Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Dec 9.
Published in final edited form as: Methods Enzymol. 2022 Apr 8;667:403–426. doi: 10.1016/bs.mie.2022.03.040

Computational tools and resources for pseudokinase research

Brady O’Boyle 1, Safal Shrestha 2, Krzysztof Kochut 3, Patrick A Eyers 4, Natarajan Kannan 1,2,*
PMCID: PMC9733567  NIHMSID: NIHMS1854066  PMID: 35525549

Abstract

Pseudokinases regulate diverse cellular processes associated with normal cellular functions and disease. They are defined bioinformatically based on the absence of one or more catalytic residues that are required for canonical protein kinase functions. The ability to define pseudokinases based on primary sequence comparison has enabled the systematic mapping and cataloguing of pseudokinase orthologs across the tree of life. While these sequences contain critical information regarding pseudokinase evolution and functional specialization, extracting this information and generating testable hypotheses based on integrative mining of sequence and structural data requires specialized computational tools and resources. In this chapter, we review recent advances in the development and application of open-source tools and resources for pseudokinase research. Specifically, we describe the application of an interactive data analytics framework, KinView, for visualizing the patterns of conservation and variation in the catalytic domain motifs of pseudokinases and evolutionarily related canonical kinases using a consistent set of curated alignments organized based on the widely used kinome evolutionary hierarchy. We also demonstrate the application of an integrated Protein Kinase Ontology (ProKinO) and an interactive viewer, ProtVista, for mapping and analyzing primary sequence motifs and annotations in the context of 3D structures and AlphaFold2 models. We provide examples and protocols for generating testable hypotheses on pseudokinase functions both for bench biologists and advanced users.

Keywords: Evolutionary constraints, pseudoenzymes, protein structure, function, regulation, webserver, computational biology

1. Introduction

Protein phosphorylation is one of the most widespread mechanisms for regulating cellular functions in both eukaryotes and prokaryotes. The enzymes that catalyze protein phosphorylation, protein kinases, are ubiquitous across the tree of life and constitute nearly 2% of most eukaryotic genomes. Protein kinases are also prevalent in prokaryotes, viruses, and pathogens (Kannan, Taylor, Zhai, Venter, & Manning, 2007; Kwon et al., 2019; Manning et al., 2011; Olson, Wang, Rico, & Wiebe, 2019; Talevich & Kannan, 2013; Wang, Zhao, Savas, Zhang, & Feng, 2020) and are defined at the sequence level based on the presence of key hallmark motifs that define three-dimensional structure and catalytic function (Hanks & Hunter, 1995; Knighton et al., 1991; Manning, Whyte, Martinez, Hunter, & Sudarsanam, 2002). These hallmark motifs include: the glycine rich loop, the ATP-binding β3-Lys, the catalytic aspartate and the cation-coordinating asparagine within the catalytic HRDXXXN loop motif, and the metal binding aspartate of the DFG motif at the beginning of the activation loop (Eyers & Murphy, 2013).

Pseudokinases are bioinformatically defined based on the absence of one or more of the canonical active site residues (β3-Lys, HRD-Asp, DFG-Asp) (Figure 1A), and the ability to define pseudokinases based on primary sequence comparison has enabled their systematic mapping across the tree of life (Kwon et al., 2019). The original cataloguing of the human kinome in 2002 revealed that about 10% of human kinases are pseudokinases (Manning, Whyte, Martinez, Hunter, & Sudarsanam, 2002). The pseudokinases identified within the human genome were not confined to a single kinase family, but rather spread across all major kinase groups. Recent efforts to identify and annotate pseudokinases using profile-based methods have revealed not only the extent of the prevalence of pseudokinases but also taxonomic specific expansions across the tree of life. A survey of the more than 10,000 reference proteomes from the tree of life suggested that around 10% of the protein kinase (PK) members in each genome are pseudokinases (Kwon et al., 2019). Moreover, the study also identified several family-specific expansions such as interleukin-1 receptor–associated kinase (IRAK)–like pseudokinases in plants and tyrosine kinase–like (TKL) pseudokinases in fungi. Within prokaryote kinomes, pseudokinases are not ubiquitously present as seen in eukaryotes, however, a major expansion of M. tuberculosis protein kinase B (pknB)-like pseudokinases was identified within many bacteria. In terms of pathogens, an expansion of pseudokinases was revealed within eukaryotic pathogens, including Plasmodium falciparum (Kwon et al., 2019; Manning et al., 2011) and Giardia lamblia (Kwon et al., 2019; Ward, Equinet, Packer, & Doerig, 2004). While there have not been any substantial efforts into studying the pseudokinomes of viruses, the presence of viral pseudokinases was established with the discovery of the B12 pseudokinase in vaccinia virus (Olson, Wang, Rico, & Wiebe, 2019; Petrie et al, 2019; Wang, Zhao, Savas, Zhang, & Feng, 2020). Indeed, the widespread presence of pseudokinases across the tree of life suggests multiple independent emergences of pseudokinases throughout kinase evolution.

Figure 1. Pseudokinases within the human kinome.

Figure 1.

Open in a new tab

(A) Key catalytic motifs in the Protein Kinase (PK) fold enzymes. (Top) Crystal structure of the prototypical protein kinase PKA (PDB ID: 1ATP; Zheng et al., 1993) is shown as cartoon and colored in gray. (Bottom) Zoomed in view of the catalytic pocket with β3-Lys (K72), catalytic HRD-Asp (D166), and metal binding DFG-Asp (D184) shown as sticks and colored in magenta. Manganese ions (Mn) are shown as spheres and colored in teal. ATP is shown as sticks and colored in black. (B) Pseudokinases are present within the human kinome. Pseudokinases present within the human kinome are shown in cyan. Dark pseudokinases are highlighted with a red circle (Berginski et al, 2020) The tree figure was generated using Coral (Metz et al., 2018)

Identification of pseudokinase orthologs provides a valuable resource for understanding pseudokinase evolution; however, a common issue during ortholog identification is the level of sequence divergence that is present when working with large sequence datasets. Sequence dissimilarities because of insertions, deletions, and other sequence restructuring events can make identification and classification of pseudokinase orthologs difficult without curated alignments (Neuwald, 2009). However, alignment curation is a difficult, time-consuming, and error-prone process that can result in inconsistencies in ortholog classification and subsequently inaccurate comparative and structural analyses. More recently, a human kinome centric ortholog detection method termed KinOrtho (Huang et al., 2021) was developed, which employs a graph-clustering approach using both primary sequence similarity and domain organization to help define human kinase orthologs across fully-sequenced proteomes. Recent application of this method across 17,000 proteomes identified 75 million overlapping kinase orthologous relationships. In comparison, other orthology inference methods only identified 35–60% of the orthologs defined by KinOrtho. KinOrtho’s ability to identify weakly-related orthologs makes it ideal for studying pseudokinases, particularly the many understudied ‘dark’ pseudokinases (Figure 1B), where orthologous relationships from other organisms can be used as features in machine learning models for function prediction (Huang et al., 2021).

