| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
Centre for Genetics of Cardiovascular Disorders, University College of London Medical School, Department of Medicine, The Rayne Institute, 5 University St., London WC1E 6JJ, UK.
a Author for correspondence. Fax + 44 171 209 6212, e-mail s.humphries{at}medicine.ucl.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
Key Words: indexing terms: microtiter array diagonal gel electrophoresis • polymerase chain reaction • LDL receptor gene • inherited disease
|
Introduction |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
|
General Principles of the SSCP Technique |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
-32P]dCTP is incorporated during the
PCR, and diluted PCR product is denatured by a brief boiling step,
after which the sample is loaded on a nondenaturing "sequencing"
(40 cm) acrylamide gel. The samples thus adopt a single-stranded
secondary structure because of the formation of interstrand base
pairing. However, although the samples have been diluted, they
concentrate upon entry into the gel, and this allows some reannealing
to take place. The considerations implicit in classical "Cot"
studies of reannealing (concentration of each single strand, DNA
complexity, and time) apply (4). Thus, to reduce the
proportions of PCR product that reform as a double strand and reduce
the amount available as single strand for polymorphism analysis, the
sample needs to be relatively dilute. When using thin gels with small
capacity for sample volume, this concentration consideration may limit
the choice of DNA detection methods, with radioisotope detection being
the method of choice because it is considerably more sensitive than
ethidium bromide or even silver-staining methods.
The "classical" SSCP protocol thus includes 32P in
the PCR mix for maximal detection sensitivity in the diluted sample,
and a gel with a long track length for maximal resolution of small
mobility differences in the DNA fragments (see Fig. 1
). However, in the interests of higher efficiency of detection,
greater convenience, or safety, several studies have been made with
other protocols. Restriction digestion of a large PCR fragment before
SSCP has been described (5). Multiplex PCR is also
possible, and SSCP can be combined with allele-specific PCR
(6)(7) to select alleles in complex sets
(e.g., HLA genes), or with dideoxy chain termination to localize the
approximate position of an SSCP variation in a sequence
(8).
|
In SSCP analysis, heteroduplex bands are often seen on the gel as a
useful byproduct of the procedure (9). Reannealing to form
double-stranded DNA has been discussed above, and when higher
concentrations of single-stranded DNA are loaded, e.g., for nonisotopic
detection methods, the double-stranded band is the predominant band.
For human DNA (except for the sex chromosomes), a PCR product usually
represents amplification from two alleles, one of each autosome of a
pair. When a sequence variation is present in the heterozygous state,
the classical SSCP picture would be of four single-stranded bands, the
sense and antisense strands of the "normal" allele, and the altered
mobility sense and antisense strands of the "variant" allele. In
addition, reannealing to double strands permits four possible products,
"normal" double strand, "variant" double strand, and two
heteroduplexes ("normal" sense/"variant" antisense; and
"normal" antisense/"variant" sense). The heteroduplexes have a
mismatch "bubble" that often alters their mobilities relative to
homoduplex, resulting in additional bands (see Fig. 1
). The combined
efficiency of SSCP plus heteroduplex analysis for detection of sequence
variations should be well in excess of the 80% estimated for each
individually (2).
Since RNA base pairing is more stable than RNA-DNA base pairing, RNA might be expected to adopt a more conformational structure and hence be more sensitive to sequence changes. Published evidence suggests that this is the case (10) with detection of up to 95% of variations, but the inconvenience of making RNA strands (which involves the complexity and expense of introducing RNA polymerase promoters and extra reaction steps) for SSCP has precluded widespread use. For FH, a suitable source of RNA would be from lymphocytes, but success is dependent upon equal representation of mRNA concentrations from the normal and mutant allele, and with premature stop codons or splice-site mutations this may not be the case (V. Gudnason, unpublished).
