Progress Report Summary
A. SPECIFIC AIMS
Computational. The Specific Aim 1 continues to be the development of
computational tools for molecular and supramolecular (Aim 1A),
microphysiological or subcellular (Aim 1B) and cellular systems (Aim 1C)
levels of complex biological processes. Our specific aims 2 and 3 also remain
unchanged as the integration of these multiscale methods, and the development
and implementa-tion of computational tools for the organization and
visualization of the data and output.
Biomedical. The three originally proposed developmental projects (DPs)
on complex cell signaling and regulatory processes remain to be the major
focus of the Biomedical Aims. The respective topics of concentration are
effect of NO on apoptotic response (DP1), DNA damage recognition and signaling
(DP2) and ligand interactions with signaling molecules (DP3).
B. STUDIES AND RESULTS
Specific Aim 1A. Progress:
(ii) the initial development of a hybrid method that combines molecular
dynamics (MD) simulations and Gaussian network model (GNM) analysis, applied
to hemoglobin T
ŕ
R transition (collaboration between Bahar (Center for Computational Biology
and Bioinformatics (CCBB), Pitt) and Ho (Biology, CMU)) (Xu et al., 2003) and
still in progress, (i) the extension of the GNM to predict the dynamics of
large assemblies, supported by its recent application to ribosomal dynamics
(Yang et al., 2004). These studies have been conducted with the partial
support from Pitt School of Medicine (SOM) provided to the CCBB, and two new
grant proposals have been submitted for pursuing these studies (See other
support pages).
Specific Aim 1B.
The software for spatially realistic simulation of microphysiology (MCell and
DReAMM;
http://www.mcell.psc.edu) were originally designed to simulate the 3-D
reaction/diffusion aspects of neurotransmission, and a critical first aim
was/is to expand MCell’s computational kernel and modeling language to more
general signaling pathways such as those of the DPs on apoptosis or DNA damage
signaling. A new prototype code base for MCell has been produced to
incorporate the interactions between different diffusing molecules, rather
than just between diffusing molecules and stationary molecules located on
surfaces representing cell and/or organelle membranes. We began preliminary
testing of molecule-molecule interactions with the new Monte Carlo (MC)
algorithms. See (Stiles et al., 2004) for more details.
Specific Aim 1C.
New features have been added to the XPP/XPPAUT software, which currently
allows for solving ~600 differential equations (http://www.math.pitt.edu/~bard/xpp/xpp.html)
(Ermentrout’s lab). Additionally, a logical network model of apoptosis has
been constructed in Ta’asan lab (CMU, Math), building on their existing
framework that allows for stochastic modeling of thousands of reactions and
population variations. The modeling is agent based where molecules are
considered at a few levels of abundance, representing no expression, low and
high expression. Likewise, reaction rates are discretized as slow, moderate
and fast. The developed model of apoptosis includes multiple compartments as
cell surface, mitochondria, and nucleus, and focuses on caspase activation
through FAS, TNFR, mitochondrial pathways including stress activation. The
apoptosis model is found to be robust to the changes in most of the
parameters, and rapid activation of caspase-3 is ensured only by a small
subset of reactions consistent with previous experimental observations (Rehm
et al., 2002; Tyas et al., 2000).
Specific Aim 2.
Major progresses towards
the integration of different tools and combined use of experimental and
computational data include; initial tests of a hybrid methodology that jointly
exploits the results from MD simulations and GNM calculations to predict
conformational changes induced by allosteric effects (Xu et al., 2004);
integration of NMR relaxation data and GNM computations to assess the
molecular mechanism of conformational transitions (Temiz et al., 2004);
assessment of stability and folding/unfolding kinetics by combined analysis of
GNM and FIRST computations (Rader & Bahar, 2004) and site-directed mutagenesis
experiments (Rader et al., 2004); comparison of dynamic MC simulations and
dielectric self energy Poisson-Nernst-Planck continuum theory (Graf et al.,
2004) towards the design of an algorithm that combines both methods; continued
development of MCell’s input language to handle the combinatorics problem of
fully general molecular complex formation.
Specific Aim 3.
