K-Map: connecting kinases with therapeutics for drug repurposing and development

Abstract

Protein kinases play important roles in regulating signal transduction in eukaryoticcells. Due to evolutionary conserved binding sites in the catalytic domain of thekinases, most inhibitors that target these sites promiscuously inhibit multiplekinases. Quantitative analysis can reveal complex and unexpected interactions betweenprotein kinases and kinase inhibitors, providing opportunities for identifyingmulti-targeted inhibitors of specific diverse kinases for drug repurposing anddevelopment. We have developed K-Map—a novel and user-friendly web-basedprogram that systematically connects a set of query kinases to kinase inhibitorsbased on quantitative profiles of the kinase inhibitor activities. Users can useK-Map to find kinase inhibitors for a set of query kinases (obtained fromhigh-throughput ‘omics’ experiments) or to reveal new interactionsbetween kinases and kinase inhibitors for rational drug combination studies.

Availability and implementation

K-Map has been implemented in python scripting language and the website is freelyavailable at: http://tanlab.ucdenver.edu/kMap.

Introduction

Protein kinases represent one of the largest ‘druggable’ and well-studiedfamilies in the human genome [1]. This class of proteins (kinome) plays a key role as regulators andtransducers of signaling in eukaryotic cells. There is an estimated >500 members of thehuman kinome which can be classified into seven different kinase families based on theirconserved catalytic domain sequences [2]. Kinases are relatively easy to target with small molecules and have beenextensively studied at the biochemical, structural, and physiological levels. In cancercells, some kinases are mutated and acquire oncogenic properties to drive tumorgenesis.Small molecules that inhibit these oncogenic kinases can effectively kill cancer cells,as demonstrated by the success story of imatinib (Gleevec®, Novartis, Basel,Switzerland) in inhibiting the activity of BCR-ABL in chronic myelogenousleukemia [3]. Imatinib also inhibits KIT and PDGFRA, which are commonlydysregulated in gastrointestinal stromal tumors [4]. The imatinib example illustrates that small-molecule kinase inhibitorsinteract with multiple protein kinase family members (BCR-ABL, KIT,PDGFRA), and understanding these complex interactions between kinases andinhibitors could be useful for drug repurposing and development. These complexinteractions could only be revealed by systematic interrogation of the small moleculesacross a large panel of kinases using quantitative assays (kinase activity profiles).Here, we have developed K-Map—a novel and user-friendly web-based program thatsystematically connects a set of query kinases to kinase inhibitors based onquantitative profiles of the kinase inhibitor activities. K-Map is motivated by the‘connectivity map’ concept [5] where gene expression changes could be used as the ‘universallanguage’ to connect between biological systems, genes, and drugs. Instead of geneexpression signatures, we used the kinase activity profiles as the‘language’ for connecting kinases and small molecules in K-Map to reveal thecomplex interactions of kinases and inhibitors.

K-Map methods and features

Quantitative kinase inhibitor selectivity data sources

Two recently published comprehensive analyses of kinase inhibitor selectivity [6, 7] were used to construct the K-Map reference database (kinase activityprofiles). The first study systematically interrogates 178 commercially availableinhibitors against a panel of 300 protein kinases using a radiometricphospho-transfer method to assess the percent kinase inhibition (IC₅₀) [6]. The second study measures inhibitor selectivity and potency of 72inhibitors across 442 kinases using direct binding affinities between inhibitors andkinases (K_d) [7]. These kinase activity profiles were converted into rank-ordered listsaccording to their inhibitions and potencies against the kinases and used as theK-Map reference profiles for matching query kinases. For each study, the kinaseactivity profiles for individual drugs were converted into rank-ordered listsaccording to their inhibitions and potencies against the kinases. As a result, wegenerated two K-Map reference databases from these two studies: one forIC₅₀ and the other one are for K_d. Both databases will beused to connect the query kinases and return the drugs in K-Map.

Pattern matching strategy

We implemented the K-Map pattern matching strategy based on the Kolmogorov-Smirnov(KS) statistics. The KS test is a nonparametric, rank-based pattern matching approachimplemented in the connectivity map [5]. The query is a list of kinases, and the goal of the algorithm is tocorrelate kinase inhibitor that enriches the same kinases based on kinase inhibitionprofiles. For every inhibitor in the reference database, the KS statistic is computedand a connectivity score is defined.

