ctakes-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From brittfi...@apache.org
Subject svn commit: r1505159 [2/3] - in /ctakes/sandbox/ctakes-scrubber-deid/data/input/pubs: txt/test_publication.txt xml/test_publication.xml
Date Sat, 20 Jul 2013 16:22:00 GMT
Added: ctakes/sandbox/ctakes-scrubber-deid/data/input/pubs/xml/test_publication.xml
URL: http://svn.apache.org/viewvc/ctakes/sandbox/ctakes-scrubber-deid/data/input/pubs/xml/test_publication.xml?rev=1505159&view=auto
==============================================================================
--- ctakes/sandbox/ctakes-scrubber-deid/data/input/pubs/xml/test_publication.xml (added)
+++ ctakes/sandbox/ctakes-scrubber-deid/data/input/pubs/xml/test_publication.xml Sat Jul 20 16:22:00 2013
@@ -0,0 +1,2028 @@
+<!DOCTYPE article
+        PUBLIC "-//NLM//DTD Journal Archiving and Interchange DTD v3.0 20080202//EN" "archivearticle3.dtd">
+<article xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML"
+         article-type="research-article"><?properties open_access?>
+    <front>
+        <journal-meta>
+            <journal-id journal-id-type="nlm-ta">Chem Biol Drug Des</journal-id>
+            <journal-id journal-id-type="publisher-id">jpp</journal-id>
+            <journal-title-group>
+                <journal-title>Chemical Biology &#x00026; Drug Design</journal-title>
+            </journal-title-group>
+            <issn pub-type="ppub">1747-0277</issn>
+            <issn pub-type="epub">1747-0285</issn>
+            <publisher>
+                <publisher-name>Blackwell Publishing Ltd</publisher-name>
+            </publisher>
+        </journal-meta>
+        <article-meta>
+            <article-id pub-id-type="pmc">2737611</article-id>
+            <article-id pub-id-type="pmid">19519740</article-id>
+            <article-id pub-id-type="doi">10.1111/j.1747-0285.2009.00832.x</article-id>
+            <article-categories>
+                <subj-group subj-group-type="heading">
+                    <subject>Research Articles</subject>
+                </subj-group>
+            </article-categories>
+            <title-group>
+                <article-title>On the Origins of Enzyme Inhibitor Selectivity and Promiscuity: A Case Study of Protein
+                    Kinase Binding to Staurosporine
+                </article-title>
+            </title-group>
+            <contrib-group>
+                <contrib contrib-type="author">
+                    <name>
+                        <surname>Tanramluk</surname>
+                        <given-names>Duangrudee</given-names>
+                    </name>
+                    <xref ref-type="aff" rid="au1">1</xref>
+                </contrib>
+                <contrib contrib-type="author">
+                    <name>
+                        <surname>Schreyer</surname>
+                        <given-names>Adrian</given-names>
+                    </name>
+                    <xref ref-type="aff" rid="au1">1</xref>
+                </contrib>
+                <contrib contrib-type="author">
+                    <name>
+                        <surname>Pitt</surname>
+                        <given-names>William R</given-names>
+                    </name>
+                    <xref ref-type="aff" rid="au1">1</xref>
+                    <xref ref-type="aff" rid="au2">2</xref>
+                </contrib>
+                <contrib contrib-type="author">
+                    <name>
+                        <surname>Blundell</surname>
+                        <given-names>Tom L</given-names>
+                    </name>
+                    <xref ref-type="aff" rid="au1">1</xref>
+                    <xref ref-type="corresp" rid="cor1">*</xref>
+                </contrib>
+                <aff id="au1">
+                    <label>1</label>
+                    <institution>Department of Biochemistry, University of Cambridge</institution>
+                    <addr-line>80 Tennis Court Road, Cambridge CB2 1GA, UK</addr-line>
+                </aff>
+                <aff id="au2">
+                    <label>2</label>
+                    <institution>UCB Celltech</institution>
+                    <addr-line>208 Bath Road, Slough, Berkshire, SL1 3WE, UK</addr-line>
+                </aff>
+            </contrib-group>
+            <author-notes>
+                <corresp id="cor1">*Corresponding author: Tom L. Blundell,
+                    <email>tom@cryst.bioc.cam.ac.uk</email>
+                </corresp>
+                <fn>
+                    <p>Re-use of this article is permitted in accordance with the Terms and Conditions set out at
+                        <ext-link ext-link-type="uri"
+                                  xlink:href="http://www3.interscience.wiley.com/authorresources/onlineopen.html">
+                            http://www3.interscience.wiley.com/authorresources/onlineopen.html
+                        </ext-link>
+                    </p>
+                </fn>
+            </author-notes>
+            <pub-date pub-type="ppub">
+                <month>7</month>
+                <year>2009</year>
+            </pub-date>
+            <volume>74</volume>
+            <issue>1</issue>
+            <fpage>16</fpage>
+            <lpage>24</lpage>
+            <history>
+                <date date-type="received">
+                    <day>24</day>
+                    <month>2</month>
+                    <year>2009</year>
+                </date>
+                <date date-type="rev-recd">
+                    <day>11</day>
+                    <month>5</month>
+                    <year>2009</year>
+                </date>
+                <date date-type="accepted">
+                    <day>11</day>
+                    <month>5</month>
+                    <year>2009</year>
+                </date>
+            </history>
+            <permissions>
+                <copyright-statement>Journal compilation &#x000a9; 2009 Blackwell Munksgaard</copyright-statement>
+                <license license-type="open-access" xlink:href="http://creativecommons.org/licenses/by/2.5/">
+                    <license-p>Re-use of this article is permitted in accordance with the Creative Commons Deed,
+                        Attribution 2.5, which does not permit commercial exploitation.
+                    </license-p>
+                </license>
+            </permissions>
+            <abstract>
+                <p>Relationships between ligand binding and the shapes of the binding sites in families of homologous
+                    enzymes are investigated by comparing matrices of distances between key binding site atoms. Multiple
+                    linear regression is used to help identify key distances that influence ligand binding affinity. In
+                    order to illustrate the utility of this generic approach, we study protein kinase binding sites for
+                    ATP and the promiscuous competitive inhibitor, staurosporine. We show that the size of the
+                    gatekeeper residue and the closure between the first glycine of the GXGXXG motif and the aspartate
+                    of the DFG loop act together to promote tight binding. Our web-based tool, &#x02018;mapping
+                    analogous hetero-atoms onto residue interactions&#x02019; (MAHORI), indicates that the greater the
+                    number of hydrogen bonds made by the kinase around the methylamine group of staurosporine, the
+                    tighter the binding. The conservation of surrounding atoms identified using our novel grid-based
+                    method clearly demonstrates that the most structurally conserved part of the binding site for
+                    staurosporine is the main chain of the hinge region. The critical role of interactions that are not
+                    dependent on side-chain identities is consistent with the promiscuous nature of this inhibitor.
+                </p>
+            </abstract>
+            <kwd-group>
+                <kwd>ATP-binding site</kwd>
+                <kwd>kinase selectivity</kwd>
+                <kwd>Mantel test</kwd>
+                <kwd>staurosporine</kwd>
+                <kwd>structural analysis</kwd>
+            </kwd-group>
+        </article-meta>
+    </front>
+    <body>
+        <p>Protein kinases catalyze the transfer of a phosphate group, usually from ATP, to a protein substrate. The
+            human genome comprises more than 500 protein kinases (<xref ref-type="bibr" rid="b1">1</xref>), which are
+            known to mediate most of signal transduction crucial to metabolism, cell proliferation and differentiation,
+            membrane transport, and apoptosis. Although they exhibit a variety of regulatory mechanisms and
+            conformational states, protein kinases share the same fold and similar ATP-binding sites (<xref
+                    ref-type="bibr" rid="b2">2</xref>).
