| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Biochemistry, University of Cambridge and MRC Dunn Human Nutrition Unit, Cambridge, U.K.
2Correspondence: Division of Endocrinology, Beth Israel Deaconess Medical Center and Harvard Medical School, 99 Brookline Ave., Boston, MA 02215, USA. E-mail: skrauss{at}caregroup.harvard.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: energy metabolism • lymphocytes • metabolic control analysis • mitogenic activation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3)
.
Signaling systems display a high degree of complexity, and it seems
difficult, if not impossible, to understand signal transfer in
biological systems on a systemic level. For instance, it is not clear
how the same receptor may induce separate responses under different
conditions (4)
. Moreover, despite the multitude of
signaling processes that are initiated during stimulation of cells,
cellular responses often appear finely tuned (5)
,
providing evidence that there must be mechanisms of intracellular
signal processing to prevent overdiversification of biological
responses. Such filtering may leave just a few pathways whose
activation is causally related to a particular cellular response.
Metabolic control analysis is a theoretical framework that has
facilitated quantitative, precise statements about regulation and
control in metabolic pathways (6)
. It is founded on
rigorously developed theory and has produced many useful practical
applications in experimental systems. It is able to deal with systems
of any complexity or architecture, and it should therefore be possible
to apply it to cellular signal transduction. In this paper we present
an experimental approach to signal transduction based on the concepts
of metabolic control analysis. We use the simple and robust model
system of mitogenic activation of energy metabolism in thymocytes.
Living cells require energy to maintain cellular integrity and basal
metabolic functions. Cellular energy metabolism has to be able to
respond rapidly to changes in a cell’s status. To achieve this, cells
need effective signaling networks that transmit external (or internal)
stimuli so that the bioenergetic machinery responds appropriately. The
regulation of energy metabolism is not simply a question of increased
ATP demand (7
, 8)
, so signals that lead to increases in
respiration and ATP production are likely to have many intracellular
targets.
Con A stimulation of thymocytes provides a simple model system in which
to quantify signal transduction. Con A, a T cell mitogen, acts as
polyclonal antigen and stimulates thymocyte steady-state respiration by
30–40% irrespective of antigen specificity (9)
. This
stimulation is stable over 25 min (10)
, so the system
changes from one pseudo steady state to another, allowing simpler
experimental and theoretical analysis. Several signaling pathways are
involved in the activation of T cells (11
, 12)
. With
respect to T cell activation, cell signaling is often discussed in the
context of cytokine biosynthesis. However, most of the signal
transduction processes activated on T cell stimulation will have a
bearing on energy metabolism. The changes in thymocyte steady-state
respiration caused by Con A are a result of a complex cascade of
events, involving recognition of cell surface receptors, engagement of
different signaling pathways, and the subsequent adjustment of process
rates of cellular targets. Any quantitative description of regulation
should be able to indicate the relative importance of each of the
various signaling routes that trigger a change in respiration. In
addition, each signaling pathway may have effects on several of the
reactions that affect steady-state respiration. A quantitative
description of regulation should also be able to indicate for each
signaling route the relative importance of activation of different
reactions in stimulating respiration. Finally, it should be possible to
integrate all of the quantitative measurements into a single coherent
overall picture of signal transduction.
One aim of this work is to show that it is possible and illuminating to measure the relative importance of different signaling pathways in a cell, and then to quantify roughly the way these signal pathways interact to regulate a designated cellular reaction. A second aim is to understand mitogenic activation of cellular respiration more fully.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). Figure 1A
shows the basic level used to analyze the
relative importance of different signaling pathways. The system
consists of a cellular variable, thymocyte respiration, which is
responsive to changes in the input level of the signal (Con A) that
triggers T cell activation. Con A affects steady-state respiration via
a number of signal transduction pathways, probably TCR-linked
(13)
. Here, we shall consider the following four signal
routes: protein kinase C, the MAPK pathway, calcineurin, and a single
block of unrelated unidentified pathways. These four pathways are each
defined by inhibitor specificity; thus, we define the calcineurin
pathway as the part of the signal that is inhibited by cypermethrin.
|
Figure 1B
shows the extended level used to analyze the way
each signaling pathway acts on each part of oxidative phosphorylation
to give the overall stimulation. The system is simplified with respect
to the number of signaling intermediates but extended to include blocks
of reactions that produce or consume mitochondrial membrane potential,
m.
