Calcium Causes a Conformational Change in Lamin A Tail Domain
that Promotes Farnesyl-Mediated Membrane Association
The MIT Faculty has made this article openly available.
Please share
how this access benefits you. Your story matters.
Citation
Kalinowski, Agnieszka, Zhao Qin, Kelli Coffey, Ravi Kodali, Markus J.
Buehler, Mathias Losche, and Kris Noel Dahl. “Calcium Causes
a Conformational Change in Lamin A Tail Domain That Promotes
Farnesyl-Mediated Membrane Association.” Biophysical Journal
104, no. 10 (May 2013): 2246–2253. © 2013 Biophysical Society
As Published
http://dx.doi.org/10.1016/j.bpj.2013.04.016
Publisher
Elsevier
Version
Final published version
Citable link
http://hdl.handle.net/1721.1/89214
Terms of Use
Article is made available in accordance with the publisher's
policy and may be subject to US copyright law. Please refer to the
publisher's site for terms of use.
Calcium Causes a Conformational Change in Lamin A Tail Domain that
Promotes Farnesyl-Mediated Membrane Association
Agnieszka Kalinowski,
†6Zhao Qin,
‡6Kelli Coffey,
†§Ravi Kodali,
{Markus J. Buehler,
‡*
Mathias Lo¨sche,
†k***
and Kris Noel Dahl
†§*
†Biomedical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania;‡Civil and Environmental Engineering, Massachusetts
Institute of Technology, Cambridge, Massachusetts;§Chemical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania;
{Department of Structural Biology, University of Pittsburgh, Pittsburgh, Pennsylvania;kDepartment of Physics, Carnegie Mellon University,
Pittsburgh, Pennsylvania; and **Center for Neutron Research, NIST, Gaithersburg, Maryland
ABSTRACT
Lamin proteins contribute to nuclear structure and function, primarily at the inner nuclear membrane. The
post-translational processing pathway of lamin A includes farnesylation of the C-terminus, likely to increase membrane association,
and subsequent proteolytic cleavage of the C-terminus. Hutchinson Gilford progeria syndrome is a premature aging disorder
wherein a mutant version of lamin A, D50 lamin A, retains its farnesylation. We report here that membrane association of
far-nesylated D50 lamin A tail domains requires calcium. Experimental evidence and molecular dynamics simulations collectively
suggest that the farnesyl group is sequestered within a hydrophobic region in the tail domain in the absence of calcium. Calcium
binds to the tail domain with an affinity
K
Dz 250 mM where it alters the structure of the Ig-fold and increases the solvent
acces-sibility of the C-terminus. In 2 mM CaCl
2, the affinity of the farnesylated protein to a synthetic membrane is
K
Dz 2 mM, as
measured with surface plasmon resonance, but showed a combination of aggregation and binding. Membrane binding in the
absence of calcium could not be detected. We suggest that a conformational change induced in D50 lamin A with divalent
cat-ions plays a regulatory role in the posttranslational processing of lamin A, which may be important in disease pathogenesis.
INTRODUCTION
A-type lamins, including lamin A (LA), contribute to a
variety of functions that include DNA transcription,
hetero-chromatin regulation, cell senescence, DNA damage repair,
and the sequestration of transcription factors (recent reviews
(
1,2
)). These type V intermediate filament proteins are also
the primary structural elements of the nucleoskeleton (
3
).
Lamins are structurally related to other intermediate
fila-ment proteins (
4
), but the C-terminal tail domain (TD) of
LA is much larger than that of other intermediate filaments
and undergoes significant posttranslational processing.
LMNA codes for the precursor prelamin A (preLA), which
is transported into the nucleus where the C-terminal CaaX
domain (specifically, Cys-Ser-Ile-Met) is farnesylated and
carboxymethylated (
5
). The modified preLA is then cleaved
by the enteroprotease ZMPSTE-1 at Y646 to produce
mature LA. Dysfunctions of this process including loss of
ZMPSTE-1 or mutations including the Y646 cleavage site
are responsible for devastating human diseases. In the
pre-mature aging syndrome Hutchinson Gilford progeria
syn-drome (HGPS), a C1824T mutation of LMNA causes the
activation of a cryptic splice site and a loss of exon 11 in
the TD (
6
). The loss of the 50 amino acid segment, which
contains the ZMPSTE-1 cleavage region, causes retention
of the farnesylation, which in turn results in changes in
ther-modynamic stability of the mutant protein,
D50LA or
progerin (
7
). At a much lower frequency than in HGPS,
acti-vation of the cryptic splice site that produces
D50LA occurs
sporadically in normal cells, and this defect accumulates
with aging (
8
).