+        </p>
+        <p>Protein kinases are of considerable interest to the pharmaceutical industry because dysfunction often results
+            in malignancy (<xref ref-type="bibr" rid="b3">3</xref>,<xref ref-type="bibr" rid="b4">4</xref>). Kinases are
+            also validated anti-inflammatory drug targets (<xref ref-type="bibr" rid="b5">5</xref>). Indeed, the success
+            of several high-affinity ATP-mimetic drugs has made the design of selective inhibitors an attractive
+            approach to useful therapeutics, particularly for oncology (<xref ref-type="bibr" rid="b6">6</xref>).
+            Although the structural conservation of the ATP-binding site can lead to off-target ligand binding, kinase
+            inhibitor design has become a promising way forward for discovery of useful therapeutic agents (<xref
+                    ref-type="bibr" rid="b7">7</xref>).
+        </p>
+        <p>The major challenge for protein kinase inhibitor design is obtaining selectivity. In order to reduce the
+            chances of undesirable side-effects, potency is usually optimized against a target kinase while reducing
+            off-target activities including those involving homologous proteins. However, the level of selectivity
+            required is dependent on the therapeutic endpoint. Indeed, in some cases, polypharmacology is an advantage,
+            although a multi-target selectivity is very difficult to achieve by design (<xref ref-type="bibr" rid="b8">
+                8</xref>).
+        </p>
+        <p>Insights into protein kinase selectivity and promiscuity are the major objectives of this article. Although
+            high-affinity targets sometimes have similar residues at positions important for binding of a given kinase
+            inhibitor, others with similar residues at these important positions can be insensitive to such inhibitors,
+            probably because of conformational differences (<xref ref-type="bibr" rid="b9">9</xref>). Therefore,
+            understanding kinase selectivity cannot be achieved only through the analysis of sequences but must also
+            consider three-dimensional structures.
+        </p>
+        <p>Here, in order to understand the relationship between the protein structure and binding affinities, we study
+            complexes of protein kinases with staurosporine (<xref ref-type="bibr" rid="b10">10</xref>), a microbial
+            alkaloid isolated from<italic>Streptomyces staurosporeus</italic>. Although staurosporine is a potent
+            inhibitor of various human protein kinases and has some antifungal activity, it is too toxic to be used as a
+            drug. However, it has been widely used in research as a universal kinase inhibitor (<xref ref-type="bibr"
+                                                                                                      rid="b11">
+                11</xref>). The three-dimensional structures of a significant number of protein kinases co-crystallized
+            with staurosporine can be found in the Protein Data Bank. They show that staurosporine mimics ATP very well,
+            in spite of the apparent lack of similarity between the two molecules. The availability of dissociation
+            constants (
+            <italic>K</italic>
+            <sub>d</sub>) for staurosporine with 119 kinases (<xref ref-type="bibr" rid="b12">12</xref>), when used in
+            combination with this structural information, allows us to relate binding affinities to differences in the
+            structures of the pockets.
+        </p>
+        <p>We have studied the relationship between the structures of protein kinases and their ability to bind
+            promiscuous ligands, which we define here as compounds that bind several and, in the case of staurosporine,
+            the majority of kinases on which they are tested. In order to investigate the factors that determine whether
+            a particular kinase will bind a particular kinase inhibitor, we have examined the regions around the pockets
+            of a set of medium-to-high-quality X-ray crystal structures; and to group kinases based on the similarity in
+            spatial arrangement of amino acid side chains, we use the Mantel test, a statistically robust method for
+            calculating correlation coefficients between distance matrices (<xref ref-type="bibr" rid="b13">13</xref>).
+            Our shape-based dendrogram shows that similarity in shape alone can sometimes determine the affinity for a
+            set of ligands, regardless of their overall sequence similarity.
+        </p>
+        <p>The study of the shapes of the pockets allows us to identify the factors required for most kinases to bind
+            the promiscuous inhibitor, staurosporine. We examine the similarities of the shapes of the pockets by
+            identifying neighboring entities that are conserved in their positions relative to staurosporine. We show
+            that some are recruited to staurosporine via an induced fit or conformational selection mechanism, which
+            contributes to staurosporine promiscuity. We focus on the differences in distance matrices that define the
+            pockets and show how the side chains affect binding affinities. Based on a quantitative structure activity
+            relationship (QSAR) approach, we select a set of distances that have an influence on binding affinities.
+            These distances indicate that tighter binding is associated with the closure of the N-lobe and C-lobe and a
+            larger size of the gatekeeper residue. The numbers of ionic interactions and hydrogen bonds around the
+            methylamine of staurosporine also affect the binding affinities.
+        </p>
+        <sec sec-type="methods">
+            <title>Methods</title>
+            <sec>
+                <title>Shape correlation by Mantel test</title>
+                <p>Protein Data Bank identification codes of structures with &#x02018;protein kinase activity&#x02019;,
+                    i.e., gene ontology ID GO:0004672 on MSDlite database (<xref ref-type="bibr" rid="b14">14</xref>),
+                    were filtered through the PISCES server in order to select representative protein crystal structures
+                    based on resolution, R-factor and completeness (<xref ref-type="bibr" rid="b15">15</xref>). The 80
+                    chosen structures with resolutions better than 3.0 &#x000c5; and R-factors less than 0.30 were then
+                    superposed onto cyclic AMP-dependent protein kinase (PKA) using the program
+                    <sc>baton</sc>
+                    based on the method developed in Comparer (<xref ref-type="bibr" rid="b16">16</xref>). PKA, the
+                    first protein kinase for which a crystal structure became available (<xref ref-type="bibr"
+                                                                                               rid="b17">17</xref>), was
+                    selected as the template because its residue nomenclature is widely used in kinase analyses (<xref
+                            ref-type="bibr" rid="b18">18</xref>). The obtained structural alignment was used to infer
+                    equivalent residues in the kinase superfamily. Distances between every residue surrounding the
+                    pocket were calculated. This was achieved by selecting a &#x02018;representative&#x02019; atom,
+                    generally located near the end of the side chain for each amino acid residue. For example, the
+                    hydroxyl oxygen was chosen for serine and the beta-carbon was chosen for threonine (see<xref
+                            ref-type="supplementary-material" rid="SD2">Appendix S2</xref>). Half-diagonal distance
+                    matrices were constructed and the correlations between the matrices were calculated by the Mantel
+                    test using program
+                    <sc>zt</sc>
+                    (<xref ref-type="bibr" rid="b19">19</xref>). Relationships between distance matrices were defined
+                    using the neighbor-joining algorithm from the program
+                    <sc>phylip</sc>
+                    (<xref ref-type="bibr" rid="b20">20</xref>) and the dendrogram was made for comparison using the
+                    program
+                    <sc>treeview</sc>
+                    (<xref ref-type="bibr" rid="b21">21</xref>). Inter-residue distances were used which characterize a
+                    particular kinase regardless of the ligand with which it was crystallized. The equivalent set of
+                    distances was also measured for another set of 35 non-redundant crystal structures of kinases, which
+                    have been assayed by Fabian
+                    <italic>et al.</italic>
+                    (<xref ref-type="bibr" rid="b12">12</xref>). These structures contain a variety of inhibitors bound
+                    in the ATP-binding site. This allows the relationship between the spatial arrangement of residues of
+                    different kinases and their binding affinities (
+                    <italic>K</italic>
+                    <sub>d</sub>) against 10 ligands to be visualized. We used the color gradient representation of the
+                    tree plotting program
+                    <sc>itol</sc>
+                    (<xref ref-type="bibr" rid="b22">22</xref>), where the intensities of the colors were proportional
+                    to &#x02212;log
+                    <sub>10</sub>
+                    <italic>K</italic>
+                    <sub>d</sub>, in order to allow comparison for millimolar to sub-nanomolar values of binding
+                    constants (
+                    <italic>K</italic>
+                    <sub>d</sub>).