In our experience there are four important variables that must be considered in designing an optimal SSCP strategy. These are the length of the PCR fragment, the effect of temperature of the gel run, the method of PCR denaturation, and the characterization of the gel. These will be examined in turn.
single-strand length
The optimal length of a single strand seems to be ~150–200
nucleotides (11). In this size range, 70–90% of single
base substitutions are apparent on SSCPs. Presumably longer strands
exhibit relatively less conformational change by a single base
substitution, and shorter strands adopt less conformation in the first
place. On average, the reverse complement of a particular four-base
sequence will occur once every 4 x 4 x 4 x 4 =
256 nucleotides; thus it would be expected that a PCR single strand
would contain quite a few possible sequences to form four-base
double-stranded stems, but few significantly longer stems form
perfect-match duplexes. Such segments would be expected to melt and
hence lose any sequence-specific conformation below the melting
temperatures at which 10–15-mer oligonucleotides dissociate from their
target (12). The behavior of stem–loop folds would also
be expected to depend on guanosine plus cytosine content, since the
three-hydrogen-bond base pair, GC, is more heat-stable than the
two-hydrogen-bond base pair AT. We have observed for exon 4 mutations
in the LDL receptor gene that mobility shifts are more frequently
apparent in the upper strand, and one explanation for this might be
that the faster-migrating single strand, which is assumed to be more
compact, has sufficient stability so as not to be much affected by a
base change at the temperatures used. A better understanding of these
factors would be of value to be able to make predictive analysis of
specific sequences to maximize the mutation-detection efficiency of
SSCP in a fashion similar to the preplanning of denaturing gradient gel
electrophoresis experiments (13).
effect of temperature
Several studies have reported the use of different temperatures
for running SSCP, and typical conditions are either at room
temperature, or 4 °C with 50 or 100 mL/L glycerol. As discussed
above, the small regions of base pairing that are responsible for the
conformation of the single strands and thus the potential polymorphisms
are likely to have a melting temperature at or below the UK average
room temperature, and our experience is that during hot weather the
migration of bands changes considerably such that certain SSCPs are no
longer detected. The use of 4 °C standardizes these conditions, but
although we have not systematically investigated the effect of
temperature control on band sharpness or detection rate of SSCPs, we
and others have not found many additional SSCPs in samples run at other
than our standard conditions (14)(15). Gels
are routinely run at 5V/cm overnight at "room" temperature in an
air-conditioned room, which is typically ~22 °C, with minima and
maxima at 20 °C and 25 °C respectively. At this voltage, power
per gel is ~4 W, and gel warming is insignificant. A more expensive
option would be the use of a water-jacketed gel plate with a
circulating water bath.
pcr product denaturation
For convenience, PCR products are used without purification, but
spurious bands may result if the number of cycles is excessive or if
there is excessive residual PCR primer that may anneal to single
strands (16). By using high detection sensitivity for DNA
(i.e., 32P), the number of PCR reaction cycles can be
reduced (e.g., to 20) and the sample can be diluted 10–30-fold, which
will minimize annealing between single strands or between single
strands and PCR oligonucleotides. However, if less sensitive detection
methods are used (see below), less dilution is possible; stronger
denaturants added to the sample may help. Formamide, sodium hydroxide,
urea, and methylmercuric hydroxide (17) have been used.
Although toxic and requiring a fume hood, methylmercuric hydroxide
appears to be the most effective. Most protocols involve heating the
sample, immediate chilling on ice, then loading onto an apparatus
between 4 °C and 25 °C. A top layer of gel with formamide
incorporated has also been proposed to aid sample denaturation
(18), but this does not avoid the strand reannealing that
will occur when the single strands first enter the nondenaturing gel.
characteristics of the gel
Acrylamide is the commonly used matrix for DNA fragments in the
SSCP size range. The ratio of acrylamide to bisacrylamide cross-linker,
and the total acrylamide percentage, determine the sieving properties
of the gel. The buffer conductivity and concentration also influence
SSCP mobility, as do gel temperature and other additives such as
glycerol. Several publications have detailed the different effects that
these different conditions can have on resolving a particular sequence
variation. Reduced cross-linker ratio (bis:acrylamide 1:49) and 50–100
mL/L glycerol are popular (19), although other protocols
such as high-percentage gels can be useful (20). We have
found (Gudnason, unpublished) that the latter is true for a 340-bp PCR
fragment representing the 5' end of exon 4 of the LDL receptor gene.
However, there is no adequate theoretical basis to explain the
substantial influence of particular conditions in resolving certain
SSCPs, although some studies of folding and single-strand mobility have
been performed under nondenaturing and denaturing conditions (see
references in refs. 21 and 22). Hydrolink is an
acrylamide-like matrix polymerized by
N,N,N',N'-tetramethylethylenediamine (TEMED) and ammonium persulfate,
which is reported to have a more uniform pore size and is claimed to
give narrower bands and hence better resolution than acrylamide
(9).