The biochemistry of LA TD processing has been
deter-mined, but the biological benefits of the complex processing
pathway and the regulation of its individual steps are unclear
(
9
). Although farnesyl transferases have been found in both
the endoplasmic reticulum and the nucleus, farnesylation
of the protein likely occurs in the nucleus (
5
). The farnesyl
modification is suggested to increase the association of
preLA with the inner nuclear membrane (
10,11
). Similarly,
the pathologies associated with the expression of
D50LA in
HGPS result, in part, from the farnesylation causing
overac-cumulation of structural proteins at the inner nuclear
membrane (
12,13
). Indeed, mislocalization of LA and
LA-associated proteins at the inner nuclear membrane results
in altered DNA repair, transcription, and signaling (
14
), as
well as changes in protein-protein interactions (
15
). On the
other hand, a single farnesylation is insufficient to confer
permanent membrane association of proteins in other
sys-tems (
16–18
). Therefore, it is unclear if the presence of the
farnesyl is sufficient for the cellular pathological phenotype
or if additional factors, including altered proteprotein
in-teractions (
15
) or altered protein packing (
7,19
) are involved.
For the full-length lamin protein, the rod domain
func-tions in dimer formation and the head and tail domains
asso-ciate with the rod to form filaments; full-length filaments
form aggregates and paracrystals in vitro (
1–4
). The TD is
Submitted November 29, 2012, and accepted for publication April 8, 2013.6Agnieszka Kalinowski and Zhao Qin contributed equally to this work.
*Correspondence:mbuehler@mit.eduorquench@cmu.eduorkrisdahl@ cmu.edu
Editor: David Odde.
responsible for association with other proteins and
mem-brane localization. For these reasons, we limited our study
to the TD to isolate the interaction between the protein
and the membrane (
2,10
). Here, we quantify
protein-mem-brane binding of farnesylated
D50LA (fn-D50LA) by
inves-tigating purified LA TDs in solution and on model lipid
bilayer membranes. We find that the binding is weak
(K
D~2
mM) and that free Ca
2þis required for binding.
The conformational response of the protein to Ca
2þis
tracked by monitoring its intrinsic fluorescence. Molecular
simulations suggest that the Ca
2þ-induced conformational
change opens the Ig-fold, increases the flexibility of the
terminal CaaX motif, and exposes the farnesyl group, which
is sequestered within the TD in the absence of the ion.
MATERIALS AND METHODS
Protein production, farnesylation,
and characterization
Constructs for the expression of LA TDs were created using polymerase chain reaction from plasmids kindly provided by Misteli at NCI/NIH (12). Plasmids were expressed in Escherichia coli, purified using gluta-thione-s-transferase fusion peptides and farnesylated in vitro using farnesyl-transferase, resulting in a population of ~50% farnesylated protein (see Methods in theSupporting Material). Purified proteins were verified for size analysis using liquid chromatography electron spray ionization mass spectroscopy, for purity using 14% SDS-PAGE, for aggregation by dynamic light scattering (DLS) (nanosizer). Bradford assay (Coomaise Plus, Pierce) was used to determine protein concentration (see Methods).
Solution conditions that minimize aggregation
of
D50LA and preLA TDs
Recombinant LA-TD variants aggregate in water, therefore accurate quan-tification of protein-membrane interactions required stabilization of mono-mer protein. We determined the minimal salt (NaCl) concentrations in the buffer necessary to suppress aggregation ofD50LA-TD and preLA-TD by DLS. Although deconvoluted DLS spectra were not able to determine aggregation form, it clearly showed large-scale aggregation under low-salt conditions (Fig. S1in theSupporting Material). Characteristic DLS peaks of d ~10 nm were regularly observed under high-salt conditions ([NaCl]> 50 mM), and we interpreted a 10 nm peak as a protein monomer. At low salt ([NaCl]< 50 mM), proteins aggregated (d > 100 nm). This analysis showed different minimal salt concentrations for the different lamin A variant TDs studied here, including with farnesylation(fn). D50LA-TD required 25 mM NaCl to maintain a monomeric solution of pro-tein, whereas fn-D50LA-TD required 50 mM NaCl (Fig. S1A). In contrast, preLA-TDs required a minimum of 500 mM NaCl to maintain dispersed solutions (Fig. S1B). The addition of 2 mM CaCl2did not alter the stability
ofD50LA-TD and fn-D50LA-TD (stable in 50 mM NaCl and 2 mM CaCl2, Fig. S1C) or preLA-TD and fn-preLA-TD (stable in 500 mM NaCl and 2 mM CaCl2, Fig. S1 D). Aggregation could also be suppressed in
10 mM sodium phosphate buffer but could not be used in these experiments because phosphate chelates calcium. Other salts, such as KCl or NaHCO3,
were not tried as membrane behavior under those conditions.
Tryptophan fluorescence
The tryptophan amino acid residue has inherent fluorescent properties that have long been used as probes to reflect protein conformation (20). LA-TD
has four tryptophan residues all of which are located on or near the Ig-fold. The peak emission intensity is at 342 nm, after excitation at 298 nm to mini-mize excitation of other aromatic residues. A peak emission at 342 nm indicates that the tryptophan residues are neither in a completely hydropho-bic environment, nor are they fully solvent exposed (20,21).
The peak emission intensity at 342 nm (with the background intensity from the buffer subtracted) was used for analysis and normalized by the peak intensity without Ca2þfor each sample. The binding affinity and stoichiometry of Ca2þassociation with the LA TD was determined by fitting the change in fluorescence intensity at 342 nm as a function of Ca2þconcentration to a Hill model. For our system where the ligand is the Ca2þand the conformational change is assumed to be proportional to the change in fluorescence intensity at the emission peak, DI (342 nm), normalized by the peak emission at 342 nm without Ca2þ in the solution, Io, the Hill equation is DI/Io ¼ [Ca2þ]
n
/(KDþ[Ca2þ] n
) gives a binding affinity of the protein to calcium KD, as well as the
stoichiometry of binding, n (i.e., the number of Ca2þthat associate with one protein molecule).
