Note: Descriptions are shown in the official language in which they were submitted.
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
IMPROVED METHOD OF IDENTIFYING AND LOCATING
IMMUNOBIOLOGICALLY-ACTIVE LINEAR PEPTIDES
TECHNICAL FIELD
The present invention relates to locating protein epitopes and more
particularly to
novel methods for identifying, determining the location, and the optimal
length of
immunobiologically active amino acid sequences.
BACKGROUND OF INVENTION
Epitopes or antigenic determinants of a protein antigen represent the sites
that are
recognized as binding sites by certain immune components such as antibodies or
immunocompetent cells. While epitopes are defined only in a functional sense
i.e. by their
ability to bind to antibodies or immunocompetent cells , it is usually
accepted that there is a
structural basis for their immunological reactivity.
Epitopes are classified as either being continuous and discontinuous (Atassi
and
Smith, 1978, Immunochemisty, vol 15 p. 609). Discontinuous epitopes are
composed of
sequences of amino acids throughout an antigen and rely on the tertiary
structure or folding
of the protein to bring the sequences together and form the epitope. In
contrast, continuous
epitopes are linear peptide fragments of the antigen that are able to bind to
antibodies raised
against the intact antigen.
Many antigens have been studied as possible serum markers for different types
of
cancer because the serum concentration of the specific antigen may be an
indication of the
cancer stage in an untreated person. As such, it would be very advantageous to
develop
immunological reagents that react with the antigen, and more specifically,
with the epitopes
of the protein antigen.
To date, methods using physical-chemical scales have attempted to determine
the
location of probable peptide epitopes which includes looking at the primary
structure, that
being the amino acid sequence, secondary structure such as turns, helices, and
even the
folding of the protein in the tertiary structure. Continuous epitopes are
structurally less
-1-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
complicated and therefore may be easier to locate, however, the ability to
predict the
location, length and potency of the site is limited.
Various methods have been used to identify and predict the location of
continuous
epitopes in proteins by analyzing certain features of their primary structure.
For example,
parameters such as hydrophilicity, accessibility, and mobility of short
segments of polypeptide
chains have been correlated with the location of epitopes (see Pellequer et
al. 1991, Method
in Enrymolo~, vol 203, p. 176-201).
Hydrophilicity, has been used as the basis for determining protein epitopes by
analyzing an amino acid sequence in order to find the point of greatest local
hydrophilicity
as disclosed in U.S. Patent No. 4,554, 101. Hopp and Woods (See Proc. Natl.
Acad. Sci.
USA, vol. 78, No. 6, pp. 3824-3828, Jun. 1981) have shown that by assigning
each amino
acid a relative hydrophilicity numerical value and then averaging local
hydrophilicity so that ,
the location of the highest local average hydrophilicity values represent the
locations of the
continuous epitopes. However, this method does not provide any information as
to the
optimal length of the continuous epitope.
Likewise, the amino acid sequence of a protein as measured by the Kyte-
Doolittle
(Kyte and Doolittle, 1982, J. Mol. Biol. vol. 72, p. 105) scale, is commonly
used to
evaluate the hydrophilic and hydrophobic tendencies of polypeptide chains by
using a
hydropathy scale. Each amino acid in the polypeptide chain is assigned a value
reflecting its
- relative hydrophilicity and hydrophobicity which are averaged across a
moving section of the
sequence. This method offers a graphic visualization of the hydropathic
character of the
amino acid chain. It is theorized that by using the hydropathic character of
the sequence,
interior sequence regions which are usually composed of hydrophobic amino
acids can be
distinguished from hydrophilic exterior sequence regions. This information
offers the ability
to evaluate the possible secondary structure. However this model, does not
predict the
optimal length of the epitope or indicate if the effective size of epitopes is
unique for each
protein molecule.
Accordingly, what is needed is a simple method to identify immunobiologically-
active
peptide epitopes, determine their optimal length, and locations of these
epitopes within a
polypeptide.
-2-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
SUMMARY OF THE INVENTION
In accordance with this invention there is provided methods for identifying
immunobiologically-active linear peptide epitopes of a protein antigen and
determining the
optimal length of amino acid residues of the epitope.
TERMS
For purposes of this invention, the terms and expressions below, appearing in
the
specification and claims, are intended to have the following meanings:
"Window" as used herein means the number of amino acid residues in a curve
segment.
"Lagging" as used herein means to move across the entire amino acid residues
sequence increasing by one (1) in each step.
"Period number" as used herein means the number of amino acids assigned as the
period between -180° to + 180° in the negative cosine fi.~nction
plot.
"Fit-Correlation Value" as used herein means a numerical value which is
indicative
of the fit between the hydropathy plot curve and a negative cosine function
wherein the value
may be positive or negative depending on the fit. The better the fit the more
positive the
value.
"Epitope"as used herein means the portion of an antigen that binds
specifically with
the binding site of an antibody or a receptor on a lymphocyte.
"Potential Ho-Hi-Ho epitope" as used herein means an epitope wherein the curve
segment of the hydrophilicity plot correlates with the negative cosine
function giving a fit-
correlation value.
"Potential Ho-Hi-Ho epitope set" as used herein means a set of epitopes having
a
positive fit-correlation value for a specific period assigned to the negative
cosine curve.
"Ho-Hi-Ho theoretical epitopes" as used herein means the epitopes in the
potential
epitope set that have ranking values that exhibit the most oscillating
behavior about an
equilibrium position and either converge towards or diverge away from this
equilibrium
position and are deemed the most immunobiologically-active linear peptides.
"Number Range" as used herein means the numerated amino acid sequence number
-3-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
region of the amino acid sequence having a length equal to a period number,
i.e. if the period
is 10, then the sequence number ranges could be 1-10, 2-11, 3-12 and so on
until (n-(»t-1~~
where n is equal to the number of amino acid residues in the entire
polypeptide and m is the
period number.
Immune responses arise as a result of exposure to foreign stimuli. The
compound
that evokes the response is referred to as antigen or as immunogen. An
immunogen is any
agent capable of inducing an immune response. In contrast, an antigen is any
agent capable
of binding specifically to components of the immune response, such as
lymphocytes and
antibodies. The smallest unit of an antigen that is capable of binding with
various immune
components, either cells ,such as T and B lymphocytes, or antibodies, is
called an epitope.
Compounds may have one or more epitopes capable of reacting with immune
components.
The methods of the present inventions provide an in silica methodology for
determining the
antigen-binding site of an antibody or a receptor on a lymphocyte that has a
unique structure
that allows a complementary "fit" to some structural aspect of the specific
antigen.
Thus understood, a primary object ofthe present invention is to provide a
method for
determining immunobiologically-active linear peptide epitopes and their
optimal length.
Another object of the present invention is to identify immunobiologically-
active linear
peptide epitopes without the need for time consuming and expensive testing
regimes to
determine immunogenic activity, such as in vivo animal testing and/or in vitro
assay testing.
A further object of this invention is to determine the immunopotency of an
epitope
and provide a ranking system delineating between dominant and subdominant
epitopes.
