratryl-
benzyl)phenoxyl]tert-butylacetate (8). To a solution of
0.0461g (0.12mmol) of 7, and 0.0248g (0.18mmol) of
K2CO3 in 10mL of DMF was added 0.0331g
(0.12mmol) of nitroveratryl bromide. The resulting
solution was allowed to stir at room temperature for
24h. The reaction mixture was extracted 3 · 10mL
methylene chloride and dried over MgSO4 and concen-
trated in vacuo. Purified by flash chromatography hex-
ane/ethyl acetate (80:20) to yield 0.628g (91.1%) as a
white crystalline.
ꢀ
ꢁ
d0t
d0t
ks1ðeꢀk þ aÞks2
ks1ðeꢀk þ aÞks2
d2t
PðtÞ ¼
ꢀ
ꢀ P0 eꢀk
kd1kd2
kd1kd2
ꢀ
ꢁꢀ
ꢁ
ks1ðeꢀk þ aÞ
eꢀk ꢀ eꢀk
d0t
dst
d1t
ꢀ ks2
ꢀ R0
kd1
kd1 ꢀ kd2
Non-linear regression analysis using the parameters de-
rived above, were used to determine the apparent first
order rates of decay for RAR and TR (0.18dꢀ1 and
0.025dꢀ1, respectively).
1H NMR (400MHz, CDCl3): d 1.27 (s, 9H), 1.55 (d,
J = 8.10Hz, 6H), 2.21 (s, 6H), 3.40 (qt, J = 8.35Hz,
1H), 3.92 (s, 2H), 3.98 (d, 6H), 4.50 (s, 2H), 5.44 (s,
2H), 6.62 (s, 1H), 6.69 (d, J = 8.24Hz, 1H), 6.80 (d,
J = 8.35Hz, 1H), 6.98 (s, 1H), 7.48 (s, 1H), 7.78 (s,
1H); 13C NMR (100MHz, CDCl3): d 169.10, 168.52,
156.04, 152.70, 141.16, 138.61, 137.60, 133.37, 129.48,
128.64, 126.10, 125.60, 114.20, 112.5, 109.4, 107.9, 82,
55, 34.01, 29.29, 28.28, 27.18, 23.01, 20.73.
Acknowledgements
Expression vector PSG5hRARc was kindly provided by
Professor Pierre Chambon (IGBMC). This work was
supported by National Institutes of Health Grant R01

K. H. Link et al. / Bioorg. Med. Chem. 12 (2004) 5949–5959
5959
27. Cruz, F. G.; Koh, J. T.; Link, K. H. J. Am. Chem. Soc.
2000, 122, 8777–8778.
28. Lin, W. Y.; Albanese, C.; Pestell, R. G.; Lawrence, D. S.
Chem. Biol. 2002, 9, 1347–1353.
29. Costet, P.; Lalanne, F.; Gerbod-Giannone, M. C.; Molina,
J. R.; Fu, X.; Lund, E. G.; Gudas, L. J.; Tall, A. R. Mol.
Cell. Biol. 2003, 23, 7756–7766.
30. Ishii, H.; Tezuka, T.; Ishikawa, H.; Takada, K.; Oida, K.;
Horie, S. Blood 2003, 101, 4765–4774.
DK54257, R01NS49523 and a focused giving grant
from Johnson and Johnson. F.G.C. was supported by
a research supplement for underrepresented minorities
from the National Institutes of Health, DK54257-
01S1. K.E.O. was supported in part by the Undergrad-
uate Sciences Education Program through a grant from
the Howard Hughes Medical Institute.
31. Ijpenberg, A.; Jeannin, E.; Wahli, W.; Desvergne, B. J.
Biol. Chem. 1997, 272, 20108–20117.
References and notes
32. Sanguedolce, M. V.; Leblanc, B. P.; Betz, J. L.; Stunnen-
berg, H. G. Embo J. 1997, 16, 2861–2873.
33. Charpentier, B.; Bernardon, J.-M.; Eustache, J.; Millois,
C.; Martin, B.; Michel, S.; Shroot, B. J. Med. Chem. 1995,
38, 4993–5006.
34. Chiellini, G.; Apriletti, J. W.; al Yoshihara, H.; Baxter, J.
D.; Ribeiro, R. C.; Scanlan, T. S. Chem. Biol. 1998, 5,
299–306.
1. Yarranton, G. T. Curr. Opin. Biotechnol. 1992, 3, 506–511.
2. Gossen, M.; Bonin, A. L.; Bujard, H. Trends Biochem. Sci.
1993, 18, 471–475.
3. Allgood, V. E.; Eastman, E. M. Annu. Rep. Med. Chem.
1997, 32, 231–239.
4. Allgood, V. E.; Eastman, E. M. Curr. Opin. Biotechnol.
1997, 8, 474–479.
5. Damke, H.; Gossen, M.; Freundlieb, S.; Bujard, H.;
Schmid, S. Methods Enzymol. 1995, 257, 209–220.
6. No, D.; Yao, T. P.; Evans, R. M. Proc. Natl. Acad. Sci.
U.S.A. 1996, 93, 3346–3351.
7. Werstuck, G.; Green, M. R. Science 1998, 282, 296–298.
8. Spencer, D. M.; Wandless, T. J.; Schreiber, S. L.;
Crabtree, G. R. Science 1993, 262, 1019–1024.