Three major developments
are: (1)
A new website has been
constructed for automated release and visualization of the results from GNM
analysis of PDB structures (http://ignm.ccbb.pitt.edu/). (2)
To optimize DReAMM for models composed of thousands of mesh objects, we
have begun modifying the underlying OpenDX (http://www.opendx.org)
source code (~one million C/C++ lines) to improve a number of critical user
interface design features, and we have also written a set of critical C code
modules that now handle manipulation of meshes, wireframes, boundaries, and
molecules, as well as custom depth cueing and color mapping of the different
objects. A new prototype version of DReAMM incorporates these changes, and is
orders of magnitude faster than the original version. (3) A new detailed
tutorial has been posted on the XPP/XPPAUT website (http://www.math.pitt.edu/~bard/bardware/tut/start.html)
Developmental Project 1
(DP1). We started
with the apoptosis model proposed by Fussenegger et al. (2000) and modified it
to improve the realism of cytochrome c release, and include p53
interactions and truncation of Bid to tBid. We also created the first
generation model of the reaction pathways of NO, coupled to apoptosis, which
includes the formation of peroxynitrites, p53 induction via DNA damage,
formation of S-nitrosative species and their inhibition of caspases, and
inhibition of cyt c release by cGMP. Our mathematical model yielded
results in reasonable agreement with experimental data (Billiar’s group) on
the dependence of NO effects on cellular environment. It also suggested that
the induction of apoptosis in macrophages as opposed to suppression in
hepatocytes is related to the different iron levels in these two cell types.
The results are summarized in a recent review (Vodovotz et al., 2004).
Developmental Project 2
(DP2). Wood’s
group recently isolated DNA polymerase
q,
a new enzyme encoded by the human POLQ gene, with an exceptional
ability to replicate past an AP site, inserting A with 22% of the efficiency
of a normal template, and then continuing extension as avidly as with a
normally-paired base (Seki et al., 2003). POLQ preferentially incorporates A
opposite an AP-site. On non-damaged templates POLQ makes frequent errors,
incorporating G or T opposite T about 1% of the time. This low fidelity
distinguishes POLQ from other A-family polymerases. Mammalian POLQ has three
unusual sequence insertions. Comparative modeling of POLQ by Bahar’s group
suggested that one insert of ~22 residues at the tip of the polymerase thumb
domain confers considerable flexibility and additional DNA contacts to reduce
the enzyme fidelity and enhance its processive ability, as summarized in a
manuscript (Seki et al., 2004).
Developmental Project 3
(DP3). Using high throughput screening, Lazo’s group
discovered a novel inhibitor of CDC25B (referred to as
5169131) with an IC50 of 10.4
mM. The inhibitor shows competitive inhibition with the
CDC25B substrate, OMFP. Co-crystallization of the protein with the inhibitor
is difficult since the inhibitor tends to precipitate. Bahar’s group performed
simulations for docking the inhibitor on the protein using MOE package. In
addition, a model for CDCD25B complexed with its substrate was generated.
Superimposition of the CDC25B-inhibitor and CDC25B-OMFP models showed that the
protein sites that bind the inhibitor and the substrate overlap, consistent
with the experimentally observed competitive inhibition. The results from this
collaborative study between Lazo, Wipf and Bahar labs are summarized in
(Brisson et al, 2004).
C. SIGNIFICANCE.
Quantitative modeling of the complex interactions involved in cell signaling
and regulation is important in understanding the origin and mechanism of
deregulation processes and identifying new targets for molecular therapies. A
productive cooperation has been initiated between scientists specialized in
different aspects of these complex processes, evidenced by several manuscripts
recently submitted/accepted, which support the synergistic effect of
coordinating experimental and computational studies.
D. PLANS
Specific Aim 1a. Three future directions of research are: (1) further
developing and testing hybrid models that combine different methods. We will
devise models where different structural regions are represented at different
levels of complexity (i.e. catalytic regions at atomic details, rigidly moving
domains as elastic networks, etc.). (2) loop structure prediction and
flexible binding, particularly at DNA/drug binding regions, in the
light of the problems that emerged in modeling damaged DNA recognition (DP2)
and competitive binding of inhibitors (DP3).