A still fizrther object is to provide monoclonal and polyclonal antibodies
highly
specific for the peptide epitopes of the present invention which may be
utilized in diagnostic
testing procedures to determine the presence of an antigen is serum.
Yet another object of the present invention is to provide for synthetic
peptides from
a protein having the specific amino acid sequence and length determined by the
methods
herein that may be used in an immunization regime wherein the synthetic
peptides are
recognized by the body's immune system and induce production of immune
components such
as antibodies and/or immunocompetent cells, i.e. B and T cells that will react
with the peptide
or the entire protein.
Another object of the present invention is to provide a method to determine
the
-4-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
optimal length of a peptide that binds to antibodies and/or immunocompetent
cells.
Still another object is to provide for nucleic acid molecules encoding for the
immunobiologically-active linear peptide epitopes having an optimal length
found by the
methods disclosed herein.
The foregoing objects are achieved by fitting a hydrophilicity and/or
hydrophobicity
plot generated for the amino acid linear sequence of a polypeptide to a
mathematically
generated continuous curve which has at least a maximum positive value thereby
generating
potential epitope sets which include ranked potential epitopes which contain a
specific
number of amino acid residues. These sets of ranked potential epitopes may be
used to
determine immunobiologically-active linear peptides by comparison methods,
such as a
comparison between the sets to determine the set exhibiting the greatest
amount of oscillating
behavior about an equilibrium position; comparing the ranked potential
epitopes with other
epitopes generated by propensity scales; comparing with a previously generated
plot such as
hydrophilicity, accessibility, hydrophobicity and the like; and/or
combinations thereof.
Preferably, the set of potential epitopes that exhibit the most alternating
positioning about
an equilibrium position when juxtaposed on the hydrophilicity and/or
hydrophobicity plot are
deemed the immunobiologically-active epitopes. Their optimal length
corresponds to the
specific number of amino acid residues in the set of ranked potential
epitopes.
This invention relates to an improved method for determining the optimal
length of
an immunobiologically active epitope that does not require either in vivo
animal testing or
in vitro immunoassay testing regimes. Unexpectedly it has been discovered by
this inventor
that an alternating rhythmic pattern in the ranked potential epitopes provides
the necessary
information to determine the optimal length.
The method for determining the optimal length of an immunobiologically-active
linear
peptide epitope comprises the following steps:
a) providing a curve characterizing the hydrophilicity and/or
hydrophobicity of the linear sequence of amino acid residues of a polypeptide;
b) generating at least one potential epitope set comprising at least
one potential epitope by fitting a window of the curve of step (a) to a
mathematically generated continuous curve, the continuous curve having
repeating values at regular intervals with at least a maximum positive value,
-5-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
the window containing a specific number of amino acid residues and the
window is lagged through the curve of step (a);
c) increasing the number of residues in the window after each
lagging;
d) determining and ranking potential epitopes for each set by
selecting potential epitopes having a positive-fit correlation value
determined
by fitting curves in step (b) thereby providing a set of ranked potential
epitopes for each window of residues used in step (b), the most positive-fit
correlation value ranked first in each potential epitope set;
e) examining the positioning of at least the highest ranked
potential epitopes of each set relative to the plot of step (a) to determine
at
least one set of potential epitopes that exhibit alternating positioning about
an equilibrium position wherein the ranking values of the potential epitopes
converge towards or diverge away from the equilibrium position; and
f) designating the potential epitopes of the set having the most
alternating ranking values that converge or diverge as the immunologically
active epitopes which have an optimal length equating to numeric value of
amino acid residues in the potential epitopes.
Preferably, the potential epitopes are generated by fitting a hydrophilicity
curve
generated by plotting hydropathy values according to the prediction method of
Kyte
Doolittle and correlating this curve to a negative cosine fiznction thereby
generating Ho-Hi
Ho theoretical epitopes.
The method of the present invention may be used to determine the length of a
contiguous amino acid sequence of a polypeptide characterized by a hydrophobic
hydrophilic-hydrophobic motif, the method comprising the steps of
a) assigning an average hydropathy value to each amino acid of the
polypeptide;
b) generating a hydrophilicity plot using the average hydropathy
value of each amino acid;
c) fitting a curve segment of the hydrophilicity plot to a negative
cosine fixnction, wherein a specific period number value of the negative
cosine
-6-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
function equates to the number of amino acids in the curve segment, the period
number increasing within a predetermined chosen period number range after
each sequential lagging through the hydrophilicity plot thereby providing fit
correlation values for each curve segment across the linear sequence when
using
the specific period number value;
d) generating a potential Ho-Hi-Ho epitope set for each specific
period number value within the chosen period number range, wherein each
potential Ho-I~-Ho epitope set contains potential Ho-Hi-Ho epitopes that have
a fit- correlation value;
e) ranking each potential Ho-Hi-Ho epitope in the potential Ho-Hi-
Ho epitope set according to positive fit-correlation values wherein the
epitope
having highest positive-fit correlation value is ranked number one thereby
providing ranked Ho-Hi-Ho potential epitopes for each specific period number
value;
f) examining the positioning of at least the highest ranked Ho-Hi-Ho
potential epitopes of each set relative to the linear sequence of the
generated plot
in step (a) to determine at least one set of Ho-Hi-Ho potential epitopes that
exhibits alternating positioning about an equilibrium position wherein the
ranking values of the Ho-Hi-Ho potential epitopes converge towards or diverge
away from the equilibrium position; and
g) designating the Ho-Hi-Ho potential epitopes of the set having the
most alternating ranking values that converge or diverge as the
immunologically
active epitopes which have an optimal length equating to numeric value of
amino
acid residues in the potential epitopes.
The present invention further provides for a Ho-Hi-Ho epitope of contiguous
amino
acid residues from a polypeptide wherein the Ho-Hi-Ho epitope is defined by a
motif of two
hydrophobic and one hydrophilic regions arranged in the following manner
hydrophobic - hydrophilic - hydrophobic
and characterized by an approximated -180° to +180° negative
cosine hydrophilicity pattern
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
wherein said Ho-Hi-Ho epitope peptide has an optimal length of amino acid
residues from
about 3 to about 250. The optimal length of amino acid residues is determined
by the
methods of the present invention.
Also provided is an antisera specific for a Ho-Hi-Ho epitope of contiguous
amino
acid residues from a polypeptide wherein the Ho-Hi-Ho epitope is characterized
by a
hydrophobic-hydrophilic-hydrophobic motif and an approximated -180° to
+180° negative
cosine hydrophilicity pattern having an optimal length of amino acid residues
from about 3
about 250. Additionally, the optimal length may be determined by the method
disclosed in
the present invention.
There is also provided an antigenic composition comprising a Ho-Hi-Ho epitope
of
contiguous amino acid residues from a polypeptide wherein the Ho-Hi-Ho epitope
is
characterized by a hydrophobic-hydrophilic-hydrophobic motif and an
approximated -180°
to +180° negative cosine hydrophilicity pattern having an optimal
length of amino acid
residues from about 3 to about 250.