Reported gel lengths range between 5 and 50 cm. At present, most results are read by eye and therefore visible resolution is necessary. Although clear-cut mobility shifts (e.g., 10%) are demonstrable on short gels, a long electrophoresis may be necessary to resolve 0.5% mobility difference. Long electrophoresis has the disadvantages of needing a large apparatus, a higher voltage power pack, and more complex arrangements to set up and control temperature. A longer run broadens a band in accord with basic theory (effects of an imperfect matrix and diffusion), and this may be compounded for SSCP without proper temperature control (i.e., nonuniform conformation and hence nonuniform mobility). Nevertheless, when it is important to avoid false-negative results, long track length is advisable.
|
Detection Methods for SSCP Single Strands |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
-32P]dNTP (or as
[
-32P]ATP end-labeling of primers or of PCR product)
involves the hazards and inconvenience of radioisotope usage. The main
options that have been explored are silver staining
(23)(24) and ethidium bromide fluorescence
(18). Ethidium bromide intercalates with 5–10-fold higher
affinity in double-stranded DNA, and is therefore not well suited to
single strand detection. Nevertheless, ethidium bromide staining is a
one-step process, and conditions to load sufficient single-stranded DNA
have been achieved (17)(18). Sensitive
silver-staining protocols are available for DNA, with detection down to
1–10 pg/mm2 (25). The catalytic process of
silver reduction is initiated on DNA bases. A typical protocol involves
deposition of silver nitrate on bands, then reduction of silver by
formaldehyde and sodium carbonate to give a brown color. The reaction
is stopped by acetic acid.
Large gels (e.g., 30 x 40 x 0.04 cm) are difficult to
handle for postelectrophoretic staining. Binding the gel to one glass
plate with -methacryloxypropyltrimethoxysilane is done for silver
staining, but this renders gel recovery (as a dried image on Whatman
3-mm paper) very difficult. A photographic film-based process can take
a direct imprint from the gel while simultaneously amplifying the
signal (Promega, Southampton, UK). We find that our staining protocol
can be used, with complete adherence of the gel to a glass plate, by
using 4 µL/cm2 of 0.5%
-methacryloxypropyltrimethoxysilane:0.5% glacial acetic
acid:ethanol, and that the gel can then be reliably recovered onto
Whatman 3-mm paper (26). The single-strand band intensity
depends on PCR fragment sequence and size (some reanneal more readily
than others) and tends to be faint.
A further possibility not yet fully explored is to blot the gel, most conveniently by "direct blotting." DNA is electrophoresed off the end of the gel onto a revolving blot (27), and then one of several methods of detection by hybridization can be used.
Automated DNA sequencers perform electrophoresis and detection on gels similar to manual sequencing gels. However, the detection system is usually fluorescence, either with oligonucleotides with fluorescent labels attached, or incorporating fluorescent nucleotides during polymerase reactions. Such apparatus can also be applied to SSCP by using fluorescent labels (28). The arguments pertinent to throughput, sensitivity of detection, and resolution are similar to those for automated sequencing (29). The main shortfalls are that one sequencer can only run one gel at a time, that there is high capital expenditure, that access to "primary data" is impossible if computer corrections for differential effect of different dyes on mobility are involved, and that there is the need for additional sophisticated workstations for secondary editing and interpretation of the data.
|
Rapid Throughput Modifications for Sample Handling |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
The setup of these PCRs is extremely simple, since the sample has zero volume, so that unless the exact quantity of template DNA is critical, any volume of a PCR master mix containing all components except template DNA can be added to the well. The imprecision introduced by pipetting is also minimized. The making and distributing of a PCR mix and oil to all wells of a dried plate with a repeating dispenser takes one worker ~10 min. We have also recently described, for PCR checking and other analyses, a system for preparing and stacking open-faced horizontal polyacrylamide gels, in which the wells retain the 96-well array [microtiter array diagonal gel electrophoresis (MADGE)] (31).