Membrane bilayer preparation and surface
plasmon resonance experiments (SPR)
SPR was used to quantify the membrane association of the LA-TD to teth-ered bilayer lipid membranes (tBLMs), which provide a fluid membrane fixed to a surface for SPR to monitor membrane association without the necessity of using a protein label. The tBLM model membrane system consists of a planar lipid bilayer suspended to the surface by a minimal number of functionalized lipopolymer interspersed with b-mercaptoetha-nol to create a submembrane hydration layer (see Methods). The ~1.5 nm hydration layer allows the membrane to retain its fluidity and reduce the effects of the substrate on the membrane. See Methods for full details. The molecular structure, functionality, and in-plane dynamics of such tBLMs have been established by the Lo¨sche group in extensive studies involving electrochemical impedance spectroscopy, neutron scat-tering, fluorescence microscopy, and fluorescence correlation spectroscopy (22–24).
Replica exchange molecular dynamics (REMD)
simulations
Protein structure predictions were carried out by the REMD method (25) implemented in the CHARMM c35b1 (26) software. Replica management and conformation analysis are done using the MMTSB script package (27). REMD is a simulation method with atomic detail to improve the dynamic properties of the Monte Carlo method, aiming at a global-minimum free energy state. It samples a wide conformation space without getting trapped in local-minimum free energy states because of the thermal stimulation by high-temperature replicas, and we can find stable structures at the replica with the lowest temperature. The REMD inputs include the peptide sequence and an initial guess of conformation (see Methods).
Modeling of farnesyl diphosphate
The structure and interaction scheme of the farnesyl diphosphate is not defined in the standard CHARMM27 force field. We setup the initial conformation of farnesyl diphosphate according to its atomic coordinates as found in the protein data bank (with PDB ID 1KZO). The interactions (covalent bond, angle, dihedral, electrostatic, and vdW interactions) among the atoms in farnesyl diphosphate are defined in the CHARMM General force field 2b6 (28). We updated the CHARMM27 force field to incorporate the section of the definition of farnesyl diphosphate (see Methods for details of the modeling and validation against the atomic structures identified from experiment).
Biophysical Journal 104(10) 2246–2253
RESULTS
Membrane binding of fn-
D50LA-TD depends on
calcium
To determine the membrane affinity of fn-
D50LA-TD, we
added
purified
monomeric
farnesylated
protein
(see
Methods and
Supporting Material
information for
produc-tion, characterizaproduc-tion, and monomeric stability) to tBLMs
and measured binding using SPR. The tBLMs are
appro-priate for precise SPR binding measurements because
bila-yers are virtually defect free, thus avoiding nonspecific
protein adsorption (
29
) and tightly coupled to gold-coated
solid substrates via hydrated oligo(ethylene oxide) spacers
(
30
). Bilayer membranes were composed of
dioleoyl-phos-phatidylcholine and the charged
dioleoyl-phosphatidylser-ine (DOPC/DOPS
¼ 7:3) with 3 mol% cholesterol to
mimic lipid composition and charge of cellular membranes.
With or without farnesylation,
D50LA-TD associated
minimally with the membrane in NaCl buffer (
Fig. 1
A).
However, the presence of physiological levels of free
Ca
2þin the buffer facilitated protein-membrane interaction.
fn-
D50LA-TD bound stably to tBLMs with 2 mM CaCl
2in
the buffer, but
D50LA-TD binding was not detected under
otherwise identical conditions (
Fig. 1
, A and B). Increasing
concentrations of fn-
D50LA-TD in the Ca
2þ-containing
buffer increased membrane binding (
Fig. 1
C) and showed
a binding affinity, K
D~2
mM (assuming that only the
farne-sylated proteins associated with the membrane, i.e.,
approx-imately half the concentration of the protein added) when fit
to a Langmuir model (
Fig. 1
C). We note that the membrane
association is a combination of binding and aggregation,
which impacts the K
D: instantaneous high levels of
fn-
D50LA-TD do not show the same levels of binding as
increased amounts of fn-
D50LA-TD (
Fig. 1
B vs.
Fig. 1
C). Furthermore, protein is not easily dissociated from the
interface. These and other aspects of interfacial aggregation
and binding are the topic of future work. Unfortunately,
the high salt required to maintain monomeric stability of
fn-preLA-TD (see Methods) interfered with membrane
stability, so that fn-preLA-TD binding to membranes could
not be quantified.
Calcium-dependent conformational changes
To investigate the mechanism by which Ca
2þinduces
mem-brane association, we considered Ca
2þ-induced
conforma-tional changes of the TD. The TD is intrinsically
disordered except for the s-type Ig-fold (
31,32
), therefore,
direct structural examination via traditional methods (e.g.,
crystallography) are not possible. Conformational changes
of
D50LA-TD and preLA-TD were followed by monitoring
the fluorescence of their four Trp residues located within
and near the Ig-fold (W467, W498, W514, W520; see full
sequence in
Table S1
). The chemical environment of Trp,
including its solvent accessibility, affects the emission
quan-tum yield, and exposure to the solvent has been shown to
quench Trp fluorescence (
20,21
). Previously, we
investi-gated thermal denaturation of LA-TD and
D50LA-TD
ter-tiary structure by monitoring Trp fluorescence, confirming
that protein denaturation reduces Trp fluorescence (
7
).