+                </p>
+            </sec>
+            <sec>
+                <title>Position-specific interactions</title>
+                <p>The structures of 20 staurosporine&#x02013;kinase complexes were superposed on the indolocarbazole
+                    moiety of staurosporine in PDB ID 1stc in order to compare them with the structures of 24 adenosine
+                    phosphate&#x02013;kinase complexes, which were superposed on the adenine ring from ATP in PDB ID
+                    1atp. The position-specific interactions were considered at two levels: the atom type and the
+                    residue type. For the atom-based arrays, atoms in the PDB file were assigned an atom type according
+                    to the simplified approach used in the AMBER force field, which has been developed specially for
+                    molecular mechanics calculations of proteins and nucleic acids (<xref ref-type="bibr" rid="b23">
+                        23</xref>) (<xref ref-type="supplementary-material" rid="SD1">Appendix S1</xref>). For the
+                    residue-based arrays, only &#x02018;representative&#x02019; atoms and oxygen of water molecules were
+                    assigned a residue type (see<xref ref-type="supplementary-material" rid="SD2">Appendix S2</xref>).
+                    We constructed a three-dimensional grid with 1 &#x000c5; dimensions around the rigid part of the
+                    superposed ligands and collected occupancy in the grid boxes based on residue or atom types and x,
+                    y, and z coordinates for non-redundant protein kinase structures. For the atomic level, the grid was
+                    stored in PDB file format and occupancies of boxes were contoured using the module color_b of
+                    program
+                    <sc>pymol</sc>
+                    (<xref ref-type="bibr" rid="b24">24</xref>) to obtain a transparent surface with the intensity of
+                    the color corresponding to the frequency with which the grid boxes are populated. For the residue
+                    level, relative positions of neighboring residues for different ligands were observed by superposing
+                    the majority of the residue clusters surrounding adenine and staurosporine, and then comparing the
+                    positions of these clusters for adenine complexes and staurosporine complexes. For visualizing the
+                    cluster, a bond is drawn automatically when two atoms come closer than about 2 &#x000c5; so the
+                    cluster can be easily found. Colors of frequently occurring residues are assigned according to the
+                    type of amino acid found in the array.
+                </p>
+            </sec>
+            <sec>
+                <title>Distance-based QSAR</title>
+                <p>We filtered the PDB (<xref ref-type="bibr" rid="b25">25</xref>) for X-ray structures of the kinases
+                    for which Fabian
+                    <italic>et al.</italic>
+                    (<xref ref-type="bibr" rid="b12">12</xref>) report a
+                    <italic>K</italic>
+                    <sub>d</sub>
+                    for staurosporine (
+                    <italic>K</italic>
+                    <sub>d,STU</sub>), and selected only those structures that are co-crystallized as either
+                    staurosporine or adenosine phosphate complexes (40 structures). As in the case of shape analysis,
+                    distances from representative atoms between 15 residues surrounding the pocket were measured to find
+                    the best correlation of distances with
+                    <italic>K</italic>
+                    <sub>d,STU</sub>. Multiple linear regression was performed using the program
+                    <sc>xlstat</sc>
+                    (<xref ref-type="bibr" rid="b26">26</xref>) to find the best equation relating the distances
+                    measured between the centre points near the ends of the side chains (see<xref
+                            ref-type="supplementary-material" rid="SD2">Appendix S2</xref>) and log
+                    <sub>10</sub>
+                    <italic>K</italic>
+                    <sub>d,STU</sub>. Amino acid residues used to create these distance descriptors were extracted by
+                    referring to PKA equivalent residues from the ClustalX (<xref ref-type="bibr" rid="b27">27</xref>)
+                    multiple sequence alignment of the 113 kinases in Fabian&#x02019;s data set (<xref ref-type="bibr"
+                                                                                                       rid="b12">
+                        12</xref>). The relationships of these influential residues were drawn from the neighbor-joining
+                    algorithm in ClustalX (<xref ref-type="bibr" rid="b27">27</xref>). The abilities of kinases to bind
+                    the inhibitors staurosporine, LY-333531, SU11248, and ZD-6474 were calculated from &#x02212;log
+                    <sub>10</sub>
+                    <italic>K</italic>
+                    <sub>d,</sub>
+                    and a dendrogram was produced with gradient color representation using program
+                    <sc>itol</sc>
+                    (<xref ref-type="bibr" rid="b22">22</xref>) in order to reflect these values. The selected
+                    inhibitors are among the most promiscuous ligands in the Fabian data set, so that the number of the
+                    kinases they can bind is sufficient to demonstrate trends in binding affinities.
+                </p>
+                <p>Because the experimental data depend not only on the method used but also the experimentalist who
+                    reports the values, we choose dissociation constants of staurosporine (
+                    <italic>K</italic>
+                    <sub>d,STU</sub>) from Fabian
+                    <italic>et al.</italic>
+                    (<xref ref-type="bibr" rid="b12">12</xref>) as the sole source of our experimental binding data.
+                    Structures in this training set structure have
+                    <italic>K</italic>
+                    <sub>d,STU</sub>
+                    between 0.5 and 870 n
+                    <sc>m</sc>
+                    and both adenine-containing and staurosporine-bound structures are considered. We include adenine
+                    ring-containing structures in the data set on the assumption that the rigid part of the pockets that
+                    harbor adenosine or staurosporine share similar conformations and electronic features. The advantage
+                    of assuming that the structure of adenosine phosphate-bound complex resembles that of the same
+                    enzyme in the staurosporine-bound complex is that there are many structures in complex with
+                    adenosine containing compounds. The greater number of structures with available
+                    <italic>K</italic>
+                    <sub>d,STU</sub>
+                    allowed us then to test our equation by predicting
+                    <italic>K</italic>
+                    <sub>d,STU</sub>
+                    for further kinase structures co-crystallized with adenine ring-containing ligands. We discarded
+                    residue points that differ in position when found in contact with ATP or staurosporine, so that we
+                    could be sure that the differences in distances were independent of the ligand bound.
+                </p>
+            </sec>
+            <sec>
+                <title>Chemical interactions of staurosporine with various kinases</title>
+                <p>MAHORI (Mapping analogous hetero-atoms onto residue interactions) is our web-based tool designed to
+                    observe interactions surrounding specific parts of a ligand<xref ref-type="fn" rid="fn1">a</xref>.
+                    The website allows for various types of ligand query e.g., chemical drawing, compound name, PDB
+                    ligand three-letter code and SMILES string. Given the PDB ligand three-letter hetID of staurosporine
+                    (STU), MAHORI searches the PDB for the staurosporine complexes against the whole PDB using the
+                    PDB-ligand database CREDO (<xref ref-type="bibr" rid="b28">28</xref>). This database stores protein&#x02013;ligand
+                    interactions using criteria adapted from Marcou and Rognan (<xref ref-type="bibr" rid="b29">
+                        29</xref>). All atoms from the ligand and their contacting neighbor atoms from the protein have
+                    predefined types. Every contacting atom pair is assigned an interaction type by considering the
+                    type, the geometry and the threshold distance. The details of residues that interact with the
+                    queried atoms are presented according to the type of interaction, so that molecular interactions of
+                    multiple structures can be compared at the level of ligand substructure.
+                </p>
+            </sec>
+        </sec>
+        <sec>
+            <title>Results and Discussion</title>
+            <sec>
+                <title>Ability to bind inhibitors correlates with spatial arrangement of residues in the pocket</title>
+                <p>We wished to investigate whether the spatial arrangement of residues in the ATP-binding pocket has an
+                    influence on which inhibitor the kinase recognizes. In order to avoid comparing extremely variable
+                    regions of the pocket, we focused only on protein structures with staurosporine or adenine
+                    ring-containing compounds bound. We selected a set of points to represent common features of the
+                    pocket and used the Mantel test to distinguish the pockets of different kinases based on the
+                    assumption that the matrix of distances between points surrounding the adenosine pocket can reflect
+                    key features of the pocket shape in multi-dimensions; we call a matrix of this sort a &#x02018;quasi-shape&#x02019;
+                    (<xref ref-type="fig" rid="fig01">Figure 1A and 1B</xref>). We used calculated correlation
+                    coefficients among distance matrices of the same size and order of elements to estimate the
+                    similarities in the spatial arrangements of side chains and hence the relationships between shape
+                    and the ability of parent kinases to bind various inhibitors.