We have also modified the gel system to allow a 5–10-fold higher
throughput, and these modifications have been described in detail
(30). Briefly, two gels are poured between one large glass
sequencing plate (33 x 42 cm) and two outer smaller plates
(33 x 39 cm), so that two gels can be subjected to
electrophoresis simultaneously on one apparatus. Second, after first
loading the gel, the samples are electrophoresed at 400 V until the
bromphenol blue has run 2.5–3 cm, at which point the electrophoresis
is stopped and a second set of SSCP samples is loaded in the same way
as the first samples. Before loading the second time, the sharktooth
comb is removed and repositioned half a well across from its original
position. This enables the first and second loading bands to be
distinguished more easily. The first and second loading single and
double strands need to be nonoverlapping if the gel is to be
informative. The calculation of the timing of the second loading is
possible if the relative mobilities of the two single strands, one
double strand, and marker dyes are already known for the set of
conditions to be used. These mobilities were determined in a prior
experiment with single loading. For repetitive loading to be applied to
both the same and different exons, we prepared a computer simulation
that allows the solution of equations relating relative mobilities for
boundary conditions, including timing of loading, range of migration
distance desired, and time delays between loadings (available from G.P.
Weavind, 73064,3063@compuserve.com by e-mail request). Double
loading the gel means that a whole 96-well microtiter array of samples
may be analyzed on one gel. Finally, the sharktooth combs used are
custom cut with 4.5-mm spacing tooth-to-tooth to allow the use of a
multichannel pipette compatible with microtiter plates (9-mm
well-to-well spacing) for loading the SSCP gels. This results in every
second well being loaded with a multichannel pipette so that two
adjacent columns or rows of a microtiter plate are interleaved (see
Fig. 2
).
|
|
Familial Hypercholesterolemia |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
FH results from different genetic defects in a cell surface receptor, which normally controls the uptake of plasma LDL (1), or, in a small proportion of patients, one particular defect in the apolipoprotein (apo) B, the ligand for the receptor (37). This disorder, which is called familial defective apo B, has been reported to occur in ~3% of FH patients in the UK, and the mutation in the apo B gene causing it can be easily detected by PCR and allele-specific oligonucleotides.
For many heterozygous FH individuals, a clear diagnosis can be made on the basis of increased plasma cholesterol. However, several studies (e.g., 38) have shown that measures of total cholesterol or LDL cholesterol do not allow unequivocal diagnosis of FH in 10–15% of cases, and even measures of LDL receptor function on patients' monocytes show overlap between the values obtained for some "normal" individuals and patients with a defect in the receptor (39). By contrast, a genetic approach to identify LDL receptor defects gives an unequivocal result, and there is no doubt that the equivocal nature of the tests currently available to identify children with FH is one of the factors that deter some clinicians from actively pursuing such diagnosis in the relatives of a patient with FH. The advantage of an unequivocal DNA test would be both to allay fears for half of the relatives, and to identify children for whom monitoring dietary intake and appropriate therapy should be started.
|
The LDL Receptor and LDL Receptor Gene Mutations |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
, spans 45 kb and comprises 18 exons and 17 introns. Five
classes of mutations at the LDL receptor locus have been identified on
the basis of the phenotypic behavior of the mutant protein, and to date
there have been >200 different mutations of the LDL receptor gene
characterized at the DNA level (reviewed in 40 and
41). They have given valuable insights into the function of
the different domains in the LDL receptor.
|
Within a geographically or culturally isolated population, or where a large proportion of people are related by descent because of migration, there may be a single mutation causing FH in many of the patients (for refs. see 40 and 41). In the UK, where there is a very heterogeneous population, a priori it is unlikely that any mutations will be present at a high frequency, and so far no mutation detected has been present at a frequency >2–3% (41). Because of the genetic diversity present, this is likely also to be the case in most countries in Europe or in the US. Although it would be feasible to develop methods to screen for reported mutations, our calculations demonstrate that it is more cost and time effective to use an approach such as SSCP that will allow any mutation in the LDL receptor gene to be detected, rather than to screen specifically for all mutations.
|
Current Progress of SSCP Analysis on FH Research |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
In some situations, we wish to undertake a large number of different
PCRs simultaneously on the same DNA sample or on a small set, e.g.,
samples from four newly diagnosed FH patients. Instead of drying a
different template in each well, drying a different premixed pair of
PCR primers in each well is equally efficient in PCR yield
(42). This suggests that little of the oligonucleotide is
irreversibly bound to the plastic. The PCR premix is arranged to
contain a template DNA instead of a primer pair, and an example of an
SSCP autoradiograph from the simplified setup of 22 PCRs from the LDL
receptor gene in four patients with FH is shown in Fig. 4
. Typical PCRs in this laboratory are currently 20 µL, with 40
ng of template DNA and 8 pmol of each PCR primer. Eight picomoles of a
20-mer primer represents ~48 ng, so that a dried DNA template plate
contains 40 ng of DNA per well and a dried primer-pair plate contains
96 ng of oligonucleotide per well. Thus, on a weight basis, the layer
of DNA dried onto the plastic is of the same order for the two
strategies.
|
|
Interpretation of DNA Results |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
threonine change, with the threonine allele
occurring in 10% of healthy individuals and being associated with at
best a mild phenotype in men (43). Therefore, such
conservative missense mutations should be given a diagnosis of
probably causing FH in the patient, unless they have been
expressed in vitro and shown to cause either a defect in binding or in
the number of LDL receptor molecules on the surface of the cell.