Integrative mining of the evolutionary information encoded in primary sequences with structural and functional data has provided new insights into the allosteric and non-catalytic functions of various pseudokinases (Eyers, Keeshan, & Kannan, 2017; Foulkes et al, 2018; Kwon et al., 2019; Murphy et al, 2015; Preuss et al., 2020; Shrestha, Byrne, Harris, Kannan, & Eyers, 2020). This approach was used to identify functionally constrained residues within the activation loop of the TRIB pseudokinase family as well as constraints within the MEK1 and COP1 binding sites of the C-tail (Eyers, Keeshan, & Kannan, 2017). Mapping of these constraints to the TRIB1 crystal structure (Jamieson et al, 2018; Murphy et al, 2015) revealed potential allosteric coupling that drives conformational changes regulating substrate binding to the C-lobe (Eyers, Keeshan, & Kannan, 2017). Consistently, C-tail tethering to the pseudokinase domain renders the protein incapable of substrate binding (Eyers, 2015; Murphy et al, 2015). Additionally, functional constraints within a leucine-rich repeat receptor-like plant pseudokinase family, termed LRRVI-2, and a receptor-like cytoplasmic kinase family, termed RLCKXII-1, were mapped to Zea mays LRRVI-2 (PDB ID: 6CPY; Aquino et al, 2018) and BSK8 (PDB ID: 4I92; Grütter, Sreeramulu, Sessa, & Rauh, 2013) crystal structures, respectively (Kwon et al., 2019). Integration of constraints and three-dimensional (3D) structure revealed surface exposed residues in LRRVI-2 involved in the unique tethering of the C-terminal tail to the kinase domain. Likewise, RLCKXII-1 specific constraints stabilize a unique inactive conformation involving β3-αC loop and the αC-helix glutamate residue.

Several new tools are available for the broader scientific community to access, visualize, and mine the evolutionary, structural, and functional information on pseudokinases. KinOrtho defined orthologs and their corresponding alignments for the core kinase domains are available through KinView (McSkimming et al., 2016), a sequence logo-based visualization tool (Crooks, Hon, Chandonia, & Brenner, 2004; Schneider & Stephens, 1990). In addition, UniProtKB, a comprehensive database for protein sequences and their functional annotations (Zaru, Magrane, Orchard, & Consortium, 2020), in conjunction with Mechanism and Catalytic Site Atlas (M-CSA), a database for enzyme active site and mechanism (Ribeiro et al, 2018), can be used as an orthogonal approach to identify enzymatically inactive kinase orthologs (Ribeiro, Tyzack, Borkakoti, & Thornton, 2020). These resources can serve as a conceptual starting point for generating new testable hypotheses on unique catalytic and non-catalytic mechanisms through integration of sequence data with 3D models and structures. Recent advancements in deep learning-based protein structure prediction, namely AlphaFold2 (Jumper et al., 2021), have generated a vast number of structural models for full-length kinases and pseudokinases, which are available for open-access analysis (https://alphafold.ebi.ac.uk/). Valuable information from these models can only be extracted by mapping important sequence, structural, and functional annotations to 3D models. ProKinO (Gosal, Kochut, & Kannan, 2011; McSkimming et al., 2015), a protein kinase-specific ontology framework, was previously developed to unify various protein kinase annotations, including data on kinase mutations and pathways. Recently, ProtVista viewer (Watkins, Garcia, Pundir, Martin & Consortium, 2017), a graphical representation to interactively map sequence annotations on 3D structures, has been integrated into the ProKinO browser. In this chapter, we describe these computational tools (KinView and ProtVista) and describe their contribution to previously published pseudokinase research pertaining to two understudied pseudokinases, Unc-51-Like Kinase 4 (ULK4) and Protein Serine Kinase H2 (PSKH2). In addition, we also provide step-by-step instruction to use these to study and generate testable hypotheses for the dark pseudokinase, Vaccinia Related Kinase 3 (VRK3). Pseudokinases regulate diverse cellular processes associated with normal cellular functions and disease. They are defined bioinformatically based on the absence of one or more catalytic residues that are required for canonical protein kinase functions. The ability to define pseudokinases based on primary sequence comparison has enabled the systematic mapping and cataloging of pseudokinase orthologs across the tree of life. While these sequences contain critical information regarding pseudokinase evolution and functional specialization, extracting this information and generating testable hypotheses based on integrative mining of sequence and structural data requires specialized computational tools and resources. In this chapter, we review recent advances in the development and application of open-source tools and resources for pseudokinase research. Specifically, we describe the application of an interactive data analytics framework, KinView, for visualizing the patterns of conservation and variation in the catalytic domain motifs of pseudokinases and evolutionarily related canonical kinases using a consistent set of curated alignments organized based on the widely used kinome evolutionary hierarchy. We also demonstrate the application of an integrated Protein Kinase Ontology (ProKinO) and an interactive viewer, ProtVista, for mapping and analyzing primary sequence motifs and annotations in the context of 3D structures and AlphaFold2 models. We provide examples and protocols for generating testable hypotheses on pseudokinase functions both for bench biologists and advanced users.

2. Investigating pseudokinase functional specialization using comparative kinomics

2.1. KinView

Accurate multiple sequence alignments (MSAs) of protein kinase sequences from diverse organisms are valuable resources for investigating the relationships connecting pseudokinase sequence, structure, function, and regulation (Eyers, Keeshan, & Kannan, 2017; Foulkes et al., 2018; Jamieson et al., 2018; Preuss et al., 2020). MSAs can be used to identify functionally constrained residues in pseudokinases of interest, especially when compared to an evolutionarily similar protein, such as an active canonical counterpart (Chatzou et al., 2016; Morrison, 2006). However, taking a manual approach to identify constrained residues in an alignment is an error-prone, time-consuming endeavor. Furthermore, inconsistencies in MSAs based on the alignment method of choice, input sequences, and parameters pose additional challenges in data interpretation, reproducibility, and sharing (Pervez et al., 2014). Therefore, KinView was developed to resolve such issues and prevent inconsistencies. KinView is a visual comparative sequence analysis tool that provides sequence logo-based graphical representations of large multiple sequence alignments for protein kinases (McSkimming et al., 2016). It enables comparative analysis and visualization of the patterns of conservation and variation at each aligned position in the kinase domain using sequence logos, which provide a graphical representation of the amino acid frequencies at each aligned position (Figure 2A) (Capra & Singh, 2007; Kullback & Leibler, 1951). For more details on KinView features, see Protocol section 4.1.