Specific Aim 1b. Further development MCell and DReAMM to bridge between
molecular and cellular simulations is a key component of the multiscale
computations aims of our pre-Center activities. MCell models can easily grow
to include hundreds to thousands of separate mesh objects representing
different cells or parts of cells, and thus significant scalability issues
arise. We will adopt a hierarchical molecule/complex naming convention
combined with a binary notation for the presence and state of different
binding sites, and use wild-card characters to enable different binding events
in any arbitrary order.
Specific Aim 1c.
We are now taking a more detailed approach in which
concentrations and rate constants reported in previous studies are
incorporated. An important improvement will be to restore the resistance of
the cells to small pro-apoptotic perturbations (i.e. maintain the stability of
the resting state (Siehl et al., 2002). To this end, we will assume Hill-type
cooperative kinetics for the binding of cyt c or procaspase-9 to
Apaf-1. Preliminary studies already show that this leads to a bistable
behavior depending on low or high apoptotic stimuli (e.g. death ligand
concentration), consistent with the concept of a threshold value to induce
apoptosis.
Specific Aim 2. Perhaps the most challenging aspect is a real integration
of the activities conducted by different groups. We will continue the
progress made already in new tool development, and work toward integrating new
algorithms at coarse-grained molecular scales (Bahar), microphysiological
scales (Stiles), and cellular/tissue scales (Ta’asan and Ermentrout).
Specific Aim 3. Spatially realistic models at different scales present
substantial difficulties at the stage of interactive user design,
visualization, and animation, so the pre-Center activities on the development
of visualization and model design tools will be continued, some of which are
already ongoing with new funding (Stiles).
DP1-3. Significant progress leading to publishable results was made during the
preliminary studies performed within the scope of the three DPs, each
constituting a good starting point for future progress in a
Center of Excellence, and
each being an example of a newly initiated productive collaboration between
experimental and computational labs. We will pursue these studies towards
a molecular understanding of factors affecting the response of cells in
response to different pro- or anti-apoptotic stimuli (DP1), the DNA damage
recognition or by-passing properties of DDB proteins (DP2), the competitive
binding of different inhibitors to phosphatases (DP3). The current apoptosis
model will be expanded to include heat shock proteins, the effects of
different proteins inhibition and DNA damage, in accord with the aims of DP2
and DP3.
Other DPs. In addition to these three DPs, four projects are
illustrative of new research areas that are directly related to the goals of
the pre-Center, and could also evolve into driving projects for a full center:
(1) Mathematical analysis (binomial probability analysis) of single
pixel calcium imaging data at neuromuscular junctions, undertaken by J.
Stiles in collaboration with S. Meriney (Neuroscience, Pitt), to predict the
number and average opening probability of voltage-gated calcium channels at
the frog neuromuscular junction, which already led to a publication (Washman
et al., 2004); (2) Monte Carlo simulation of presynaptic calcium
dynamics and neurotransmitter release, directed by J. Stiles and carried
out by John Pattillo, a post-doc in Stiles lab. Potential-activation of
voltage-gated calcium channels, stochastic calcium ion entry and diffusion,
calcium binding to sensor sites on arrays of synaptic vesicles, and vesicle
fusion and resulting transmitter release are simulated to obtain novel
predictions for the number of calcium-binding sites on synaptic vesicles.
(3) Dynamics of ligand-gated ion channels that contribute to neuronal
function, with focus on glycine receptor (GlyR) as a prototype. These
channels play a fundamental role in fast electrical signaling in the nervous
system, and channel dysfunction or pharmacological modulation by drugs or
anesthetics have profound effects. A multi-faceted collaboration is conducted
with the leadership of Coalson (Chem, Pitt), Kurnikova (Chem, CMU) and Cascio
(Mol Gen & Biochem, SOM, Pitt) labs, with both experimental and
theoretical/modeling components, to elucidate structure/function relations in
the GlyR. (4) Mechanism of inhibition of HIV-1 reverse transcriptase
(RT), a structural, computational and experimental study of the
NNRTI-induced conformational changes in HIV-1 RT conducted by Bahar (CCBB,
Pitt), Madura (Chem & Biochem, Duquesne) and Sluis-Kremer (Medicine, Pitt)
labs (Sluis-Kremer et al., 2004; Zhou & Madura, 2004a; 2004b)
E. PUBLICATIONS.
Publications related to
this grant are presented in Appendix I. A total of 22 publications (21 papers
+ 1 Book chapter) are listed, of which 15 are published/accepted and 7
submitted. Soft copies of most of the published/accepted papers are made
accessible (Appendix I).
