Ion Transit Pathways and Gating in ClC Chloride Channels

1
Ion Transit Pathways and Gating in ClC Chloride
Channels
Thomas L. Beck Department of Chemistry University
of Cincinnati thomas.beck_at_uc.edu
Acknowledgments
NSF AFOSR DoD/MURI
People
John Cuppoletti, Danuta Malinowska Rob Coalson,
Guogang Feng Achi Brandt, Jian Yin, Zhifeng
Kuang, Uma Mahankali, Anping Liu
2
Computational Methods for Ion Channels

• Full MD
• Brownian dynamics (hybrid)
• Electrodiffusion Poisson-Nernst-Planck
  (continuum theory, PMF?) Multigrid solvers
• Multiscale MC simulation and gating?
• Homology models
• Transport pathway search (TransPath)
• The potential distribution theorem for ion PMFs
  and pKas quasi-chemical theory

Length and time scales? ? Multiscale methods for
solving PDEs or simulation of large-amplitude
molecular motions (folding)
3
Rev. Mod. Phys. 72, 1041 (2000)
Multiscale Methods Multigrid V-cycle
MG accelerates convergence by decimating error
components with all wavelengths!
2 relaxations per level
Correct, relax
Restrict, relax
Coarsen
FMG
Same principle for polymers ? Hc
4
Molecular Coarsening Example
Modeling of large loop conformational transitions
with Monte Carlo simulations
Coarse point Center of gravity of 4 fine
points, e.g.
Hc on coarse scale ? PMF
Bai and Brandt (2001)
Fine scale corrections?
5
(CUP, in press)
Excess ? r(r)
Ideal
Uncoupled
Probability of observing interaction energy e,
fully coupled
6
Liquid chromatography and the hydrophobic
effect Oil/Water Interfaces
JACS 127, 2808 (2005)
90/10 water/methanol
7
Density profiles
Alkanes
Water
MeOH
90/10 water/methanol
8
Excess chemical potential for HS solute (R2 Ang)
Pure water
kcal/mol
50/50 Water/MeOH
90 MeOH
9
Proximal g(r) ?
10
No water pullout from the hydrophobic alkanes,
yet excess free volume at interface due to
fluctuations water near membrane channels,
hydrophilic and hydrophobic domains and ligand
binding?
11
Mammalian Acid secretory mechanism
Stomach (106 proton gradient)
Pump
K
Cl- (hClC-2)
H
pH sensor
C terminus
Cell
K
H/K-ATPase
K channel
Cl channel
12
Bacterial ClC Structure
Dutzler, et al. (2002)
13
Electrostatic Potential Contours of X-ray Dimer
EcClC in water

Using Connolly Surface, grid size
140 (0.5 Angstrom) x 199 x 139
E148 Gate?
Proton path in prokaryotes? 3 acidic residues.
Antiport behavior (Miller). This domain removed
in ClC-0. ClC-4,5 ? Recently shown ClC-4, 5 are
antiporters also (Pusch, Jentsch, Nature, 2005).
14
HOLEm pores, closed
O.S. Smart, J.M. Goodfellow and B.A. Wallace
(1993) The Pore Dimensions of Gramicidin A
Biophysical Journal 652455-2460. Uses van der
Waals radii for protein.
x-ray (3356,Rmin0.621 3851,Rmin0.736)
Glu148 gate
Filter Ile356 Phe357 Ser107 Tyr445
Difficulties with curved pores!
15
New Pathway Search Algorithm TransPath

• Utilize Monte Carlo methods to generate
  transmembrane spanning trajectories
• Incorporate both geometric and electrostatic
  information to bias the random walks
• Once the trajectories are generated, then anneal
  to find the geometric pore center of the
  protopath
• Obtain pore radius and potential profile
• No real Cl ion searches geometry and
  electrostatics
• No prior pore information required
• Searches for transit paths for positive or
  negative ions

