Dual-Specificity, Tyrosine Phosphorylation-Regulated Kinases (DYRKs) and cdc2-Like Kinases (CLKs) in Human Disease, an Overview

Dual-specificity tyrosine phosphorylation-regulated kinases (DYRK1A, 1B, 2-4) and cdc2-like kinases (CLK1-4) belong to the CMGC group of serine/threonine kinases. These protein kinases are involved in multiple cellular functions, including intracellular signaling, mRNA splicing, chromatin transcription, DNA damage repair, cell survival, cell cycle control, differentiation, homocysteine/methionine/folate regulation, body temperature regulation, endocytosis, neuronal development, synaptic plasticity, etc. Abnormal expression and/or activity of some of these kinases, DYRK1A in particular, is seen in many human nervous system diseases, such as cognitive deficits associated with Down syndrome, Alzheimer’s disease and related diseases, tauopathies, dementia, Pick’s disease, Parkinson’s disease and other neurodegenerative diseases, Phelan-McDermid syndrome, autism, and CDKL5 deficiency disorder. DYRKs and CLKs are also involved in diabetes, abnormal folate/methionine metabolism, osteoarthritis, several solid cancers (glioblastoma, breast, and pancreatic cancers) and leukemias (acute lymphoblastic leukemia, acute megakaryoblastic leukemia), viral infections (influenza, HIV-1, HCMV, HCV, CMV, HPV), as well as infections caused by unicellular parasites (Leishmania, Trypanosoma, Plasmodium). This variety of pathological implications calls for (1) a better understanding of the regulations and substrates of DYRKs and CLKs and (2) the development of potent and selective inhibitors of these kinases and their evaluation as therapeutic drugs. This article briefly reviews the current knowledge about DYRK/CLK kinases and their implications in human disease.


Protein Phosphorylation, Protein Kinases, Kinase Inhibitors, and Human Disease
Protein phosphorylation is probably one of the most important and most studied mechanism used by cells to regulate their proteins in terms of enzymatic activity, functions, localization, half-life, interactions with other proteins or other ligands, etc. It is also a key mechanism for signal transduction between cells and within cells. Protein phosphorylation occupies a central place in the scientific literature with 337,916 references (as of 1 June 2021). Protein phosphorylation on serine, threonine, and tyrosine residues is carried out by protein kinases, a family of enzymes known as the human kinome, comprising at least 538 members [1,2] divided into tyrosine kinases and serine/threonine kinases (some of the latter are so-called dual specificity, as they also phosphorylate tyrosine residues), histidine kinases, and pseudo-kinases (protein kinases: 573,472 references (as of 1 June 2021), i.e., one article published every 7 min for the last five years). Quite uniquely, four different Nobel Prizes in medicine or physiology have been awarded to this field (1989,1992,2000,2001) (Figure 1).

Figure 1.
Four Nobel Prizes in Physiology or Medicine awarded in the field of protein phosphorylation and protein kinases. Protein kinases catalyze the transfer of the γ-phosphate of ATP to the hydroxyl substituents of serine, threonine, or tyrosine residues in proteins, thereby altering the physiological properties of their protein substrates. The human kinome comprises 538 protein kinases. Michael Bishop and Harold E. Varmus received the Nobel Prize 1989 "for their discovery of the cellular origin of retroviral oncogenes" (src, the first described oncogene, which encodes a tyrosine kinase). Edmond H. Fischer and Edwin G. Krebs received the Nobel Prize 1992 "for their discoveries concerning reversible protein phosphorylation as a biological regulatory mechanism" (they are the true discoverers of protein kinases). The Nobel Prize 2000 was awarded jointly to Arvid Carlsson, Paul Greengard, and Eric R. Kandel "for their discoveries concerning signal transduction in the nervous system" (Paul Greengard investigated the mechanism of signal transduction of neurotransmitters in the central nervous system and demonstrated the key importance of phosphorylation by kinases such as CDK5, PKA, CK1, and CK2 and Eric Kandel the importance of PKA in memory in Aplysia). The Nobel Prize 2001 was awarded jointly to Leland H. Hartwell, Tim Hunt, and Paul M. Nurse "for their discoveries of key regulators of the cell cycle" (using yeast or sea urchin embryos, they discovered how the cell division cycle is regulated by CDKs).
Since protein phosphorylation is involved in essentially all physiological events, abnormal phosphorylation is implicated in many human diseases. Abnormally expressed or abnormally active kinases represent the most frequent situation. Consequently, inhibiting disease-relevant kinases or normalizing their activities constitutes a rational approach to tackle numerous diseases. This is why protein kinases have become, in a few decades after their initial discovery [3], the first therapeutic targets-before G-protein-coupled receptors-in the pharmaceutical industry's search for novel drug candidates (reviews: [2,[4][5][6][7][8]). As of early February 2021, 62 kinase inhibitors have reached the market, mostly for the treatment of various cancer indications [9][10][11].
Since protein phosphorylation is involved in essentially all physiological events, abnormal phosphorylation is implicated in many human diseases. Abnormally expressed or abnormally active kinases represent the most frequent situation. Consequently, inhibiting disease-relevant kinases or normalizing their activities constitutes a rational approach to tackle numerous diseases. This is why protein kinases have become, in a few decades after their initial discovery [3], the first therapeutic targets-before G-protein-coupled receptorsin the pharmaceutical industry's search for novel drug candidates (reviews: [2,[4][5][6][7][8]). As of early February 2021, 62 kinase inhibitors have reached the market, mostly for the treatment of various cancer indications [9][10][11].