Additionally, the optimal length may be determined by the method disclosed in
the
present invention.
Still further provided is a diagnostic testing method comprising the steps of
(i) providing a sample;
(ii) contacting the sample with antisera specific for a Ho-Hi-Ho
epitope of contiguous amino acid residues from a
polypeptide wherein the Ho-Hi-Ho epitope is characterized
by a hydrophobic-hydrophilic-hydrophobic motif having an
optimal length of amino acid residues from about 3 to about
250 determined by the methods of the present invention; and
(iii) detecting binding of the antisera to a polypeptide in the
sample.
Also provided is a diagnostic testing method comprising the steps of
(i) providing an antisera sample
(ii) contacting said antisera sample with at least one Ho-Hi-Ho epitope
having an optimal length determined by the present methods; and
(iii) detecting the binding said Ho-Hi-Ho epitope to said antisera sample.
_g_
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
Alternatively, the above diagnostic testing method may include a tissue sample
which
may be contacted with at least one Ho-Hi-Ho epitope.
The present invention also provides for isolated nucleic acid molecules that
encode
for the Ho-Hi-Ho immunobiologically active epitope having an optimal length
determined by
the methods of the present invention. The nucleic acid molecule may include; a
cDNA
molecule comprising the nucleotide sequence of the coding region of the
epitope, isolated
DNA or RNA molecule or a genetic variant thereof which encodes the
immunobiologically
active epitope.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 shows the hydropathy plot for the amino acid sequence of Prostate
Specific
Antigen (PSA) and the oscillating behavior of the Ho-Hi-Ho theoretical
rankings.
Figure 2 shows the hydropathy plot for the amino acid sequence of Gelonin and
the
oscillating behavior of the Ho-Hi-Ho theoretical rankings.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
The present invention is concerned with providing methods for identifying
immunobiologically-active linear epitopes, determining the length of
continuous amino acid
residues of the identified epitopes and locating their position in a protein
antigen.
The method to identify immunobiologically-active linear epitopes, and
particularly
epitopes characterized by a hydrophobic-hydrophilic-hydrophobic motif,
includes generating
average propensity values for each amino acid of the protein sequence. These
average values
may be determined from propensity scales that describe the tendency of each
residue to be
associated with properties such as accessibility, hydrophilicity ,
hydrophobicity and/or
mobility. Preferably, the average value is determined by a hydrophilicity
parameter. These
average values may then be plotted. The average values of amino acids can be
obtained
from any of the methods well known in the art including, but not limited to
Kyte-Doolittle
tables (Kyte and Doolittle, 1982, J. Mol. Biol., vol 72, p. 105) which are
based on solubility
of amino acids in water vapors, Hopp-Woods (Hope and Woods, 1981, Proc. Natl.
Acad.
-9-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
Sci., vol. 78, p. 3824) values which are based on the ability of amino acids
to bind to a C18
HPLC column and/or Parker-Hodge (J.M.D. Parker, D. Guo, and R. S. Hodges,
1986,
Biochemistry 25, 5425) which is based on peptide retention times during high-
performance
liquid chromatography.
Preferably the Kyte-Doolittle measurement scale is used wherein a hydropathy
value
is assigned to each natural amino acid based on side chain (i) interior-
exterior distribution and
(ii) water-vapor transfer free energy as determined by water-vapor partition
coefficients. The
Kyte-Doolittle hydropathy index values include the following:
Isoleucine (9.5), Valine (4.2), Leucine (3.8), Phenylalanine (2.8),
Cysteine/cystine (2.5), Methionine (1.9), Alanine (1.8), Glycine (-0.4),
Threonine (-0.7), Tryptophan (-0.9), Serine (-0.8), Tyrosine (-1.3), Proline
(-1.6), Histidine (-3.2), Glutamic acid (-3.5), Glutamine (-3.5), Aspartic
acid
(-3.5), Asparagine (-3.5), Lysine (-3.9), Arginine (-4.5).
NOTE: The above values when used for plotting a curve will provide a
hydrophobicity curve. To generate a hydrophilicity curve the sign of the
index values must be reversed, e.g., Isoleucine becomes (-9.5).
The average hydropathy value of each amino acid is accomplished by averaging
the
hydropathy values of the amino acid residues within a predetermined segment.
The
segment may include any number, however, in a preferred embodiment the length
of the
segment is 5 amino acids. A window average hydropathy value is calculated for
each amino
acid residue by assigning the average hydropathy value to the amino acid at
the center point
of each of the moving segments. Average hydropathy values are obtained by
shifting the
segment by a single amino acid along the entire amino acid sequence of the
protein as it
advances from the amino to the carboxyl terminus. This is repeated until each
amino acid
residue is the center point of a segment has been assigned a average
hydropathy value. A
hydrophilicity an/or hydrophobicity plot of these average hydropathy values is
then
generated. The plot can be obtained manually, any commercially available or
shareware
software, or the source code for a custom computer program included in the
above-identified
reference by Kyte and Doolittle. The hydropathy plot may be generated by the
software
package "Wisconsin Package v4" commercially available from Genetics Computer
Group,
Inc., Madison, WI. Figure 1 and Figure 2 are representative examples of a
hydropathy plot
-10-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
for prostate specific antigen (PSA) and gelonin, a plant toxin, respectively.
The resulting curve is then fitted to a mathematically generated continuous
curve
wherein the curve has repeating values at regular intervals with a maximum
positive value.
The mathematically generated curves may include, but is not limited to
trigonometric curves,
such as sine, cosine, negative cosine curves, and other curve such as gaussian
curves and the
like. Preferably, the trigonometric function is a negative cosine function
which will identify
curve regions representing areas having a hydrophobic-hydrophilic-hydrophobic
(Ho-Hi-Ho)
pattern. The definition of the negative cosine curve is described according to
Abramowitz
and Stegun, Eds., HANDBOOK OF MATHEMATICAL FUNCTIONS WITH
FORMULAS, GRAPHS AND MATHEMATICAL TABLES, National Bureau of Standards
and Applied Mathematics, Series #55, June 1964, p. 71-79. Additionally, the
specific
definition of the negative cosine curve provided in the Microsoft Fortran
Library, version 5.1.
Preferably, successive segments of a protein Kyte-Doolittle hydropathy curve
are
fitted with the negative cosine curve fiznction using custom software with the
source code
defined in Appendix A. The custom software determines a fit-correlation value
for
sequential regions of amino acid residues of the protein. The fit-correlation
values are
dependent upon the period number of the negative cosine curve function which
determines
the assigned number of amino acids in each region (window). In other words,
the assigned
number of amino acids in a curve segment (window) is equivalent to the period
number used
in the negative cosine function. The period number represents the length of
amino acid
residues in the hydropathy curve segment that will be analyzed. For each
period number
specified in the software input, one set (containing of negative cosine
fi~nction-hydropathy
curve region fit-correlation values is generated specific to that period
number. The set of
fit-correlation values will contain (n-(m-1)~ values, where n is the number of
amino acids
in the protein and m is the period number used in the negative cosine curve
fixnction.