The relative importance of different signaling pathways:
quantitative topology
Regulation of respiration by signal transduction pathways can be
quantified using inhibitors of selected signaling pathways. We have
chosen three pathways thought to play a pivotal role in T cell
stimulation and for which specific inhibitors are available, including
PKC, the MAPK pathway, and calcineurin. Although interpretation is
easier if inhibitors with absolute specificity are used, this is not
strictly necessary, as the signaling pathways are defined by the
specificity of the inhibitors used. By using inhibitors alone and in
combinations and measuring their effects on the Con A-induced increase
in respiration, we can work out how much of the signal passes through
each inhibitable pathway and which pathways overlap. Thus a
quantitative topology can be established.
Basic concepts of metabolic control analysis
The central concepts in metabolic control analysis are
represented by three coefficients. They are control, elasticity, and
response coefficients (for more detailed descriptions and further
references, see ref 6
). A flux control coefficient quantifies the
amount of control that a step or a block of reactions exerts on a
pathway flux. It normally assumes values between 0 (no control) and 1
(total control). Elasticity coefficients are more central to the work
described here. They describe how a step in a pathway or a block of
steps vary with the concentration of a metabolite that affects it
(either a metabolic intermediate or an effector external to the
system). Variations in an intermediate or an external effector will
cause steps that are sensitive to their levels to change (as quantified
by the elasticities), and this effect will be relayed through a
metabolic network according to the control that the steps within the
network exert on the network fluxes. Such adjustments in response to
internal or external effectors are quantified by response coefficients.
The extended level of analysis: regulation of processes producing
and consuming a bioenergetic intermediate- mitochondrial membrane
potential
We can analyze how strongly each signaling pathway affects
respiration through the two blocks of reactions that produce or consume
m. To do this we multiply the effect of the
signaling pathway on each block (IE) by the control exerted
by the block over respiration rate (C), using two equations
given in ref 14
 |
 |
q),
Ji/Ji
is the fractional change in the rate of block i,
is the elasticity of block
i to intermediate x,
x/x is the fractional change in
x, IR is the integrated response of variable
a to q, and C is the
control coefficient of i over process a. The
integrated elasticity
(equation 1)
quantifies the direct modulation of a block by a step
change in effector q, which here is the action of a
signaling pathway. The term ‘integrated’ reflects the fact that large
changes in flux are achieved as a result of manipulation of signal
transduction pathways (14)
. To determine the response of
respiration to the change in each signaling pathway activity, we can
use equation 2
, which states that the response of the system is the sum
of the weighted change in activity of each block as a result of the
parameter change. The weighting factor is the control coefficient of
the block over respiration.
Thus, the total response can be partitioned into individual effects
that the parameter change has on the subsystems or blocks. Assume, for
instance, that Con A affects respiration through PKC by acting on the
m producers and the
m consumers. Starting from a stimulated
state, the following analysis could be made. Measure the effect of
addition of a PKC inhibitor on respiration and
m. In separate experiments, determine the
elasticities of each of the blocks to m by
a suitable method (e.g., performing titrations with myxothiazol or
oligomycin) (10)
(see Fig. 3
: the elasticities of the
blocks to the intermediate in the stimulated state are given by scaling
the slopes of lines 1 and 2 in panel A). Correct for the
indirect effects of the PKC signal on each block through changes in
mitochondrial potential by calculating the integrated elasticity of
each block to the PKC signal using equation 1
. The control coefficients
of the blocks over respiration can be determined from the elasticities
to m as described in ref 15
. With these
data, the response of respiration to Con A via PKC can be determined
and broken down into partial integrated responses via each of the
blocks using equation 2
. There are assumptions made in the calculation
of the partial responses that were discussed in detail in ref 14
.