The observed fluorescence emission band between 310
and 400 nm was broad, presumably because multiple Trp
residues in different microenvironments contribute to the
band shape. We observed that the emission decreased with
CaCl
2concentration (
Fig. 2
, A and B A).), and saturated for
[CaCl
2]
z 1 mM at ~30% of the original intensity. Fitting
to a two-parameter Hill model indicated a Ca
2þaffinity of
261
5 49 mM for D50LA-TD and 286 5 27 mM for preLA
TD, both with 1:1 stoichiometries (
510%) (
Fig. 2
C). These
results suggest that Trp residues of the two TDs become more
exposed to buffer upon Ca
2þbinding. High concentrations of
NaCl (
>200 mM) also reduced Trp emission (
Fig. S2
), but
Ca
2þ-induced effects occur at six orders of magnitude lower
salt concentration (100
mM), independent of NaCl.
10−2 10−1 100 101
0 0.5 1
[fn−Δ50LA−TD], μM
Protein adsorbed (saturated fraction)
2 6 10 0 100 200 Time, hr Density (ng/cm 2) 0 3 6 0 100 200 Time, hr Density (ng/cm 2) 0 1 2 0 100 200 Time, hr Density (ng/cm 2)
A
B
C
D
.01F 0 .05F 0.1F 0.5F 1F 2F 0 1 0.1F 1F 2F 0 1.5 4 3.2F +Ca2+ +Ca2+−Ca2+ +Ca2+ FIGURE 1 Binding of fn-D50LA-TD to a model
membrane. (A) 1 mM D50LA-TD (gray) did not associate with the tBLM in 50 mM NaCl. Further-more, 0.1mM, 1 mM, or 2 mM fn-D50LA-TD in 50 mM NaCl (0.1 F, 1 F, and 2 F in blue) did not associate with the membrane (B) For a similar tBLM and buffer conditions plus 2 mM CaCl2,
1.5mM or 4 mM D50LA-TD (gray) did not asso-ciate with the membrane. However, 3.2 mM fn-D50LA-TD (3.2 F in red) did show association. (C) Stepwise addition to the solution phase of 0.01mM, 0.05 mM, 0.1 mM, 0.5 mM, 1 mM, and 2mM fn-D50LA-TD (red) showed increased mem-brane association with 50 mM NaCl and 2 mM CaCl2. (D) From this association we can produce
a binding curve to determine the KDof the protein
to the membrane. (tBLM composition: 30:70:þ3 DOPS/DOPC/Chol)
Fluctuations of the Ig-fold of
D50LA-TD
and preLA-TD in the presence of calcium
To obtain a deeper insight into the structural responses of the
LA TDs to Ca
2þ, we performed molecular dynamics (MD)
simulations of the proteins in buffer. To deal with the high
degree of intrinsic disorder, we simulated multiple
low-energy, pseudoequilibrated structures with REMD (
25,33
).
Starting with amino acid chains with extreme initial
struc-tures (straight or fully coiled) except for the structured
Ig-fold region taken from the NMR structure in the protein
database (
31
), we allowed the protein to relax freely in an
implicit solvent environment. To enhance the sampling
speed as well as to ensure that structures reached relevant
local energy minima, we investigated numerous
indepen-dent structures with implicit-solvent REMD to find as
many low-energy states as possible (see Methods). Results
were categorized according to their structural similarity,
and the four most significant conformations were chosen
for further studies of the structural impact of ions (
7
). These
four conformations were further refined in REMD
simula-tions with explicit solvent under various ion condisimula-tions
(see Methods). To quantify the TD’s structural response to
ion exposure, we focused on four peripheral amino acids
(S429, S437, L478, and I497), which form orthogonal
ex-tremes of the two
b-sheets that form the Ig-fold (
Fig. 3
A).
From the positions of these four amino acids, we determined
an orientation angle between the two
b-sheets, q (
Fig. 3
A).
The Ig-fold is slightly more open (
q is systematically larger)
in the presence of Ca
2þ(
Fig. 3
B).
We computationally included a farnesyl group to the
LA-TDs. For an initial conformation, we used the
pseudoe-quilibrium structure of the unfarnesylated LA-TDs and
embedded the added C-terminal farnesyl group within the
hydrophobic core of the Ig-fold (see Methods for force
field). We then allowed the new structure to come to a
new equilibrium with explicit solvent. The Ig-fold of both
D50LA-TD and preLA-TD were more open if farnesylated,
and exposure to Ca
2þopened the fold further up by a larger
angle than with the unfarnesylated proteins (
Fig. 3
C).