Specifically, when utilizing the custom software, if tJ. l is equal to the
Kyte-Doolittle
hydropathy average value (using a 5-amino acid segment as mentioned above) at
the amino
acid residue or lag point l, where l = 1,...,n designates the amino acid
residue of an amino
acid chain containing n amino acids, then
-11-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
~o~yl+k yl~ ' Cck- c~
! ~~ ~~l + k ~I / 2 / ' ~ ~ Ck c
is the hydropathy curve-negative cosine curve fi,~nction fit-correlation ~ at
lag point l of
period number m where
c =-cos ~2~ck~~m- 1
'~
is the negative cosine curve fi~nction of period number m, and where
c=~l c~m and y=~l y ~m
k-p k 1 k-p !+k
are the respective means.
The fit-correlation process is lagged (shifted) over the entire range of amino
acids in
the polypeptide by increasing the value of l by one (1) until the value (n-(m-
1)) is reached.
Subsequently, the period number m of the negative cosine curve fi.~nction is
increased by
one (1) in order to generate the next potential Ho-Hi-Ho epitope set. The
numerical value
for m may be any number greater than 2 extending to the number of amino acid
residues in
the polypeptide, and preferably, between 3 and 50 thereby creating 48
potential Ho-Hi-Ho
epitope sets. Each potential epitope set varies slightly in location as the
negative cosine
fiznction period number used to generate each set is changed; accordingly, the
fit-correlation
values vary slightly. By changing the period number of the applied negative
cosine fi~nction,
as one would change the aperture of a camera lens, the mathematical
perspective of the
negative cosine function curve-fit algorithm is altered. This enables the
algorithm to detect
sequential amino acid hydrophobic-hydrophilic-hydrophobic patterns of a
particular length
not readily distinguished visually.
Listed in the output of the specifically designed software are the amino acid
sequence
number ranges that project a hydropathy curve segment having a fit correlation
with the
-12-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
negative cosine curve function and are considered the potential Ho-Hi-Ho
epitopes. A
positive-fit correlation value indicates the potential presence of a
immunobiologically-active
linear epitope in the corresponding amino acid sequence number range, i.e. a
hydrophobic-
hydrophilic-hydrophobic sequence with dominant (high positive-fit correlation)
or
subdominant (low positive-fit correlation) immunobiological epitope activity.
For each
period number m, a set of fit-correlation values is generated. For example, if
period number
m of the negative cosine curve function is chosen from 3 to 50 then there will
be 48 different
potential Ho-Hi-Ho epitope sets wherein each set represent a hydropathy curve-
negative
cosine curve function fit analysis for the entire protein antigen. Each one of
these sets has
different amino acid sequence number ranges because the period number is
changed for each
set. For example, the amino acid number ranges for a period number (m) of 10
may include
amino acid residues in the number ranges 1-10, 2-11, 3-12, 4-13, and the
average hydropathy
value for each amino acid in the curve segment (period number range) is
inputted into the
software program until l is equal to (n-(m-1)). Also, the output will give a
fit-correlation
value for each one of the number ranges such as, 1-10, 2-11, 3-12. More
specifically, when
using a protein antigen which has 237 amino acid residues in the sequence, l
will increase
by one until number range 228-237 is inputted into the program. A period
number (m) of
11 will include amino acid numbers from 1-11, 2-12, 3-13, 4-14 until l is
equal to 227 and
number range 227-237 is reached. A set of fit-correlation values from each
period number
m spans the entire protein antigen and provides a potential Ho-Hi-Ho epitope
set.
In each one of the potential Ho-Hi-Ho epitope sets the potential epitopes are
ranked
according to the magnitude of the positive-fit correlation values. The epitope
with the
highest fit-correlation value is assigned the number one (1) ranking in each
set. This is
repeated for each of the sets, that is for each set generated by one of the 48
period numbers
utilized by the negative cosine fitting custom software in the range from 3 to
50. The
number of amino acid residues in the ranked Ho-Hi-Ho potential epitopes
corresponds to
the period number m used in the negative cosine function which generated the
original
potential Ho-Hi-Ho epitope set.
To determine the optimal length of the immunobiologically-active epitope and
the
position of the continuous epitope in a polypeptide, it has been discovered
that a recurrent
-13-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
pattern provides the necessary information. Specifically, the ranked potential
epitopes for
each set are superimposed on the hydrophilicity plot so that the sequence of
amino acid
residues in the potential epitopes are juxtaposed on the plot to correspond to
the linear
sequence of the polypeptide as shown in Figures 1 and 2.
Each of the generated sets of ranked Ho-Hi-Ho potential epitopes are plotted
on the
generated hydrophilicity curve thereby providing a plurality of dii~erent
plots, each one
representing a different period number m. Each of the different plots are
reviewed to
determine which of the plots exhibit an alternating rhythmicity wherein the
highest rankings
of the potential epitopes oscillate about an equilibrium position and either
converge towards
or diverge away from this centralized position with the concomitant increasing
of the
rankings.
This oscillating of the ranking values of the positioned potential epitopes
about an
equilibrium position may be exhibited in several different plots but the set
of potential
epitopes having the greatest number of epitopes that exhibit the oscillating
behavior
provides information for the optimal length. The period number m that was used
to generate
the set of potential epitopes is consider the optimal number of amino acid
residues in an
immunobiologically active epitope.
Additionally, it has been found that if more than one plot, having a different
period
number m, exhibit the same oscillating rhythmicity, then the plot generated by
m having
the highest fit-correlation values between the hydrophilicity curve and the
negative cosine
function is considered the potential set having the most immunobiologically-
active epitopes
and their optimal length is determined by the number of amino acid residues in
the ranked
potential epitopes.
The disclosed method of generating a plurality of potential epitope sets (for
a
polypeptide) by fitting a hydrophilicity curve to the curve generated by a
negative cosine
function may be used with other data to determine and/or verify the optimal
length. For
instance, the ranked potential epitopes for each set, having a specific length
of amino acid
residues and a Ho-Hi-Ho motif may be compared or correlated with other ranked
epitopes
(for the polypeptide in question found) by well known propensity scales that
are based on
accessibility, hydrophilicity, flexibility, and the like. Along this line,
statistical methods may
be used to determine the highest correlation coefficient between the rankings
of potential
-14-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
epitopes and epitopes found by propensity scales. Likewise, the potential
epitope sets may
be fitted or juxtaposed on other generated plots including hydrophobicity,
The method of the present invention can be used to select immunobiologically-
active
linear peptide epitopes from a variety of polypeptides once the amino acid
sequence of the
S polypeptide is determined. Any method know in the art which can determine
the amino acid
sequence of a protein may be used in the present invention. A preferred method
is briefly
explained. The first step in the sequence determination of a protein is to
cleave the
polypeptide chain into smaller peptides and then separate homogeneous samples
of these
peptides. Trypsin is especially useful for this initial cleavage, because of
its specificity for
lysine and arginine residues. A polypeptide chain containing five such
residues, for example,
will be cleaved by trypsin into six shorter peptides. The shorter peptides are
separated and
analyzed. The amino acid sequence of the isolated peptides is then determined
by the
sequential cleavage of amino acids from the carboxyl-terminal and amino-
terminal ends of
each peptide. This can be accomplished by the use of exopeptidases which are
specific for
the amino- or carboxyl-terminal ends of the peptide chain, or by chemical
methods.
