| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
* Dipartimento di Fisiologia e Biochimica Generali, Ia Università di Milano,
CNR, Centro di Farmacologia Cellulare e Molecolare e Dibit, Instituto Scientifico San Raffaele, San Raffaele, Italy; and
Dipartimento di Biologia Cellulare e dello Sviluppo, Università `La Sapienza', Roma, Italy
1Correspondence: Department Biologia Cellulare e dello Sviluppo, Univ. `La Sapienze', piazalle A. Moro, 5, 00185 Roma, Italy.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: NPCs • nuclei isolation • histogram analysis • single-channel currents
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3)
. The
envelope itself functions as a semipermeable device, dividing
nucleoplasm from cytosol. In a single lipid bilayer, proteins like
ionic channels or transporters are sufficient to ensure the
communication; however, in a system composed of two biological
membranes, a more complex communication mechanism is probably required.
Many recent studies deal with the control of intracellular traffic in
and out the nucleus (4
, 5
), and it is now clear that the movement of
molecules and RNAs is regulated in a precise manner (6
, 7
). It is also
clear that the main route for particles to cross the envelope are the
nuclear pore complexes
(NPC)2
(5)
. The architecture of the pore complex allows the free diffusion of
small particles and ions. Proteins above 40 kDa and RNA molecules
actively shuttle in and out the nucleus; most of the proteins that
permeate the nuclear envelope must expose a nuclear localization
sequence, a sort of molecular key (3)
.
The view of the pore as a crystallized aperture unable to modify its
diameter has recently been challenged by patch-clamp experiments
showing ionic channel activity on the surface of isolated nuclei (8
,
9
10
11
12
13
14
15
16
17
18
). Along with ionic pathways classified as reticulum channels and
thought to reflect the typical ion channel endowment of ER cisternae, a
large ionic pathway with a conductance ranging between 300 and 400 pS
(depending on animal species and experimental conditions) has been
discovered. Great interest was elicited by the presence of this large
ionic conductance, a distinctive feature of nuclei soon to be
considered as the sign of the presence of NPCs. However, the
morphofunctional correlation between the number of current levels
corresponding to the 300–400 pS conductance and the number of pores
per patch failed to show adequate correspondence (19)
. This discrepancy
led to the hypothesis that under particular conditions, NPCs might open
to a different extent and even close (20)
. Based on the known structure
of the pore complex, at least two mechanisms can be hypothesized to
account for the control of the conduit: either the cooperative
association of several small channels or regulation of pore size by its
subunits acting as a variable iris (21)
. This would imply that under
certain conditions, the pores could control not only the permeability
of large proteins and RNAs, but also small molecules and ions. Recently
it was shown that intracellular calcium elevation above 300 nM reduces
the diffusion of the divalent ion in the nucleoplasm (22)
and
down-regulates the big nuclear ionic channel (20)
. Also, ATP appears to
play a modulatory role in the permeability of the pore complex. A
reduction of its concentration below 500 nM was reported to lower the
open probability of the channel (13)
. The study of NPC structure by
atomic force microscopy showed a conformational change on ATP addition
(23
, 24
). Finally, the large nuclear channel is also regulated by
cytoskeletal components. It has long been known that actin and myosin
are integral components of the pore complex (25
, 26
), and the
assembly/disassembly of nuclear actin filaments has recently been
proposed to be directly involved in the modulation of the
single-channel current (27)
.
