A time lapse study of NF-jB translocation in 10 000 cells showed discernible oscillations in levels of nuclear NF-jB amongst cells when stimulated with interleukin IL-1a, which suggests
Trang 1of the nuclear factor-jB signaling pathway by in vitro
studies of IKK2
Adaoha E C Ihekwaba1,4, Stephen J Wilkinson1, Dominic Waithe3, David S Broomhead2,
Peter Li1, Rachel L Grimley3 and Neil Benson3
1 School of Chemistry, The University of Manchester, Faraday Building, UK
2 School of Mathematics, The University of Manchester, UK
3 Pfizer Global Research and Development, Sandwich, UK
4 VBI, Virginia Tech, Blacksburg, VA, USA
In silicoanalysis of complex cellular processes (whether
for data description, drug discovery, genetic
engineer-ing or scientific discovery) with its focus on elucidatengineer-ing
system mechanisms, has become critical for progress in
biology [1–5] Detailed computational models can
reveal complex behavior [6] in signaling pathways [7–
9] For example, under certain conditions, signaling
molecules can undergo periodic translocation between
different cellular compartments resulting in sustained
oscillations of their local concentrations [10–12] This
has been demonstrated for the nuclear transcription
factor nuclear factor (NF-jB), whose nuclear
concen-tration has been shown to oscillate due to
transloca-tion to⁄ from the cytoplasm
For the oscillations to be observable in a cell popula-tion rather than a single cell, they need to be largely synchronous [13–15] Of course, with the more recent availability of experimental capabilities to inspect single cells dynamically [16], more and more cells have been seen to exhibit asynchronous oscillations [11,12,17] Intact cells like yeast cells can synchronize their oscilla-tions with each other [14], and theoretical studies have demonstrated synchronization (of e.g metabolic path-ways) in communicating cells [15]
Experimental observations of oscillations have also been made for the p53 [18,19] and mitogen-activated protein kinase [9] signaling pathways, and can also be seen in mathematical models of such processes
Keywords
enzyme kinetics; in silico; in vitro; nuclear
factor-jB regulation; signal transduction
Correspondence
A E C Ihekwaba, VBI, Virginia Tech,
Washington Street, Blacksburg, VA, USA
Fax: +1 540 2312606
Tel: +1 540 2310795
E-mail: ihekwaba@vbi.vt.edu
(Received 13 July 2006, revised 19
Decem-ber 2006, accepted 22 January 2007)
doi:10.1111/j.1742-4658.2007.05713.x
Previously, we have shown by sensitivity analysis, that the oscillatory behavior of nuclear factor (NF-jB) is coupled to free IkappaB kinase-2 (IKK2) and IkappaBalpha(IjBa), and that the phosphorylation of IjBa by IKK influences the amplitude of NF-jB oscillations We have performed further analyses of the behavior of NF-jB and its signal transduction net-work to understand the dynamics of this system A time lapse study of NF-jB translocation in 10 000 cells showed discernible oscillations in levels
of nuclear NF-jB amongst cells when stimulated with interleukin (IL-1a), which suggests a small degree of synchronization amongst the cell popula-tion When the kinetics for the phosphorylation of IjBa by IKK were measured, we found that the values for the affinity and catalytic efficiency
of IKK2 for IjBa were dependent on assay conditions The application of these kinetic parameters in our computational model of the NF-jB path-way resulted in significant differences in the oscillatory patterns of NF-jB depending on the rate constant value used Hence, interpretation of in silico models should be made in the context of this uncertainty
Abbreviations
IKK, IkappaB kinase; IL-1a, interleukin-1a; MeOH ⁄ EtOH ⁄ PEG, methanol ⁄ ethanol ⁄ polyethylene glycol; NF-jB, nuclear factor kappa B; SC-514, 4-amino-2,3¢-bithiophene-5-carboxamide.
Trang 2[7,9,18,20,21] It is clear that the complexity of
biologi-cal systems and the difficulty (or at least infrequency)
of obtaining kinetic parameters require the
develop-ment of new analytical methods for both in vitro and
in silico biology [22] A still bigger challenge is the
measurement of in vivo values of kinetic constants,
which may differ critically from their in vitro
biochemi-cal counterparts
Oscillations have been demonstrated in a variety of
components of the NF-jB signaling pathway in single
cells [11,12] These results agreed with in silico
simula-tions of the downstream region of the pathway that
was modeled, as well as suggesting that the oscillatory
frequency has functional significance for downstream
events, such that the signal is not simply encoded in its
amplitude [23]
Activation of the transcription factor NF-jB can be
triggered by exposure of cells to a multitude of external
stimuli, including the cytokines tumor necrosis factor
(TNF-a) and interleukin-1a (IL-1a), thus initiating
numerous and diverse intracellular signaling cascades,
most of which activate the IkappaB kinase (IKK)
com-plex This crucial component in the NF-jB activation
cascade typically consists of two catalytic subunits
[24,25], IKKa (IKK1) and IKKb (IKK2) [26–29] and a
regulatory unit NF-jB essential modulator (NEMO,
IKKc) [30–33] The cytoplasmic inhibitors of NF-jB
(the IjBs [34,35]) are phosphorylated by activated IKK
at specific N-terminal residues, tagging them for
poly-ubiquitination and rapid proteasomal degradation This
allows NF-jB to be released upon activation and it
translocates to the nucleus where it induces the
tran-scription of a large number of target genes encoding
reg-ulators of immune and inflammatory responses and also
genes involved in apoptosis and cell proliferation [36]
In this paper, we report the results of cell-based,
in vitroand in silico experiments on the NF-jB pathway
First, we demonstrate oscillations in a population of