Figure 2. Comparative kinomics using KinView identifies key amino acid differences in several pseudokinases.

Figure 2.

Open in a new tab

(A) Snapshot of sequence logos of key catalytic motifs in various dark pseudokinases in KinView (https://prokino.uga.edu/kinview/). Relative entropy (RE) was calculated using a BLOSUM62 as the background and matrix with a window size of 3 and sequence weighting enabled (Capra & Singh, 2007). Kinase domain sequence alignments were obtained from KinOrtho. PSKH2 mutations of the DFG motif found in cancer and their frequency are shown. (B) Presence of pseudo/canonical ULK4 orthologs mapped to SwissTree species tree (https://swisstree.sib.swiss/cgi-bin/swisst?page=species_tree). Key catalytic residue losses are shown as unfilled shapes. The lengths of the activation loop as compared to the core kinase domain are shown as bar charts and colored in green. (C) PSKH2 orthologs mapped to the SwissTree species tree. Absence and presence of PSKH2 orthologs are colored gray and blue, respectively.

An example showing the sequence logos for the ATP-binding motif (β3-Lys), the catalytic motif (HRD-Asp), and the metal binding motif (DFG-Asp) in an alignment of 95591 annotated protein kinases (PK) is shown in Figure 2A. This PK sequence logo can serve as a reference for investigating pseudokinase divergence at catalytic positions. Sequence logos for some of the dark pseudokinases (ULK4, VRK3, and PSKH2) are shown next to the PK sequence logo (Figure 2A). Differences in the distribution of different amino-acid residues in pseudokinases are immediately apparent. For ULK4 (Eyers, 2020), the β3-Lys and DFG-Asp are less constrained, whereas the HRD-Asp remains highly conserved in ULK4 across diverse organisms (Figure 2A, B). Although most of the chordate ULK4 (including humans) are missing canonical residues at β3-Lys and DFG-Asp positions, all ULK4 orthologs in plants possess all three canonical residues (Figure 2B). Moreover, the increase in the length of the activation loop seems to correlate with the emergence of ULK4 pseudokinases in higher organisms (Figure 2B). Despite the lack of β3-Lys (L33) and DFG-Asp (N139), human ULK4 has been convincingly shown to bind ATP in a metal-independent manner (Khamrui, Ung, Secor, Schlessinger, & Lazarus, 2019; Murphy et al., 2014; Preuss et al., 2020). The sequence basis for these adaptations can be further probed by using the advanced statistical methods described below (Section 4.3) (Preuss et al., 2020).

In the case of PSKH2, an asparagine is observed at the catalytic aspartate (HRD-Asp) position (Figure 2A). This substitution is selectively observed in primates (N) (Figure 2C) (Reiterer, Eyers, & Farhan, 2014; Shrestha, Byrne, Harris, Kannan, & Eyers, 2020), suggesting primate-specific emergence of PSKH2 pseudokinases. Another feature of KinView is a visual mapping of cancer mutations and post-translational modifications (PTMs). Interestingly, in a few sequenced cancer samples, the DFG-Asp is substituted to an asparagine (N) (Figure 2A), although whether this a driver or bystander mutation requires further investigation. The presence of asparagine in place of DFG-Asp in pseudokinases nearly always abolishes cation binding (Murphy et al., 2014), and consequently, experimentally mutating DFG-Asp to an asparagine in canonical kinases also prevents metal-ion binding (Haydon et al, 2003; Tsuchiya et al., 2020) generating questions as to the specific nature of the cancer variant and whether it modifies the ability of PSKH2 to bind to metal ions.

In VRK3 pseudokinases, the metal binding aspartate (DFG-Asp) is replaced by a glycine (G) whereas the catalytic aspartate (HRD-Asp) is present but only partially conserved (Figure 2A). In humans and a subset of VRK3 orthologs, the catalytic aspartate is substituted to an asparagine (N) (Figure 2A), resulting in a degraded nucleotide binding site. This suggested that VRK3 is incapable of binding and hydrolyzing ATP in a metal dependent manner, which was supported by biochemical and biophysical studies (Murphy et al., 2014; Nichols & Traktman, 2004; Scheeff, Eswaran, Bunkoczi, Knapp, & Manning, 2009). However, more recently, recombinant GST-VRK3 or FLAG-VRK3 were reported to phosphorylate the nuclear envelope protein barrier-to-autointegration factor (BAF) in vitro (Park et al, 2015). The N-terminal flanking region of VRK3, which was absent in recombinant VRK3 when the kinase domain was crystallized in an inactive form containing an occluded ATP site, was required for phosphorylation of BAF. Thus, a potential for a compensatory catalytic residue and an undescribed flanking region interaction with the kinase domain warrants more thorough investigation in VRK3 and many other pseudokinases. In this context, it is worth noting that purified VRK3 can be destabilized by non-specific kinase inhibitors such as DAP and VI16832 in vitro (Murphy et al., 2014), suggesting atypical effects of small molecules that might involve the highly degraded ‘nucleotide-binding’ site.

2.2. Bayesian Partitioning with Pattern Selection (BPPS)

While active site variations can be visually examined using KinView, identifying allosteric networks that couple the active site to distal regulatory sites requires more advanced statistical analysis of sequence datasets. Bayesian Partitioning with Pattern Selection (BPPS) is one such approach that has been successfully employed to investigate pseudokinase functional specialization (Eyers, Keeshan, & Kannan, 2017; Kwon et al., 2019; Preuss et al., 2020; Shrestha, Byrne, Harris, Kannan, & Eyers, 2020). Conceptually, BPPS clusters aligned sequences into functionally divergent subgroups based on correlated residue patterns that are highly conserved within subgroup sequences but divergent in sequences outside of the subgroup (Neuwald, 2011; Neuwald, 2014). The pattern-based classification method identified a conserved surface patch in the N-terminal ATP-binding lobe of ULK4 pseudokinase that distinguishes it from other canonical ULK members (Preuss et al, 2020). Likewise, BPPS analysis revealed a subset of co-conserved residues distinguishing the pseudokinase PSKH2 (His 291, Leu303, His307, and Asp316) from closely related canonical PSKH1. These co-conserved residues map to the substrate-binding C-lobe of PSKH2 kinase domain model, presumably conferring specificity in protein-protein interactions (Shrestha, Byrne, Harris, Kannan, & Eyers, 2020). For a step-by-step protocol to perform constraint-based analysis using BPPS, see Protocol section 4.3.