DYRKs and CLKs: Structure, Activation, Interactors, and Substrates
Among serine/threonine kinases, DYRKs and CLKs (Figures 2-4) belong to a family of 62 kinases known as the CMGC group, which also includes mitogen-activated protein kinases (MAPKs), cyclin-dependent kinases (CDKs), and the glycogen synthase kinases 3 (GSK3) family. DYRKs and CLKs are two highly related and conserved kinase families (Table 1), usually sensitive to the same pharmacological inhibitors. The DYRK family comprises 5 members: DYRK1A and DYRK1B (class 1 DYRKs) and DYRK2, 3, and 4 (class 2 DYRKs) (reviews: [12][13][14]). The CLK family comprises 4 members: CLK1, 2, 3, and 4 (review: [15]).     1A  48  51  50  48  100  85  45  43  45  1B  49  49  48  48  93  100  45  44  45  2  55  52  53  53  60  61  100  79  59  3  55  51  52  54  63  64  89  100  57  4  54  56  55  55  62  63  74  72  100 Alignment of DYRKs and CLKs sequences shows the classical central kinase catalytic domain flanked by N-terminal and C-terminal extensions (Figures 3 and 4). The N-terminal domain of all DYRKs displays a conserved DYRK homology box (DH) [16] that contributes to autophosphorylation of a conserved tyrosine in the kinase domain (Tyr321 in DYRK1A) during maturation of the kinase [17,18]. Autophosphorylation on the tyrosine residue is preceded by hydroxylation of a proline residue by the PHD1 prolyl hydroxylase, an absolute requirement for catalytic activation of the kinase [19]. The Nterminal domain of all DYRKs except DYRK3 contains a nuclear localization signal domain (NLS) [20]. DYRK2, DYRK3, and DYRK4 contain a conserved N-terminal autophosphorylation accessory (NAPA) domain essential for autophosphorylation of the activation loop tyrosine [21]. The C-terminal domain of DYRK1A and DYRK1B displays a region enriched in proline, glutamic acid, serine, and threonine known as a PEST sequence, which favors rapid degradation [22]. A region containing 13 consecutive histidine residues is present in the C-terminal region of DYRK1A but not in other DYRKs or CLKs. A comprehensive analysis of the human proteome revealed that only 86 proteins display such a histidine repeat stretch (5 or more histidines) [23]. The presence of a homopolymeric histidine repeat in nuclear proteins appears to be involved in the targeting/localization of these proteins to the nuclear speckles compartment. Many of these polyhistidine sequence-bearing proteins are expressed in the nervous system [23]. The unique polyhistidine sequence provides a natural His-tag which allows the purification/enrichment of DYRK1A using immobilized metal-affinity chromatography (IMAC) (nickel, cobalt) [24,25] [Sévère et al., unpublished]. DYRKs and CLKs have been highly conserved throughout evolution, and orthologs are found in yeast [26,27], plants [28][29][30][31][32], unicellular algae [33,34], and unicellular parasites such as Trypanosoma [35][36][37], Leishmania [38][39][40], and Plasmodium [41][42][43][44][45][46].  Alignment of DYRKs and CLKs sequences shows the classical central kinase catalytic domain flanked by N-terminal and C-terminal extensions (Figures 3 and 4). The N-terminal domain of all DYRKs displays a conserved DYRK homology box (DH) [16] that contributes to autophosphorylation of a conserved tyrosine in the kinase domain (Tyr321 in DYRK1A) during maturation of the kinase [17,18]. Autophosphorylation on the tyrosine residue is preceded by hydroxylation of a proline residue by the PHD1 prolyl hydroxylase, an absolute requirement for catalytic activation of the kinase [19]. The N-terminal domain of all DYRKs except DYRK3 contains a nuclear localization signal domain (NLS) [20]. DYRK2, DYRK3, and DYRK4 contain a conserved N-terminal autophosphorylation accessory (NAPA) domain essential for autophosphorylation of the activation loop tyrosine [21]. The C-terminal domain of DYRK1A and DYRK1B displays a region enriched in proline, glutamic acid, serine, and threonine known as a PEST sequence, which favors rapid degradation [22]. A region containing 13 consecutive histi- Crystal structures of various DYRKs and CLKs, alone or in complex with inhibitors, have been solved (Table 2). These structures have allowed a detailed understanding of the mechanism of activation of DYRKs by autophosporylation on the tyrosine residue