Because the average distance between I497 and L478
10−1 100 0 0.5 1 log [Ca2+], mM Δ I/I0 ( λ = 342 nm) 315 355 395 0.5 1.0 λ (nm) I/I0 (a.u.) 315 355 395 0.5 1.0 λ (nm) Δ50LA−TD preLA−TD preLA−TD R2 = 0.99 increasing [Ca2+] increasing [Ca2+] Δ50LA−TD
A
C
B
FIGURE 2 Calcium caused a conformational change toD50LA-TD and preLA-TD. The emis-sion spectra of tryptophan residues for purified (A)D50LA-TD in 50 mM NaCl and (B) preLA-TD in 500 mM NaCl excited at 295 nm were measured for increasing Ca2þconcentration. For Ca2þ concentration above 1 mM, there was no change in intensity, indicating no more substantial conformational change induced by Ca2þ. (C) From plots of emission intensity at 342 nm versus Ca2þ concentration, we determined Ca2þ dissociation constant (KD~250300 mM) and binding
stoichi-ometry (1:1). The DI/I0 (342 nm) is averaged
for 2–3 samples and in NaCl concentrations 50–700 mM NaCl for a given CaCl2concentration.
The Ca2þdissociation constant did not change with the NaCl concentration of the solution.
FIGURE 3 Altered Ig-fold structures of preLA-TD and D50LA-TD exposed to calcium. (A) We determined the dihedral angle between the twob-sheets of the Ig-fold, q, from the a-carbon of orthogonal amino acids (S429, S437, I497, and L478) within the full TD. (B) We tracked the fluc-tuations of the dihedral angleq for the preLA-TA conformation in the replica with lowest temperature after each exchange event in pseudoequili-brium. The angle was always larger in the presence of Ca2þ. (C) Statistical comparison of the dihedral angle of the Ig-fold for numerous pseudoequili-brated protein structures for preLA-TD, fn-preLA-TD,D50LA-TD, and fn-D50LA-TD showed increases in q with Ca2þ.
Biophysical Journal 104(10) 2246–2253
is
>1.7 nm, an increase in the orientation angle q indicates
that the space between the two
b-sheets accommodates
more water (with a van der Waals radius of 0.14 nm) in
the presence of the ion than in its absence.
Fluctuations of the C-terminus of
D50LA-TD
and preLA-TD with calcium
REMD simulations also provided insight into the
mecha-nism by which calcium triggers membrane binding of
farne-sylated lamin A tail domains. We observed that Ca
2þbound
spontaneously to the C-terminal CaaX motif, and the
inter-action persisted through the entire REMD simulation (cyan
Ca
2þto red CSIM in
Fig. 4
A,
Fig. S3
). Generally, Ca
2þbinding increased the fluctuation and mobility of the
C-terminal region for both
D50LA-TD (beyond 610
0; the
prime refers to the new amino acid number for the deletion)
and preLA-TD (beyond 660) by 30% and 39%, respectively
(
Fig. 4
B).
C-terminal motility was reduced when the protein was
farnesylated (
Fig. 4
B), and the Ca
2þ-mediated motility
increase was enhanced for farnesylated proteins (
Fig. 4
B)
by 89% for fn-preLA-TD and 120% for fn-
D50LA-TD. In
the absence of Ca
2þ, the farnesyl (
Fig. 5
, red) maintained
association with the hydrophobic core (
Fig. 5
A,
Ca
2þ).
In the presence of Ca
2þ, the farnesyl group escaped
the hydrophobic interior of the protein and was found at
the periphery of the pseudoequilibrated structures (
Fig.
5
A,
þCa
2þ). This was quantified by the increase in solvent
available surface area of the farnesyl group in the presence
of Ca
2þ(
Fig. 5
B) and increased end-to-end length of the
TDs (
Fig. 5
C). We suggest that the farnesyl group
associ-ates with the hydrophobic protein core, and the association
of Ca
2þto the CaaX causes increased probability of farnesyl
solvent exposure for membrane association.
DISCUSSION
Membrane binding of farnesylated lamin A tail
requires calcium
We find that fn-
D50LA-TD binds to purified membrane
bilayers with a K
D~2
mM. This binding coefficient is
mark-edly weaker than most biological binding coefficients;
spe-cific protein-protein binding typically occurs on the order
of 10s to 100s of nM. We suggest that this weak binding
affinity allows protein-membrane association while not
causing irreversible association. In the case of lamin A,
we suggest that the farnesyl group weakly associates the
protein to the membrane, and transmembrane lamin-binding
proteins then more permanently connect lamin A to the
inner nuclear membrane. This weak binding is similar to
qualitative observations that farnesyl groups are insufficient
to stably associate proteins with membranes (
16–18
).
Dur-ing normal processDur-ing, lamin TDs, and other CaaX-domain
proteins, are farnesylated, the aaX residues are cleaved by
the ZMPSTE-1 protease, and the terminus is methylated
in the nucleus (
34
). The proteolysis and methylation steps
may play an additional role in stabilizing membrane
associ-ation by increasing hydrophobicity. However, there have
been conflicting reports regarding the additional stability
conferred by methylation (
35–37
).