3. Structural informatics and visualization tools for pseudokinase research

3D structures of proteins are critical for investigating the unique non-catalytic functions of pseudokinases. Indeed, analyzing the active site variations and co-variations considering available crystal structures, homology models, and molecular dynamics (MD) simulations have provided unique insights into non-catalytic functions of various pseudokinases (Eyers, Keeshan, & Kannan, 2017; Foulkes et al., 2018; Jamieson et al., 2018; Kwon et al., 2019; Preuss et al., 2020; Shrestha, Byrne, Harris, Kannan, & Eyers, 2020; Talevich & Kannan, 2013). However, the lack of structural modelling data for the vast majority of pseudokinases has hindered functional insights. Recent advancement in the field of protein structure prediction using Artificial Intelligence (AI) is now partially alleviating this issue. AlphaFold2 (Jumper et al., 2021) is an AI-based protein structure prediction tool developed by DeepMind. Based on Critical Assessment of protein Structure Prediction 14 (CASP14) measures, AlphaFold2 can predict protein structures to near experimental accuracy (median backbone accuracy of 0.96 Å r.m.s.d. as compared to 2.8 Å r.m.s.d. for the next best performing method). The method employs novel neural network based on evolutionary, physical, and geometric constraints. It takes the protein sequence and multiple sequence alignment of related sequences as inputs and directly predicts the 3D structure. Through a collaboration between DeepMind and EMBL-EBI, structures predicted by AlphaFold2 for the human proteome and 20 other model organisms have been made available online (https://alphafold.ebi.ac.uk/).

The Protein Kinase Ontology (ProKinO) is an integrated knowledge graph capturing the relationships connecting sequence, structure, function, mutation, and pathways on protein kinases. We have recently integrated the ProtVista viewer within ProKinO to provide 3D context for sequence annotations. We demonstrate the utility of this viewer in analyzing 3D structures and predicted structural models of pseudokinases using human ULK4 full length AlphaFold2 model (UniProt ID: Q96C45) as an example (Figure 3A). ULK4 lacks the conserved salt bridge between the lysine (K72PKA) in the β3 strand and glutamate (E92PKA) in the αC-helix. Identifying these variations in the active site typically requires structural and sequence alignment of ULK4 with canonical protein kinase such as Protein Kinase A (PKA). However, using the ProtVista viewer, such variations can be rapidly identified simply by selecting the residues annotated as “KE Salt bridge” under the “Structural Motifs” tab in the sequence viewer. Selection of these residues highlights L33 and W46 in the ULK4 structural model, providing insights into ULK4’s unique mode of ATP binding. Similarly, the DFG-Asp is substituted to an asparagine (N139) (Figure 3B) in ULK4, visualized in the context of 3D models or crystal structures (PDB ID: 6U5L (Khamrui, Ung, Secor, Schlessinger, & Lazarus, 2020) and 6TSZ (Preuss et al., 2020)) co-crystallized with an inhibitor and ATP-analog (AGS). The phosphates of the AGS molecules interact with positively charged residues such as K39 and R125 (Figure 3B). Moreover, L128, which is annotated as catalytic spine 6 (CS6) residue, forms van der Waals interaction with the adenine ring. This interactive mining of the sequence motifs and annotations in the context of 3D models allows one to further explore the binding modes. The ProtVista viewer can also be used to define domain boundaries of the catalytic domain and visualize interactions between the kinase domain and flanking regulatory segments, as predicted by AlphaFold2 models. This is illustrated for the pseudokinase PSKH2, showing predicted interactions between the kinase domain and flanking N- (1–62) and C- (321–385) terminal flanking regions (shown in green and orange respectively) (Figure 3C). For detailed instruction on how to perform similar analysis, refer to Protocol section 4.2.

Figure 3. Integration of AlphaFold2 models and ProtVista viewer in ProKinO for pseudokinase research.

Figure 3.

Open in a new tab

(A) Snapshot of the ProKinO page for human ULK4 (UniProt ID: Q96C45). Various sequence and structural annotations can be accessed and easily mapped to the AlphaFold2 model or crystal structures in the ProKinO browser. (B) Cartoon representation of the kinase domain of human ULK4 co-crystallized with AGS, an ATP analog (PDB ID: 6TSZ; Preuss et al, 2020), as shown using the ProtVista viewer. (Right) Zoomed in view of the interactions made by AGS with the kinase domain. (C) Multiple AlphaFold2 predicted models of full length human PSKH2. The models are shown as cartoons with N- (1–62) and C- (321–385) terminal regions colored in green and orange, respectively. PyMOL version 2.3.2 was used to generate the figure.

4. Protocols

4.1. Comparative kinomics of VRK family members using KinView (Figure 4)

Figure 4. The KinView web interface.

Figure 4.

Open in a new tab

(A) The ProKinO Browser. Buttons to go to the online version or to download the offline version of KinView are indicated by numbers 1 and 2, respectively. (B) Comparison of VRK paralogs on KinView. The kinase family of interests can be toggled on/off using the selection on the left. The multiple sequence alignments for VRK1–3 are represented as sequence logos. The sequence and structural annotations are at the top. HRD and DFG motifs are highlighted with Protein Kinase (PK) as reference.

1.1 Open a web browser (Chrome, Firefox, Safari, Edge) on a computer connected to the internet.

1.2 Open the URL https://prokino.uga.edu/. KinView can be accessed through the ProKinO browser (Figure 4A). To access the KinView app on the web, click “Launch” button under KinView (1). For convenience, users can download an offline version by clicking on “Launch” button under “Downloads.” KinView can be installed on any standard operating system (Linux, Mac, Windows 10) and requires an installation of “node js” (https://nodejs.org/en/). For offline use, it is recommended to have at least 16 GB of RAM and 10 GB of free disk space.

(2). Alternatively, KinView can be accessed through this link: https://prokino.uga.edu/kinview/.

1.3 Once the KinView app is launched, toggle selection for the kinase families of interest by clicking on the boxes near the names (Figure 4B). Here, pseudokinase VRK3 as well as its paralogs, VRK1 and VRK2, are selected (3). Protein Kinase (PK) has been selected as reference (4). The search bar can be used to directly select the family of interest (5).

1.4 Turn on the kinase-specific sequence motifs by clicking on the “Motif” button under “Settings” (6). The motifs such as HRD, DFG, APE and so on are shown in graphical format (7). To get to these C-lobe motifs, the users will need to scroll to the right using the scroll bar at the bottom of the page.

1.5 Turn on the secondary structure representation by clicking on the “Domain Structure” button under “Settings” (8). The secondary structures, such as β6–9 and αF, are shown in graphical format (9).

1.6 Sequence logos for the selected families (3,4) are shown in the sequence logo section (Figure 4B). By default, the “Residue” option is turned on (10). Turning off the “Residue” option will hide the sequence logos.

1.7 Post Translational Modifications (PTM) type and counts can be toggled on using the “PTM” button (11). Cancer mutations in the form of a sequence logo (12) or mutation counts grouped by amino acid properties can also be toggled on to aid in visualizing documented mutations (13).