10 000 A549 cells, which is consistent with
synchron-ous behavior Secondly, we present in vitro kinetic
measurements of IKK2 protein kinase We
demon-strate that the assay conditions can affect substantially
the apparent Kmand kcatvalues of this reaction, whose
parameters are known to be important in an existing
computational model of the NF-jB pathway Thirdly,
we use the aforementioned computational model
[10,37] to analyze in silico the effect of the parameter
variation discussed above The parameter values
cho-sen for this reaction have a significant effect on the
amplitude (but not the frequency) of the oscillations
Finally, we extend this in silico and in vitro strategy to
a cell-based approach to analyze the effect of a known
inhibitor of IKK2,
4-amino-2,3¢-bithiophene-5-carbox-amide (SC-514) [38] We initially performed an in vitro study which confirmed the competitive nature of the inhibition and the published IC50 value Surprisingly,
we then found that cells pretreated with inhibitor dis-played oscillations of a similar strength (frequency) to those observed in the untreated cell population In order to shed light on the cause of this result, we car-ried out an in silico analysis by incorporating the inhi-bition kinetics within the existing computational model This showed that the SC-514 inhibitor has lim-ited impact on the dynamics of NF-jB activation
Results and Discussion
Immunocytochemistry Immunocytochemical staining of cell-based NF-jB pro-teins was used to study oscillatory patterns in NF-jB nuclear⁄ cytoplasmic localization in A549 cells In a pre-vious study, an EC50 of 0.340 ngÆmL)1 for IL-1a was established (Fig 1A,B; data not published); in this case, EC50 is a measure of the IL-1a concentration required to produce 50% of maximal response To investigate if different fixatives and types of the culture substrate have an effect on the intensity of nuclear NF-jB observed, a population time lapse study of nuc-lear NF-jB translocation following cell stimulation with 8 ngÆmL)1 IL-1a was examined Firstly, we com-pared the use of a 96-well plastic-bottomed plate with methanol⁄ ethanol ⁄ polyethylene glycol (MeOH ⁄ EtOH ⁄ PEG) fixative (with 12 repeats for each time point to minimize error as a result of background noise) and glass-bottomed plates with 3.7% formaldehyde fixative affected the quality of the stained images Using im-munocytochemical analysis, significant differences in the peak intensity was observed between the two assays The comparison at the 30-min time point revealed peak intensity of nuclear NF-jB to be 2.94 arbitrary units for the MeOH⁄ EtOH ⁄ PEG fixative and plastic culture plate combination (Fig 1G) and 5.75 arbitrary units for the formaldehyde fixative and glass culture plate combination (Fig 1F) These results indi-cated that the use of a combination of formaldehyde fixative and glass culture plates produces a better resolved image (higher signal⁄ noise ratio), giving a bet-ter dynamic range of output values when compared with the use of the alcohol-based fixative and plastic culture plates
We recently showed asynchronous oscillation follow-ing cell stimulation across four sfollow-ingle cells [12,39] It has been previously suggested that population-based analyses may not always reveal oscillatory behavior that is occurring on the single-cell level, because
Trang 3pro-tein extracts average potentially asynchronous
responses of individual cells [40] Despite this, we
observed discernible oscillations in the overall levels
of NF-jB activation in a population of 10 000 A549
cells suggesting a significant degree of synchronicity
(Fig 1F,G) Note that the immunocytochemical
approach used does not facilitate the tracking of
individ-ual cells over time [40]
Having previously shown by sensitivity analysis
[37,41] that the oscillatory behavior of nuclear NF-jB
is tightly coupled to just two participating species, free
IKK2 and free IjBa, and that reactions such as the phosphorylation of IjBa by IKK exerted a major con-trolling influence on the amplitude [42] of the oscilla-tions in the computational model [12,37], we next studied the rate of IjBa phosphorylation by IKK
Enzyme kinetics of rhIKK2 for glutathione-S-transferase-IjBa
We investigated the kinetics of rhIKK2 Figure 2A shows a typical progress curve for the IKK catalyzed
0
0
1
0
2
0
3
0
4
0
5
0 1 0 1
1 0 1 0 1
0
.
0
L –1 ) m
· g
c o a p l a -L I
L–1 m
· g 4 0 0 C
1 2 3 4 5 6 7
0 1 0 1
1 0 1 0 1 0 0
c o a p l a -L I
L –1 m
· g 8 1 0
= 0 C
5 1 7 1
9 1 1 2 3 2 5 2 7 2 9 2 1 3
0 0 0 0 0 0 0 0 0 0
n) i m ( m i T
- κκ
G E / H t E / H e M c it s l p
1
.
4
3
.
4
5
.
4
7
.
4
9
.
4
1
.
5
3
.
5
5
.
5
7
.
5
9
.
5
0 0 0 0 0 0 0 0 0
0
n) i m ( e m i T
Fκκ
e h l a m r o
f s l g
Fig 1 Immunocytochemical staining of A549 cells and analysis of the dynamics of NF-jB nuclear translocation (A,B) Dose–response data from A549 stimulated with IL-1a and fixed with formaldehyde (A) and MeOH ⁄ EtOH ⁄ PEG (B) (data unpublished) (C–E) Cell images showing nuclear cytoplasmic localization of NF-jB in stimulated (D) and stimulated after pretreatment with SC-514 inhibitor following 30 min of IL-1a exposure Cytoplasmic localization of NF-jB in nonstimulated cells is shown in (C), whereas in (D), localization of NF-jB is primarily in the nucleus Arrows draw attention to the localization of NF-jB In (E), NF-jB is observed in both the nuclei and the cytoplasm of the cells (F,G) Time course plot of A549 cells stimulation with 8 ngÆmL)1IL-1a generated on glass-bottomed plates with 3.7% formaldehyde fixative (F) and clear-bottomed plastic plates with MeOH ⁄ EtOH ⁄ PEG fixative (G) The peaks are the fluorescent intensity of nuclear NF-jB when com-pared with cytoplasmic NF-jB The error bars in (A), (B), (F) and (G) display standard deviations.