as well as an understanding of the binding mode of numerous inhibitors, providing very useful information for the structure-guided synthesis of improved pharmacological inhibitors. Multiple sequence alignment of the canonical sequences DYRK and CLK members was performed using Clustal Omega [41] (https://www.ebi.ac.uk) (accessed on 12 April 20 and edited using Jalview [42]. Each residue in the alignment is assigned a colour if the amino acid profile of the alignme at that position meets some minimum criteria specific for the residue type (Clustal X Colour Schem http://www.jalview.org/help/html/colourSchemes/clustal.html) (accessed on 12 April 2021). Distinct sequences are in cated: Activation loop and tyrosine residue that is autophosphorylated (Yn); DH, DYRK homology box; His domain, consecutive histidine residues region; kinase domain; NAPA, N-terminal autophosphorylation accessory domain; NL nuclear localization signal domain (NB: a NLS sequence is only found in isoform 4 of DYRK4, not in the canonical quence); PEST, proline (P), glutamic acid (E), serine (S), and threonine (T)-rich domain; S/T, serine, and thr nine-enriched domain; WDR68 binding domain. . Sequence alignment of human DYRKs and CLKs. Multiple sequence alignment of the canonical sequences of DYRK and CLK members was performed using Clustal Omega [41] (https://www.ebi.ac.uk) (accessed on 12 April 2021) and edited using Jalview [42]. Each residue in the alignment is assigned a colour if the amino acid profile of the alignment at that position meets some minimum criteria specific for the residue type (Clustal X Colour Scheme, http://www.jalview.org/help/html/colourSchemes/clustal.html) (accessed on 12 April 2021). Distinct sequences are indicated: Activation loop and tyrosine residue that is autophosphorylated (Yn); DH, DYRK homology box; His domain, 13 consecutive histidine residues region; kinase domain; NAPA, N-terminal autophosphorylation accessory domain; NLS, nuclear localization signal domain (NB: a NLS sequence is only found in isoform 4 of DYRK4, not in the canonical sequence); PEST, proline (P), glutamic acid (E), serine (S), and threonine (T)-rich domain; S/T, serine, and threonineenriched domain; WDR68 binding domain.  CLK4 CX-4945 6FYV [65] The nuclear interactome of DYRK1A is highly enriched in DNA damage repair factors (RNF169), transcriptional elongation factors, and E3 ubiquitin ligases [72][73][74]. The interactome of all CMGC kinases, including DYRKs and CLKs, has been extensively studied [75]. Other large-scale interactome studies provide information on proteins binding to DYRKs and CLKs [76,77]. A detailed description of the DYRKs and CLKs interactomes is beyond the scope of this review. However, we would like to mention WDR68, also known as DCAF7 (DDB1-associated and CUL4-associated factor 7) or HAN11 (Human homolog of the Petunia hybrida an11 gene), a scaffolding protein of the WD40-repeat protein family [78] that binds class 1 DYRKs and HIPK2 (Homeodomain-interacting protein kinase 2). The interaction between WDR68 and DYRK1A/DYRK1B has been extensively studied [79][80][81]: it involves a conserved 12 amino acid sequence located in the N-terminal domain of DYRK1A/1B. This interaction mediates binding to other proteins, such as the adenovirus E1A oncoprotein [81] and RNA polymerase II [82], thereby probably favoring substrate recruitment for DYRK1A/1B and HIPK2. WDR68 is essential for craniofacial development, a process involving DYRK1A [83,84]. DYRK1A regulates the interaction between WDR68 and Huntington-associated protein 1 (Hap1), which may contribute to postnatal growth retardation in Down syndrome (DS) [85]. Expression of WDR68 regulates the level of expression of DYRK1A and DYRK1B [86].
DYRK and CLK kinases phosphorylate many substrates involved in signaling pathways, mRNA splicing, chromatin transcription, DNA damage repair, cell survival, cell cycle control, differentiation, homocysteine/methionine/folate regulation, endocytosis, neuronal development and functions, synaptic plasticity, etc. Reviewing substrates and cellular functions of all DYRKs and CLKs is beyond the scope of this brief review, although phosphorylation of substrates and their cellular and physiological consequences underlie normal functioning and pathological conditions.