1.8 The residue numbering below the sequence logos for each family can be changed by selecting the options under “Reference Position” (14). Alignment numbering can be set to “VRK3”, which is the residue numbering within VRK3, or to “PKA”, which changes the residue numbering in relation to the alignment numbering of PKA. Setting the numbering to “PKA” allows for position specific comparisons across kinases. In this example, VRK3 numbering is used for the VRK3 sequence logo.

1.9 To download the alignment in the FASTA format, click on the floppy disk icon (15). Cancer mutations and PTM data are also downloadable options (15). Kinase domain or full-length domains can be downloaded by clicking on the button next to “Ortholog sequences” (16).

1.10 The sequence logos can then be used to compare key sequence motifs. HRD and DFG motifs are highlighted (Figure 4B, Red Boxes). Only a subset of VRK3 members conserve the HRD-Asp. In some VRK3 sequences it is replaced by an asparagine (N) (17), whereas VRK1 and VRK2 sequences strictly conserve an Asp at this position (17). Another key catalytic residue, the DFG-Asp is a strictly a glycine (G, Gly) in VRK3 sequences (18). Moreover, the DFG-Phe is strictly conserved as a tyrosine (Y, Tyr) residue (19). One can see that the same change is seen in a subset of VRK1 sequences and all VRK2 sequences.

4.2. Structural insights into VRK3 active site using ProtVista viewer in ProKinO browser (Figure 5)

Figure 5. The ProtVista Viewer on ProKinO.

Figure 5.

Open in a new tab

(A) Home page of the ProKinO Browser (B) ProKinO page for the Human_VRK3 showing the Structure section. (C) ProtVista viewer view of the AlphaFold2 model for human VRK3 full length protein. The different annotations are also shown. (D) Snapshot of the human VRK3 AlphaFold2 model. Kinase domain is colored green with N-terminal region colored in gray. The residues are shown as Ball & Stick with carbon atoms colored in gray.

2.1 Open a web browser (Chrome, Firefox, Safari, Edge) on a computer connected to the internet.

2.2 Open the URL https://prokino.uga.edu/browser/. ProKinO homepage with the search bar is shown in Figure 5A.

2.3 Type “VRK3” in the search bar (1) as indicated Figure 5A to search for VRK3. You will see multiple results. Click on the first link “prokino: Human_VRK3”.

2.4 You will get to the ProKinO page for Human VRK3 (UniProt Id: Q8IV63). Here, you will find multiple information about the protein such as cellular localization, functional domains, sequence, disease annotations, and so on. You can access both crystal structures and AlphaFold2 models under “Structure” (2) (Figure 5B). For this example, click on the AlphaFold2 model “AlphaFold: AF_Q8IV63_F1_model-v1.cif, uniprot: Q8IV63, organism: Homo Sapiens (Human)”.

2.5 The ProtVista viewer is loaded, and you will see the AlphaFold2 model of full length human VRK3 (3) (Figure 5C).

2.6 The sequence view is shown (4). Human VRK3 is 474 residues long. You can zoom to residue level resolution by hovering over it and using the scroll wheel on the mouse.

2.7 Under the sequence view, you will find “Model Confidence” (5). It is an AlphaFold2 calculated per-residue confidence score (pLDDT). Blue represents “Very High” confidence while orange represents “Very Low” confidence.

2.8 To map the KE salt bridge, go to “Structural Motifs” (6). The residues in the salt bridge can then be directly selected and shown as Ball & Stick representation by clicking on the vertical bars (7).

2.9 To map sequence motifs to the AlphaFold2 model, go to “Sequence Motifs” (8). The motifs DFG, HRD, αC-Glu, and β3-Lys can be viewed in the structure by clicking on the vertical bars 9–12 respectively.

2.10 To make publication quality figures (Figure 5D), use the different ProtVista viewer tools (13–18). The functions of the tools are as follows:

  • 13: Reset Camera

  • 14: Screenshot: HD (1280 × 720), Full HD (1920×1080), Ultra HD (3840 × 2160) and custom resolutions available for download. Ultra HD resolution was used for snapshots in Figure 5D.

  • 15. Structure Tools: You can make measurements as well as create new selections and representations.

  • 16. Expanded Viewport: Allows user to toggle on/off full screen mode.

  • 17. Settings: You can change the background color, lighting, clipping and so on.

  • 18. Selection: By toggling on this tool, you can select residues of interest, create selection objects, and color them.

2.11 The snapshots in Figure 5D were generated using ProtVista tools. The selection tool (18) was used to select the N-terminal region colored in gray (Figure 5D, Top).

2.12 Using the selection tool (18) and the annotations (9–12), the key catalytic residues as well as αC-Glu were mapped and shown as Ball & Stick representation (Figure 5D, Bottom Left, Red box).

2.13 The viewer also provides an interaction network for residues within 5 Å of the selected residue. N306 (HRD-Asp equivalent residue in human VRK3) is selected and interactions with residues within 5 Å are shown (Figure 5D, Bottom Right, Blue box). The side-chain carbonyl of N306 forms a hydrogen bond with D353 backbone amide.

4.3. Comparative kinomics of VRK family members using BPPS (Figure 6)

Figure 6. Mammalian VRK3-specific sequence constraints.

Figure 6.

Open in a new tab

(A) Sequence constraints in the glycine-rich loop and catalytic loop of mammalian VRK3 sequences (foreground) when compared to the Set 3 sequences (background). Set 3 contains VRK1, VRK2 and VRK3 sequences that are clustered together based on shared constraints identified by BPPS. The alignment shows mammalian VRK3 sequences from diverse organisms. The histogram in red above the residue indicates the extent to which the foreground alignment diverges from the corresponding position in the background alignment. Black dots mark the alignment positions used by the BPPS procedure when distinguishing the mammalian VRK3 sequences from Set 3 sequences. (B) AlphaFold2 model for full length human VRK3. The glycine-rich loop and the catalytic loop are colored in magenta and cyan, respectively. The N-terminal region is colored darker gray as compared to the kinase domain. (C) Zoomed in view showing the sequence constraints identified in (A). The residues are shown as sticks and colored orange. glycine-rich loop and the catalytic loop are colored as in (B). A co-conserved aspartate within the activation loop, D353, is also shown.

3.1 Open a web browser (Chrome, Firefox, Safari, Edge) on a computer connected to the internet. A Linux-based operating system, such as RedHat or Ubuntu, is recommended when running BPPS and no non-standard hardware is required.

3.2 Navigate to https://www.igs.umaryland.edu/labs/neuwald/software/bpps/ and left click “bpps1.1.6.tar” to download. Move the downloaded file, bpps1.1.6.tar.gz, to the desired folder. Extract the contents of the bpps1.1.6.tar.gz file.