F. SPECIAL REQUIREMENTS
F.1. Description of
activities.
Research activities have been described above. Organizational and educational
activities are described in F.2 and F.4, respectively.
F.2. Organizational
structure implemented.
The organizational
structure and activities are:
1. Formation of an
Executive Committee (EC) also serving as an Internal Scientific Committee,
composed of Bahar
(Chair, Comp Biol & Bioinformatics, Pitt), Stiles (Co-Chair, PSC/CMU), Brown
(Co-Chair, Biology, CMU), Madura (Co-Chair, Chem & Biochem, Duquesne),
Ermentrout (Mathematics, Pitt), Rosenberg (Biology, Pitt), and Ho (Biology,
CMU). These individuals are committed to these positions for the duration of
the pre-NPEBC. The PI's of the DPs, Billiar (Surgery, Pitt), Wood (Molecular
Oncology, UPCI), and Lazo (Pharmacology, Pitt), are serving as adjunct
members. The EC Chair and/or Co-Chairs serve as a liaison with the Chairs (or
Directors) of the participating Departments (or Centers). Quarterly meetings
were held by the EC members.
2. The organizational structure of research activities
exactly follows the diagram presented in the original proposal (Figure B.2.)
with the only exception of B. Ermentrout replacing C. Chow (who moved to NIH)
as the computational group leader in DP1.
3. The selection of development projects was already
made in the original submission. We also proposed a mechanism for the review
and elimination/selection of old/new DPs. The progress made in each of the
original DPs supports their continuation, while other collaborations (other
DPs; see above) have also emerged in the past year. The limited budget of the
pre-NPEBC does not presently permit us to allocate more funds to support these
other DPs and these are currently being conducted using other sources. Our
plan is to apply for a full Center in the coming year, and include/request
support for these newly emerging collaborative projects, as appropriate.
4. External Advisory Committee. Ronald Levy (Rutgers),
Douglas Lauffenberger (MIT), Angela Gronenborn (NIH) and Robert L. Jernigan
(Iowa State U) were contacted to serve as members of the EAC, who kindly
accepted. The visits of the EAC members to Pittsburgh has been coordinated
with the pre-NPEBC workshop and seminar series (see below).
5. The first Pre-NPEBC workshop was held on
Feb 3, 2004 at the
Pittsburgh Supercomputing
Center, and was attended by over 50 researchers (PIs, postdoctoral fellows,
and students) from Pitt, CMU, PSC, and Duquesne. The goal was to bring
together researchers within the local community and to present the types of
existing problems and data/methods that could benefit from an efficient
collaboration between the computational and experimental groups. The workshop
was organized in five 1-hour sessions, one on each of the existing
developmental projects, one on “other” possible projects (with emphasis on
membranes, signaling and neurobiology problems), and one on the integration of
computational and mathematical models and methods. It concluded with breakout
sessions for brainstorming within groups, and a final summary session where
plans for advancing/continuing collaborative research were outlined. An
outline of the workshop program can be found in Appendix III. A second
pre-NPEBC workshop has been scheduled for Fall 2004, where speakers from
universities outside the Pittsburgh area, including the EAC members will be
invited to present their research and discuss the state-of-the-art research in
multiscale modeling of complex biological processes.
6. New pre-NPEBC Investigators. New Faculty that
contributed to pre-NPEBC activities are J. Klein-Seetharaman
(Pharmacology, SOM, Pitt), D. Zuckerman (CCBB, Pitt) who joined Pitt
after the submission of the revised pre-NPRBC proposal, N. Sluis-Kremer
(Medicine, SOM, Pitt), S. Meriney (Neuroscience, Pitt), R. Rosenfeld
(CS, CMU), and M. Kurnikova (Chem, CMU).