DYRKs and Human Disease
There is growing evidence for the involvement of various DYRKs in human disease. We will briefly review these accumulating data (Table 3 and Figure 5A).

DYRK1A and Down Syndrome (DS)
The gene encoding DYRK1A is located on chromosome 21, within the Down syndrome critical region (DSCR), the triploidy of which is responsible for most DS-associated deficiencies (reviews: [13,14]) ( Table 3 for more details). There is considerable genetical and pharmacological evidence showing that the mere 1.5-fold overexpression of DYRK1A is responsible for most cognitive deficits observed in DS patients (reviews: [14,[87][88][89][90][91][92]). Genetical normalization of DYRK1A levels or pharmacological inhibition of its catalytic activity restores cognitive functions (Table 3 for specific references). The development of pharmacological inhibitors of DYRK1A is a major avenue for the treatment of cognitive deficits associated with DS (and Alzheimer's disease) (reviews: [88,89,93]).
terfering with its folding process [254], all reported inhibitors appear to act by competing with ATP in its binding to the catalytic site of the kinases (as demonstrated by enzymological studies as well as by co-crystallization with their kinase targets ( Table 2)). Several DYRK1A inhibitors have been reported in recent years (reviews: [88,93,95]) which, like Leucettines and Leucettine L41 in particular, correct cognition deficits in DS and AD animal models [127,128,148].

DYRK1A and Parkinson's Disease (PD)
Genome-wide association studies (GWAS) have revealed that DYRK1A is a risk factor for PD [99]. DYRK1A phosphorylates key factors for PD such as parkin, septin 4, and α-synuclein. Upregulation of micro-RNAs specific for PD targets DYRK1A expression [100]. There is further evidence that DYRK1A expression is increased in PD and in Pick's disease [101].

DYRK1A and Diabetes
There is a growing body of evidence showing that DYRK1A/1B inhibitors induce the proliferation of insulin-producing pancreatic β-cells, making DYRK1A/1B kinases attractive therapeutic targets for β cell regeneration for both type 1 and type 2 diabetes [105,106] (Table 3 for more details).

DYRK1B is involved in the replication of various viruses including HCV, Chikungunya virus, Dengue virus, SARS coronavirus, HCMV, and human papillomavirus (HPV).
Like with DYRK1A, DYRK1B inhibition leads to the proliferation of pancreatic, insulinproducing β-cells. DYRK1B is involved in neuroinflammation [115]. Targeting DYRK1B provides a new rationale for treatment of various solid cancers such as liposarcoma or breast cancers (reviews: [116,117]) as well as in chronic myeloid leukemia (CML).

CLKs and Human Disease
The data supporting the involvement of various CLKs in human disease is briefly described below and in Table 4 and Figure 5B.
CLKs play essential functions in alternative splicing. CLKs act as a body-temperature sensors, which globally control alternative splicing and gene expression. The activity of CLKs is indeed highly responsive to physiological temperature changes, which is conferred by structural rearrangements within the kinase activation segment [57]. CLK1 triggers periodic alternative splicing during the cell division cycle [222]. CLK1 regulates influenza A virus mRNA splicing, and its inhibition prevents viral replication. CLK1 and CLK2 also regulate HIV-1 gene expression. CLK1 is an autophagy inducer. CLK1 inhibition may prevent chemoresistance in glioma, and CLK1 inhibition by TG693 allows the skipping of mutated exon 31 of the dystrophin gene in Duchenne Muscular Dystrophy. CLK1 autoregulates itself through exon skipping and intron retention [223].
Inhibition of CLK2 has been proposed as a way to improve neuronal functions and combat intellectual disability and autism in Phelan-McDermid syndrome (PMDS) [65]. Alternative splicing of Tau exon 10 is regulated by CLK2 and other CLKs, leading to changes in the 3R/4R isoform ratio and neurodegeneration in sporadic AD [224,225]. Dual inhibitions of CLK2 and DYRK1A by Lorecivivint (SM04690) and by its analogue SM04755 are potential disease-modifying approaches for knee osteoarthritis [193,194] and for tendinopathy, respectively [195]. CLK2 inhibition compromises MYC-driven breast tumors, triple-negative breast cancer, and glioblastoma. Inhibition of CLK2, CLK3, and/or CLK4 blocks HIV-1 production.