Trang 4phophorylation of GST-IjBa Figure 2B shows the
corresponding control without GST-IjBa These data
are consistent with limited autophosphorylation
Fig-ure 2C shows plots of rhIKK2 velocity as a function
of varying concentration of ATP (0.47–60 lm) at eight
fixed concentrations of GST-IjBa (0.12–15.33 lm) We found rhIKK2 displayed standard Michaelis-Menten kinetics at each GST-IjBa concentration with an apparent Km,ATP value of 9.6 ± 3.5 lm (Fig 2C and Table 1) We further examined the kinase activity of
[ATP] (µ M )
0 0
0 0
0
0
0
0
0
0
1
I T G [ κBα ( µ M )
I T G [ κBα ( µ M )
6 4 2 0 8 6 4 2 0
0 0 0 0 0
0 1
[ATP] (µ M )
0 0
0 0
2
4
6
8
0
2
4
6
8
0
2
6 4 2 0 8 6 4 2 0
0 2 4 6 8 0 2 4 6 8 0
0 1 0 1 0 0 0 0 0
0
2
.
0
4
.
0
0 1 0 1 0 0 0 0 0
2 0 0 4 0 0 6 0 0 8 0 0 1 0 2 0 0 4 0 0 6 0 0 8 0 0 2 0 2 0 0 4 0 0 6 0 0 8 0 0
F E
Fig 2 Enzyme kinetics of rhIKK2 for substrate ATP and GST-IjBa (A) Interaction between rhIKK2 and GST-IjBa (rhIKK2 in vitro kinase assay coupled with GST-IjBa as described in Experimental procedures); Vmax is 1.11 · 10)3 l M Æmin)1, Ks is 5 · 10)3± 1.4 · 10)3 l M (B) The control rhIKK2 in vitro kinase assay with no GST-IjBa substrate The time (in minutes) on the abscissa indicates the time the reac-tions were stopped with trichloroacetic acid and the plot shows the number of repeats Phosphorylation of tagged IjBa (d; A) and auto-phosphorylated rhIKK2 (j; B) is shown In (A), s is the control assay, and in (B), s and D represent the control repeats, and j represents the average of the two Kinase activities were recorded as incorporation of c- 33 P (countsÆmin)1) into GST-IjBa (A) and IKK2 (B) (C, D) Micha-elis–Menten plots generated by varying [ATP][60 l M (s), 30 l M (d), 15 l M (h), 7.5 l M (j), 3.75 l M (D), 1.88 l M (m), 0.94 l M (), 0.47 (.)]
at fixed [GST-IjBa] (C), and varying [GST-IjBa] [15.33 l M (s), 7.67 l M (d), 3.83 l M (h), 1.92 l M (j), 0.96 l M (D), 0.48 l M (m), 0.24 l M (), 0.12 l M (.) at a fixed [ATP] (D) Reactions (45 lL, plate assay) were performed at room temperature for 70 min with [c- 33 P]ATP (2.4 lCi).(E, F) Enzyme kinetics of recombinant IKK2 for substrate ATP and GST-IjBa with MnCl2in Tris ⁄ HCl ⁄ MgCl2 kinase buffer Michaelis–Menten plots generated by varying [ATP] at 15.33 l M GST-IjBa (E), and varying [GST-IjBa] at 60 l M ATP (F) Km,ATP, Km,GST-IjBa, kcat and Vmax were 2.3 ± 0.6 l M , 3.7 ± 0.9 l M , 1.51 · 10)3s)1 and 18.7 n M Æmin)1, respectively, in kinase buffer Tris ⁄ HCl ⁄ MgCl2 ⁄ MnCl2 and 2.5 ± 1.2 l M , 6.1 ± 1.3 l M , 2.15 · 10)3s)1and 16.2 n M Æmin)1in kinase buffer Hepes ⁄ MgCl2⁄ MnCl2 (data not shown).