Since ATP and calcium are molecules that could interact with the cytoskeleton elements, our investigation is based on the idea that NPC aperture is controlled by the structures linked to the nucleoskeletal network. The function of nucleoskeletal proteins associated with the nuclear pore complex, depending on the concentration of several soluble molecules, could determine the state of nucleocytoplasmic communication.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Patch-clamp recording
The patch electrodes were pulled from hard borosilicate glass
(Corning 7052) on a Brown-Flaming P-87 puller (Sutter
Instrument, Movato, Calif.). The pipettes were coated with Sylgard (Dow
Corning, Midland, Mich.) and fire polished to an external tip diameter
of 1–1.5 µm. These electrodes had resistances of 7–10 M
. We
applied the standard patch-clamp technique to obtain nucleus-attached
patches between 20 and 50 G in resistance and single-channel
recordings. In all the experiments, the patch pipette contained 120 KCl
solution (above). Single-channel currents were recorded with an
Axopatch-1D (Axon Instrument, Novato, Calif.) patch-clamp amplifier and
were digitized and stored on a VCR (Panasonic, Milan, Italy). Data were
analyzed on a Mitsuba computer after filtering at 1000 Hz using custom
made programs developed by W. Gooldby, Department of Anatomy and Cell
Biology, Emory University, Atlanta Georgia.
Nuclear patch area
Measurement of the area isolated by the patch electrode on the
nuclear membrane is very important in the present study. Our pipettes
have a diameter of 1–1.5 µm and a maximum calculated area of 1.76
µm2. Direct measurements on nuclear envelope of
the area with patch capacitance gave values of around 15
µm2, comparable to measurements obtained on the
cell membrane (28)
. This high value could be due to two factors: the
double membrane of the envelope and the omega shape of the membrane
patch in the pipette. Using AMF microscopy, we recently obtained an
average patch area of around 5 µm2 (29)
with a
larger pipette (5 M). Since our pipettes are on average half this
size, we assume that 2–3 µm2 of membrane would
be encircled by the patch electrode. Furthermore, according to the
classical description of patch geometry (30)
, we presume that of the
membrane sucked into the electrode tip, only one-third could be active,
allowing current flow. The rest of the membrane inside the electrode
tip is laid against the pipette wall, providing a value of 1
µm2 or less as a working area.
Freeze-fracture
Livers from adult and 2-day-old rats were fixed in 1%
glutaraldehyde (Fluka, Buchs, Switzerland) and 0.5% formaldehyde
(freshly prepared from paraformaldehyde) in 0.1 M phosphate buffer. pH
7.2. After a few minutes, small pieces were cut from the livers and
maintained for an additional hour in the fixative solution. At the end
of this period, the specimens were glycerinated (final concentration
30% in 0.1 M phosphate buffer), sandwiched between two copper holders,
and frozen in nitrogen slush (31)
. They were stored in liquid nitrogen
and finally fractured at -110°C and 5 x
10-8 Torr with a complementary replica device in
a freeze fracture apparatus (Cryofract 190, Reichert Jung S.A., Paris).
The platinum-carbon replicas were cleaned in sodium hypochlorite, taken
up on uncoated grids, and examined with a Hitachi H 7000 (Hitachi Ltd,
Tokyo, Japan) electron microscope.
Morphometric analysis
Electron micrographs of replicas of nuclei were printed at a
final magnification of x34,000. The diameter and the density of NPCs
were measured and counted only on the flat portions of the replicas,
i.e., on adequately metallized regions where the angle of shadowing was
constant. The means were compared by Student's t test for
unpaired observation. A probability level (P >0.05) was
used to define significance throughout the study.
Immunoblot analysis
Proteins from each fraction (50 µg) were resolved by 12%
sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE)
(32)
and transferred overnight to a nitrocellulose sheet according to
ref 33
. After blocking nonspecific sites with blocking solution
provided by Boehringer Mannheim (Mannheim, Germany), the sheet was
incubated overnight with 2.5 µg/ml rabbit monoclonal anti-actin
(Boehringer Mannheim) or polyclonal anti-myosin nonmuscle (Biomedical
Technology) antibodies. A peroxidase-labeled secondary antibody and the
chemiluminescent substrate luminol were used to detect the first
antibody according to the manufacturer's instructions (BM,
Chemiluminescence Western blotting kit, Boehringer Mannheim). The
molecular weight was determined using Bio-Rad standard proteins. Equal
loading of protein on the gel was verified by 12% SDS-PAGE stained
with Coomassie brilliant blue R250. The results were analyzed by
densitometric analysis using an ImageMaster software (Pharmacia,
Piscataway, N.J.).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 35
). In light of the structural connections between
the nucleus and cytoskeleton, the idea is to investigate whether there
is a functional relation between these structures—in particular,
whether cytoskeletal elements are involved in a mechanism possibly
controlling the aperture of the pores. In fact, we know not only that
the pore complex contains actin and myosin (25
, 26
), but also that the
fibrous network of the lamina (36)
interconnects the pores.