|
Isolation and incubation of cells
Small thymocytes were isolated from female white Wistar rats
(4–8 wk old). Animals were killed by cervical dislocation. The thymus
was removed, immediately transferred and washed in 10 ml RPMI 1640
medium/1% FCS, and pressed through a nylon mesh into 3 ml of RPMI 1640
medium/1% FCS. The crude cell suspension was allowed to stand for a
few minutes, and small thymocytes were retrieved by density gradient
centrifugation as described in ref 16
. The final cell density was
5 x 107 cells/ml. Cell incubation
medium was RPMI 1640 with glutamine, without glucose (Sigma, St. Louis,
Mo.). For density gradient centrifugation, RPMI 1640 medium with
glutamine and glucose was used (ICN). Medium also contained
NaHCO3 (25 mM), HEPES (10 mM), and gentamicin (20
µg/ml). Before use, media were filtered through 0.22 µm cellulose
acetate discs (Millipore, Bedford, Mass.), and adjusted to pH 7.4 with
2.5M HCl.
Measurement of oxygen consumption and mitochondrial membrane
potential
Thymocytes were transferred to 4-ml plastic vials mounted on
submersible stirrers and kept at 37°C under 5%
CO2/95% air. Cells were pretreated with
inhibitors (as appropriate) as follows: Cypermethrin (2 µM, 30 min)
to inhibit calcineurin (17)
, bisindolylmaleimide I (1
µM, 30 min) to inhibit protein kinase C (18)
, and PD
98059 (20 µM, 60 min) to inhibit MAPKK (19)
. Twenty
micrograms of Con A per 5 x 107 cells and,
where appropriate, myxothiazol (8 nM) and oligomycin (2.5 ng/ml) were
added to the cells. Oxygen consumption was measured in Clarke-type
electrodes and m was determined using
[3H]TPMP+ as described in
ref 10
.
Statistical analysis
Respiration and membrane potential values were normalized for
every preparation to an appropriate reference (quiescent or stimulated
respiration), and data from different preparations were expressed as
means ± SE. Because the raw data were used more than
once in the calculations, and the errors in the calculated coefficients
were not normally distributed, there were no obvious simple statistical
tests that could be used to give a quantitative statement about the
reliability of the final results. Therefore, we used Monte Carlo
analysis to assess significance of values of responses to Con A via
particular signaling pathways (see refs 10
, 20
).
P1 values were assigned to indicate
how many simulated values in the distribution for a particular response
had the opposite sign to the one calculated directly from averaged
data. In a similar way, 
P2 values
were used to express the significance of the difference between two
values.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
30%.
The effects of inhibiting signal transduction pathways on respiration
in Con A-stimulated thymocytes are shown in Figure 2
. Inhibition of PKC, MAPKK, or calcineurin depressed the stimulation by
Con A to different degrees. PKC inhibition suggests that it carries
55% of the signal. Pathways related to the MAPK cascade and
calcineurin showed slightly weaker contributions (40 and 30%,
respectively). However, the signal reduction achieved by combining the
three inhibitors was 80% (Fig. 2A
). This is inconsistent
with an independent role of all three intermediates in transmitting the
signal.
|
Figure 2B
shows effects on Con A-stimulated respiration of
pairs of inhibitors. PKC and calcineurin together accounted for 80%
of the transmitted signal, which is close to the sum of individual
effects (54 and 31%, respectively), suggesting that the two pathways
are independent. Concomitant inhibition of the PKC and MAPK signaling
pathways indicates that 50% of the signal goes through these
pathways. This is not significantly different from the signal through
PKC alone, showing that the MAPK signal is entirely contained within
the PKC pathway.
MAPK and calcineurin accounted for 70% of the signal. Individual
inhibition of these pathways indicates that they account for 40 and
30% of the Con A signal, respectively. Again, the sum is close to the
combined effect that was observed when inhibitors were combined,
suggesting independent pathways.