+                </p>
+                <fig id="fig01" position="float">
+                    <label>Figure 1</label>
+                    <caption>
+                        <p>Correlation between the shapes of the kinase ATP-binding pockets and their binding affinities
+                            to inhibitors. (A, B) The quasi-shape of the adenine-binding pocket of cAMP-dependent
+                            protein kinase (PDB ID:1stc, chain E) in complex with staurosporine (orange stick) is
+                            illustrated by drawing purple lines between the &#x02018;representative&#x02019; atoms of
+                            the 17 residues surrounding the pocket. (C) The shape-based dendrogram shows the matrix
+                            correlation between the shapes of the kinases and their binding affinities to 10 inhibitors.
+                            The method employs 17 active-site residues to construct the distance matrices for each
+                            kinase and then find the correlations between them. The pocket of STK10, an STE kinase,
+                            shows the greatest similarity in shape to that of LCK, a tyrosine kinase. Their inhibition
+                            profiles appear very similar (lower left). (D) The classic dendrogram based on sequence
+                            similarity from structure-based sequence alignment using program<sc>baton</sc>. The
+                            intensity of the color is proportional to log
+                            <sub>10</sub>
+                            <italic>K</italic>d of the inhibitor. It clusters the kinase with similar performance to the
+                            shape-based dendrogram. Staurosporine is the most promiscuous inhibitor in this data set
+                            (red).
+                        </p>
+                    </caption>
+                    <graphic xlink:href="jpp0074-0016-f1"/>
+                </fig>
+                <p>Our preliminary results suggest that Mantel test correlations between the matrices derived from a
+                    small set of inter-atomic distances can separate the majority of staurosporine complexes from the
+                    adenine-containing complexes. This means that there are observable differences in spatial
+                    arrangement of these atoms when staurosporine and adenine are bound. Thus, the Mantel Test appears
+                    to work well for classifying different three-dimensional geometric shapes. However, the same kinase
+                    in different crystal forms can be scattered throughout the resulting shape-based dendrogram,
+                    implying that similarities between conserved atoms are not able to identify identical kinases in
+                    different conformational states. Therefore, we investigated the use of &#x02018;representative&#x02019;
+                    atoms as the centers for distance measurements in the construction of the distance matrix, thus
+                    allowing the derivation of a quasi-shape from each PDB file. The choice of these atoms, which
+                    depends on their residue type, is shown in<xref ref-type="supplementary-material" rid="SD2">Appendix
+                        S2</xref>. By gradually increasing the number of residue points, we were able to cluster the
+                    same kinase in different crystal forms and different complexes into the same branch of the
+                    dendrogram. The most useful shape-based dendrogram is constructed from 17 representative points from
+                    17 residues. These are equivalent to the following residues in PKA: LEU 49, GLY 50, VAL 57, ALA 70,
+                    MET 71, LYS 72, VAL 104, MET 120, GLU 121, TYR 122, VAL 123, GLU 170, ASN 171, THR 183, ASP 184, GLU
+                    127, LEU 173. This dendrogram places the same type of kinase in different complexes in the same
+                    branch regardless of the bound ligand, and the staurosporine binding structures were clustered into
+                    one-half of the tree (<xref ref-type="supplementary-material" rid="SD3">Appendix S3</xref>).
+                </p>
+                <p>We then applied this method to a set of 35 non-redundant protein kinases and 10 inhibitors from the
+                    Fabian
+                    <italic>et al.</italic>
+                    data set (<xref ref-type="bibr" rid="b12">12</xref>) and obtained a dendrogram that characterizes
+                    the ability to bind 10 ligands, based on the similarity in quasi-shape (<xref ref-type="fig"
+                                                                                                  rid="fig01">Figure
+                        1C</xref>). The general sequence-based dendrogram is shown for comparison (<xref ref-type="fig"
+                                                                                                         rid="fig01">
+                        Figure 1D</xref>). It is evident that kinases with similar pocket quasi-shapes are likely to
+                    have similar inhibitor binding profiles, regardless of their family membership. A nice example is
+                    serine/threonine kinase 10 (STK10), which is clustered in the sequence-based dendrogram as an STE
+                    kinase, or Homolog of yeast Sterile 7, 11, 20 kinases, as defined in the protein kinase phylogenetic
+                    tree by Manning
+                    <italic>et al.</italic>
+                    (<xref ref-type="bibr" rid="b1">1</xref>). When considering STK10 in terms of similarity in spatial
+                    arrangement of residues, it is instead paired with leukocyte-specific protein tyrosine kinase (LCK)
+                    which is a tyrosine kinase. The sequences are quite different, but the quasi-shape of the pockets
+                    and their abilities to bind seven inhibitors are very similar. Many kinases with similar sequences,
+                    for example CDK2 and CDK5 or DAPK2 and DAPK3, also have very similar quasi-shapes and inhibition
+                    profiles. This quasi-shape-based dendrogram provides a way of visualizing relationships among
+                    kinases, complementing that of the classical sequence-based dendrogram. Our dendrogram demonstrates
+                    that the similarity in quasi-shape can sometimes explain the ability to bind a set of ligands
+                    regardless of the overall sequence identity.
+                </p>
+            </sec>
+            <sec>
+                <title>Conserved interaction shows induced-fit mechanism upon staurosporine binding</title>
+                <p>We hypothesize that if protein features surrounding a particular ligand remain conserved both in atom
+                    type and position in complexes of different kinases, they may be required for binding the ligand. We
+                    have developed software to extract generalized features that are frequently found in protein kinase
+                    structures by constructing four-dimensional arrays to capture different entities that are conserved
+                    in atomic position on structure superposition of staurosporine complexes. The array collects
+                    occupancies of atoms from superposed structures that satisfy the four criteria, i.e., x,y,z
+                    coordinates and atom type. This approach increases signal to noise by superposing a significant
+                    number of structures (20 structures of staurosporine complexes).
+                </p>
+                <p>The staurosporine molecule is quite rigid as it contains very few rotatable bonds; hence, we may
+                    observe interaction partners that are position-specific by superposing the kinases onto its lactam
+                    and indolocarbazole rings. In a similar manner, interactions around the adenosine phosphate
+                    complexes can be compared by superposing the non-redundant kinase structures onto the adenine ring
+                    (24 structures of adenine-containing complexes). The conserved atomic environment can be found by
+                    observing the frequently occurring atoms at a particular location defined by a 1 &#x000c5; grid box
+                    (<xref ref-type="fig" rid="fig02">Figure 2A and 2B</xref>).
+                </p>
+                <fig id="fig02" position="float">
+                    <label>Figure 2</label>
+                    <caption>
+                        <p>Frequently occurring atoms and residues around the ATP- and staurosporine-binding sites. The
+                            hinge region is on the left, the N-terminal lobe at the top and the C-terminal lobe at the
+                            bottom of the figure. (A,B) Color is related to the frequency of finding atoms at a position
+                            (CT = sp
+                            <sup>3</sup>
+                            carbon, C = carbonyl sp
+                            <sup>2</sup>
+                            carbon, N = sp
+                            <sup>2</sup>
+                            amide nitrogen, O = sp
+                            <sup>2</sup>
+                            oxygen). Both staurosporine (white stick) and the adenine ring (yellow stick) are recognized
+                            by the main chain atoms in the hinge region on the left. The atomic environment of adenine
+                            (A) shows more variability than that of staurosporine (B) as shown by the maximum occupancy
+                            of frequently occurring atoms (58%); this is smaller than that of staurosporine (75%). (C)
+                            The first glycine of the GXGXXG motif in purple moves within a rhombohedron-shaped volume
+                            when located near the adenine ring. This glycine retains its position in the same grid in
+                            75% of staurosporine structures (15 from 20 structures) indicating induced fit of the
+                            glycine-rich loop when located near staurosporine (D).