Therapeutic Potential of DYRK and CLK Inhibitors
Abnormal activities in DYRKs and CLKs have motivated numerous groups to search for, optimize, and characterize pharmacological inhibitors of these kinases for their use in various indications (reviews: [88,89,93]) ( Figure 5). There is particular interest in the development of DYRKs/CLKs inhibitors as potential drug candidates to address cognitive deficits in DS and AD as well as to increase the pancreatic β-cell mass in both type 1 and type 2 diabetes (review: [106]) or to inhibit several cancers and leukemias by inhibiting cell proliferation. A few representative inhibitors are shown in Figure 6. Most DYRK1A inhibitors also inhibit, to various extent, DYRK1B, 2, 3, and 4 as well as the closely related CLK1, 2, 3, and 4 [93]. Apart from FINDY, which inhibits DYRK1A by interfering with its folding process [254], all reported inhibitors appear to act by competing with ATP in its binding to the catalytic site of the kinases (as demonstrated by enzymological studies as well as by co-crystallization with their kinase targets ( Table 2)). Several DYRK1A inhibitors have been reported in recent years (reviews: [88,93,95]) which, like Leucettines and Leucettine L41 in particular, correct cognition deficits in DS and AD animal models [127,128,148].  . DYRK and CLK inhibitors. A few representative pharmacological inhibitors: AnnH75 [51], EGCG [255], EHT-1610 [55], Harmine [256], INDY [44], Leucettine L41 [43,257], Lorecivivint [258,259], GNF4877 [170,179], MU1210 [68], and TCMDC-135051 [251,252]. Numbers under each structure indicates IC50 values (expressed in µM) towards DYRK1A, CLK1, and GSK3β (33PanQinase™ assay, Reaction Biology Corp.).

Conclusions
The limited studies that have been carried out so far with DYRKs and CLKs have opened up new avenues in our understanding of their regulation and functions. Yet, a great deal of work remains to be done to fully understand the cellular and physiological functions of each member of the DYRK and CLK families. Tissular and cellular distribution, polymorphism and mutations, regulation of expression levels, and post-translational modifications are just a few of the parameters that need to be investigated in detail. Conditional knock-out/ knock-in and overexpression models will also contribute to the understanding of the unique roles of each of these kinases and their eventual redundancy. Very precious tools-antibodies, affinity reagents, pharmacologi- Figure 6. DYRK and CLK inhibitors. A few representative pharmacological inhibitors: AnnH75 [51], EGCG [255], EHT-1610 [55], Harmine [256], INDY [44], Leucettine L41 [43,257], Lorecivivint [258,259], GNF4877 [170,179], MU1210 [68], and TCMDC-135051 [251,252]. Numbers under each structure indicates IC50 values (expressed in µM) towards DYRK1A, CLK1, and GSK3β (33PanQinase™ assay, Reaction Biology Corp.).

Conclusions
The limited studies that have been carried out so far with DYRKs and CLKs have opened up new avenues in our understanding of their regulation and functions. Yet, a great deal of work remains to be done to fully understand the cellular and physiological functions of each member of the DYRK and CLK families. Tissular and cellular distribution, polymorphism and mutations, regulation of expression levels, and post-translational modifications are just a few of the parameters that need to be investigated in detail. Conditional knock-out/knock-in and overexpression models will also contribute to the understanding of the unique roles of each of these kinases and their eventual redundancy. Very precious tools-antibodies, affinity reagents, pharmacological inhibitors, kinase inactive mutants, transgenic animals-have been developed, yet DYRK1A has been mostly studied, and other DYRKs and CLKs will require the development of specific tools.
The currently available data demonstrate major implications of several protein kinases of the DYRK and CLK families in several human diseases. The first inhibitors are reaching regulatory preclinical studies and early clinical studies. The next few years will certainly see the validation of specific DYRKs and CLKs inhibitors for specific clinical indications. It is still a bit early to speculate which one these will be. Clearly though, cognition in DS and AD, diabetes, cancers, and osteoarthritis are the most advanced examples of potential applications, but viral and unicellular parasite infections will certainly gain momentum as potential therapeutic indications for DYRKs/CLKs inhibitors. Higher potency and higher selectivity will also emerge in the near future. We can clearly anticipate that, as fundamental knowledge will accumulate on these protein kinases, more applied pharmaceutical work will result in well characterized, selective, and potent inhibitors leading to significant clinical improvements for patients.