9. Panman, L.; te Welscher, P.; Soete, G.; Michos, O.; Zeller,
R.; Zuniga, A. Dev. Biol. 2002, 247, 475–475.
10. Sanz-Ezquerro, J. J.; Tickle, C. J. Anat. 2003, 202, 51–58.
11. Zernicka-Goetz, M. Development 2002, 129, 815–829.
12. Curley, K.; Lawrence, D. S. J. Am. Chem. Soc. 1998, 120,
8573–8574.
35. Calkhoven, C. F.; Ab, G. Biochem. J. 1996, 317, 329–
342.
36. Gennis, R. B. In Biomembranes; Molecular Structure and
Function; Springer: NewYork, 1989; pp 235–269.
37. Hargrove, J. L.; Schmidt, F. H. Faseb J. 1989, 3, 2360–
2370.
38. Thompson, J. F.; Hayes, L. S.; Lloyd, D. B. Gene 1991,
103, 171–177.
39. Bronstein, I.; Fortin, J.; Stanley, P. E.; Stewart, G. S.;
Kricka, L. J. Anal. Biochem. 1994, 219, 169–181.
40. McNally, J. G.; Muller, W. G.; Walker, D.; Wolford, R.;
Hager, G. L. Science 2000, 287, 1262–1265.
41. Reid, G.; Hubner, M. R.; Metivier, R.; Brand, H.; Denger,
S.; Manu, D.; Beaudouin, J.; Ellenberg, J.; Gannon, F.
Mol. Cell. 2003, 11, 695–707.
13. Wood, J. S.; Koszelak, M.; Liu, J.; Lawrence, D. S. J. Am.
Chem. Soc. 1998, 120, 7145–7146.
14. Curley, K.; Lawrence, D. S. Pharmacol. Ther. 1999, 82,
347–354.
42. Reid, G.; Denger, S.; Kos, M.; Gannon, R. Cell. Mol. Life
Sci. 2002, 59, 821–831.
15. Curley, K.; Lawrence, D. S. Curr. Opin. Chem. Biol. 1999,
3, 84–88.
43. Freeman, B. C.; Yamamoto, K. R. Trends Biochem. Sci.
2001, 26, 285–290.
16. Monroe, W. T.; McQuain, M. M.; Chang, M. S.;
Alexander, J. S.; Haselton, F. R. J. Biol. Chem. 1999,
274, 20895–20900.
17. Asanuma, H.; Liang, X.; Yoshida, T.; Komiyama, M.
Chem. Biochem. 2001, 2, 39–44.
18. Shimizu-Sato, S.; Huq, E.; Tepperman, J. M.; Quail, P. H.
Nat. Biotechnol. 2002, 20, 1041–1044.
19. Molecular Biology of the Steroid and Nuclear Hormone
Receptors; Freedman, L. P., Ed.; Birkhauser: Boston,
1998.
44. Dace, A.; Zhao, L.; Park, K. S.; Furuno, T.; Takamura,
N.; Nakanishi, M.; West, B. L.; Hanover, J. A.; Cheng, S.
Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 8985–8990.
45. Nawaz, Z.; Lonard, D. M.; Dennis, A. P.; Smith, C. L.;
OÕMalley, B. W. Proc. Natl. Acad. Sci. U.S.A. 1999, 96,
1858–1862.
46. Osburn, D. L.; Shao, G.; Seidel, H. M.; Schulman, I. G.
Mol. Cell. Biol. 2001, 21, 4909–4918.
47. El Khissiin, A.; Leclercq, G. FEBS Lett. 1999, 448, 160–
166.
20. Tsai, M.-J.; OÕMalley, B. W. Annu. Rev. Biochem. 1994,
63, 451–486.
48. Wijayaratne, A. L.; McDonnell, D. P. J. Biol. Chem. 2001,
276, 35684–35692.
21. Shi, Y.; Koh, J. T. Chem. Biol. 2001, 8, 501–510.
22. Shi, Y. H.; Koh, J. T. J. Am. Chem. Soc. 2002, 124, 6921–
6928.
23. Koh, J. T.; Putnam, M.; Tomic-Canic, M.; McDaniel, C.
M. J. Am. Chem. Soc. 1999, 121, 1984–1985.
24. Koh, J. T. Chem. Biol. 2002, 9, 17–23.
25. Tedesco, R.; Thomas, J. A.; Katzenellenbogen, B. S.;
Katzenellenbogen, J. A. Chem. Biol. 2001, 8, 277–287.
26. Miller, N.; Whelan, J. J. Steroid Biochem. Mol. Biol. 1998,
64, 129–135.
49. Qiu, M.; Lange, C. A. J. Steroid Biochem. Mol. Biol. 2003,
85, 147–157.
50. Samuels, H. H.; Stanley, F.; Casanova, J. Endocrinology
1979, 105, 80–85.
51. Rochel, N.; Renaud, J.-P.; Ruff, M.; Vivat, V.; Granger,
F.; Bonnier, D.; Lerouge, T.; Chambon, P.; Gronemeyer,
H.; Moras, D. Biochem. Biophys. Res. Commun. 1997, 230,
293–296.
52. Wecksler, W. R.; Norman, A. W. Methods Enzymol. 1980,
67, 494–500.