Figure 1
represents the starting point of our study. We performed on-nucleus
patch-clamp experiments on isolated nuclei from 2- and 18-day mouse
livers. The current pattern of younger nuclei is greater at all
potentials (n=25). Adult nuclear single-channel recordings
have on average two or three current levels. Some reach 300 pS
conductances, but most show intermediate states. Neonatal nuclei, on
the contrary, show many more current levels; for the inward current in
particular, they appear to be equally spaced.
|
Amplitude histogram analysis of inward and outward neonatal nuclear
channels reveals several conductive states, most having conductances of
300 pS. Previous studies have suggested that the big ionic conductance
identified with patch-clamp on isolated nuclei is indeed the NPC acting
like a conventional ionic channel (8
9
10
11
12
13
, 15
, 18
). If this is true, one
of the most probable differences between adult and neonatal nuclei
could be the density of NPCs; however, this is not the case.
Figure 2
presents freeze-fracture experiments on both types of nuclei in which
the number of pores per unit area is not significantly different. In
adult mouse, as in adult rat (19)
, the average pore density is 12 ± 1.8 pores/µm2 (n=6), not
significantly different from neonatal nuclei (13±2.5
pores/µm2; n=8). The morphological
data obtained from neonatal nuclei are extremely useful since the
number of pores/unit area matches, in the first approximation, the
measured 300 pS current levels in patch-clamp on-nucleus experiments.
Figure 3
depicts the amplitude histogram (top left) of a current recordings (top
right) obtained from neonatal animal liver nuclei at 30 mV voltage step
applied to the recording pipette. The experimental single-channel
traces show several current levels, equally spaced. Since the
channel(s) never comes to a completely closed state, it is problematic
to subtract the leakage current. In any case, we assume only the
histogram peak values as channel openings. Over 14 experiments, we
averaged a number of 10 ± 3.4 apparent ionic pathways having
~300 pS conductance. In adult preparations, we often observed
openings corresponding to ~300 pS, but never above two or three units
and usually composed of smaller current sublevels from 25, 50, or 100
pS already characterized in previous investigations (20)
. The number of
300 pS current levels recorded in adult nuclei is generally 2.8 ±
1.2 (n=52). It appears, therefore, that the 300 pS channels
in neonatal nuclei are very close in number to the pore complex units
present on the nuclear membrane. The problem remains that in adult
preparations, even if the morphology is comparable to neonatal nuclei,
current levels are limited to few occurrences. So far, the only way
known to increase the patch current in adult preparation is to add ATP
in the external and the pipette solutions (13)
.
|
|
Figure 4
shows the effect of 0.5 mM ATP in the pipette solution during
on-nucleus experiments. Amplitude histograms obtained from 5 s
current recordings at -30 (top) and -40 mV (bottom) applied in the
electrode show the increase in 300 pS current levels upon application
of 1 mM ATP perfused in the external solution. Each histogram depicts
an example of the raw data (see insert). ATP was able to increase the
number of active channels, but the increments were always limited and
never able to come close to the number of pores present in the patch.
Increasing ATP in the external solution limited the ability to observe
single-channel events (20)
. Since the number of pores is not a
limitation in adult cell nuclei, the absence of multiple 300 pS
single-channel openings could be due to a down-regulation of the
mechanism maintaining the channel/pore in the open state. A possible
explanation could be found in the organization of the cytoskeletal
network. Actin was demonstrated to be involved in isolated nuclei
single-channel modulation (27)
.