Trang 5rhIKK2 as a function of varying concentrations of
GST-IjBa (0.12–15.33 lm) at eight fixed
concentra-tions of ATP (0.47–60 lm) Analyses showed the
apparent Km,GST-IjBa, value to be 3.8 ± 1.7 lm
(Fig 2D and Table 1), at saturated ATP concentration
(six-fold of Km,ATP) We also found the apparent
maximal turnover rates (kcat) for rhIKK2 to be
1.13· 10)2s)1 at room temperature under the same
conditions
Previously determined kinetics for IKK2 (Table 2),
revealed a 10–52-fold and 30–140-fold variation in
the Km values estimated for IjBa and ATP,
respect-ively The wide variation in these reported Km values
may be attributed to the use of rhIKK2, nonrhIKK2
or IKK complex, and also different experimental
conditions The Km that we determined for
GST-IjBa is comparable to a number of previously
pub-lished values within this wide range [28,43–47]
Simi-larly, our result for Km,ATP is in agreement with
some of the values reported in the literature
[43,47,48]
A noteworthy difference in the previously reported experiments is the presence [33,38,45,46,49–51] or absence [43,44,47,48] of MnCl2 in the assay conditions (i.e MgCl2 with MnCl2 vs MgCl2only) We therefore decided to perform a second investigation of the Km
values in the presence of MnCl2, but with all other experimental conditions constant Comparison with the values already obtained in the absence of MnCl2would therefore enable us to quantify this effect on two key parameters (as determined by us [37]) in our in silico model
Kinetic analysis showed Km,ATP, Km,GST-IjBa,kcatand
Vmax to have values of 2.3 ± 0.6 lm, 3.7 ± 0.9 lm, 1.51· 10)3s)1 and 18.7 nmÆmin)1, respectively, using Tris⁄ HCl ⁄ MgCl2⁄ MnCl2 buffer and 2.5 ± 1.2 lm, 6.1 ± 1.3 lm, 2.15 · 10)3s)1and 16.2 nmÆmin)1in the kinase assay using Hepes⁄ MgCl2⁄ MnCl2 buffer These findings confirmed the importance of kinase conditions used for determining kinetic values (Fig 2E,F) A list
of previously established kinetic values for IKK2 is reviewed in Table 2
Having shown that the disparity in the experimental kinetic results is dependent on the kinase condition used, we next studied how the experimental kinetic data reported here affected the NF-jB model previ-ously described [12,37,39] Substitution of the rates with the kinetic values reported in this section and Table 2 showed a more damped oscillatory pattern, similar to (see Fig 3H in [52]) and with comparable frequency to the original model (see Fig 3D)
These findings indicate that substituting previously reported kinetic data in the original model with the experimental data determined here results in an oscilla-tory pattern analogous to that seen in the population time study of the A549 cells (Fig 1F,G, where the
Table 2 A list of kinetic constants for IjBa and ATP substrates with IKK2 rh, recombinant human IKK2; nonrh, nonrecombinant human IKK2; norm, IKK complex; Y, present.
Km (ATP)
(l M )
Km (IjBa)
(l M )
kcat (s)1) · 10)3
Type of
Table 1 Michaelis–Menten kinetics, maximal turnover rates for
rhIKK2, the limiting maximal velocity and the ratio of apparent
dis-sociation constants for binding GST-IjBa in the presence and
absence of ATP K app
m is the apparent dissociation constant for full
length GST-IjBa substrate at saturation concentration of 60 l M
ATP, and the dissociation constant for ATP at saturation
concentra-tion of 15.33 l M GST-IjBa The apparent Vmax(V app
max ) at 60 l M ATP and 50 n M IKK2 is 136 ± 4.2 n M Æmin)1.
kcat
(s)1) · 10)2
Kappm (l M )
V maxapp (n M Æmin)1) a
GST-IjBa 1.13 ± 0.016 3.8 ± 1.7 136 ± 4.2 0.9 ± 0.5
Trang 6amplitudes are damped) We have thus far
demonstra-ted that kinase assay conditions affect the experimental
rate values We have also substantiated that the
oscilla-tory pattern of the model is affected when the new data
is implemented in the model We next studied the impact
of a rhIKK2 inhibitor on the oscillatory pattern of both
cell-based and in silico nuclear NF-jB translocation
Effect of SC-514 Inhibitor on cell-based nuclear NF-jB translocation
Kishore et al [38] first characterized the selective inhibitor SC-514 in 2003, and showed that it inhibited all forms of recombinant human IKK2 including rhIKK2 homodimer and rhIKK1⁄ IKK2 complex
6 3
1 4
6 4
1 5
6 5
1 6
0 4 0 4 0 0 3 0 2 0 2 0 0 1 0 0
Time (min)
- κ
d t a l u m it s 9 5 A d t a r t n s ll e c
d t a l u m it s 9 5 A d t a r t -e r p s ll e c
0
1
.
0
2
.
0
3
.
0
4
.
0
5
.
0
6
.
0
7
.
0
8
.
0
9
.
0
1
.
0
0 0 0 0 0 0 0 0 0
0
Time (min)
- κ
l e d m l a n i g i r O l e d m w e N
0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0
0 0 0 0 0 0 0 0 0 0
Time (min)
- κ
l e m l a i g i r O l e m w e N l e m r o t b i h I
0 0 0
1 0
1 1
0
0 0 0 0 0
0 1
0 0 0
1 0
1 1
.
0
0
2
.
0
4
.
0
6
.
0
8
.
0
1
C
Fig 3 Effect of SC-514 inhibitor on the activity of rhIKK2 homodimer (A,B) Different concentrations of SC-514 inhibitor was incubated with recombinant IKK2, and an IC50 experiment was undertaken using 10 l M (A, h), 1 l M (B, d) and 0.1 l M (B, s) ATP as described in Experi-mental procedures (C) Time lapse of nuclear cytoplasmic localization of NF-jB in 10 000 A549 cells These cells were dispensed onto a Whatman 96 glass bottomed plates and treated with 8 ngÆmL)1 IL-1a in the presence and absence of the SC-514 inhibitor (D, E) Time course plot of nuclear NF-jB from in vitro and in silico analysis of the data The plot shows nuclear NF-jB oscillation in the original model and in the updated model with newly measured kr1,ka1and kd1for IKKIjBa complex (D) In the original model kr1(kcat) for IKKIjBa was 4.07 · 10)3s)1 In the updated model, the original values are replaced with 1.13 · 10)2s)1(D) (E) shows a plot of the original and the updated model with the inclusion of newly measured Ki in the updated model.