Carboxypeptidase successively cleaves amino acids from the carboxyl-terminal
end of the
peptide and it is possible to determine the sequence of the amino acids by
following the time
course for the release of the amino acids. The most useful chemical method for
the analysis
of peptide sequences is the reaction of N-terminal amino acids with
phenylisothiocyanate.
This reaction removes amino acids sequentially from the N-terminal end of the
chain as their
phenylthiohydantoin (PTH) derivatives. In the first step of the reaction,
isothiocyanate
undergoes nucleophilic attack by the terminal amino group of the peptide to
give a
substituted thiourea. This step is carried out in dilute base. Upon treatment
with a weak
acid, the terminal amino group of the thiourea attacks the peptide bond of the
terminal amino
acid to give the phenylthiohydantoin derivative of the original N-terminal
amino acid. This
amino acid may be identified by chromatography and by comparing with standard
phenylthiohydantoin derivatives of known amino acids. Cleavage of the peptide
bond gives
a new N-terminal amino acid that may be identified by repetition of the whole
process.
Additionally, the method of the present invention may be used to select Ho-Hi-
Ho
epitopes from cancer cells, viral, microbial, and other molecules of basic and
clinical research
interest including, but not limited to examples provided below:
-15-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
Lymphokines and Interferons:
IL,-1, IL-2, IL,-3, IL,-4, IL,-S, IL,-6, IL,-7, IL,-8, IL,-9, IL-10, IL-11, IL-
12, IFN-a, IFN-(3, IFN-
Y
S Cluster Diil'erentiation Antigens and MHC Antigens:
CD2, CD3, CD4, CDS, CD8, CD 11 a, CD 1 1b, CD 11 c, CD 16, CD18, CD21, CD28,
CD32,
CD34, CD35, CD40, CD44, CD45, CD54, CD56, K2, K1, P~3, Oa, Ma, M~i2,
M~31,LMP1,
TAP2, LMP7, TAP1, O~i, IA(3, IAa, IE(3, IE(32,IEa, CYP21, C4B, CYP21P, C4A, Bf
C2,
HSP, G7a/b, TNF-a, TNF-Vii, D, L, Qa, Tla, COL11A2, DP(32, DPa2, DP(31, DPal,
DNa,
DMa, DM(3, LMP2, TAPiI, LMP7, D0~3, DQ(32, DQa2, DQ~33, DQ~il, DQal, DR~3,
DRa,
HSP-70, HLA-B, HLA-C, HLA-X, HLA-E, HLA-J, HLA-A, HLA-H, HLA-G,HLA-F.
Hormones and Growth Factors:
nerve growth factor, somatotropin, somatomedins, parathormone, FSH, LH, EGF,
TSH
THS-releasing factor, HGH, GRHR, PDGF, IGF-I, IGF-II, TGF-Vii, GM-CSF, M-CSF,
G-
CSF1, erythropoietin.
Tumor Markers and Tumor Suppressors:
~3-HCG, 4-N-acetylgalactosaminyltransferase, GM2, GD2, GD3, MAGE-l, MAGE-2,
MAGE-3, MUC-1, MUC-2, MUC-3, MUC-4, MUC-18, ICAM-1, C-CAM, V-CAM,
ELAM, NM23, EGFR, E-cadherin, N-CAM, CEA, DCC, PSA, Her2-neu, UTAA, melanoma
antigen p75, K19, HKer 8, pMel 17, tyrosinase related proteins 1 and 2, p97,
p53, RB, APC,
DCC, NF-1, NF-2, WT-1, MEN-I, MEN-II, BRCA1, VHL, FCC and MCC.
Oncogenes:
ras, myc, neu, raf, erb, src, f'ms jun, trk ret, gsp, hst, bcl arid abil.
Complement Cascade Proteins and Receptors:
C 1 q, C 1 r, C 1 s, C4, C2, Factor D, Factor B, properdin, C3, C5, C6, C7,
C8, C9, C lInh,
Factor H, C4b-binding protein, DAF, membrane cofactor protein, anaphylatoxin
inactivator
S protein, HRF, MIRL, CR1, CR2, CR3, CR4, C3a/C4a receptor, CSa receptor.
Viral Antigens:
HIV (gag, pol, qp4l, gp120, vif, tat, rev, nef, vpr, vpu, vpx), HSV
(ribonucleotide reductase,
a-TIF, ICP4, ICPB, 1CP35, LAT-related proteins, gB, gC, gD, gE, gH, gI, gJ),
influenza
(hemagluttinin, neuraminidase, PB1, PB2, PA, NP, M,, MZ, NS,, NSZ),
papillomaviruses (El,
-16-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
E2, E3, E4, ESa, ESb, E6, E7, E8, L1, L2) adenovirus (ElA, E1B, E2, E3, E4,
E5, L1, L2,
L3, L4, L5), Epstein-Barn Virus (EBNA), Hepatitis B Virus (gp27s, gp36S,
gp42', p22', pol,
x).
Nuclear Matrix Proteins.
The Ho-I~-Ho epitopes of the present invention can be used in diagnostic
tests, such
as immunoassays, to detect viruses, microbes and malignant cells.
Immunoassays, in their
most simple and direct sense, are binding assays. Certain preferred
immunoassays are various
types of enzyme linked immunosorbent assays, radioimmunoassays,
immunofluorescence and
surface plasmon resonance. Immunohistochemical detection using tissue sections
is also
particularly useful. However, it should be appreciated that detection methods
are not limited
to such techniques, and Western blotting, dot blotting, FACS analyses, and the
like may be
used.
After identifying the Ho-Hi-Ho epitopes and determining the optimal length of
amino
acid residue sequence, peptides can be synthesized that correspond to the
exact amino acid
sequence and length of residues. In turn, polyclonal antibodies or monoclonal
antibodies can
be generated specific for a peptide.