The association of the fn-
D50LA-TD is dependent on
Ca
2þ. There have been previous studies examining
quali-tative membrane association of fn-preLA (
38
), but we
note that the buffer used contained the divalent cation
magnesium. Similarly, recoverin has a Ca
2þ-induced
A
B
FIGURE 4 Dynamics of the C-terminus of preLA-TD andD50LA-TD exposed to calcium. (A) We examined all pseudoequilibrated conformations of unfarnesylated preLA-TD produced in explicit-solvent REMD simula-tions and acquired snaphopys of the lowest energy conformasimula-tions (overlaid). In the presence of explicit water and sodium, the C-terminus CSIM (red) was located inside the preLA tail structure. With the addition of Ca2þ(cyan), the C-terminus pointed away from the protein. (B) The C-terminal amino acids for each of the four TDs for 10 ns of MD simulation showed altered fluctuations or root mean-square displacement from initial coordinates in the presence of Ca2þ. For each TD and ionic condition we repeated the simulation for the four lowest energy conformations and statis-tically compared the results. For the last 3 to ~6 residues, Ca2þcaused a substantial increase in amino acid fluctuation. The Ca2þ-induced fluctua-tion changes were more substantial for farnesylated proteins, which showed reduced C-terminal fluctuations in the absence of Ca2þ.
conformational switch to expose the myristoyl group, which
is responsible for membrane association (
39
). Unlike
recov-erin, lamin A tail domains are mostly disordered, therefore
the conformational switch is much less obvious. We
demon-strated through experiments and simulations that
D50LA-TD and pre-LA D50LA-TD undergo a conformational change
induced by Ca
2þthat opens the Ig-fold and increases the
exposure and mobility of the farnesylated C-terminus.
The Ca
2þ-induced conformational change may also
regu-late protein-membrane interactions, protein-DNA, or
pro-tein-protein interactions. The large size, intrinsic disorder,
and Ig-fold binding pocket of the tail domain may all allow
for promiscuous binding and may be at least partially
responsible for the ability of lamin A tail to bind different
partners (
1
). Here, we have shown Ca
2þ-induced
conforma-tional changes, and protein binding to the lamin A of one of
the many lamin A-associated proteins may also induce
conformational changes in the tail domain. We find that,
similar to Ca
2þ, Mg
2þcan facilitate the membrane
associa-tion of fn-
D50LA-TD, and is the subject of future work.
Membrane association of farnesylated lamin A tail
domain includes binding and aggregation
Simulations of the
D50LA-TD structure show the protein
diameter to be 5.2
5 0.2 nm, and a complete monolayer
of protein coverage is reached at ~1
mM of solution protein
concentration. However, above 1
mM we still observe
addi-tional, stable membrane association. This additional
associ-ation is likely due to a combinassoci-ation of aggregassoci-ation coupled
with binding. Because high salt levels are required to
pre-vent aggregation of fn-preLA-TD, and proteins show no
membrane association as monomers, we suggest that the
protein-protein association of fn-preLA-TD overwhelms
the protein-membrane association with excess protein
(
40
). This highlights that fn-
D50LA-TD has an apparent
greater affinity for the membrane and enhanced membrane
association of
D50LA may contribute to the pathological
membrane association (
12,13
).
Lifetime of wild-type lamin A and pathological
D50LA
Most contemporary literature suggests that prelamin A
far-nesylation is responsible for the association of protein to the
inner nuclear membrane for further incorporation into the
lamina (
10,11
). It is unclear why lamins do not embed
them-selves into membrane structures outside the nucleus because
farnesyltransferases have been located in the cytoplasm near
the endoplasmic reticulum and inside the nucleus.
From the results presented here, we suggest changes in
the tail domain of preLA from the time it is produced
from LMNA mRNA until it is processed to form mature
lamin A. We speculate that after transcription, the
C-termi-nal CaaX domain is hidden inside the TD of preLA. The
preLA (probably as a dimer) is transported into the nucleus
via the TD nuclear localization sequence (NLS). In the
nucleus, the higher levels of charge from DNA and counter
ions induce a conformational change that exposes the
CaaX for farnesylation and other modifications leading to
FIGURE 5 Simulation of farnesyl dynamics of preLA andD50LA TD. We preequilibrated the far-nesyl group within the hydrophobic interior of the respective TDs and allowed equilibration in explicit solvent with 50 mM NaCl or 50 mM NaCl þ 2 mM CaCl2. (A) Comparison of thefour most significant pseudoequilibrated structures of fn-preLA-TD. With no Ca2þthe farnesyl group (red) was found within the molecule. In the pres-ence of Ca2þ, the farnesyl group escaped from the TD. (B) The solvent accessible surface area of the farnesyl group, given by the number of water molecules in contacting with the surface of farnesyl group (threshold distance of 0.38 nm), of fn-D50LA-TD and fn-preLA-TD increased in Ca2þ. (C) The end-to-end length of the entire TD of fn-D50LA-TD and fn-preLA-TD also increased, suggesting that the Ca2þ makes the C-terminus more extended.
Biophysical Journal 104(10) 2246–2253
insertion into the inner nuclear membrane. There are high
levels of divalent cations including calcium, magnesium,
and zinc inside the nucleus (
41
), but exact levels of free
divalent cations in the nucleus are difficult to determine
because free ions are buffered by surrounding chromatin
structure. It is also possible that small amounts of preLA
may be farnesylated by cytoplasmic farnesyltransferases.