From these considerations, a weighted topological picture emerges (Fig. 2C
). About 54% of the signal is transmitted through the PKC
pool of signaling intermediates. The MAPK pathway participates in the
stimulatory event, but is downstream of PKC. Calcineurin mediates 30%
of the signal transfer independently of the other systems. Less than
20% of the signal must be ascribed to unidentified signaling systems.
The relatively strong role of calcineurin in transducing the Con A
signal was unexpected. One of calcineurin’s main roles is thought to
be the dephosphorylation of NFAT (21
, 22)
, and the high
contribution to the increase in oxygen consumption rate during early
stimulation is surprising. Cypermethrin might target other phosphatases
or unidentified processes. However, there is growing evidence that
calcineurin is linked to mitochondrial functions (23
, 24)
,
and the present result could point toward a role of calcineurin in
bioenergetic mitochondrial activities that remains as yet obscure.
Extended system: how each signaling pathway acts on different parts
of oxidative phosphorylation to give the overall stimulation
The results shown in Figure 3
were used to measure the effects of the PKC and calcineurin signaling
pathways on different blocks of energy-transducing reactions so that we
could calculate how the signals caused the observed stimulation of
respiration. Results were normalized to the condition with Con A
present (central filled squares), which therefore has respiration and
mitochondrial membrane potential values of 100%. Relative to this
point, quiescent cells (central open squares) respired at 77% but had
the same mitochondrial potential. The inhibition of PKC and calcineurin
(independently in Fig. 3A
, B
and in combination
in Fig. 3C
) depressed respiration to intermediate values
(central circles). In agreement with the results reported above, 50%
of the response of oxygen consumption to Con A was via PKC (Fig. 3A
) and 27% was via calcineurin (Fig. 3B
).
Quiescent and stimulated cells had similar values of
m. When the PKC or calcineurin pathways
were inhibited, m was slightly (but not
significantly) elevated compared with the stimulated reference value.
This indicates that PKC and calcineurin might act to decrease
m during stimulation, implying that they
may stimulate the consumers and/or inhibit the producers of
m slightly. If so, their effect is
counteracted by unidentified pathways resulting in no change in
m between quiescent and activated cells.
Top-down elasticity analysis: transition from the quiescent to the
stimulated state
Decrease of the mitochondrial potential by addition of myxothiazol
(to inhibit electron transport) allowed measurement of the overall
kinetic response of the m-consuming
pathways to m under the different
conditions in Figure 3
. In a similar way, increase of the mitochondrial
potential by addition of oligomycin (an inhibitor of the ATP synthase,
part of the m-consuming pathways) allowed
measurement of the overall kinetic response of the
m-producing pathways to
m (10
, 25)
. The slopes of
these relationships were used to calculate elasticity and control
coefficients. The transition from the quiescent state (open squares in
Fig. 3
) to the stimulated (reference) state (filled squares) was
characterized around the quiescent value of
m by the following two major effects:
1) the m-consuming pathways were
directly stimulated, because at any given potential their rate was
higher with Con A present; and 2) the
m-producing pathways were directly
stimulated, because their rate was also higher with Con A present.
As the potential did not change significantly when Con A was added, the
direct stimulation of the producers was equal to the direct stimulation
of the consumers, leading to no change in m
at this concentration of Con A. It is also apparent that the elasticity
of the m-producers to
m increased in the presence of Con A (i.e.,
the slope of line 2 was steeper in the stimulated condition).