Briefly, monoclonal antibodies are produced by immunizing animals, such as
rats or
mice with the peptide antigen of choice. Once the animals are making a good
antibody
response the spleens or lymph node cells are removed and a cell suspension
prepared. These
cells are fused with a myeloma cell line by the addition of polyethylene
glycol (PEG) which
promotes membrane fusion. Only a small proportion of the cells fuse
successfully. The
fusion mixture is then set up in a culture with medium containing "HAT". HAT
is a mixture
of Hypoxanthine, Aminopterin and Thymidine. Aminopterin is a powerful toxin
which blocks
a metabolic pathway. This pathway can be bypassed if the cell is provided with
the
intermediate metabolites hypoxanthine and thymidine. Thus, spleen cells can
grow in HAT
medium, but the myeloma cells die in HAT medium because they have a metabolic
defect and
cannot use the bypass pathway. When the culture is set up in the HAT medium it
contains
spleen cells, myeloma cells and fused cells. The spleen cells die in culture
naturally after 1-2
weeks and the myeloma cells are killed by the HAT medium. Only fused cells
survive
because they have the immortality of the myeloma cells and the metabolic
bypass of the
spleen cells. Some of the fused cells will have the antibody producing
capacity of spleen
-17-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
cells. The wells containing growing cells are tested for production of the
desired antibody
(often by RIA or ELISA) and, if positive, the cultures are cloned, that is,
plated out so that
only one cell is in each well. This process produces a clone of cells derived
from a single
progenitor, which is both immortal and produces monoclonal antibody. These
highly
specific, monoclonal antibodies may be used as reagents for numerous
applications ranging
from specific diagnostic tests to "magic bullets" in immunotherapy of
different types of
cancer. In immunotherapy, various drugs or toxins may be conjugated to the
monoclonal
antibodies and delivered to the tumor cells against which the antibodies are
specific.
The Ho-Hi-Ho epitopes of the present invention can also be used in
prophylactic or
therapeutic vaccines to elicit immune responses. Vaccines produced by
microorganism such
as yeast, through recombinant DNA technology provide another area that may be
benefltted
by the present invention. The DNA that codes for a Ho-Hi-Ho epitope can be
spliced into
the DNA of yeast, which, in turn can produce copies of the peptide. In this
regard,
production of vaccines against hepatitis B may provide greater quantities of a
safer vaccine
than the vaccine prepared from blood plasma of humans.
Synthetic vaccine can be prepared by chemically synthesizing a chain of amino
acids
corresponding to the sequence of amino acids of the Ho-Hi-Ho epitopes. The
amino acid
chain containing the Ho-Hi-Ho epitopes is disposed on a physiologically
acceptable carrier
and diluted with an acceptable medium. The synthetic vaccines may contain one
or a
plurality of Ho-Hi-Ho epitopes of at least one antigen. Vaccines are
contemplated for the
following antigens, including, but not limited to Hepatitis B surface antigen
histocompatibility
antigens, influenza hemagglutinin, fowl plague virus hemagglutinin and rag
weed allergens
Ra3 and RaS. Also, vaccines are contemplated for the antigens of the following
viruses
including, but not limited to vaccinia, Epstein Barr virus, polio, rubella,
cytomegalovirus,
small pox, herpes, simplex types I and II, yellow fever, and many others.
Antigen compositions are contemplated by the present invention which include
antibodies specific for peptides with a hydrophobic-hydrophilic-hydrophobic
motif having a
length of amino acid residues determined by the method of the present
invention and which
may be administered in the form of injectable, pharmaceutical compositions. A
typical
composition for such a purpose comprises a pharmaceutically acceptable Garner.
For
instance, the composition may contain about 10 mg of human serum albumin and
from about
-18-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
20 to 200 micrograms of the labeled monoclonal antibody or fragment thereof
per milliliter
of phosphate buffer containing NaCI. Other pharmaceutically acceptable corners
include
aqueous solution, non-toxic excipients, including salts, preservative, buffers
and the like.
Examples of non-aqueous solvents are propylene glycol, polyethylene glycol,
vegetable oil
and injectable organic esters such as ethyloleate. Aqueous carrier include
water,
alcoholic/aqueous solutions, saline solutions, parenteral vehicles such as
sodium chloride,
Ringer's dextrose, etc. Intravenous vehicles include fluid and nutrient
replenishers. The pH
and exact concentration of the various components in the pharmaceutical
composition are
adjusted according to routine skills in the art.
It is further contemplated that a chain of nucleotides specific to code for a
preferred
Ho-Hi-Ho epitope may be used for immunization compositions. Recently,
immunization
techniques in which DNA constructs are introduced directly into mammalian
tissue in vivo
have been developed. Known as DNA vaccines, they use eukaryotic expression
vectors to
produce immunizing proteins in the vaccinated host. Methods of delivery
include
intramuscular and intradermal saline injections of DNA or gene gun bombardment
of skin
with DNA-coated gold beads. Mechanistically, gene gun-delivered DNA initiates
responses
by transfected or antigen-bearing epidermal Langerhans cells that move in
lymph from
bombarded skin to the draining lymph nodes. Following intramuscular
injections, the
functional DNA appears to move as free DNA through blood to the spleen where
professional antigen presenting cells initiate responses. These methods are
described inter
olio in Robinson, Sources in Immunology, 9(5): 271-283, (1997 Oct) and Fynan
et al,Proc.
Natl. Acad. Sci. USA" 90:11478-11482 (1993) and incorporated herein by
reference.
In another embodiment of this invention, the method can be used to test the
potential
antigenicity of a peptide antigen prior to being used to generate bulk
antisera for vaccines.
The Ho-I~-Ho epitope of a test antigen can be compared to its standard Ho-Hi-
Ho epitope
(obtained when the antigen was known to generate ei~cacious vaccine). Any
deviations from
the standard values may indicate alteration or denaturation of the antigen.
This is also
applicable not just for peptide antigens but for any protein. Specifically, if
the m-value is
determined by the methods of the present invention for a protein, then this
value can be used
as a comparative value used to determine if a protein used for immunization is
viable. For
instance, if a protein is used to immunize a subject and the anti-protein
antisera does not
-19-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
correctly describe the determined m-value then the protein may have been
denatured before
the immunization. This knowledge may cause the re-immunize a subject to ensure
a
su~cient and correct immunological response to the protein.
In yet another embodiment of the invention, the method can be used to
determine Ho
Hi-Ho epitopes involved in enzyme-substrate interaction, in protein-protein
interaction,
protein-nucleic acid interactions, protein-lipid interactions, protein-
carbohydrate interactions
and the like.
The methods of the present invention may also be used to alter the
immunogenicity
of a Ho-Hi-Ho epitope, once it has been determined by the methods of the
present
invention, by altering the amino acid composition therein. Specifically,
certain amino acids
within the Ho-I~-Ho epitope may be replaced thereby either increasing or
decreasing the fit
between the negative cosine curve and generated hydrophilicity curve. By
altering the
immunogenicity of the epitope, affinity for the epitope binding site by either
an antibody or
receptor on a lymphocyte can be increased or decreased.
The following examples using prostate specific antigen as a polypeptide having
immunobiologically active linear epitopes will help to illustrate the present
invention.
EXAMPLE 1
Hvdr~athv Plots for PSA and Gelonin
To generate a hydrophilicity plot for prostate specific antigen (PSA), the
hydropathy
values according to the method of Kyte and Doolittle, were assigned to each
amino acid
residue. The sign of each value was changed from positive to negative or vice
versa
dependent upon the original sign. (See Hentuu and Vihko, 1989, Biochem.