Trang 7[38,53,54] A comparable IC50 value for rhIKK2 was
obtained in the present study to that previously
repor-ted by Kishore et al [38] We obtained values of
0.13 ± 0.06 lm for 0.1 lm ATP, 0.17 ± 0.08 lm for
1 lm ATP and 5.61 ± 0.65 lm for 10 lm ATP
(Fig 3A,B) This shift in IC50 values confirms the
competitive nature of this SC-514 inhibitor It was also
observed that a concentration of 100 lm of SC-514 is
sufficient to completely inhibit IjBa degradation by
IKK2 in vitro (Fig 3A,B); this was also as established
by Kishore et al [38] Having shown SC-514 to inhibit
IjBa phosphorylation, thus demonstrating an
inhibi-tion of IKK2 activity in vitro, we next determined
whe-ther SC-514 would inhibit activated native IKK
complex in IL-1a-stimulated A549 cells
To test whether these in vitro data were also found in
in vitro cell cultures, a nuclear NF-jB translocation
assay was performed where the cells were pretreated
with 100 lm of the SC-514 inhibitor We examined the
effect of SC-514 treatment on NF-jB activation by
stimulating A549 cells with IL-1a for 400 min In the
presence of SC-514, the kinetics of NF-jB activation
and inactivation with the IL-1a was observed
Immu-nofluorescence analysis showed that following a 6-h
exposure with IL-1a, NF-jB translocated from the
cytoplasm to the nucleus in the entire A549 population,
irrespective of their pretreatment with SC-514 inhibitor
(Fig 3C) Figure 3C displays time-course plots of
nuc-lear NF-jB dynamics for pretreated and untreated
A549-stimulated cells Interestingly, the pretreated cells
displayed clearly discernible oscillations that closely
fol-lowed those of the untreated cells in terms of their
fre-quency The effects of exposing the cells to the
inhibitor were slight, amounting to a modest reduction
in amplitude and a delay in the first oscillatory peak
It is interesting to speculate on the failure of the
inhibitor to eliminate the oscillations or at least
sub-stantially dampen them A similar apparent discrepancy
between in vitro and cell-based results was also reported
by Kishore et al [38], who reported some
phosphory-lation of IjBa even after SC-514 pretreatment at a level
(100 lm) that caused complete in vitro inhibition One
possible explanation is that this inhibitor does not block
the activity of another IKK isoform, IKK1
Conse-quently, IjBa may still be phosphorylated by the IKK1
isoform when the IKK1 isoform is present and
activ-ated in the system Another could be that the
intracellu-lar concentrations of ATP (Mg) are high enough to
attenuate observed inhibition Alternatively, it may well
be the case that IKK is not the only point of regulation
in the NF-jB pathway [55], and that IjBa
phosphory-lation and degradation, and the subsequent
transloca-tion of NF-jB into the nucleus may be mediated by
another mechanism Such a mechanism could be derived from the theory underlying metabolic control analysis, which states that the control exerted by indi-vidual parameters depends not only on their own magnitude but also on that of all the others [56,57]
Effect of SC-514 Inhibitor on in silico nuclear NF-jB translocation
To investigate whether the observed in vitro and cell-based inhibition data translated to an in silico effect,
we examined the impact of the determined experimen-tal inhibition kinetic data on the same model previ-ously described by Ihekwaba et al [37] Inclusion of our experimentally determined rate constant (Ki 0.114 lm; kcat 11.3· 10)3s)1) resulted in a dampened oscillatory pattern with a frequency similar to that of the original model (Fig 3E)
Interestingly, the inclusion of IKK2 inhibition by SC-514 with our experimentally determined rate con-stants in the original model resulted in a delay in simu-lated peak 1 and damping of subsequent peaks (Fig 3E), a feature also observed in the cell-based assay (Fig 3C) One implication of this finding is that the effects of future inhibitors designed for this NF-jB signaling pathway should be tested not just in vitro and cell based but also simulated in silico Combining this method of analysis (in vitro, cell-based and in silico analysis) will facilitate systematic understanding of the underlying properties of this signaling pathway
To summarize Activation of cells via stimuli, TNF-a [12] and IL-1a [58] induces activation of the NF-jB transcription fac-tor The consequences of how changes in external stim-uli influenced a cascade of co-operative events were assessed in vitro, in cell cultures, and also in silico in this study
Previous work [12] demonstrated oscillatory behav-ior in the levels of nuclear NF-jB in single cell studies
In our cell-based experiments, a population of
10 000 A549 cells was observed to undergo similar oscillatory behavior to that discovered in single cells in terms of the peak periods and frequency This clearly demonstrated that these cells have the ability to syn-chronize their oscillations with each other
The study of signaling pathway dynamics requires detailed cell-based measurements of time-varying phe-nomena, in this case, oscillatory variations of nuc-lear levels of NF-jB In order to attain this degree of precision, optimization of experimental conditions and techniques is required In the past three decades,