Finally, with mutations in splice junctions, at intron–exon
boundaries, within the promoter or upstream region, within introns,
within the wobble position of amino acids, or within the 3'
untranslated region, formally, all such mutation must be treated as
possible or probable FH until expressed. However, where such a mutation
has been detected in more than one patient with FH, and where the
mutation occurs on different haplotypes (i.e., is an independent
mutation), and is not detected in the general population, such a
mutation can safely be designated as definite FH, but the predicted
effect on LDL receptor function must be made with care. Thus a mutation
in the conserved AG sequence that occurs within the intron side of the
exon–intron boundary is extremely likely to be causing a splice
defect, but it may not be possible to predict the exact effect on mRNA
splicing and therefore LDL receptor function. |
Future Developments in SSCP Mutation Screening for FH |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
A proportion of the patients with no detected SSCP may have been misdiagnosed as having FH, but it is also possible that a (unknown) proportion of patients may have hypercholesterolemia caused by a defect in another gene. In such families it would be possible to use DNA polymorphisms and a "cosegregation" approach to confirm or exclude the inheritance of the LDL receptor gene with the hyperlipidemia phenotype, and only depending on the results to identify relatives who have inherited the defective allele of the LDL receptor gene. There are several bi-allelic PCRable restriction fragment length polymorphisms that can be used for this purpose (46), as well as some hypervariable repeat polymorphisms (47). These polymorphisms are also useful in haplotyping to prove recurrent mutations.
At the present time the majority of detected mutations are novel, indicating that general screening methods such as SSCP will continue to be useful for FH for some time. However, in patients in the UK, ~30% of all detected SSCPs occur in exon 4 and 18% in exon 3 (41). This suggests that a useful strategy will focus on familial defective apo B and exons 3 and 4, and that this would lead to the rapid detection of the specific mutation in 11–14% of patients. In other parts of the world where founder mutations have been identified, or if common mutations or "region-specific" mutations are detected, it is possible that, as with mutations causing thalassemia, a targeted or sequential approach to specific mutation testing will be useful in the future.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionGeneral Principles of the...Detection Methods for SSCP...Rapid Throughput Modifications...Familial HypercholesterolemiaThe LDL Receptor and...Current Progress of SSCP...Interpretation of DNA ResultsFuture Developments in SSCP...References |
|---|
Stop. Hum Genet 1993;92:6-10.
[Web of Science][Medline]
[Order article via Infotrieve]
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  J. Vera-Otarola, M. I. Barria, U. Leon, P. Carvallo, A. Soza, and M. Lopez-Lastra Is Single-Strand Conformation Polymorphism Analysis of the Full 5' Untranslated Region an Adequate Approach To Study Hepatitis C Virus Quasispecies Distribution? J. Virol., September 1, 2009; 83(17): 9018 - 9021. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Ellis, D. A. Evans, C. R. H. Martin, and J. A. Hartley A 96-well DNase I footprinting screen for drug-DNA interactions Nucleic Acids Res., June 22, 2007; (2007) gkm467v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Kumar and P. K. Shukla Use of PCR Targeting of Internal Transcribed Spacer Regions and Single-Stranded Conformation Polymorphism Analysis of Sequence Variation in Different Regions of rRNA Genes in Fungi for Rapid Diagnosis of Mycotic Keratitis J. Clin. Microbiol., February 1, 2005; 43(2): 662 - 668. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A Gupta and S Agarwal Efficiency and cost effectiveness: PAGE-SSCP versus MDE and Phast gels for the identification of unknown {beta} thalassaemia mutations Mol. Pathol., August 1, 2003; 56(4): 237 - 239. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. G. Henderson, P. R. Wenham, J. P. Ashby, and G. Blundell Detecting familial defective apolipoprotein B-100: three molecular scanning methods compared Clin. Chem., September 1, 1997; 43(9): 1630 - 1634. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