3.3 Download the VRK kinase domain ortholog sequences from KinView using the protocol described above (Section 4.1). As well, download CSNK1A1 kinase domain ortholog sequences from KinView to serve as an outgroup. Merge all the FASTA files together so that the sequences are all in FASTA format and contained within a single file.

cat *.fasta >> vrk_kd_unaligned.fasta

3.4 Align the sequences using MUSCLE (https://github.com/rcedgar/muscle/releases/tag/v5.0.1428) to generate FASTA formatted alignments of ortholog sequences: 

muscle -in vrk_kd_unaligned.fasta -out vrk_kd_aligned.fasta

3.5 The aligned FASTA file must be converted to a CMA file for BPPS input. Download ConvertMSA from https://www.igs.umaryland.edu/labs/neuwald/software/auxiliary/. Unzip the file to and run the following line to convert the FASTA to CMA format: 

ConvertMSA fa2cma vrk_kd_aligned.fasta > vrk_kd.cma

3.6 Open command prompt or terminal and navigate to the folder containing the BPPS executable file and aligned sequences in CMA format. Run BPPS 1 using the command below. Note that the file extension is not included when running BPPS, just the file name. Output files will be .cmd, .chk, _user.sma, _bst.out, and _aln.rtf.

bpps 1 vrk_kd [options]

3.7 Run BPPS 2 using the output of BPPS 1 by using the file name (but not the extension) as the input for BPPS 2. Output files will be _himsa.cma, _himsa.tpl, _himsa.sets, _himsa.dft, _himsa.hpt, and _himsa.ph.

bpps 2 vrk_kd [options]

3.8 BPPS 3 can be used to generate a PyMOL (Schrödinger, 2019) session file that will overlay constrained residue patterns onto a 3D structure. A pattern set must first be selected from the vrk_kd_himsa.sets file. The set that contains the human VRK3 sequence will be used to generate a PyMOL session file. The set of interest can be identified by searching the vrk_kd_himsa.sets file for the human VRK3 uniProt ID (Q8IV63). First run the following command to separate the sets file into searchable files:

tweakcma vrk_kd_himsa -write

3.9 Several vrk_kd_himsa_Set files will have been generated. Run the following command to locate the set that the human VRK3 sequence belongs to: 

grep ‘Q8IV63’ *Set*

3.10 The human VRK3 sequence is present in Sets 1, 3, and 7. The number associated with the pattern will be different for different runs, but the patterns identified will generally remain consistent. Set 1 consists mainly of VRK sequences whereas Set 3 consists of VRK3 sequences. Set 7 is a subgroup of Set 3 and contains only mammalian VRK3 sequences.

3.11 Open vrk_kd_himsa.hpt file. This file contains hierarchal grouping of the sequences, with each group being defined by the constrained residues that are shared by a set of sequences. Identify the hierarchical grouping number associated with Set 7, in this case it is 9. This grouping number will be used to run BPPS 3.

3.12 Search online for an rtf-pdf converter to turn vrk_kd_aln.rtf to a pdf (https://online2pdf.com/convert-rtf-to-pdf is free and easy-to-use). This file indicates the constrained residue patterns for each hierarchal group in comparison with the residues present at the corresponding positions of the parent hierarchal group (Figure 6A). It also includes the frequency of amino acids in the foreground and the background of each group. The statistical significance of a residue within a pattern is indicated by a red bar above the corresponding position. A larger bar indicates a residue is more highly conserved within the foreground when compared to the residues at the same position within the background sequences.

3.13 Download a 3D structure of human VRK3. This can be obtained from AlphaFold2 (https://alphafold.ebi.ac.uk/) by searching for “VRK3”, selecting the “Homo sapiens” protein and clicking the “PDB File” button. If a crystal structure is available, then it can be obtained from RCSB (https://www.rcsb.org/) by searching “VRK3,” selecting the VRK3 structure, clicking “Download Files” and then selecting PDB format. Save the structure as VRK3_H.pdb.

3.14 Make a text file that indicates the path to the pdb file and save it as path_to_pdb.txt. The path to a file can generally be obtained by right-clicking on the file and opening the file properties. This can then be copied and pasted in the text file and should appear as shown below: 

this/is/the/path/to/your/file/path_to_pdb.txt

3.15 Generate a PyMOL session by running BPPS 3 using the following command: 

bpps 3 vrk_kd_himsa 9 -pdb=path_to_pdb.txt -pml

3.16 Download PyMOL (https://pymol.org/2/) and open it on the computer. Click “File” at the top left, select “Open” and select vrk_kd_himsa_6_1_1.pml. This will open the 3D structure file that was provided in BPPS 3 (Figure 6B) and annotate the constrained residue patterns from Set 7, as well as the patterns from the parent hierarchal groups.

3.17 On the right side of the PyMOL window, there are classes that correspond to each group’s constrained residue patterns. Class Y contains the residue patterns of Set 1, Class R contains the residue patterns of Set 3 and Class O contains the residue patterns of Set 7. Patterns can be hidden or shown by clicking the class name on the right side of the screen.

3.18 Side chain interactions can be predicted using the tools available in PyMOL. Some useful plugins available for PyMOL are: APBS for electrostatic surface visualization (Baker, Sept, Joseph, Holst, & McCammon, 2001), Autodock for protein-ligand docking (Trott & Olson, 2010), GROMACS for molecular dynamics simulation (Makarewicz & Kaźmierkiewicz, 2013) and PyMod for homology modeling (Janson & Paiardini, 2021). Distance between atoms and polar/hydrophobic contacts can also be determined using tools available on the right side of the PyMOL window. Residue side chains of the G-loop and those of the catalytic loop are shown as reference in Figure 6C.

5. Summary

Advances in open-source computational tools for pseudokinase analysis allow for tracing of robust evolutionary histories and are paving the way for a more thorough understanding of their structural adaptations. KinOrtho’s ability to accurately identify and classify pseudokinase orthologs provides a means to utilize the vast wealth of exponentially increasing sequence data. With these orthologs, tools such as KinView and BPPS can identify constrained residues through comparative sequence analysis. The conservation of these residues will provide the foundation for generating and subsequently testing new hypotheses on pseudokinase functional specialization. Tools such as AlphaFold2 and ProKinO can then be utilized to integrate these evolutionary data with structural models and provide context for the conserved residue patterns. While these tools are valuable resources for the pseudokinase community, additional work is needed to further enhance the utility of these tools. In particular, families that are selectively expanded in plants, prokaryotes, fungi and viruses need to be systematically classified and incorporated in the protein kinase ontology and kinase evolutionary hierarchy to enable evolutionary comparisons. Likewise, models of protein-protein interactions will significantly enhance the value of ProtVista in pseudokinase function prediction and annotation.