16
TransPath Details

• Random origin points throughout protein center
• Grow polymer with configurational bias MC
• (Siepmann and Frenkel, polymer equilibrium
  method)
• Lattice or continuum
• Excluded volume within chain and chain/polymer
• Geometric and/or electrostatic (protein) weights
• Forward bias 5/8 of sphere
• Solve Poisson once for protein potential
• All exiting trajectories averaged (sorted)
• Calculate mex
• Yields a string, then typical HOLE anneal

p
17
TransPath
OTT mutant (open)
18
1OTT Cl Pathway (OPEN)
In
Out
Radius
Potl
S(bs)
R120 Helix D
S(int)
S(ext)
S(cen)
Agrees with proposed MacKinnon pathway
19
Mutant Structures (E148A, E148Q Open)
Cl ion at Scen
20
1OTS closed structure (gating and proton access?)
Green Cl- Purple H Blue H
21
pKa shift of Glu148
under different Cl ion binding conditions Chen
and Chen pKa 5.3, i.e. shift of 1
Shifts from roughly 4.3
Cl ion required at Scen for physical pKa of E148,
and possibly at Sbs
22
Conclusions on pathways and gating

• Glu148 is gate as proposed by MacKinnon, 1OTT and
  1OTU are open relative to wild-type structure
  (recent Chung simulations?)
• Single Cl path (possible alternative on
  intracellular side)
• Multiple proton access paths from external side?
• External Cl affects gating through binding near
  Arg147
• Cl at bind site shifts Glu148 pKa and alters
  proton path potentials
• Internal Cl resides in selectivity filter,
  required for necessary pKa shift of Glu148
• Two gating processes as proposed by Chen and
  Chen Cl dependent opening at depolarization Glu
  charged, pH dependent at hyperpolarization Glu
  neutral
• MD simulations, if E148p ? opens within 100 ps

23
Simulation system closed bacterial ClC? opens
upon protonation of E148(but can open as E148-
also at pH9.6)
24
Average trajectory of Cl- starting from Scen 11
independent MD runs (10 out of 11 open w/in 100ps)
Sext
Glu148-
Glu148P
Scen
25
The open state vs. closed state
Closed
Open
E148p
PMF for opening/closing of E148p and E148-
(replica exchange MD)?
26
Excess chemical potential for Cl- in the filter
PMF(z)
Interaction energy distribution of Cl at Scen
Glu148P
Glu148
100 ps
27
Solvation Energy of in Water at 298K
Gaussian model Asthagiri and Pratt (2005)
Probability
ex
(Betas2/268.4)
Energy kcal/mol
Lit. -66 to 79 kcal/mol
2.5 ns
28
PMF for Cl through bacterial ClC(1OTU)
Also want PMF for gating E148p and E148-
using Replica exchange MD
29
Multigrid electronic structure for solvation QCA
Cl(H2O)n, n3,4
Beck, Paulaitis, and Pratt (2005)
Table 5 Cl(H2O)3 cluster result.
kJ/mol
Table 6 Cl(H2O)4 cluster result. b bottom water
molecules t top water molecule.
pKas also for E148?
Note Gaussian03 is run in 6-31(d,p) basis. Our
fine grid is 81.
30
The Study of Mutant E. Coli ClC channel
homologues Modeling of Eukaryotic
Channels Intracellular domains of ClC-0 E111,
R451A/T452K vs. ClC-0 expts. Chen and Chen, 2003
31
The chloride transit paths in ClC-0 homology
model (OTU)
T452
R
N
D
E148Q
E111
F
F
R
N
D
R451
P1 orange P2 purple
P1 P2
A View from dimer interface B Turnover of A
three red spheres represent three chloride ions.
32
Conclusions vs. Expts. (intracellular
electrostatics) ?

• E127 (E111 bact) and K519 (K452 bact) exert large
  electrostatic influence on Sint. Balance of
  forces.
• Move from R451 (bact) to K452 (K519 ClC-0)
  (R451I/T452K) may explain increased conductivity
  in eukaryotic channels vs. low conductivity in
  prokaryotes. This move creates larger potl at
  Sint (Accardi and Miller, 2004). Note also
  bacterial structure is a transporter!
• Window of potl at Sint to allow transport
  (roughly 0.2-0.6V). (- -) and ( ) mutations
  exceed those limits. Appears to be largely
  electrostatic control. But very little room for
  surface charge effect (screening), since right
  near the entrance to the filter.