Effect of the inhibitors bisindolylmaleimide I and cypermethrin
When thymocytes were pretreated with inhibitors of PKC or MAPKK,
the kinetic profiles of the m-producing and
consuming blocks changed during stimulation. Figure 3A
shows
that, in the presence of bisindolylmaleimide I, the consumers (line 1,
open circles) were depressed compared with the stimulated reference
state (filled squares), but the m-producers
(line 2, open circles) were unchanged. Pretreatment of thymocytes with
cypermethrin to inhibit calcineurin (Fig. 3B
) allows similar
conclusions: the consumers (line 1, open circles) were depressed
compared with the stimulated reference state (filled squares). The
m-producers (line 2, open circles) were
unchanged or may even have been stimulated. When both
bisindolylmaleimide I and cypermethrin were present (Fig. 3C
), a strong inhibitory effect on the consumers was
apparent (line 1, open circles), but it is unclear whether the
m-producers (line 2, open circles) were
different from the stimulated reference state. We conclude that the
m-consumers are stimulated by PKC and
calcineurin, in comparison the m-producers
are probably unaffected or maybe slightly inhibited through these
signaling pathways.
The importance of the remaining unidentified pathways becomes evident
from the transition between the quiescent state (open squares) and
state with Con A, bisindolylmaleimide I and cypermethrin present (open
circles in Fig. 3C
). Both the
m-producing and
m-consuming blocks were stimulated by these
unidentified pathways, but the stimulation of the
m-consuming reactions may not have been
significant. The m-producers were
stimulated significantly compared with the quiescent state and were
stimulated significantly more than the
m-consumers.
Flux control coefficients of the m-producing and
consuming blocks of reactions
The elasticities of the m-producing
and consuming blocks to m were measured
from Figure 3
and used to calculate the flux control coefficients of
the two blocks over respiration. These coefficients describe how
strongly each block controls respiration rate (15)
.
Respiration was largely controlled by the
m-consumers (control coefficients 0.6–0.9
under all conditions) with less control exerted by the
m producers (control coefficients
0.1–0.4). These values were then used in Equation 2
to calculate the
partial integrated responses that describe how strongly the different
signaling pathways activate respiration by acting on each of the
reaction blocks.
PKC and Con A stimulation of thymocyte respiration
Figure 4A
shows the partial integrated responses that quantify how
PKC contributes to the stimulation of respiration. The signal through
PKC (50% of the total Con A signal) is partitioned into two unequal
subsignals. Most of the PKC signal acts on respiration rate directly
through the m-consuming block, and little
or none acts directly and independently through the
m-producing block. This is consistent with
the observation (Fig. 3)
that during PKC inhibition in stimulated
cells, m may increase slightly.
|
Calcineurin and Con A stimulation of thymocyte respiration
Fig. 4B
shows how calcineurin contributes to the
stimulation of respiration. Calcineurin conveys 27% of the Con A
stimulus. The signal via calcineurin has two components: a stimulatory
effect on the consumers and a (smaller) inhibitory effect on the
producers of m. This is consistent with the
observation (Fig. 3)
that during calcineurin inhibition in stimulated
cells, m may increase slightly.
Unidentified pathways and Con A stimulation of thymocyte
respiration
In principle, we can calculate the contribution of the
unidentified pathways of signal transduction using the data in Figure 3
. Instead of measuring the effects of inhibition of PKC and
calcineurin, we can calculate by difference the properties of the
components that are insensitive to inhibition. Although calculation by
difference is prone to several types of error and the conclusions are
therefore not reliable, we found that these unidentified pathways
stimulate the m-producers more strongly
than the consumers (26)
.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
summarizes our conclusions and provides an approximate quantitative
topology of the regulation of respiration in thymocytes by Con A. It
shows the relative importance of different signaling pathways:
PKC-related pathways carry 54% of the Con A signal,
calcineurin-related pathways carry 30%, and unidentified signaling
pathways carry the remainder. It also shows the interactions of these
signaling mechanisms with metabolic reactions: the stimulatory effects
of Con A on respiration acting through PKC and calcineurin mostly
target the consumers of m and the
m-producers are much less important
targets. As the intermediate, mitochondrial membrane potential, does
not change on Con A stimulation, the stimulatory effects of PKC and
calcineurin working through the m-consumers
must be balanced by an approximately equal activation of the
m-producers by the unidentified pathways.