Acknowledgements

We acknowledge members of the Kannan lab for valuable comments and suggestions. Funding from NIH (R35GM139656 and U01CA239106) is acknowledged.

References:

  1. Aquino B, Counago RM, Godoi PHC, Massirer KB, Elkins JM, Arruda P, et al. (2018). Structure of apo GRMZM2G135359 pseudokinase doi: 10.2210/pdb6cpy/pdb. [DOI] [Google Scholar]
  2. Baker NA, Sept D, Joseph S, Holst MJ, & McCammon JA (2001). Electrostatics of nanosystems: application to microtubules and the ribosome. Proceedings of the National Academy of Sciences, 98(18), 10037–10041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Berginski ME, Moret N, Liu C, Goldfarb D, Sorger PK & Gomez SM (2020). The Dark Kinase Knowledgebase: an online compendium of knowledge and experimental results of understudied kinases. Nucleic Acids Research, 49(D1), D529–D535. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Capra JA & Singh M (2007). Predicting functionally important residues from sequence conservation. Bioinformatics, 23(15), 1875–1882. [DOI] [PubMed] [Google Scholar]
  5. Chatzou M, Magis C, Chang JM, Kemena C, Bussotti G, Erb I, et al. (2016). Multiple sequence alignment modeling: methods and applications. Briefings in bioinformatics, 17(6), 1009–1023. [DOI] [PubMed] [Google Scholar]
  6. Crooks GE, Hon G, Chandonia J-M & Brenner SE (2004). WebLogo: A Sequence Logo Generator. Genome Research, 14(6), 1188–1190. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Eyers PA (2015). Tribbles: A twist in the pseudokinase tail. Structure, 23(11), 1974–1976. [DOI] [PubMed] [Google Scholar]
  8. Eyers PA (2020). Marveling at the Incredible ULK4. Structure, 28(11), 1181–1183. [DOI] [PubMed] [Google Scholar]
  9. Eyers PA, & Murphy JM (2013). Dawn of the dead: protein pseudokinases signal new adventures in cell biology. Biochemical Society transactions, 41(4), 969–974. [DOI] [PubMed] [Google Scholar]
  10. Eyers PA, Keeshan K & Kannan N (2017). Tribbles in the 21st Century: The Evolving Roles of Tribbles Pseudokinases in Biology and Disease. Trends in Cell Biology, 27(4), 284–298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Foulkes DM, Byrne DP, Yeung W, Shrestha S, Bailey FP, Ferries S, et al. (2018). Covalent inhibitors of EGFR family protein kinases induce degradation of human Tribbles 2 (TRIB2) pseudokinase in cancer cells. Sci. Signal, 11(549). [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gosal G, Kochut KJ & Kannan N (2011). ProKinO: An Ontology for Integrative Analysis of Protein Kinases in Cancer. PLoS ONE, 6(12), e28782. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Grütter C, Sreeramulu S, Sessa G, & Rauh D (2013). Structural characterization of the RLCK family member BSK8: a pseudokinase with an unprecedented architecture. Journal of molecular biology, 425(22), 4455–4467. [DOI] [PubMed] [Google Scholar]
  14. Hanks SK & Hunter T (1995). The eukaryotic protein kinase superfamily: kinase (catalytic) domain structure and classification1. The FASEB Journal, 9(8), 576–596. [PubMed] [Google Scholar]
  15. Haydon CE, Eyers PA, Aveline-Wolf LD, Resing KA, Maller JL, & Ahn NG (2003). Identification of novel phosphorylation sites on Xenopus laevis Aurora A and analysis of phosphopeptide enrichment by immobilized metal-affinity chromatography. Molecular & cellular proteomics: MCP, 2(10), 1055–1067. [DOI] [PubMed] [Google Scholar]
  16. Huang L-C, Taujale R, Gravel N, Venkat A, Yeung W, Byrne DP, et al. (2021). KinOrtho: a method for mapping human kinase orthologs across the tree of life and illuminating understudied kinases. BMC Bioinformatics, 22(1), 446. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Jamieson SA, Ruan Z, Burgess AE, Curry JR, McMillan HD, Brewster JL, et al. (2018). Substrate binding allosterically relieves autoinhibition of the pseudokinase TRIB1. Science Signaling, 11(549). [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Janson G, & Paiardini A (2021). PyMod 3: a complete suite for structural bioinformatics in PyMOL. Bioinformatics, 37(10), 1471–1472. [DOI] [PubMed] [Google Scholar]
  19. Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, et al. (2021). Highly accurate protein structure prediction with AlphaFold. Nature, 596(7873), 583–589. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kannan N, Taylor SS, Zhai Y, Venter JC, & Manning G (2007). Structural and functional diversity of the microbial kinome. PLoS biology, 5(3), e17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Khamrui S, Ung PMU, Secor C, Schlessinger A & Lazarus MB (2019). High-Resolution Structure and Inhibition of the Schizophrenia-Linked Pseudokinase ULK4. Journal of the American Chemical Society, 142(1), 33–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Knighton DR, Zheng J, Eyck LFT, Ashford VA, Xuong N-H, Taylor SS, et al. (1991). Crystal Structure of the Catalytic Subunit of Cyclic Adenosine Monophosphate-Dependent Protein Kinase. Science, 253(5018), 407–414. [DOI] [PubMed] [Google Scholar]
  23. Kullback S & Leibler RA (1951). On Information and Sufficiency. The Annals of Mathematical Statistics, 22(1), 79–86. [Google Scholar]
  24. Kwon A, Scott S, Taujale R, Yeung W, Kochut KJ, Eyers PA, et al. (2019). Tracing the origin and evolution of pseudokinases across the tree of life. Sci. Signal, 12(578). [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Makarewicz T & Kaźmierkiewicz R Molecular Dynamics Simulation by GROMACS Using GUI Plugin for PyMOL. Journal of Chemical Information and Modeling, 2013. 53 (5), 1229–1234. [DOI] [PubMed] [Google Scholar]
  26. Manning G, Reiner DS, Lauwaet T, Dacre M, Smith A, Zhai Y, et al. (2011). The minimal kinome of Giardia lamblia illuminates early kinase evolution and unique parasite biology. Genome Biology, 12(7), 1–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Manning G, Whyte DB, Martinez R, Hunter T & Sudarsanam S (2002). The Protein Kinase Complement of the Human Genome. Science, 298(5600), 1912–1934. [DOI] [PubMed] [Google Scholar]