Similar to the connectivity map approach [5], to compute the connectivity score, let N be the number ofkinases in the reference database and M be the number of query kinases. Forevery drug in the reference database, we can compute the rank-ordered list Rfor all kinases (1, 2, …, N) based on the drug inhibitions andpotencies against the kinases. For a list of query kinases of j, wherej = 1, 2, …, M, compute the following two values for eachdrug i in the reference database:

$$a=\begin{array}{c}\hfill M\hfill \\ \hfill \underset{j=1}{\max }\hfill \end{array}\left[\frac{j}{M}-\frac{R\left(j\right)}{N}\right]$$

and

$$b=\begin{array}{c}\hfill M\hfill \\ \hfill \underset{j=1}{\max }\hfill \end{array}\left[\frac{R\left(j\right)}{N}-\frac{\left(j-1\right)}{M}\right]$$

Let KSⁱ be the KS score for drug i,

$$K{S}^i=\left\{\begin{array}{c}\hfill \begin{array}{cc}\hfill a,\hfill & \hfill \mathrm{if}a>b;\hfill \end{array}\hfill \\ \hfill \begin{array}{cc}\hfill -b,\hfill & \hfill \mathrm{if}b>a.\hfill \end{array}\hfill \end{array}\right.$$

Finally, to compute the connectivity score (Sⁱ) for drugi in a reference database, let P =max(KSⁱ) and Q =min(KSⁱ),

$$S^i=\left\{\begin{array}{c}\hfill \begin{array}{cc}\hfill \frac{K{S}^i}{P},\hfill & \hfill \mathrm{if}K{S}^i>0;\hfill \end{array}\hfill \\ \hfill \begin{array}{cc}\hfill -\left(\frac{K{S}^i}{Q}\right),\hfill & \hfill \mathrm{if}K{S}^i<0.\hfill \end{array}\hfill \end{array}\right.$$

The connectivity score (Sⁱ) for every drug is reportedas the ‘Score’ in the results page. A positive score represents that theinhibitor has a similar rank order as the query kinases, indicating that theinhibitor is more specific in inhibiting the query kinases. A negative scorerepresents that the inhibitor has a reverse rank order as the query, hence notspecific in inhibiting the query kinases. Connectivity scores for each inhibitor werenormalized to yield a score ranging 0 to 1, and inhibitors were ranked based on thisnormalized score. We also computed the running sum of the connectivity score for eachinhibitor. The maximum value of the running sum is equivalent to the connectivityscore of each inhibitor. Since the query kinases are unitless, K-Map can be appliedto any technology platform.

Computing the permutation pvalue

To estimate the p value for each drug i, we perform a permutationtest by randomly selecting the number of M instances from the rank-ordereddrug i kinase profiles. Let t = 1, 2, …, T trials;the same procedure as computing the KSⁱ for drugi is performed T times and is denoted as $K{S}_t^i$. Let $K{S}_0^i$ denote the actual KSⁱ fordrug i. Count the number of times f where

$${\left[\left|K{S}_t^i\right|\ge \left|K{S}_0^i\right|\right]}_{t=1}^T$$

is true. The frequency of this event (f/T) is estimated as a(two-sided) p value. This procedure is similar to the implementation ofpermutation test by the connectivity map [5]. The p value reported in the results page of K-Map is computed by500 permutations.

Query features

K-Map implements three query functions: users can (1) directly enter query kinases inthe query text box or upload a list of query kinases (Figure 1 (A)) in the K-Map tab, (2) select kinases from the kinase family(Figure 1 (B)) in the K-Map (by family) tab, or (3)query a set of kinases involved in certain biological processes according to GeneOntology (Figure 1 (C)) in the K-Map (by GO) tab. Usersalso need to define which database they would like their query kinases to connectwith (IC₅₀ or K_d). All inhibitors available in the K-Map couldbe browsed under the Drug Info tab (Figure 1 (D)). Underthe Download tab, users can search and download kinase-inhibitor relationships. TheHelp tab provides user guide to query and navigate the K-Map. The user manual forK-Map is available at http://tanlab.ucdenver.edu/kMap.