6. New Recruitments. A major development was the
recruitment of new Faculty who would contribute to the pre-NPEBC efforts. Our
first goal was to strengthen the systems biology group, and we have been
successful in recruiting an excellent scientist, Dr. Ivan V. Maly (PhD
at Northwestern U with G. Borisy and postdoc with Doug Lauffenberger at MIT),
specialized in cellular self-assembly and its coupling to cellular networks.
Dr. Maly will be joining the CCBB as a tenure-tract Faculty in
July 1, 2004, and will be an
active member of the pre-NPEBC multiscale modeling efforts. Another Faculty
recruitment is Dr. David Swigon (Rutgers), by the Math Dept at Pitt,
originally a mathematician, specialized in modeling DNA elasticity and the
regulation of transcription, who is expected to be involved in DP2. The CCBB
is in also at the final stages of recruiting Dr. Carlos Camacho (Res
Assoc in Vajda’s lab, Boston U) at Associate Prof level, an expert in
structure prediction and protein-protein docking, who will contribute to
DP3.
7. Starting a new
Department of Computational Biology at Pitt SOM. The CCBB at Pitt SOM
recently submitted a proposal for becoming a department at the SOM, Pitt,
which has already been approved by the SOM Executive Committee and is
currently awaiting the approval of Planning and Budgetary Committee. It is
expected that the organizational and educational activities of the pre-NPEBC
and the possible future Center will significantly benefit from the
establishment of a department (to be chaired by Dr. Bahar) fully dedicated to
the development, implementation and integration of computational and
theoretical methodologies in cooperation and coordination with experimental
groups/studies.
F.3. DPs that have been
initiated and progress made in each of them.
See the paragraphs DP1,
DP2 and DP3 in Section
B
(Studies and Results), and DP1-3 and Other DPs in Section D.
F.4. Progress toward
developing educational opportunities in biomedical computing.
1. Planning a
cross-campus PhD program in Computational Biology.
A cross-campus PhD program
in computational biology is planned between Pitt and CMU with the leadership
of Drs. Bahar (CCBB, SOM, Pitt), Carbonell (Director of Language Techn
Institute, Comp Sci, CMU) and Stiles (PSC/CMU). The vision is to offer a
first-class program that capitalizes on the excellence of CMU in computer
science and on the expertise and strength of Pitt in biomedical sciences.
While the organization and administration of a joint program is a challenging
task, both groups are enthusiastic and encouraged by productive collaboration
already initiated within the scope of the present pre-NPEBC award and an
NSF-ITR grant (PIs: Raj Reddy (CS, CMU) and J. Klein-Seetharaman (SOM, Pitt;
Bahar and Carbonell are co-PIs ). Three major tracts are conceived:
bioinformatics, computational structural/molecular biology and system biology.
2. Biomedical and
Bioinformatics Summer Institute (BBSI). This is an NSF/NIH funded program
that started in Sept 2002, offering a 10-week summer program to undergraduate
(junior and senior) and graduate (1st and 2ndyear)
students interested in computational biology. The major theme of the program
is ‘Multiscale Modeling’ in line with the pre-NPEBC goals, and six PI playing
a key role in the pre-NPEBC (Bahar, Stiles, Coalson, Madura, Ermentrout and
Meirovitch) serve as the course instructors and co-PIs of the program. 20+
Faculty from Pitt, CMU, PSC and Duquesne take part in mentoring the research
activities of the students.
3. Other programs and
courses. A new undergraduate degree-granting program in
Bioinformatics
has been designed with the leadership of Pitt CS and Biology departments,
which is expected to admit students in Fall 2006. A PhD program in Molecular
Biophysics with a strong component in Comp Bio has been approved at Pitt and
CMU, starting in Fall 2004. CMU is currently offering B.S and M.S. programs in
Comp Bio & Chem.
A new graduate course has been designed at Duquesne,
entitled “Simulation and
Visualization”, and another graduate course “Simulation Methods” jointly
taught by Duquesne Chem and CS departments will be offered in Spring 2005.
4. Postdocs and
students are involved in the pre-NPEBC activities, and thus have the
opportunity to directly learn/apply up-to-date concepts and methods in
biomedical computing, and take part in multidisciplinary research activities.
See Appendix IV for the list of postdocs/students who have actively
contributed to the computational and experimental activities of the pre-NPEBC
in the past year, all being funded by ‘other’ sources.