  28. McSkimming DI, Dastgheib S, Baffi TR, Byrne DP, Ferries S, Scott ST, et al. (2016). KinView: a visual comparative sequence analysis tool for integrated kinome research. Molecular BioSystems, 12(12), 3651–3665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. McSkimming DI, Dastgheib S, Talevich E, Narayanan A, Katiyar S, Taylor SS, et al. (2015). ProKinO: A Unified Resource for Mining the Cancer Kinome. Human Mutation, 36(2), 175–186. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Metz KS, Deoudes EM, Berginski ME, Jimenez-Ruiz I, Aksoy BA, Hammerbacher J, et al. (2018). Coral: Clear and Customizable Visualization of Human Kinome Data. Cell Systems, 7(3), 347–350. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Morrison DA (2006). Multiple sequence alignment for phylogenetic purposes. Australian Systematic Botany, 19(6), 479–539. [Google Scholar]
  32. Murphy JM, Nakatani Y, Jamieson SA, Dai W, Lucet IS & Mace PD (2015). Molecular Mechanism of CCAAT-Enhancer Binding Protein Recruitment by the TRIB1 Pseudokinase. Structure, 23(11), 2111–2121. [DOI] [PubMed] [Google Scholar]
  33. Murphy JM, Zhang Q, Young SN, Reese ML, Bailey FP, Eyers PA, et al. (2014). A robust methodology to subclassify pseudokinases based on their nucleotide-binding properties. Biochemical Journal, 457(2), 323–334. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Neuwald AF (2009). Rapid detection, classification and accurate alignment of up to a million or more related protein sequences. Bioinformatics, 25(15), 1869–1875. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Neuwald AF (2011). Surveying the manifold divergence of an entire protein class for statistical clues to underlying biochemical mechanisms. Statistical applications in genetics and molecular biology, 10(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Neuwald AF (2014). A Bayesian Sampler for Optimization of Protein Domain Hierarchies. Journal of Computational Biology, 21(3), 269–286. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Nichols RJ & Traktman P (2004). Characterization of Three Paralogous Members of the Mammalian Vaccinia Related Kinase Family. Journal of Biological Chemistry, 279(9), 7934–7946. [DOI] [PubMed] [Google Scholar]
  38. Olson AT, Wang Z, Rico AB & Wiebe MS (2019). A poxvirus pseudokinase represses viral DNA replication via a pathway antagonized by its paralog kinase. PLoS Pathogens, 15(2), e1007608. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Park C-H, Ryu HG, Kim S-H, Lee D, Song H & Kim K-T (2015). Presumed pseudokinase VRK3 functions as a BAF kinase. Biochimica et Biophysica Acta (BBA) - Molecular Cell Research, 1853(7), 1738–1748. [DOI] [PubMed] [Google Scholar]
  40. Pervez MT, Babar ME, Nadeem A, Aslam M, Awan AR, Aslam N, et al. (2014). Evaluating the Accuracy and Efficiency of Multiple Sequence Alignment Methods. Evolutionary Bioinformatics Online, 10, 205–217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Petrie EJ, Sandow JJ, Lehmann WI, Liang LY, Coursier D, Young SN, et al. (2019). Viral MLKL homologs subvert necroptotic cell death by sequestering cellular RIPK3. Cell reports, 28(13), 3309–3319. [DOI] [PubMed] [Google Scholar]
  42. Preuss F, Chatterjee D, Mathea S, Shrestha S, St-Germain J, Saha M, et al. (2020). Nucleotide Binding, Evolutionary Insights, and Interaction Partners of the Pseudokinase Unc-51-like Kinase 4. Structure, 28(11), 1184–1196. [DOI] [PubMed] [Google Scholar]
  43. Reiterer V, Eyers PA & Farhan H (2014). Day of the dead: pseudokinases and pseudophosphatases in physiology and disease. Trends in Cell Biology, 24(9), 489–505. [DOI] [PubMed] [Google Scholar]
  44. Ribeiro AJM, Holliday GL, Furnham N, Tyzack JD, Ferris K & Thornton JM (2018). Mechanism and Catalytic Site Atlas (M-CSA): a database of enzyme reaction mechanisms and active sites. Nucleic Acids Research, 46(Database issue), D618–D623. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Ribeiro AJM, Tyzack JD, Borkakoti N & Thornton JM (2020). Identifying pseudoenzymes using functional annotation: pitfalls of common practice. The FEBS Journal, 287(19), 4128–4140. [DOI] [PubMed] [Google Scholar]
  46. Scheeff ED, Eswaran J, Bunkoczi G, Knapp S & Manning G (2009). Structure of the Pseudokinase VRK3 Reveals a Degraded Catalytic Site, a Highly Conserved Kinase Fold, and a Putative Regulatory Binding Site. Structure, 17(1), 128–138. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Schneider TD, & Stephens RM (1990). Sequence logos: a new way to display consensus sequences. Nucleic acids research, 18(20), 6097–6100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Schrödinger, LLC, The PyMOL Molecular Graphics System, Version 2.3.2 (2019) [Google Scholar]
  49. Shrestha S, Byrne DP, Harris JA, Kannan N & Eyers PA (2020). Cataloguing the dead: breathing new life into pseudokinase research. The FEBS Journal, 287(19), 4150–4169. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Talevich E & Kannan N (2013). Structural and evolutionary adaptation of rhoptry kinases and pseudokinases, a family of coccidian virulence factors. BMC Evolutionary Biology, 13(1), 1–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Trott O, & Olson AJ (2010). AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. Journal of computational chemistry, 31(2), 455–461. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Tsuchiya Y, Byrne DP, Burgess SG, Bormann J, Baković J, Huang Y, et al. (2020). Covalent Aurora A regulation by the metabolic integrator coenzyme A. Redox biology, 28, 101318. 10.1016/j.redox.2019.101318 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Wang T, Zhao J, Savas AC, Zhang S & Feng P (2020). Viral pseudoenzymes in infection and immunity. The FEBS Journal, 287(19), 4300–4309. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Ward P, Equinet L, Packer J & Doerig C (2004). Protein kinases of the human malaria parasite Plasmodium falciparum: the kinome of a divergent eukaryote. BMC Genomics, 5(1), 1–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Watkins X, Garcia LJ, Pundir S, Martin MJ, & UniProt Consortium. (2017). ProtVista: visualization of protein sequence annotations. Bioinformatics, 33(13), 2040–2041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Zaru R, Magrane M, Orchard S, & UniProt Consortium. (2020). Challenges in the annotation of pseudoenzymes in databases: the UniProtKB approach. The FEBS journal, 287(19), 4114–4127. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Zheng J, Trafny EA, Knighton DR, Xuong N, Taylor SS, Eyck LFT, et al. (1993). 2.2 Å refined crystal structure of the catalytic subunit of cAMP-dependent protein kinase complexed with MnATP and a peptide inhibitor. Acta Crystallographica Section D: Biological Crystallography, 49(3), 362–365. [DOI] [PubMed] [Google Scholar]

RESOURCES