We suggest that these farnesyl groups may incorporate
into the Ig-fold until they are exposed to high Ca
2þlevels.
Ig-folds and similar
b-sheet structures have been shown to
act as solubilization agents for lipidated proteins (
42–44
).
In the case of fn-preLA, the farnesylation could associate
with its own Ig-fold or that of a neighbor.
SUPPORTING MATERIAL
Five figures and supplemental methods are available at http://www. biophysj.org/biophysj/supplemental/S0006-3495(13)00445-1.
We gratefully acknowledge the assistance by Matthew Biegler (CMU MSE) and Peter Yaron (CMU ChemE) and we are grateful to Katherine Wilson and Susan Michaelis (Johns Hopkins Department of Cell Biology) for care-ful reading of the manuscript.
Z.Q. and M.J.B. received support from National Science Foundation (NSF) (CMMI-0642545) and Office of Naval Research-Presidential Early Career Award for Scientists and Engineers (ONR-PECASE) (N00014-10-1-0562). K.N.D. and A.K. received support from the Progeria Research Foundation, NSF (CBET- 0954421 CAREER) and National Institutes of Health (NIH) National Institute on Aging (NIA) (NRSA F30-AG030905 to A.K.). M.L received support from NIH (1R01-GM101647).
REFERENCES
1. Simon, D. N., and K. L. Wilson. 2011. The nucleoskeleton as a genome-associated dynamic ‘network of networks’. Nat. Rev. Mol. Cell Biol. 12:695–708.
2. Dittmer, T. A., and T. Misteli. 2011. The lamin protein family. Genome Biol. 12:222–304.
3. Dahl, K. N., and A. Kalinowski. 2011. Nucleoskeleton mechanics at a glance. J. Cell Sci. 124:675–678.
4. Herrmann, H., H. Ba¨r,., U. Aebi. 2007. Intermediate filaments: from cell architecture to nanomechanics. Nat. Rev. Mol. Cell Biol. 8:562–573.
5. Barrowman, J., C. Hamblet,., S. Michaelis. 2008. Analysis of prela-min A biogenesis reveals the nucleus to be a CaaX processing compart-ment. Mol. Biol. Cell. 19:5398–5408.
6. Eriksson, M., W. T. Brown,., F. S. Collins. 2003. Recurrent de novo point mutations in lamin A cause Hutchinson-Gilford progeria syn-drome. Nature. 423:293–298.
7. Qin, Z., A. Kalinowski,., M. J. Buehler. 2011. Structure and stability of the lamin A tail domain and HGPS mutant. J. Struct. Biol. 175:425–433.
8. Scaffidi, P., and T. Misteli. 2006. Lamin A-dependent nuclear defects in human aging. Science. 312:1059–1063.
9. Davies, B. S., C. Coffinier,., L. G. Fong. 2011. Investigating the pur-pose of prelamin A processing. Nucleus. 2:4–9.
10. Hennekes, H., and E. A. Nigg. 1994. The role of isoprenylation in membrane attachment of nuclear lamins. A single point mutation pre-vents proteolytic cleavage of the lamin A precursor and confers mem-brane binding properties. J. Cell Sci. 107:1019–1029.
11. Lutz, R. J., M. A. Trujillo,., M. Sinensky. 1992. Nucleoplasmic local-ization of prelamin A: implications for prenylation-dependent lamin A assembly into the nuclear lamina. Proc. Natl. Acad. Sci. USA. 89:3000– 3004.
12. Dahl, K. N., P. Scaffidi,., T. Misteli. 2006. Distinct structural and mechanical properties of the nuclear lamina in Hutchinson-Gilford progeria syndrome. Proc. Natl. Acad. Sci. USA. 103:10271–10276. 13. Goldman, R. D., D. K. Shumaker,., F. S. Collins. 2004.
Accumula-tion of mutant lamin A causes progressive changes in nuclear architec-ture in Hutchinson-Gilford progeria syndrome. Proc. Natl. Acad. Sci. USA. 101:8963–8968.
14. Reddy, S., and L. Comai. 2012. Lamin A, farnesylation and aging. Exp. Cell Res. 318:1–7.
15. Kubben, N., M. Adriaens,., T. Misteli. 2012. Mapping of lamin A-and progerin-interacting genome regions. Chromosoma. 121:447–464. 16. Silvius, J. R., and F. l’Heureux. 1994. Fluorimetric evaluation of the affinities of isoprenylated peptides for lipid bilayers. Biochemistry. 33:3014–3022.
17. Nadolski, M. J., and M. E. Linder. 2007. Protein lipidation. FEBS J. 274:5202–5210.
18. Philips, M. R. 2012. Ras hitchhikes on PDE6d. Nat. Cell Biol. 14:128–129.
19. Taimen, P., K. Pfleghaar,., R. D. Goldman. 2009. A progeria muta-tion reveals funcmuta-tions for lamin A in nuclear assembly, architecture, and chromosome organization. Proc. Natl. Acad. Sci. USA. 106:20788–20793.
20. Vivian, J. T., and P. R. Callis. 2001. Mechanisms of tryptophan fluores-cence shifts in proteins. Biophys. J. 80:2093–2109.