|
Experimental control analyses have so far not considered signal
transduction pathways. Signal transduction constitutes an important
feature of biological systems: it provides the bridge between external
stimuli and cellular reactions. However, signal transduction pathways
are often extremely complex. Metabolic pathways can be viewed as a
number of processes that interact via linking intermediates. They are
characterized by mass exchange and are often at steady state. If an
intermediate is changed (for instance by changing the kinetic
properties of a block of reactions), the system will usually settle to
a new steady state. For signaling pathways, however, it appears much
more difficult to formulate such a clear conceptual view. Often, there
is no mass exchange, and several systems that are characterized by
different temporal and spatial aspects have superimposing effects. The
number of mechanisms seems endless; protein kinase cascades may work as
rapid switches that produce all-or-nothing responses (27)
;
calcium signals encode and transmit information in frequency- or
amplitude-modulation mode (28)
; feedback in signaling
systems can build up over wide time frames, spanning 30 min and more
(29)
, and the same signal may have a multitude of
temporally distinct effects (13)
.
The approach used here gathered large parts of metabolism or signal
transduction into black-box groups of reactions (30)
. It
simplified the application of control analysis to a system comprising
both metabolic and signaling components. The explicit treatment of
temporal and spatial aspects of certain signals was circumvented by
defining the time frame of observation and by relating signal
transduction events to cellular variables in systems that are, to a
first approximation, at steady state. In our experiments, we chose to
consider only a small number of signals and targets, but by using more
inhibitors and measuring more intermediates the resolution of our
method could be enhanced. The resolution of the method as we have used
it here depends only on the availability of specific inhibitors and on
achievable experimental accuracy.
Our results clearly distinguish the analysis from classical studies. For example, it is firmly established that mitogen action increases cellular oxygen consumption, but this is where conventional analyses stop. The use of control analysis allows us to go beyond such a qualitative characterization: it is possible to compare the relative importance of, for example, PKC in the stimulatory process with calcineurin and other unidentified pathways. Calcineurin was not expected to carry as much as 30% of the signal, showing how the analysis forces us to think in terms of topologies that are defined by the system itself and not by our expectations. In addition, the presence of unidentified signaling reactions can be discovered. Even without knowing their identities, the system properties and quantitative contributions of the unidentified reactions can be determined (albeit crudely) by difference.
The approach introduced here and applied experimentally to a model system allows a quantitative description of signal transduction that gives a novel and revealing insight into signal transduction and offers a starting point for the exploration of new relationships between signaling pathways and their cellular targets.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication February 21, 2000.
Revision received May 24, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. J. Lambert and B. J. Merry Effect of caloric restriction on mitochondrial reactive oxygen species production and bioenergetics: reversal by insulin Am J Physiol Regulatory Integrative Comp Physiol, January 1, 2004; 286(1): R71 - R79. [Abstract] [Full Text]
|
|
|
|
 |
|
 J. Wang, R. S. Al-Lamki, H. Zhang, N. Kirkiles-Smith, M. L. Gaeta, S. Thiru, J. S. Pober, and J. R. Bradley Histamine Antagonizes Tumor Necrosis Factor (TNF) Signaling by Stimulating TNF Receptor Shedding from the Cell Surface and Golgi Storage Pool J. Biol. Chem., June 6, 2003; 278(24): 21751 - 21760. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A Kuhnke, G-R Burmester, S Krauss, and F Buttgereit Bioenergetics of immune cells to assess rheumatic disease activity and efficacy of glucocorticoid treatment Ann Rheum Dis, February 1, 2003; 62(2): 133 - 139. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. N. Kholodenko, A. Kiyatkin, F. J. Bruggeman, E. Sontag, H. V. Westerhoff, and J. B. Hoek Untangling the wires: A strategy to trace functional interactions in signaling and gene networks PNAS, October 1, 2002; 99(20): 12841 - 12846. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Pfitzer Signal Transduction in Smooth Muscle: Invited Review: Regulation of myosin phosphorylation in smooth muscle J Appl Physiol, July 1, 2001; 91(1): 497 - 503. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