Biophys. Res.
Comm, vol. 160, p. 903-910 for the amino acid sequence of the protein). The
window
average hydropathy values were then plotted for the entire amino acid sequence
of PSA. The
plot was generated with the software package "The Wisconsin Package v4"
commercially
available from Genetics computer Group, Inc., Madison, WI. and shown in Figure
2.
Likewise, a similar plot was generated for Gelonin and shown in Figure 3. (For
sequence,
see Rosenblum et al, 1995, J. Interferon-Cytokine Res. vol. 15, p. 547).
-20-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
EXAMPLE 2
Determination of Hydrophobic-Hydrophilic-Hydrophobic Regions
The negative cosine curve function of a specific period number was fitted with
custom
software using the source code disclosed in Appendix A to successive segments
of the PSA
and gelonin Kyte-Doolittle hydropathy curve. Each point along the hydropathy
curve
obtained in Example 1 was fitted to a negative cosine curve function from -
180° to +180°.
The period number of the negative cosine curve function was changed from 8 to
40
producing a series of 33 potential Ho-Hi-Ho epitope sets. A fit-correlation
value was
obtained for each lag point l along the amino acid sequence in each chosen
period number
m. Number ranges having a positive-fit correlation value represented
hydrophobic
hydrophilic-hydrophobic regions in the amino acid sequences and these
sequences are
deemed ranked theoretical epitopes. The period number m of the negative cosine
curve
function represented the size of the hydrophobic-hydrophilic-hydrophobic
regions, that being,
1 S the number of amino acids in the Ho-Hi-Ho epitopes.
EXAMPLE 3
Oscillating Behavior of Ranked Potential E i~toye
The Ho-Hi-Ho potential epitopes in each set were determined and ranked
according
to the positivity of the correlation between the hydrophilicity curve and a
curve generated by
the negative cosine function wherein the period numbers m = 8-40 were used.
The ranked potential epitopes for each set were juxtaposed on the
hydrophilicity plot
so that the sequence of amino acid residues in the ranked theoretical epitopes
corresponded
to the linear sequence of the polypeptides of PSA and Gelonin. Thus
understood, the amino
acid sequence of each ranked Ho-Hi-Ho potential epitope had a specific
location
corresponding to the placement of the same amino acid sequence found in the
polypeptide.
Each of the 33 sets of potential epitopes for PSA and gelonin, which contained
the
ranked Ho-Hi-Ho potential epitopes, were plotted on the generated
hydrophilicity curve
thereby providing 33 different plots for each polypeptide. It was discovered
in reviewing
-21-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
the plots that the rankings of the potential epitopes were either randomly
positioned on the
respective plots or the rankings alternated or oscillated about an equilibrium
position. This
equilibrium position was not necessarily in the center of the linear sequence
of the
polypeptide. Specifically, in PSA (Figure 1)the plot which contained the
ranked potential
epitopes generated when m=19 showed an alternating rhythmicity wherein the
highest
rankings (1-6) of the positioned Ho-Hi-Ho potential epitopes alternated about
a centralized
position and converged towards this region. Likewise in Figure 2 for gelonin
it is evident
that the highest rankings (1-6) of the potential epitopes exhibit an
alternating rhythmicity and
diverge from a centralized region between the potential epitopes when the
theoretical
epitopes were generated using m=31.
Results: It was determined that the immunobiologically- active epitopes are
those ranked
Ho-I-F-Ho potential epitopes that exhibit the most oscillating behavior about
an equilibrium
position that either converges to or diverges away from this position. The
number of amino
acid residues in these ranked potential epitopes was assigned to be the
optimal length of the
immunobiologically-active epitope. It may be concluded from this example that
several
amino acid regions in PSA and gelonin adhered strongly to the hydrophobic-
hydrophilic-
hydrophobic amino acid hydropathy pattern of the protein Ho-Hi-Ho theoretical
epitope.
This local rhythmic hydropathy pattern enables a protein-specific number of
amino acids in
the region to act as an immunobiologically active epitope. The epitope length
indicated by
the optimal negative cosine function period number is specific for PSA (19
amino acids) and
for gelonin (31 amino acids). It is theorized that Ho-Hi-Ho theoretical
epitopes and their
specific length are biochemical entities inherent in a protein. Also, the
primary amino acid
sequence thus plays a vital role in determining the location, length and
immunobiological
potency of protein Ho-Hi-Ho theoretical epitopes.
-22-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
APPENDIX A
FORTRAN PROGRAM FOR FITTING HYDROPATHY PLOT
TO NEGATIVE COSINE FUNCTION
15
25
-23-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
program lagfcn
parameter (mseql=1000,m1en=50)
dimension a(mlen),b(mseql),c(mseql,mlen)
character*30 fileout,fileb,filedat
character*80 forseq
character*1 seq(mseql),target
logical first, last
mseq=1000
write(*,'(" Lag E~nction Program -- Enter output file ")')
read(*,1) fileout
1 format(a30)
write(*,'(" Enter min length to max length " )')
11 read(*,*) istart,istop
if (istop.gt.mlen) then
write(*,'(" Sequence Length Greater than " ,i5)') mlen
write(*,'(" Try again or enter -1 -1 to stop " )')
go to 11
else
if (istop.lt.l) go to 999
end if
write(*,'(" Enter the sequence filename ")')
read(*,1) fileb
write(*,'(" Enter the output data filename ")')
read(*,1) filedat
open(unit=l, file=fileb,status='OLD')
inunit=1
open(unit=7,file=fileout,status='UNKNOWN')
open(unit=8,file=filedat,status='UNKNOWN')
write(*,'(" Enter length of sequence to be lagged on " )')
read(*,*) lenseq
write(*,'(" Enter target " )')
read(*,3) target
3 format(80a1)
write(*,'( " Enter 1 to input sequence " )')
write(*,'( " Enter 2 to input sequence and hydro. " )')
read(*,*) inptype
if (inptype .eq. 2) go to 500
call kytedoo(length,lenseq,seq,b,mseq,inunit)
go to 60
500 write(*,'(" Enter sequence format -- seq,b ")')
read(*,2) forseq
2 format(a80)
do 50 1-l,lenseq
do 25 i=istart,istop
25 c(l,i)=0.0
50 read(l,forseq,end=55) seq(1),b(1)
length=lenseq
go to 60
55 write ( 7, 54 ) 1
write(*,54) 1
54 format(' Sequence terminated short of end ',i5)
-24-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
length=1
60 continue
write ( 7, 51 ) (b ( 1 ) ,1=1, lenseq)
51 format(' Sequence to lag over fcn '/(lx,8f9.5))
write(*,'( " Current function is -cosine**power " )')
write(*,'(" Enter the integer power, sign and cycles ")')
read(*,*) npower,sig,cycles
do 100 i=istart,istop
write(7,52) i
write(*,52) i
52 format(' Lag ',i5,' Calculate Et~action ')
call fcn(a,i,npower,sig,cycles)
write(7,53)
write(*,53)
write(7, *) (a(j),j=l,i)
c write(*, *) (a(j),j=l,i)
53 format(' Calculate lags ')
len=lenseq-i
call lagl(lenseq,b,i,a,c(l,i),l,len)
c hin=i/4
c hax=3*(i+1)/4
first=.true.
last=.false.
if(cycles.gt.l.) then
hain=1
hax=i
else
do 65 j=l,i
if (first .and. -sign(l,sig)*a(j).gtØ) then
first=. false.
hn.in=J
end if
if (.not. first .and. .not. last .and.