(submitted, Biophys. J., 2005)
33
Cl- moving along z-direction
Steered MD of full homology model of ClC-0
Good pore
34
Cl- Potential of Mean Force ClC-0
Scen
Sext
B
B
(recent Chung gating proposal?)
35
StClC Dimer and hClC-2 models (front)using
Modeller v6.2
StClC X-ray
pH Sensor loop
hClC-2 model (model3_07AB_BL020001)
hClC-2 model (model3_07AB_BL020002)
36
Replica Exchange Monte Carlo Simulations of
the30-residue pH sensor loop in hClC-2Multiple
temperatures (100-1000K)
37
Monte Carlo

• Ten temperatures were used 100K,200K,,1000K
• For the lowest temperature the equilibration
  takes at least 34000 mc sweeps
• At each temperature data is based on 160
  configurations saved from
• every 100th sweep from the last 16,000 MC sweeps
  of
• the total of 50,000 sweeps.

Energy Time Series
38
Average energy and specific heat
Cv

• specific heat var(E)/(k T2), where k0.001987
  kcal/(molK)
• transition at 600/-50K

39
Secondary Structure Prediction

• ECEPP/3 force-field with implicit solvent model
• (modified OONS-parameter set 1) implemented in
• SMMP-program 2
• run4 T300K, mc50,000
• run4b T300K, mc50,000
• Knowledge-based approach, implemented in
• Sable-program 3

good prediction
Insertion vs. bacterial structure
Sequence FDNRTWVRQGLVEELEPPSTSQAWNPPRANVFLTL SM
MP,Run4b ---CCCCCCHHHHHHHCCCCCCCCCCCHHHHHH-- SMMP
,Run4 ---HHHHHHTTTTCCCCCCCCCCCCHHHHHHCC-- Sable
HCCCCHHHHCCCCCCCCCCCCCCCCCCCCHHHHHH
1. U.H.E. Hansmann, Phys.Rev.E 70, 012902 (2004)
2. F. Eisenmenger, U.H.E. Hansmann, Sh. Hayryan,
C.-K. Hu, SMMP A Modern Package for Simulation
of Proteins, Comp. Phys. Comm (2001), 138
192-212. 3. J. Meller et al, http//sable.cchmc.o
rg
more analysis needed
40
Most populated structures

• Two independent runs 4 and 4b
• Run4b, red E-192 kcal/mol
• Run4, blue E-198 kcal/mol
• RMSD7Å

beginning
end
41
Some other low-energy structures
Run4, E-175kcal/mol
T100K
Run3b, E-250kcal/mol
T300K
Run4, E-240kcal/mol
Run4, E-204kcal/mol
42
Loop in hClC-2
side view
top view
helix A at the beginning of the loop
helix B at the end of the loop

• left loop is SMMP prediction, right Modeller
• due to secondary structures loop is smaller
• according to J.Meller helix B continues on the
  protein

43
Loop in hClC-2

• helix C restricts the motion to
• the right
• helix C loop also very closely
• interacts with our loop

helix C
44
Multiscale Polyethylene chain
Probability
Probability
Dihedral angle
End-to-end distance (Angstrom)
The red curves are obtained from coarsened fine
chain while the green curves are from actual
coarse level simulations. The blue curve is the
the end-to-end distance of the fine chain. Coarse
level simulations were performed with no
correction term to the Hamiltonian. These
results were reproduced by following the method
in D. Bai and A. Brandt Multiscale Computation of
Polymer Models, Multiscale Computational Methods
in Chemistry and Physics, A. Brandt et al (Eds)
IOS Press, 2001.
Working on coarse?fine scale correction scheme
45
In progress

• Proton access to the gate and detailed gating
  mechanism of transporter two gates?
• ClC-0 MD simulations do we have a viable open
  pore? Pore fluctuation with ion passage?
• Fluctuations in ClC-0 filter with Cl occupancy
  pore radius and potential via TransPath? How
  rigid is the pore?
• Gating mechanism in ClC-0
• Extracellular loop in ClC-2 pH sensor domain
  conformation charged and neutral in presence of
  channel
• Experiment/theory test of proposed proton paths
  in ClC-0,2 by mutation of E414 homologue, etc.
  ClC-4,5 antiporters also?
• Cl ion solvation in water vs. in the channel
  quasi-chemical theory for PMFs and pKas.
• Effects of high pressure pH sensor loop, pore
  geometry, etc.

46
pH, voltage, Cl dependence of gating?
Chen and Chen, J. Gen. Physiol. 118, 23 (2001)
Extracellular Cl dependence of gating prob (ClC-0)
hi Cl
lo Cl
47
pH dependence of gating
5.6
7.1
9.6
48
Sbs
P1
P2
P3
E414
pK3.4 epsp20 pK6.2 epsp4