+                        </p>
+                    </caption>
+                    <graphic xlink:href="jpp0074-0016-f2"/>
+                </fig>
+                <p>The atom type categorization is based on the assumption that atoms around the side chains that have
+                    the same functional group can be classified as the same atom type, e.g., carboxylate oxygens of Asp
+                    and Glu have sp<sup>2</sup>-oxygen atom types (<xref ref-type="supplementary-material" rid="SD1">
+                        Appendix S1</xref>). By this approach, we can capture similar interactions in the pocket made by
+                    similar parts of ligands. The resulting atomic level arrays suggest that for both adenine and
+                    staurosporine ligand complexes (<xref ref-type="fig" rid="fig02">Figure 2A and 2B</xref>), the main
+                    chain atoms of the kinase make the most conserved interactions in terms of type and position, which
+                    explains why staurosporine, mimicking ATP, can bind to most of the kinases.
+                </p>
+                <p>The conserved neighboring atoms of ATP are more variable in position than those of staurosporine (38&#x02013;58%
+                    conservation in
+                    <xref ref-type="fig" rid="fig02">Figure 2A</xref>
+                    versus 50&#x02013;75% in<xref ref-type="fig" rid="fig02">Figure 2B</xref>). This supports the idea
+                    that the two ligands require a different degree of flexibility within the active site of the kinase.
+                    The greater number of rotatable bonds in the adenosine phosphate results in lower degree of
+                    conservation of neighboring atoms in these complexes than in staurosporine complexes. The ribose
+                    moiety of the adenosine complex can adopt several conformations, so the frequently occurring atoms
+                    fall in several grid boxes. The main chains of protein kinases in the hinge regions interact with
+                    the amine group of the adenine rings and are the most conserved parts of the adenine complexes. In a
+                    similar way, the main chains of the hinge regions, which interact with the lactam oxygen, and the
+                    &#x003b1;-carbons of the first glycines of the GXGXXG motifs, which interact with the
+                    tetrahydropyran, are the most conserved parts for staurosporine complexes. Furthermore, for 15 out
+                    of 20 staurosporine-bound structures, the main chain alpha carbons from the first glycine of the
+                    G-rich loop fall in the same 1 &#x000c5; grid box (<xref ref-type="fig" rid="fig02">Figure 2D</xref>
+                    ). When superposed on the adenine ring, these atoms are distributed within a rhombohedron-shaped
+                    volume (see<xref ref-type="fig" rid="fig02">Figure 2C</xref>). It is perhaps surprising that this
+                    glycine can become fixed in position upon staurosporine binding because the glycine-rich loop is
+                    generally believed to be highly flexible (<xref ref-type="bibr" rid="b30">30</xref>,<xref
+                            ref-type="bibr" rid="b31">31</xref>). The relatively well-conserved position of this
+                    glycine, but not for adenine complexes, suggests that the staurosporine causes an induced fit or
+                    conformational selection in the kinases upon binding. However, further evidence from apo structures
+                    is required to confirm this observation.
+                </p>
+                <p>The residue arrays serve to complement the pictorial representations of the atomic environment. For
+                    most of the residues, we chose the penultimate atoms of the side chains to represent the identities
+                    and the positions of the residues (<xref ref-type="supplementary-material" rid="SD2">Appendix
+                        S2</xref>). In this way, residues with a similar functional group at the end of the side chain
+                    in different kinases can be captured as points at similar locations in the superposed structures.
+                    For instance, C
+                    <sub>&#x003b2;</sub>
+                    of valine and C
+                    <sub>&#x003b3;</sub>
+                    of leucine at the active sites occupy the same or close-by grid boxes in the superposed structures.
+                    We observe that some clusters of amino acids preserve their functional groups in most of the
+                    staurosporine and the adenine complexes; examples are the side chains of glutamate and aspartate of
+                    the salt-bridges which flank the pocket on both sides, the residues that are equivalent to Ala
+                    <sub>70</sub>
+                    in PKA which acts as the ceiling of the cleft, and again Gly
+                    <sub>50</sub>
+                    which interacts with the ether oxygen of the ribose (<xref ref-type="fig" rid="fig02">Figure 2C and
+                        2D</xref>). Several clusters of amino acids are seen to have moved, leading to contraction and
+                    expansion of the residues in the pocket to accommodate staurosporine.
+                </p>
+                <p>The ends of some similar hydrophobic side chains, including Cys and Met, or Ala, Val and Leu, which
+                    surround the planar indolocarbazole ring, have equivalent positions, implying that these amino acids
+                    perform the same function in that part of the active site. On the contrary, the amino acids that
+                    make contact around the methyl amino and methoxy groups of staurosporine demonstrate that remarkably
+                    different functional groups can occupy the same position and carry out the same structural role.
+                </p>
+            </sec>
+            <sec>
+                <title>Distance descriptors and interaction types correlate with the magnitudes of binding affinities
+                </title>
+                <p>Staurosporine has very few strongly electrostatic features. Its major interactions with protein
+                    kinases are largely steric with non-polar groups. Thus, we hypothesized that the tightness of
+                    inhibitor binding might be determined by the compactness of residues in the pocket. In order to test
+                    this idea and to predict binding affinities from the structures, we assumed that good binding
+                    requires certain geometric restraints and investigated which distance descriptors correlate well
+                    with the dissociation constants. Thus, if distances are shorter for most of the structures with low
+                    binding constants, we investigate whether contraction along that direction is required for tight
+                    binding. This approach resembles QSAR, but all the input parameters are measured from the structure
+                    in terms of distances that constitute the quasi-shape of the ATP-binding pocket. Although QSAR
+                    methodologies have been widely used to try to understand binding affinities through various
+                    parameters related to lipophilicity, charge and hydrogen bonding character (<xref ref-type="bibr"
+                                                                                                      rid="b32">
+                        32</xref>), distances between certain atoms in the protein have not been used.
+                </p>
+                <p>In order to select distances from the quasi-shape defined by 15 points in contact with staurosporine
+                    (<xref ref-type="fig" rid="fig01">Figure 1A and 1B</xref>;<xref ref-type="supplementary-material"
+                                                                                    rid="SD4">Appendix S4</xref>), we
+                    carried out multiple linear regression with the equations shown in<xref ref-type="table" rid="tbl1">
+                        Table 1</xref>. We tested the predictive power of these equations by leaving out randomly
+                    selected test sets. While the purpose of using multiple linear regression in this context was simply
+                    to select the set of distances that correlate well with the binding affinities, the resulting
+                    equations suggest that predictive power might be demonstrated if a larger data set were available.
+                    All resulting equations appear to contain the same best sets of distances producing
+                    <italic>R</italic>
+                    <sup>&#x0200a;2</sup>
+                    values for the random test sets of about 0.7 for both equations (<xref
+                            ref-type="supplementary-material" rid="SD5">Appendix S5</xref>).