Figure 1

Query and results of the K-Map. Left, query features. K-Map could bequeried by: (A) direct input or directly uploading a list of kinases inthe K-Map tab, (B) by selecting kinases from the kinase family, or(C) querying by Gene Ontology Biological Processes. All druginformation is available in the Drug Info tab (D) and could bedownloaded in the Download tab. Right, connectivity results. Query kinases wereconnected to the inhibitors and sorted by normalized connectivity scores.Link-out features include PubMed, PubChem, ChEMBL, and ChemSpider for drugsources and information. K-Map also provides links to pathway, biomarker, andclinical trial information for each drug via KEGG, GDSC, and ClinicalTrials.govdatabases, respectively.

Connectivity results and linking features

The output of K-Map is a rank-ordered list of inhibitors based on the normalizedconnectivity scores, accompanied by p values and running sum plots. The 2Ddrug structure is viewable by scrolling through the drug name. Kinase inhibitorspecificity within the kinase family tree is generated under KinaseTree column wherethe red circles indicate degrees of inhibition. Linking features are available fordata source of the kinase inhibition assay (via PubMed) and three major chemicaldatabases (PubChem [8], ChEMBL [9], and ChemSpider (http://www.chemspider.com)). Additional linksto drug pathway and drug biomarkers are available through the Kyoto Encyclopedia ofGenes and Genomes (KEGG) [10] and Genomics of Drug Sensitivity in Cancer (GDSC) [11] databases, respectively. K-Map also provides link-out toClinicalTrials.gov for ongoing or completed clinical trials of these inhibitors invarious diseases. We plan to update the K-Map database every quarter to keep up withthe new data and link-out information.

Implementation

K-Map is implemented in python (v2.6) and CGI script. The kinase family tree map and2D drug structure are generated by the E.T.E. software (v2.0) and Open Babel (v2.3.1) [12], respectively. The K-Map website is freely available at:http://tanlab.ucdenver.edu/kMap.

K-Map application: case study

We have recently performed a genome-wide functional genetic screen to identifysynthetic lethality genes for Nutlin-3 (p53 inhibitor) in p53 wild-type cancer celllines [13]. From this screening, we identified MET as a synthetic lethalgene with Nutlin-3 in killing cancer cell. By querying MET in the K-Mapusing K_d database (Figure 1 (A)), fourcompounds were returned with connectivity score >0.9 (Figure 1,right side). All four compounds are specific in inhibiting METwith K_d ≤ 0.025 μM (Figure 2).Interestingly, crizotinib, a recently FDA-approved ALK inhibitor is ranked#4 (p value = 0.004). As we demonstrated, treating p53 wild-type cancercells with Nutlin-3 and crizotinib inhibits proliferation and enhances cell killingin in vitro experiments [13]. This supports the finding that the K-Map could reveal new inhibition ofkinase inhibitor (Figure 1, kinase family tree indicatesthat crizotinib shows the highest selectivity in inhibiting ALK andMET).

Figure 2

Query results for MET using K _d database. These four compounds were returned withconnectivity score >0.9, indicating that these compounds were highly specificin inhibiting MET. SGX-523 is ranked #1 and is a specific METinhibitor, with K_d = 0.0019 μM as illustrated in the kinasefamily tree. PHA-665752 is ranked #2 with K_d = 0.0027 μMagainst MET. These two compounds (SGX-523 and PHA-665752) werepreclinical compounds and have not moved into clinical trials. Forentinib is anFDA-approved drug that inhibits MET and KDR (K_d =0.014 μM against MET) and is ranked #3; however, the pvalue is high (p = 0.801). Crizotinib is also an FDA-approved drugthat inhibits ALK, and this compound has K_d = 0.021 μMagainst MET (p value = 0.004). Crizotinib was validated inin vitro experiments and showed synergistic effects when combinedwith Nutlin-3 in p53wild-type cancer cell lines [13].

Summary

K-Map is a novel and user-friendly web-based tool for connecting kinases with drugsbased on quantitative profiles of the kinase inhibitor activities. Many kinaseinhibitors could promiscuously inhibit multiple kinases due to conserved sequencesimilarity among kinase family members; we have exploited these complex andunexpected interactions between kinases and inhibitors as opportunities for drugrepurposing and development. Users can use K-Map to search kinase inhibitors for aset of query kinases (obtained from high-throughput ‘omics’ experiments)or to reveal new interactions between kinases and kinase inhibitors for rationalcombination studies. In the future, we plan to extend K-Map by including more kinaseinhibitor profiles. In summary, we believe that K-Map will be a valuablebioinformatics tool in connecting altered/mutated genes identified by next-generationsequencing with therapeutics, accelerating the process of personalized medicine.