$ -sign(l,sig)*a(j).1tØ) then
last=.true.
hoax=j -1
go to 70
end if
65 continue
end if
70 continue
call p3seq(istart,c(l,i),len,seq,lenseq,target,noin,noout,l,
$ Jnnin, hoax )
ntottar=0
do 80 1=l,lenseq
if (seq(1).eq. target) ntottar=ntottar+1
80 continue
atot=noin+noout
write(7,99) i,noin,noout,ntot,atottar, hzin,lQnax
write(8,99) i,noin,noout,ntot,ntottar,hain, hoax
write ( *, 99 ) i, noire, noout, ntot, ntottar, lQain, l~nax
99 format(7i5)
100 continue
-25-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
do 110 1=l,lenseq
write(8,101) 1,(c(l,i),i=istart,istop)
101 format(iS,lOf8.5)
110 continue
999 stop
end
subroutine p3seq(n,c,len,seq,lseq,target,ni,no,inc,
k1, ku)
character*1 seq(lseq),target
dimension c(len),f(3),x(3)
data f/ -1,2,-1 /
ni=0
no=0
x(1)=0.
write(*,1)
write(7,1)
1 format(lOx,'Position',2x,'Correlation',' Target Seq',lOx,'X')
do 100 i=l,len
x(2)~c(i)
x(3)=c(i+inc)
s=0.
do 20 j=1,3
s=s+f (j)*x(j)
20 continue
c if (s .gt. 0) then
if((x(2)-x(1) .gtØ).and.(x(2)-x(3) .gtØ)) then
kmin=i+kl
kmax=i+ku
do 40 k=kmin,kmax
if (seq(k).eq. target) go to 45
40 continue
no=no+1
k=(kmin+kmax)/2
go to 47
45 ni=ni+1
47 write(*,46) i,c(i),target,seq(k),k,x
write(7,46) i,c(i),target,seq(k),k,x
46 format(lox,i5,2x,f10.5,2(2x,a1),i4,2x,3f10.5)
50 continue
end if
if (i-inc .ge. 0) x(1)=c(i-inc+1)
100 continue
return
end
subroutine fcn(a,n,npower,sig,cycles)
dimension a(n)
pi=3.14159
twopi=pi*2
ratio=cycles*twopi/float(n-1)
do 100 i=l,n
arg=(i-1)*ratio
dat=cos(arg)
-26-
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
100 a(i)=sig*sign(l.,dat)*(abs(dat))**npower
return
end
SUBROUTINE LAG1(LA,A,LB,B,C,LSTART,LSTOP)
C
C THIS ROUTINE CALCULATES A SAMPLE CROSS-CORRELATION OF THE RECORD
C A OVER THE RECORD B WITH LAGS BETWEEN LSTART AND LSTOP AND
C STORES THE RESULT IN C
C **** CAUTION ***** THERE IS NO CHECK FOR A ZERO RECORD
C
DIMET1SION A(LA),B(LB),C(LA)
DO 50 J=LSTART,LSTOP
U=0.0
SUMA--0.0
SUMB-0.0
SA=0.0
sB-o.o
IF(LB-(LA-J+1)) 10,10,20
N=LB
GO TO 30
N=LA-J+1
IF(N.GT.O) GO TO 30
DO 25 I=J,LSTOP
C(I)=-2.
RETURN
EN=N
DO 40 I=1,N
IJ=I+J-1
SUMAgSUMA+A(IJ)
svr~=sUMB+B ( I >
SA--SA+A(IJ)*A(IJ)
SB=SB+B(I)*B(I)
U=U+A(IJ)*B(I)
SUMA=SUMA/EN
SUMB=SUMB/EN
SA=SA-SUMA*SUMA*EN
sB-sB-svMB*sur~*EN
C(J)=(U-EN*SUMA*SUMB)/SQRT(SA*SB)
RETURN
END
subroutine kytedoo(length,lenseq,seq,b,mseql,inunit)
dimension b(mseql), weights(21)
character*1 seq(mseql), name(20), buff(80), seqname(80)
character*1 gt,ast,blank
data name/'G', 'Q', 'S', 'Y', 'A', 'K', 'T', 'W',
2 'V', 'H', 'D', 'C', 'L', 'R', 'E', 'M',
3 ~Irr ~F'~r ~N~r ~P~/
data gt/'>'/, ast/'*'/, blank/' '/
data weights /-0.4,-3.5,-0.8,-1.3, 1.8,-3.9,-0.7,-0.9,
2 4.2,-3.2,-3.5, 2.5, 3.8,-4.5,-3.5, 1.9,
3 4.5, 2.8,-3.5,-1.6, 0.0/
_27_
CA 02370760 2001-10-19
WO 00/63693 PCT/US00/10585
numprot=20
1-0
read(inunit,l,end=1000) buff
1 format(80a1)
do 100 i=1,80
if (buff ( i ) . eq. gt ) go to 50
if (buff (i) .eq. ast) go to 110
if (buff (i) .eq. blank) go to 10
1=1+1
if (1 .gt. mseql) go to 1000
seq(1) = buff (i)
write(*,*) l,mseql,i,seq(1),buff(i)
go to 100
50 write(*,'(" Sequence Name ",80a1)') (buff(j),j=i+1,80)
k=0
do 60 j=i+1,80
k=k+1
60 seqname(k)=buff(j)
1=0
go to 10
100 continue
go to 10
110 length=1
write(*,2) (seq(j),j=l, length)
2 format(1x,80a1)
write(*,'(" Enter Kyte-Doolittle number to average ")')
read(*,*) 1
12=1/2
lstart-12+1
lstop=length-12
do 120 i=1,12
b(i)=0
120 b(length-i+1)=0
do 200 i=lstart,lstop
b(i)=0
do 150 j=i-12,i+12
do 130 k=l,numprot
if (seq(j) .eq. name(k)) go to 190
130 continue
write(*,131) j,seq(j)
131 format(' At ', i9, lx,al,' not recognized - weight =0')
k=21
190 b(i)=b(i)+weights(k)
150 continue
200 continue
write(*,'(" Kyte-Doolittle calculation complete ")')
write(7,'(" Kyte-Doolittle calculation complete ")')
write(7,201) (seq(i),b(i),i=l, length)
201 format (8(lx,al,lx,f6.3))
return
1000 write(*,1001) l,mseql
1001 format(' Unexpected end of file or '/
2 ' sequence length ',i5,' too long for buffer ',i5)
return
end
-28-