+                </p>
+                <table-wrap id="tbl1" position="float">
+                    <label>Table 1</label>
+                    <caption>
+                        <p>Equations correlating the influential distances with log
+                            <sub>10</sub>
+                            <italic>K</italic>
+                            <sub>d,STU</sub>
+                        </p>
+                    </caption>
+                    <table frame="hsides" rules="groups">
+                        <thead>
+                            <tr>
+                                <th align="left" rowspan="1" colspan="1">Random Test</th>
+                                <th align="center" rowspan="1" colspan="1"><italic>R</italic>&#x0200a;
+                                    <sup>2</sup>
+                                    training
+                                </th>
+                                <th align="center" rowspan="1" colspan="1"><italic>R</italic>&#x0200a;
+                                    <sup>2</sup>
+                                    test set
+                                </th>
+                                <th align="center" rowspan="1" colspan="1">Equation</th>
+                            </tr>
+                        </thead>
+                        <tbody>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">None</td>
+                                <td align="center" rowspan="1" colspan="1">0.6</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">log
+                                    <sub>10</sub>
+                                    <italic>K</italic>
+                                    <sub>d,STU</sub>
+                                    = 3.4 + 0.1D<sub>50_184</sub>&#x02212;0.4D
+                                    <sub>120_123</sub>
+                                </td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">5 Structures</td>
+                                <td align="center" rowspan="1" colspan="1">0.6</td>
+                                <td align="center" rowspan="1" colspan="1">0.7</td>
+                                <td align="center" rowspan="1" colspan="1">log
+                                    <sub>10</sub>
+                                    <italic>K</italic>
+                                    <sub>d,STU</sub>
+                                    = 3.4 + 0.1D<sub>50_184</sub>&#x02212;0.4D
+                                    <sub>120_123</sub>
+                                </td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">10 Structures</td>
+                                <td align="center" rowspan="1" colspan="1">0.7</td>
+                                <td align="center" rowspan="1" colspan="1">0.7</td>
+                                <td align="center" rowspan="1" colspan="1">log
+                                    <sub>10</sub>
+                                    <italic>K</italic>
+                                    <sub>d,STU</sub>
+                                    = 3.6 + 0.1D<sub>50_184</sub>&#x02212;0.4D
+                                    <sub>120_123</sub>
+                                </td>
+                            </tr>
+                        </tbody>
+                    </table>
+                </table-wrap>
+                <p>The distance descriptors which correlate well with binding affinities, either having positive or
+                    negative influence on
+                    <italic>K</italic>
+                    <sub>d,STU</sub>, are called the &#x02018;influential distances&#x02019;. Although correlation does
+                    not imply causation, examination of the crystal structures allows us to interpret these results in
+                    terms of structure.
+                    <xref ref-type="fig" rid="fig03">Figure 3A</xref>
+                    illustrates these influential distances in the structure of PKA, PDB ID 1stc. They can be used to
+                    describe how positions of representative side-chain atoms can influence
+                    <italic>K</italic>
+                    <sub>d,STU</sub>
+                    and also to help understand the major changes in neighboring atomic positions around the
+                    staurosporine (<xref ref-type="fig" rid="fig03">Figure 3C</xref>).
+                </p>
+                <fig id="fig03" position="float">
+                    <label>Figure 3</label>
+                    <caption>
+                        <p>The interpretation of multiple linear regression equations. (A) Interpretation of the
+                            multiple linear regression analysis shows that smaller values of
+                            <italic>K</italic>
+                            <sub>d,STU</sub>
+                            result from the larger size of side chains of the gatekeeper and gatekeeper + 3 residues,
+                            i.e., PKA equivalent residue: Met
+                            <sub>120</sub>
+                            and Val
+                            <sub>123</sub>
+                            (orange bar). The equation suggests that the closer approach between Gly
+                            <sub>50</sub>
+                            of the N-terminal lobe and Asp
+                            <sub>184</sub>
+                            of the C-terminal lobe (purple bar) correlate with tighter binding to staurosporine. (B) A
+                            dendrogram displaying relationships between 113 kinases based on neighbor-joining of the 13
+                            residues which are in contact with staurosporine and show correlation (&#x0003c;&#x02212;0.4
+                            and &#x0003e;0.4) with
+                            <italic>K</italic>
+                            <sub>d,STU</sub>. The aim is to investigate whether the similarities between these
+                            influential residues, equivalent to PKA residues 49, 50, 57, 70, 71, 72, 120, 121, 122, 123,
+                            170, 171, and 184, give rise to similar binding constants. The resulting dendrogram can
+                            cluster staurosporine tight binders into two major groups with better binding affinity to
+                            staurosporine (dark red). This group of kinases tends to have large gatekeeper residues,
+                            e.g., Phe (F), Met (M). Smaller gatekeeper residues, e.g., Thr or Leu, tend to be associated
+                            with weaker binding affinities to staurosporine. A majority of kinases which are inhibited
+                            by ZD-6474 (blue) has threonine (T) or valine (V) as a gatekeeper residue. Binding
+                            affinities to LY-333531 (green) and SU11248 (yellow) are shown for comparison. (C)
+                            Staurosporine structural components. (D) Chemical structure of staurosporine based on
+                            annotation from Zhao
+                            <italic>et al.</italic>
+                            (<xref ref-type="bibr" rid="b33">33</xref>).
+                        </p>
+                    </caption>
+                    <graphic xlink:href="jpp0074-0016-f3"/>
+                </fig>
+                <p>Of the equations shown in<xref ref-type="table" rid="tbl1">Table 1</xref>, the distance between
+                    residues 50 and 184, described in equation as D<sub>50_184</sub>, is directly proportional to the
+                    value of log
+                    <sub>10</sub>
+                    <italic>K</italic>
+                    <sub>d,STU</sub>, and the distance between residue 120 and 123, D<sub>120_123</sub>, is inversely
+                    proportional to log
+                    <sub>10</sub>
+                    <italic>K</italic>
+                    <sub>d,STU</sub>. A possible interpretation of the equation is that in kinases that are tightly
+                    bound to staurosporine, i.e., have a small log
+                    <sub>10</sub>
+                    <italic>K</italic>
+                    <sub>d,STU</sub>, there is a preference for a smaller D
+                    <sub>50_184</sub>
+                    and a larger D<sub>120_123</sub>. In PKA, the distance between residues 50 and 184 is measured
+                    between C&#x003b1; of Gly
+                    <sub>50</sub>
+                    of the GXGXXG motif in N-terminal lobe to C&#x003b3; of Asp
+                    <sub>184</sub>
+                    of the DFG loop in C-terminal lobe. Staurosporine is located between the two lobes, and the closer
+                    approach of these two motifs in a direction perpendicular to the plane of staurosporine reflects the
+                    better binding affinities presumably because of the resultant tighter binding. In contrast, the
+                    distance between residue 120 (gatekeeper) and 123 (gatekeeper + 3) implies the expansion of the
+                    pocket along this direction. The equation suggests that these two residues should move further apart
+                    to accommodate staurosporine. The gatekeeper residue points toward the plane of staurosporine, while
+                    the gatekeeper + 3 residue is located under the indolocarbazole ring. The size of the gatekeeper and
+                    the gatekeeper + 3 residues appear to have a key role in locking the lactam in the correct
+                    orientation while making optimal steric interactions with the indolocarbazole of staurosporine. The
+                    larger size of the gatekeeper residue likely results in the larger distance and correlates with good
+                    binding because the larger volumes of the side chains in the plane of the lactam ring promote
+                    favorable hydrophobic interactions in the pocket.
+                </p>
+                <p>We speculated that the interactions involving the methyl amino (N<sub>4&#x02032;</sub>) and the
+                    methoxy group (O<sub>3&#x02032;</sub>) of staurosporine should constrain the distance between the N-
+                    and C-terminal lobes to the optimal value (<xref ref-type="fig" rid="fig03">Figure 3D</xref>).
+                    Indeed, the hydrogen bonds or ionic interactions that the staurosporine can make along this
+                    direction are associated with the major differences in the binding affinities. We find that the
+                    number of hydrogen bonds made by residues around N
+                    <sub>4&#x02032;</sub>
+                    of staurosporine corresponds well with the trend in binding affinities (<xref ref-type="table"
+                                                                                                  rid="tbl2">Table
+                        2</xref>). Kinase structures that have two residues making hydrogen bonds or ionic interactions
+                    to N
+                    <sub>4&#x02032;</sub>
+                    of staurosporine, i.e., CDK2, PKA, PIM1, and LCK, have binding affinities below 51 n<sc>m</sc>. Most
+                    structures that have only one residue contributing hydrogen bonds or ionic interaction to N
+                    <sub>4&#x02032;</sub>
+                    have binding affinities between 51 and 440 n<sc>m</sc>, i.e., CSK, EGFR, FYN, M3K5. The kinase STK16
+                    which does not make any interaction with N
+                    <sub>4&#x02032;</sub>
+                    has a binding affinity of 200 n<sc>m</sc>.