|
To further investigate the relationship between cytoskeleton and
nuclear traffic, we carried out an immunoblot analysis of two
well-known cytoplasmic cytoskeletal proteins, actin and nonmuscle
myosin, in neonatal and adult nuclei from mouse liver. The localization
of both actin and myosin has been associated with the NPCs (25
, 26
),
and several hypotheses on nucleocytoplasmic traffic view the two
proteins as being involved in the functional aspect of the pore
complex. Figure 5
shows an increased expression of adult actin and nonmuscle myosin with
respect to neonatal nuclei of 87% and 57%, respectively
(n=4; actin increases 82 ± 11.5%, nonmuscle myosin
increases 61 ± 14.2%). The increase of actin contents in adult
nuclei could cause the partial closure of NPCs in our experimental
conditions due to the absence of ATP. If so, we should be able to
up-regulate channel activity by partially or totally destroying
cytoskeletal elements. Preincubation of isolated nuclei with a
cytoskeletal disrupter agent causes shrinkage of nuclear membranes,
making any attempt to obtain a tight seal between patch pipette and
nuclear envelope impossible. However, the latter experiment provided a
way to be sure that the nucleoskeletal network was intact and that
actin filaments still participated to maintain the structure of
isolated nuclei. The nucleus-attached configuration was then reached in
the control solution and the cytoskeletal disrupter agent was slowly
perfused. The electrical signal from the envelope patch was
continuously monitored to capture any transient increase in the number
of current levels. During on-nucleus experiments, in which one or two
300 pS ionic channel activity could be recorded, the whole nucleus was
superfused by a solution containing 10 µM cytochalasin. Figure 6
depicts examples, from three different nuclei, of the functional
transitions that occurred in the membrane patch current when the entire
nucleus was exposed for several minutes to cytochalasin. The pipette
voltage was held constant at 50 mV. After various times (from 1 to 6
min) in which the current recordings showed one, two, or no channel
apertures corresponding to a conductance of 300 pS each, the current
amplitude increased. During this period, several current levels that
progressively appeared were easily recognized. After a few seconds of
this massive current activation, the ionic flow eventually became too
large and went off scale. Probably the electrode lost contact with the
membrane due to shrinkage of the nucleus. We separated portions of
current recordings in control solution from the ones after cytochalasin
addition. From each trace we isolated the short period (1–2 s) in
which the current increase in amplitude and single-channel openings was
clearly visible. The amplitude histograms shown at the bottom of Fig. 6
quantify these results. The graph on the left (relative to control
conditions) shows three peaks that represent two current levels whose
conductances are 300 pS each. In the histogram on the right (relative
to the current recordings after cytochalasin addition), it is possible
to recognize several peaks (six or more). The conductance value between
adjacent peaks is constant and has a value between 300 and 320 pS.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 37
). The potential difference
recorded across the envelope (8
, 38
, 39
) can easily be explained by a
Donnan equilibrium generated by nucleoplasmic charge retention. Another
factor that contradicts the idea that NPCs could be ion selective
concerns the mechanism of channel gating. If conformational changes due
to protein charge displacement are required to open and close a few Å
holes, it is arduous to hypothesize a device able to gate a 9 nm pore.
The above considerations appear sufficient to lay at rest the
conjecture that the nuclear envelope could be a semipermeable
interface. However, increasing experimental evidence demonstrates that
under specific conditions, the nucleus is able to modulate diffusion of
solutes to and from the cytoplasm. Nuclear resting potential
measurements should be analyzed with more precision. If nuclear voltage
buildup were due to nucleoplasmic ionic retention or different
nucleocytoplasmic ion activity, it should be independent from the type,
stage, and intracellular content of the cell. On the contrary, there
are nuclei with and without measurable potential across the envelope
(38)
. Furthermore, even if the observation was not consistent (40)
, in
some cell nuclei localized disruption of the nuclear envelope cancels
the voltage difference (41)
. From this we can conclude that nuclear
resting potential is probably due to many factors, including Donnan
equilibrium, and under some conditions to changes occurring in nuclear
envelope permeability.