Trang 8advances in cell culture techniques and
immunocyto-chemistry have enabled the explanation of diverse
immunological phenomena We evaluated two
alternat-ive immunocytochemical experimental protocols for
cell assay analysis and found one to be superior for
the scope of this study Better resolved
immunocyto-chemical images were obtained using glass bottomed
plates with formaldehyde fixative rather than
plastic-bottomed plates with MeOH⁄ EtOH ⁄ PEG fixative
A quantitative assay to measure the phosphorylating
activity of rhIKK2 enzymes with GST-IjBa substrate
was described It was observed that subtle changes in
the experimental design had a profound effect on the
kinetic data obtained Kinase assay environment was
shown in this study to have a significant effect on the
Km,GST-IjBa, Km,ATP, and the Vmaxfound, and cautions
us to take appropriate measures when choosing rate
val-ues from the literature The importance of choosing the
relevant kinetic parameters when building a
computa-tional model was also demonstrated in this study
We have shown that the inhibition of IKK2 blocks
response in vitro Despite the fact that IKK2 has been
identified as a key participant in the NF-jB signaling
pathways, both our cell-based and in silico studies
revealed that this inhibition has limited impact on the
dynamics of NF-jB activation
It should be stressed that the in silico model
presen-ted here represents a considerable simplification of the
NF-jB signaling pathway For example, it does not
consider participants upstream of IKK2, or other
putative mechanisms for regulation of nuclear NF-jB
Nevertheless, the findings presented in this paper
dem-onstrate that even a simplified computational model
can give us a deeper understanding of the complex
sys-tem behavior of such signaling pathways
The key findings indicate that computational
mode-ling can be a useful complement to biochemical and
imaging experiments The results reported in this paper
should encourage further synergistic experimental and
computational studies aimed towards elucidating other
complex signaling systems
Experimental procedures
Materials
Materials and apparatus, and their suppliers, were as
follows: formaldehyde (3.7%) in NaCl⁄ Pi (internal stores,
Pfizer Global Research and Development, Sandwich, UK);
MeOH⁄ EtOH ⁄ PEG [60% v ⁄ v 95% EtOH, 20% v ⁄ v MeOH
(HPLC quality), 7% v⁄ v PEG (Sigma Aldrich, Gillingham,
UK); NaCl⁄ Pi, pH 7.2 (Invitrogen, Paisley, UK)];
poly-oxyethylene sorbitan monolaurate (Sigma); Draq 5 nuclear
stain (Biostatus Ltd., Shepshed, UK); Cellomics NF-jB Hit kit Evaluation⁄ Screening (Cellomics Inc., Pittsburg, PA, USA); DMEM (Gibco, Invitrogen); 200 mm l-glutamine; fetal bovine serum (Gibco Invitrogen, Virkon, Pfizer internal stores) IL-1a (R & D Systems, Minneapolis, MN, USA); Whatman 96-well sterile tissue culture treated glass bot-tomed plates; 96-well, clear-botbot-tomed plastic plates (Costar); Gilson p200 yellow, p1000 blue and p10 pink tips (Gilson, Middleton, WI, USA); Reagent boats and Falcon tubes Recombinant IKK2 was donated by Frank Stuhmeier of the Hit Discovery Group (HDG laboratory at Pfizer) Other reagents and apparatus used were as follows: GST-IjBa fusion protein [c-33P]ATP (Amersham Bioscience, Chalfont St Giles, UK); ATP (Roche diagnostics GmbH, Mannheim, Germany); trichloroacetic acid; 50 mm Tris⁄ HCl pH 7.5, 10 mm MgCl2; 50 mm Hepes pH 7.5,
10 mm MnCl2; NaCl⁄ Pi(Invitrogen); Microscint 40 (Pack-ard, Waltham, MA, USA); Plate seals (Packard); 96-well white microplate with bonded GF⁄ C filter [unifilter 96,
GF⁄ C (Perkin Elmer)]; microtiter plate (Millipore Corp.) All other reagents and apparatus were of high quality avail-able from Sigma sources
Cell culture
A549 cells (human lung carcinoma epithelial cell line SNB0000178-CE A549) were passaged every 4 days in DMEM (+ 5 mm l-glutamine and 5% fetal bovine serum) and maintained at 37C and 5% CO2 For translocation experiments, cells were removed with 0.05% trypsin⁄ EDTA, and plated with cell solution of 1· 106
cellsÆmL)1 (in a
50 mL flask) and grown until 80% confluency
Cellular assays
A549 cell solution (100 lL) was seeded on a plastic, flat-bot-tomed 6· 96-well- plates (Coster) at a density of 1 · 104 cells per well and incubated for an 18–24-h period at 37C and 5% CO2 A solution of IL-1a (concentration
40 ngÆmL)1 resulting in a final concentration of 8 ngÆmL)1 per well due to the 1 : 5 dilution factor) was prepared for the time-course assay Stimulation of cells was performed at 10-min intervals for 400 min with IL-1a After 400 min, the plates were inverted to remove media into a dish containing Virkon disinfectant to destroy cells not adhered to the plates MeOH⁄ EtOH ⁄ PEG fixation solution (100 lL; pre-warmed in a water bath at 37C) was dispensed into each well and incubated for 15 min (prewarming fixative is critical
to maintaining cell integrity) After 15 min, the plates were inverted to remove the fixation solution, and 100 lL of NaCl⁄ Piwas dispensed into the wells The plates were next inverted to remove the NaCl⁄ Pi wash solution, 100 lL of permeabilization buffer was then dispensed into the wells and left to incubate for 90 s at room temperature The plates were again inverted to remove the permeabilization buffer,