+                </p>
+                <table-wrap id="tbl2" position="float">
+                    <label>Table 2</label>
+                    <caption>
+                        <p>Number of interactions made by the kinases with N
+                            <sub>4&#x02032;</sub>
+                            of staurosporine
+                        </p>
+                    </caption>
+                    <table frame="hsides" rules="groups">
+                        <thead>
+                            <tr>
+                                <th align="center" rowspan="1" colspan="1"/>
+                                <th align="center" rowspan="1" colspan="1"/>
+                                <th align="center" rowspan="1" colspan="1"/>
+                                <th align="center" colspan="3" rowspan="1">Number of interactions
+                                    <hr/>
+                                </th>
+                            </tr>
+                            <tr>
+                                <th align="center" rowspan="1" colspan="1">Protein kinase</th>
+                                <th align="center" rowspan="1" colspan="1">PDB ID</th>
+                                <th align="center" rowspan="1" colspan="1">
+                                    <italic>K</italic>
+                                    <sub>d,STU</sub>
+                                    (n<sc>m</sc>)
+                                </th>
+                                <th align="left" rowspan="1" colspan="1">H-bond</th>
+                                <th align="center" rowspan="1" colspan="1">Ionic</th>
+                                <th align="center" rowspan="1" colspan="1">vdW</th>
+                            </tr>
+                        </thead>
+                        <tbody>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">CDK2</td>
+                                <td align="center" rowspan="1" colspan="1">1AQ1</td>
+                                <td align="center" rowspan="1" colspan="1">8.1</td>
+                                <td align="center" rowspan="1" colspan="1">2</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                                <td align="center" rowspan="1" colspan="1">2</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">PIM1</td>
+                                <td align="center" rowspan="1" colspan="1">1YHS</td>
+                                <td align="center" rowspan="1" colspan="1">15</td>
+                                <td align="center" rowspan="1" colspan="1">2</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                                <td align="center" rowspan="1" colspan="1">2</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">LCK</td>
+                                <td align="center" rowspan="1" colspan="1">1QPJ</td>
+                                <td align="center" rowspan="1" colspan="1">20</td>
+                                <td align="center" rowspan="1" colspan="1">2</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">2</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">PKA</td>
+                                <td align="center" rowspan="1" colspan="1">1STC</td>
+                                <td align="center" rowspan="1" colspan="1">50</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                                <td align="center" rowspan="1" colspan="1">2</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">KSYK</td>
+                                <td align="center" rowspan="1" colspan="1">1XBC</td>
+                                <td align="center" rowspan="1" colspan="1">7</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">FYN</td>
+                                <td align="center" rowspan="1" colspan="1">2DQ7</td>
+                                <td align="center" rowspan="1" colspan="1">51</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">M3K5</td>
+                                <td align="center" rowspan="1" colspan="1">2CLQ</td>
+                                <td align="center" rowspan="1" colspan="1">120</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">CSK</td>
+                                <td align="center" rowspan="1" colspan="1">1BYG</td>
+                                <td align="center" rowspan="1" colspan="1">440</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">MKNK2</td>
+                                <td align="center" rowspan="1" colspan="1">2HW7</td>
+                                <td align="center" rowspan="1" colspan="1">22</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">EGFR</td>
+                                <td align="center" rowspan="1" colspan="1">2ITU</td>
+                                <td align="center" rowspan="1" colspan="1">70</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">1</td>
+                                <td align="center" rowspan="1" colspan="1">2</td>
+                            </tr>
+                            <tr>
+                                <td align="left" rowspan="1" colspan="1">STK16</td>
+                                <td align="center" rowspan="1" colspan="1">2BUJ</td>
+                                <td align="center" rowspan="1" colspan="1">200</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                                <td align="center" rowspan="1" colspan="1">&#x02013;</td>
+                            </tr>
+                        </tbody>
+                    </table>
+                </table-wrap>
+                <p>Therefore, in order to modify staurosporine to achieve better affinity for kinases, the strategy
+                    might be to identify a residue close to the methyl amino (N<sub>4&#x02032;</sub>) and to modify the
+                    staurosporine to make another hydrogen bond. Making point mutations of active site residues in order
+                    to achieve a better binding affinity to staurosporine might also be achieved by selecting the
+                    residue type that can make an optimal hydrogen bond to N
+                    <sub>4&#x02032;</sub>
+                    and O
+                    <sub>3&#x02032;</sub>
+                    of staurosporine.
+                </p>
+            </sec>
+        </sec>
+        <sec>
+            <title>Conclusion</title>
+            <p>Simple comparison of distance matrices, generated from representative atoms toward the ends of
+                side-chains, can be used to describe the geometry of a ligand binding site and this can be related to
+                inhibitor binding. By considering the similarities and differences in the active sites, our
+                computational approach can rediscover several kinase binding determinants that have been previously
+                identified from manual and experimental analyses. We show by grouping structures with similar inhibition
+                profiles that the shape of the pocket can contribute to inhibitor selectivity. The most significant part
+                of the protein kinase structures that remains fixed in type and position for both staurosporine and
+                adenosine structures is the main chain at the beginning of the hinge region. This implies that the
+                reason that staurosporine binds to most kinases is that the lactam from staurosporine and adenine from
+                ATP recognize a similar set of atoms. Our observation gives a more precise picture of the induced fit of
+                the conserved glycine-rich loop upon binding to staurosporine. In addition, our statistical analysis
+                shows that a larger size of the gatekeeper residues normally results in tight binding to staurosporine.
+                We have also learned that the hydrogen bond and ionic interaction made with methylamine is important to
+                the tightness of the binding with staurosporine.
+            </p>
+            <p>These results indicate that by understanding differences in the active sites, we can identify residues
+                that affect the ability to bind the inhibitor and also suggest the part of the inhibitor that might be
+                modified to achieve better binding affinities. This approach offers a new perspective on computational
+                descriptions of specificity determinants when a series of different proteins with the same ligand
+                becomes available.
+            </p>
+        </sec>
+    </body>
+    <back>
+        <fn-group>
+            <fn id="fn1">
+                <label>a</label>
+                <p>
+                    <ext-link ext-link-type="uri" xlink:href="http://www-cryst.bioc.cam.ac.uk/mahori">
+                        http://www-cryst.bioc.cam.ac.uk/mahori
+                    </ext-link>
+                </p>
+            </fn>
+        </fn-group>
+        <ack>
+            <p>We thank Dr Rinaldo Wander Montalvao for recommending the Mantel test to find correlation between the
+                distance matrices, Dr David Burke for the program
+                <sc>baton</sc>
+                for multiple structural alignment, and Dr. Richard Smith for program
+                <sc>kinasemap</sc>
+                used for extracting cAMP-dependent protein kinase equivalent residues from different PDB files. DT is
+                supported by the Royal Thai Government. TLB thanks the Wellcome Trust for research support.