It is hard to imagine a semipermeable envelope without taking into
consideration the ability of NPCs to become ion selective. The major
problem in sustaining this hypothesis is the absence of direct evidence
that the large channels recorded with patch-clamp and NPCs are the same
phenomena. Here we present experiments that correlate the envelope
ionic permeability with pore distribution. In neonatal liver nuclei,
the density of NPCs is comparable to the number of current levels
detected with the patch-clamp on-nucleus technique. The homogeneity of
the single-channel amplitude, with a conductance of 300–320 pS, can be
numerically compared with the number of pores present on the membrane
area isolated by the electrode. The calculation of pore density
indicates 13 ± 2.5 pores/µm2. The working
patch area should not exceed 1 µm2 (see
Materials and Methods). We are confident that the average number of
current levels (10±3.4) underestimates the real value, since the open
probability of the nuclear channel, even if it is high, is not 100%.
We rarely observed an apparent closure of all the permeabilities
present in the membrane patch. This could certainly be due to current
leakage, even if all the patch seal resistances were over 10 G. A
second, more likely, possibility is the absence of single events at
lower current amplitude due to the high open probability of the
channels that never come to a complete closure (Fig. 1)
.
There are other ionic pathways on neonatal envelope, but they have much lower conductances, never reaching such a high number and never appearing as constant in number and amplitude from patch to patch.
Is 300 pS a compatible conductance for a NPC? According to
microscopical and biochemical measurements, the pore has an opening of
~9 nm and a calculated conductance of 850–1000 pS (8)
. Channel
openings of almost 1 nS or more have been observed (9
, 13
), but these
amplitudes are not the rule and may be the result of contemporaneous
opening of several ionic pathways probably working together. In
previous experiments, the nuclear conductances of 25 to 1000 pS have
been described (8
9
10
11
12
13
14
15
16
17
18)
. In the present study, we obtained a more
homogeneous result by using neonatal preparations. Though several pore
complexes are certainly present, we cannot say the same for all the
other endoplasmic reticulum-like channels. The constant current values
suggest a unique ionic pathway that, for the coincidence with pore
density, could be represented by the nucleocytoplasmic conduit. The
value of 300 pS is compatible with the possibility that ionic
permeability is always less than the calculated conductance obtained
from microscopical measurements. Micrographs obtained with different
techniques indicate that the diameter of the external part of the
protein complex is not necessarily similar to the internal opening
(42)
.
From these considerations, we can conclude that it is highly probable
that single-channel signals from patch-clamp on-nucleus experiments in
neonatal nuclei represent the current passing through the NPCs. If the
numbers of pores and current levels are comparable in neonatal nuclei,
the same was never demonstrated for all the adult preparations used
previously (19)
and in this work. Even if the channels recorded in
adult liver nuclei showed 300 pS conductance (Fig. 1
and Fig. 4
), their
numbers appear to be underestimated compared with the NPC density. It
is unusual in adult preparations to record more than two or three high
conductance current levels. The use of ATP in the recording electrode
and external solution is a way to increase the patch current by many
fold, to the point where it is impossible to recognize the transitions
between different levels (13)
. However, if the concentration of ATP is
carefully modulated, we are able to observe an increase in the patch
current and maintain the ability to witness singular events in which
conductance is around 300 pS (Fig. 4)
. These results suggest that in
order to obtain single-channel current traces with many levels, as in
neonatal preparations, we have to unblock an ATP-dependent mechanism
that keeps the majority of ionic conductances in the closed state. A
different way to increase single openings during patch-clamp recordings
in adult preparations is to use a cytoskeleton disrupter such as
cytochalasin (Fig. 6)
, which is able to reveal many more 300 pS
single-channel events before the entire structure collapses. The
cytoskeletal disrupter may cause an increase in the patch leakage, and
probably does so after several seconds. Our observation is based on the
first few seconds in which we assisted in the activation of several
conductances. We are confident that the gating mechanism of the
channels is preserved during this short time. The ionic pathway showed
both opening and closing kinetics, as in traces 1 and 2 of Fig. 6
,
suggesting that the seal between the pipette and the membrane was not
yet compromised.