Trang 9and washed twice with NaCl⁄ Pithereby removing wash
buf-fer by inverting the plates Rabbit polyclonal
immunoglob-ulin IgG (50 lL; primary antibody) was dispensed into each
well and left to incubate for 1 h at room temperature The
plates were inverted to remove antibody after the 1 h
incu-bation period, 100 lL of detergent [NaCl⁄ Pi and 0.1%
Tween 20 (polyoxyethylene sorbitan monolaurate)] was
dis-pensed into the wells and the plates left to incubate for
15 min The plates were inverted to remove the detergent
after the 15 min incubation period and the wells were
washed twice with wash buffer by inverting the plates
Stain-ing solution (50 lL; containStain-ing goat antirabbit IgG
conju-gated to Alexa Fluor 488 secondary antibody and Draq5
dye; or and Hoechst 33258 dye) was dispensed into each well
and left to incubate for 1 h at room temperature in the dark
The plates were inverted to remove the antibody solution
and 100 lL of detergent dispensed into the wells and left to
incubate for 10 min The plates were inverted to remove
detergent solution and 100 lL of wash solution dispensed
into the wells The plates were inverted to remove the wash
solution for the last time and replaced with 200 lL of wash
buffer The plates were sealed and analyzed on Evotec
OPERA (Evotec, Hamburg, Germany) This assay study
was also repeated with a glass flat-bottomed 6· 96-well
plates (Whatman) and 4% formaldehyde fixative
Immunocytochemical analysis
On reading a microplate using the NF-jB protocol, the
Evotec OPERA has been programmed to find the nuclei
centers of the cells by using the DRAQ5 or Hoechst 33258
nuclear stain image DRAQ5 is excited with 633 nm laser
and its peak emission is 685 nm, whilst Hoechst uses
near-UV excitation (380 nm) and gives blue emission (530 nm)
The software was used according to the manufacturer’s
instructions (Scheme 1)
Cloning, expression and purification of GST-IjBa
fusion proteins
To overexpress the protein GST-IjBa, the plasmid vectors
were transformed into BL21 (DE3) Escherichia coli strains,
and the cells were grown overnight in 10 mL LB medium
containing 100 lgÆmL)1 ampicillin A colony of E coli in
LB agar plates was inoculated into 50 mL of LB liquid med-ium and incubated on shaking platform with 200 r.p.m at
37C for 3 h The value at D600measured by spectrophotom-etry was used to indicate the bacterial concentrations Inocu-lated liquid medium (2· 25 mL) was added into a
2· 500 mL of LB liquid medium, and incubated on rotator with 200 r.p.m at 37C for 1.5 h The value at D600was again measured by spectrophotometry The glutathione-S-transferase fusion proteins were induced by 2· 500 lL of
1 mm isopropyl b-d-1-thiogalactopyranoside addition to the
E coli medium and finally incubated on a rotator with
200 r.p.m at 37C for 3 h The bacterial cells in the
2· 500 mL medium were harvested by centrifugation (27 500 r.p.m for 10 min, 4C, Beckman rotor) Collected bacteria were re-suspended in a 2· 25 mL NaCl ⁄ Pibuffer The re-suspended cell mixture was placed in a disrupter machine with NaCl⁄ Piand 2-mercaptoethanol (total collec-ted volume¼ 120 mL) Benzoase (125 unitsÆmL)1) added to the collected viscous liquid The collected liquid was centri-fuged, the separated soluble fusion protein filtered (volume collected¼ 110 mL) and purified using immobilized metal chromatography at 4 mLÆmin)1 (absorbance of collected liquid using IMAC¼ A280) The supernatant was loaded onto a glutathione affinity column according to the manufac-turer’s protocol Bound glutathione-S-transferase proteins eluted with 5 mm glutathione in NaCl⁄ Pi(and 2-mercapto-ethanol) GST-IjBa (6 mL) eluted from the column Protein concentrations measured in a Bradford (Bio-Rad, Hercules,
CA, USA) protein assay Peak fraction were pooled and sub-jected to 12% Tris-glycine SDS⁄ PAGE and western analysis
to determine the purity of the GST-IjBa Glycerol (2 mL) was added to prevent damage from freezing, and the end vol-ume was transferred into Eppendorf tubes in aliquots of
400 lL
Kinase time-course assay
Recombinant human IKK2 (rhIKK2) time-course reaction was carried out for 113 min in 50 mm Tris⁄ HCl, pH 7.5, and 10 mm MgCl2 Reactions were performed in a final vol-ume of 45 lL (15 lL of rhIKK2, 15 lL ATP [c-33P]ATP,
15 lL GST-IjBa for kinase assay and 15 lL of rhIKK2,
15 lL ATP [c-33P]ATP, 15 lL 50 mm Tris⁄ HCl pH 7.5,
10 mm MgCl2for control assay) For experiments related to
Ksdetermination of rhKK2 and GST-IjBa binding, assays were carried out with 50 nm IKK2, 1 lm GST-IjBa peptide, 0.05 lCi [c-33P] ATP (10 mCiÆmmol)1) and 0.2 lm ATP Reaction mixture was withdrawn and dispensed into a 96-well white microplate with bonded GF⁄ C filter [unifilter
96, GF⁄ C (Perkin Elmer)] Each well was successively washed five times with 100 lL of 12% w⁄ v trichloroacetic acid, once with 100 lL 2 lm ATP, twice again with 100 lL 12% w⁄ v trichloracetic acid, and once with 100 lL 50 mm Tris⁄ HCl, pH 7.5, and 10 mm MgCl2 The plate was
Intensity of cytoplasm
Intensity of nucleus
Ratio of Translocation =
Intensity of cytoplasm Intensity of nucleus
Scheme 1 This is a simplification of a cell as seen by analysis
software, where the software measures the intensity of NF-jB in
the nucleus when compared with the intensity of NF-jB in the cell.