+            </p>
+        </ack>
+        <ref-list>
+            <title>References</title>
+            <ref id="b1">
+                <label>1</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Manning</surname>
+                            <given-names>G</given-names>
+                        </name>
+                        <name>
+                            <surname>Whyte</surname>
+                            <given-names>DB</given-names>
+                        </name>
+                        <name>
+                            <surname>Martinez</surname>
+                            <given-names>R</given-names>
+                        </name>
+                        <name>
+                            <surname>Hunter</surname>
+                            <given-names>T</given-names>
+                        </name>
+                        <name>
+                            <surname>Sudarsanam</surname>
+                            <given-names>S</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>The protein kinase complement of the human genome</article-title>
+                    <source>Science</source>
+                    <year>2002</year>
+                    <volume>298</volume>
+                    <fpage>1912</fpage>
+                    <lpage>1934</lpage>
+                    <pub-id pub-id-type="pmid">12471243</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b2">
+                <label>2</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Huse</surname>
+                            <given-names>M</given-names>
+                        </name>
+                        <name>
+                            <surname>Kuriyan</surname>
+                            <given-names>J</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>The conformational plasticity of protein kinases</article-title>
+                    <source>Cell</source>
+                    <year>2002</year>
+                    <volume>109</volume>
+                    <fpage>275</fpage>
+                    <lpage>282</lpage>
+                    <pub-id pub-id-type="pmid">12015977</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b3">
+                <label>3</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Blume-Jensen</surname>
+                            <given-names>P</given-names>
+                        </name>
+                        <name>
+                            <surname>Hunter</surname>
+                            <given-names>T</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>Oncogenic kinase signalling</article-title>
+                    <source>Nature</source>
+                    <year>2001</year>
+                    <volume>411</volume>
+                    <fpage>355</fpage>
+                    <lpage>365</lpage>
+                    <pub-id pub-id-type="pmid">11357143</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b4">
+                <label>4</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Cohen</surname>
+                            <given-names>P</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>Protein kinases: the major drug targets of the twenty-first century?</article-title>
+                    <source>Nat Rev Drug Discov</source>
+                    <year>2002</year>
+                    <volume>1</volume>
+                    <fpage>309</fpage>
+                    <lpage>315</lpage>
+                    <pub-id pub-id-type="pmid">12120282</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b5">
+                <label>5</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Herlaar</surname>
+                            <given-names>E</given-names>
+                        </name>
+                        <name>
+                            <surname>Brown</surname>
+                            <given-names>Z</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>p38 MAPK signalling cascades in inflammatory disease</article-title>
+                    <source>Mol Med Today</source>
+                    <year>1999</year>
+                    <volume>5</volume>
+                    <fpage>439</fpage>
+                    <lpage>447</lpage>
+                    <pub-id pub-id-type="pmid">10498912</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b6">
+                <label>6</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Noble</surname>
+                            <given-names>ME</given-names>
+                        </name>
+                        <name>
+                            <surname>Endicott</surname>
+                            <given-names>JA</given-names>
+                        </name>
+                        <name>
+                            <surname>Johnson</surname>
+                            <given-names>LN</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>Protein kinase inhibitors: insights into drug design from structure</article-title>
+                    <source>Science</source>
+                    <year>2004</year>
+                    <volume>303</volume>
+                    <fpage>1800</fpage>
+                    <lpage>1805</lpage>
+                    <pub-id pub-id-type="pmid">15031492</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b7">
+                <label>7</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Force</surname>
+                            <given-names>T</given-names>
+                        </name>
+                        <name>
+                            <surname>Kuida</surname>
+                            <given-names>K</given-names>
+                        </name>
+                        <name>
+                            <surname>Namchuk</surname>
+                            <given-names>M</given-names>
+                        </name>
+                        <name>
+                            <surname>Parang</surname>
+                            <given-names>K</given-names>
+                        </name>
+                        <name>
+                            <surname>Kyriakis</surname>
+                            <given-names>JM</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>Inhibitors of protein kinase signaling pathways: emerging therapies for
+                        cardiovascular disease
+                    </article-title>
+                    <source>Circulation</source>
+                    <year>2004</year>
+                    <volume>109</volume>
+                    <fpage>1196</fpage>
+                    <lpage>1205</lpage>
+                    <pub-id pub-id-type="pmid">15023894</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b8">
+                <label>8</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Hopkins</surname>
+                            <given-names>AL</given-names>
+                        </name>
+                        <name>
+                            <surname>Mason</surname>
+                            <given-names>JS</given-names>
+                        </name>
+                        <name>
+                            <surname>Overington</surname>
+                            <given-names>JP</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>Can we rationally design promiscuous drugs?</article-title>
+                    <source>Curr Opin Struct Biol</source>
+                    <year>2006</year>
+                    <volume>16</volume>
+                    <fpage>127</fpage>
+                    <lpage>136</lpage>
+                    <pub-id pub-id-type="pmid">16442279</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b9">
+                <label>9</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Sheinerman</surname>
+                            <given-names>FB</given-names>
+                        </name>
+                        <name>
+                            <surname>Giraud</surname>
+                            <given-names>E</given-names>
+                        </name>
+                        <name>
+                            <surname>Laoui</surname>
+                            <given-names>A</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>High affinity targets of protein kinase inhibitors have similar residues at the
+                        positions energetically important for binding
+                    </article-title>
+                    <source>J Mol Biol</source>
+                    <year>2005</year>
+                    <volume>352</volume>
+                    <fpage>1134</fpage>
+                    <lpage>1156</lpage>
+                    <pub-id pub-id-type="pmid">16139843</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b10">
+                <label>10</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Omura</surname>
+                            <given-names>S</given-names>
+                        </name>
+                        <name>
+                            <surname>Iwai</surname>
+                            <given-names>Y</given-names>
+                        </name>
+                        <name>
+                            <surname>Hirano</surname>
+                            <given-names>A</given-names>
+                        </name>
+                        <name>
+                            <surname>Nakagawa</surname>
+                            <given-names>A</given-names>
+                        </name>
+                        <name>
+                            <surname>Awaya</surname>
+                            <given-names>J</given-names>
+                        </name>
+                        <name>
+                            <surname>Tsuchya</surname>
+                            <given-names>H</given-names>
+                        </name>
+                        <name>
+                            <surname>Takahashi</surname>
+                            <given-names>Y</given-names>
+                        </name>
+                        <name>
+                            <surname>Masuma</surname>
+                            <given-names>R</given-names>
+                        </name>
+                    </person-group>
+                    <article-title>A new alkaloid AM-2282 of
+                        <italic>Streptomyces</italic>
+                        origin. Taxonomy, fermentation, isolation and preliminary characterization
+                    </article-title>
+                    <source>J Antibiot (Tokyo)</source>
+                    <year>1977</year>
+                    <volume>30</volume>
+                    <fpage>275</fpage>
+                    <lpage>282</lpage>
+                    <pub-id pub-id-type="pmid">863788</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b11">
+                <label>11</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Meggio</surname>
+                            <given-names>F</given-names>
+                        </name>
+                        <name>
+                            <surname>Deana</surname>
+                            <given-names>AD</given-names>
+                        </name>
+                        <name>
+                            <surname>Ruzzene</surname>
+                            <given-names>M</given-names>
+                        </name>
+                        <name>
+                            <surname>Brunati</surname>
+                            <given-names>AM</given-names>
+                        </name>
+                        <name>
+                            <surname>Cesaro</surname>
+                            <given-names>L</given-names>
+                        </name>
+                        <name>
+                            <surname>Guerra</surname>
+                            <given-names>B</given-names>
+                        </name>
+                        <name>
+                            <surname>Meyer</surname>
+                            <given-names>T</given-names>
+                        </name>
+                        <name>
+                            <surname>Mett</surname>
+                            <given-names>H</given-names>
+                        </name>
+                        <name>
+                            <surname>Fabbro</surname>
+                            <given-names>D</given-names>
+                        </name>
+                        <name>
+                            <surname>Furet</surname>
+                            <given-names>P</given-names>
+                        </name>
+                        <etal/>
+                    </person-group>
+                    <article-title>Different susceptibility of protein kinases to staurosporine inhibition. Kinetic
+                        studies and molecular bases for the resistance of protein kinase CK2
+                    </article-title>
+                    <source>Eur J Biochem</source>
+                    <year>1995</year>
+                    <volume>234</volume>
+                    <fpage>317</fpage>
+                    <lpage>322</lpage>
+                    <comment>doi
+                        <ext-link ext-link-type="doi" xlink:href="10.1111/j.1432-1033.1995.317_c.x">
+                            10.1111/j.1432-1033.1995.317_c.x
+                        </ext-link>
+                    </comment>
+                    <pub-id pub-id-type="pmid">8529658</pub-id>
+                </element-citation>
+            </ref>
+            <ref id="b12">
+                <label>12</label>
+                <element-citation publication-type="journal">
+                    <person-group person-group-type="author">
+                        <name>
+                            <surname>Fabian</surname>
+                            <given-names>MA</given-names>
+                        </name>
+                        <name>
+                            <surname>Biggs</surname>
+                            <given-names>WH</given-names>
+                            <suffix>III</suffix>
+                        </name>
+                        <name>
+                            <surname>Treiber</surname>
+                            <given-names>DK</given-names>
+                        </name>
+                        <name>

[... 722 lines stripped ...]


Mime
View raw message