21. Chen, Y., and M. D. Barkley. 1998. Toward understanding tryptophan fluorescence in proteins. Biochemistry. 37:9976–9982.
22. Shenoy, S., R. Moldovan,., M. Losche. 2010. In-plane homogeneity and lipid dynamics in tethered bilayer lipid membranes (tBLMs). Soft Matter. 2010:1263–1274.
23. Vockenroth, I. K., C. Ohm,., I. Ko¨per. 2008. Stable insulating teth-ered bilayer lipid membranes. Biointerphases. 3:FA68–FA73. 24. Valincius, G., D. J. McGillivray,., M. Lo¨sche. 2006. Enzyme activity
to augment the characterization of tethered bilayer membranes. J. Phys. Chem. B. 110:10213–10216.
25. Sugita, Y., and Y. Okamoto. 1999. Replica-exchange molecular dynamics method for protein folding. Chem. Phys. Lett. 314:141–151. 26. Brooks, B. R., R. E. Bruccoleri,., M. Karplus. 1983. Charmm: a pro-gram for macromolecular energy, minimization, and dynamics calcula-tions. J. Comput. Chem. 4:187–217.
27. Feig, M., J. Karanicolas, and C. L. Brooks, 3rd. 2004. MMTSB Tool Set: enhanced sampling and multiscale modeling methods for applica-tions in structural biology. J. Mol. Graph. Model. 22:377–395. 28. Vanommeslaeghe, K., E. Hatcher, ., A. D. Mackerell, Jr. 2010.
CHARMM general force field: a force field for drug-like molecules compatible with the CHARMM all-atom additive biological force fields. J. Comput. Chem. 31:671–690.
29. Shenoy, S., P. Shekhar,., M. Lo¨sche. 2012. Membrane association of the PTEN tumor suppressor: molecular details of the protein-mem-brane complex from SPR binding studies and neutron reflection. PLoS ONE. 7:e32591.
30. McGillivray, D. J., G. Valincius,., M. Lo¨sche. 2007. Molecular-scale structural and functional characterization of sparsely tethered bilayer lipid membranes. Biointerphases. 2:21–33.
31. Dhe-Paganon, S., E. D. Werner,., S. E. Shoelson. 2002. Structure of the globular tail of nuclear lamin. J. Biol. Chem. 277:17381–17384. 32. Krimm, I., C. Ostlund,., S. Zinn-Justin. 2002. The Ig-like structure of
the C-terminal domain of lamin A/C, mutated in muscular dystrophies, cardiomyopathy, and partial lipodystrophy. Structure. 10:811–823. 33. Swendsen, R. H., and J. S. Wang. 1986. Replica Monte Carlo
34. Young, S. G., L. G. Fong, and S. Michaelis. 2005. Prelamin A, Zmpste24, misshapen cell nuclei, and progeria—new evidence sug-gesting that protein farnesylation could be important for disease path-ogenesis. J. Lipid Res. 46:2531–2558.
35. Leung, K. F., R. Baron,., M. C. Seabra. 2007. Rab GTPases contain-ing a CAAX motif are processed post-geranylgeranylation by proteol-ysis and methylation. J. Biol. Chem. 282:1487–1497.
36. Parish, C. A., A. V. Smrcka, and R. R. Rando. 1996. The role of G pro-tein methylation in the function of a geranylgeranylated beta gamma isoform. Biochemistry. 35:7499–7505.
37. Hancock, J. F., K. Cadwallader,., C. J. Marshall. 1991. A CAAX or a CAAL motif and a second signal are sufficient for plasma membrane targeting of ras proteins. EMBO J. 10:4033–4039.
38. Polychronidou, M., A. Hellwig, and J. Grosshans. 2010. Farnesylated nuclear proteins Kugelkern and lamin Dm0 affect nuclear morphology by directly interacting with the nuclear membrane. Mol. Biol. Cell. 21:3409–3420.
39. Xu, X., R. Ishima, and J. B. Ames. 2011. Conformational dynamics of recoverin’s Ca2þ-myristoyl switch probed by 15N NMR relaxation dispersion and chemical shift analysis. Proteins. 79:1910–1922. 40. Marshall, C. J. 1993. Protein prenylation: a mediator of protein-protein
interactions. Science. 259:1865–1866.
41. Tvedt, K. E., G. Kopstad,., J. Halgunset. 1987. Subcellular concen-trations of calcium, zinc, and magnesium in benign nodular hyperplasia of the human prostate: x-ray microanalysis of freeze-dried cryosec-tions. Cancer Res. 47:323–328.
42. Rak, A., O. Pylypenko,., K. Alexandrov. 2003. Structure of Rab GDP-dissociation inhibitor in complex with prenylated YPT1 GTPase. Science. 302:646–650.
43. Tian, W., J. Yu,., X. Luo. 2010. Structural and functional analysis of the YAP-binding domain of human TEAD2. Proc. Natl. Acad. Sci. USA. 107:7293–7298.
44. Hoffman, G. R., N. Nassar, and R. A. Cerione. 2000. Structure of the Rho family GTP-binding protein Cdc42 in complex with the multifunc-tional regulator RhoGDI. Cell. 100:345–356.
Biophysical Journal 104(10) 2246–2253