Trang 10allowed to dry for 10 min in a 55C oven, and then 35 lL
of scintillation fluid (Microscint 40) was dispensed to each
well Incorporated [c-33P]ATP was measured using a Top
count NXT (Packard) The amount of IKK-catalyzed
incorporation of 33P into each peptide was quantified by
liquid scintillation counting The counts represent initial
velocity of rhIKK2-catalysed phosphorylation (< 30% of
total ATP conversion) The graphs were fitted using
gra-fitTM software, and k1 and k2 were calculated from Vmax
and Ks values expressed in unitsẳmol)1 of enzyme per min
and unitsẳmol)1, respectively
Kinase assay with Tris⁄ HCl ⁄ MgCl2
rhIKK2 kinase reactions were carried out for 70 min in
50 mm Tris⁄ HCl, pH 7.5, and 10 mm MgCl2 The amounts
of substrates ATP, [c-33P]ATP (10 mCiẳmmol)1; Amersham
Bioscience), and GST-IjBa are specified for each individual
experiment Reactions were performed in a final volume of
45 lL (15 lL of rhIKK2, 15 lL ATP, [c33
P]ATP, 15 lL GST-IjBa) For experiments related to Km determinations
of IKK2, various concentrations of ATP and GST-IjBa
peptide were used in the assay at a fixed concentration of
either GST-IjBa or ATP For GST-IjBa peptide Km,
assays were carried out with 50 nm IKK2, 60 lm ATP,
2.4 lCi [c-33P]ATP (10 mCiẳmmol)1) and GST-IjBa
pep-tide from 0.12 to 15.33 lm For ATP Km, assays were
car-ried out with 50 nm IKK2, 15.33 lm GST-IjBa peptide,
1 lCi [c-33P]ATP (10 mCiẳmmol)1) and ATP from 0.47 to
60 lm Sample was analyzed by precipitation on a
micro-titer plate (Millipore Corp) For the micromicro-titer plate assays,
45 lL of reaction sample⁄ well was precipitated with 45 lL
of 12% w⁄ v trichloroacetic acid 70 lL of the reaction
mix-ture was withdrawn and dispensed into a 96-well white
microplate with bonded GF⁄ C filter (unifilter 96, GF ⁄ C;
Perkin Elmer) Washing of precipitated sample was
per-formed using the same protocol as that described for the
kinase time-course assay The assay was again repeated
with the inclusion of 10 mm MnCl2in the kinase condition
Kinase assay with Hepes⁄ MgCl2⁄ MnCl2
rhIKK kinase reactions were carried out for 70 min in
50 mm Hepes pH 7.5, and 10 mm MgCl2 and 10 mm
MnCl2 The amounts of substrates, ATP, [c-33P]ATP
(10 mCiẳmmol)1, Amersham Bioscience) and GST-IjBa
were the same as those specified in the assay with
Tris⁄ HCl ⁄ MgCl2 Reactions were performed using the same
protocol as that described for the Tris⁄ HCl ⁄ MgCl2assay
IC50([I]0.5) doseỜresponse assay
IC50 experiments were performed in 96-well Millipore
plates The reactions were carried out for 45 min in 50 mm
Tris⁄ HCl, pH 7.5, and 10 mm MgCl2 and typically inclu-ded: 50 ng of rhIKK2; varying concentrations of SC-514 inhibitor [300Ờ0.1 lm; reconstituted 2.688 mg of SC-514 (relative molecular mass 224 g) to 1 mL of 12 000 lm stock solution in 100% dimethyl sulfoxide]; and 5.11 lm GST-IjBa peptide per well at 10 lm ATP 1 lCi [c33
P]ATP (10 mCiẳmmol)1), 1 lm ATP 0.1 lCi [c33
P]ATP (10 mCiẳmmol)1) and 0.1 lm ATP 0.05 lCi [c33
P]ATP (10 mCiẳmmol)1) separate ATP concentrations, to make a total volume of 40 lL (rhIKK2 10 lL, SC-514 10 lL, ATP
10 lL and GST-IjBa 10 lL) The reaction was run in duplicate A positive and a negative control assay were also included, where the positive control contains no inhibitor in the assay and the negative control was stopped at time zero Reaction sample (40 lLẳwell)1) was precipitated with
40 lL of 12% w⁄ v trichloroacetic acid Reactions were per-formed using the same protocol as that described for the Tris⁄ HCl ⁄ MgCl2and Hepes⁄ MgCl2⁄ MnCl2assay
Kinetic analysis
For two substrate profile analysis, initial velocity studies were performed with varying concentrations of GST-IjBa
at several fixed concentrations of ATP and vice versa (order
of binding experiments) LineweaverỜBurk double recipro-cal plots were generated by linear least squares fits of the data Replotting the slopes and the y intercepts of the lines
as function of 1⁄ [ATP] generated secondary plots Kinetic constants (Km for ATP, GST-IjBa, and Vmax) values were determined from a global fit to the database using eritha-cussoftware grafit 4- where Vmaxis the limiting maximal velocity that would be observed when all the enzyme is pre-sent as enzymeỜsubstrate ỔESỖ [rhIKK2-GST-IjBa], Km is the MichaelisỜMenten constant and the kcat is the break-down of the ES complex to E + product (P) [59] (Eqn 1) The equilibria describing competitive inhibition of the SC-514 are show in Eqn 2, where Kiis the dissociation con-stant for the enzymeỜinhibitor (EI) complex To obtain 50% (IC50) inhibition, refer to Eqn 2 [59]
E
ợợ S!
k 2
k 1
ES!kcat
Eợ P I
#" Ki EI
ơ1
KiỬ IC50=ơ1 ợ đơS=Kmỡ ơ2
For a random sequential model, values for Km,ATP,
Km,GST-IjBa, Vmax and a was determined from the global
fit The constant a is the ratio of apparent dissociation con-stants for binding GST-IjBa in the presence and absence
of ATP, and the value of a indicates whether the binding
of one substrate (ATP) affects the affinity of the enzyme for the other substrate (GST-IjBa) [59]