From this latter observation we can deduce that the difference between
adult and neonatal nuclear ionic permeability could be due to the
difference in cytoskeletal components (Fig. 5)
. This does not imply a
direct link between NPCs and the cytoskeletal network. As demonstrated
by different researchers, actin and myosin are in tight association or
are intrinsic parts of the pore complex (25
, 26
). We suggest that
cytochalasin, in addition to disassembling the nucleoskeletal network,
is also able to interact with the actin molecules associated with the
NPC that eventually are connected with the nucleoskeleton.
To hypothesize a modulatory mechanism for nuclear ionic permeability
pathways within the pore complex, we have to consider many factors:
1) adult nuclei present few current levels up-regulated by
ATP and blocked by Ca2+ (20)
; 2)
neonatal preparation have a number of single-channel openings
compatible with the NPC density; 3) actin and nonmuscle
myosin are less concentrated in neonatal nuclei (Fig. 5)
; and
4) cytoskeleton elements are nuclear channel modulators
(27)
.
From the present study and previous experiments, we propose two
distinct behaviors that can be identified during nuclear ionic
permeation. First, a mechanism involving the single pore complex, which
can open to a maximum of 300 pS conductance. On the basis of the
channel kinetics (20)
and from structural studies (43)
, we have already
hypothesized that the pore could open and close like a diaphragm (44)
.
The single pore forming an ionic conduit can show either a fully open
channel (300 pS conductance) or current substrates. This mechanism,
dependent on ATP and cytoplasmic calcium, involves the intrinsic
cytoskeletal elements, actin and nonmuscle myosin, associated with the
pore complex. A disassembly of this association would abolish the
presence of substrates in the 300 pS pathway current recordings. A
second mechanism involves adjacent nuclear pore complexes. The
interconnection between the protein complexes is well documented in the
nucleoplasmic side and is promoted by the nuclear lamina (36
, 45
).
There is the possibility, shown by analysis of electrophysiological
data (20)
, that nuclear pores are cooperative structures. This
hypothesis could explain large current jumps, multiples of 300 pS,
occurring without intermediate levels (see Figs. 1
, 3
, 4
, 6
).
The scenario emerging from these results suggests a mechanism involving
cytoskeletal structures as predominant components for nucleocytoplasmic
ionic flux modulation. Actin filaments, with the probable contribution
of myosin, are not only responsible for the nuclear structure, but also
play an active role in the permeability of the envelope. In this
respect, the more organized the cytoskeletal network, the more chance
the nucleus has to control passive diffusion when needed. NPC, or part
of it, does not behave like a conventional ionic channel. Without the
cytoskeleton elements, the pore is unable to reduce its aperture to
become ion selective. The gating mechanism is not only modulated as in
conventional ionic channels, but is directly dependent on the presence
of cytoskeletal elements. Moreover, we suggest that ATP and
Ca2+ act directly on the actin and myosin
molecules that are intrinsic components of the pore complex.
Conformational changes of pore complexes due to these two compounds
were described recently using atomic force microscopy in fixed and
fresh nuclear envelope preparations (23
, 24
). Physiological conditions
(high ATP and low Ca2+) keep the pore constantly
in the full open state, where passive diffusion is free and large
proteins are transported by the NPC machinery. A decrease in the energy
content or an uncontrolled Ca2+ rise in the
cytoplasm could induce a change in the nucleocytoplasmic cytoskeletal
elements association, causing a decrease in nuclear permeability, which
represents a dynamic protection for the nucleus.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication January 4, 1999.
Revision received February 18, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. Schermelleh, P. M. Carlton, S. Haase, L. Shao, L. Winoto, P. Kner, B. Burke, M. C. Cardoso, D. A. Agard, M. G. L. Gustafsson, et al. Subdiffraction Multicolor Imaging of the Nuclear Periphery with 3D Structured Illumination Microscopy Science, June 6, 2008; 320(5881): 1332 - 1336. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Mazzanti, J. O. Bustamante, and H. Oberleithner Electrical Dimension of the Nuclear Envelope Physiol Rev, January 1, 2001; 81(1): 1 - 19. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
