PI3Ks are ubiquitously expressed intracellular lipid kinases that phosphorylate the 3′ hydroxyl groups of phosphatidylinositol and also harbor Ser/Thr protein kinase activity. These proteins regulate several critical cellular functions including cell survival, proliferation, motility, and vesicle trafficking [13,14]. PI3Ks are divided into three classes (class I, II, and III) based on their primary structure and substrate specificity. Class-I PI3Ks are further separated into subclasses IA (PI3Kα, PI3Kβ, PI3Kδ) and IB (PI3Kγ) based on their regulatory proteins and signaling pathway involvement.

PI3Ks function as heterodimers, consisting of a regulatory subunit and a catalytic subunit that converts the phosphatidylinositol second messenger PI(3,4)P2 (PIP2) to PI(3,4,5)P3 (PIP3) in response to extracellular stimuli on the activation of upstream receptor tyrosine kinases (RTKs) and G-protein coupled receptors (GPCRs) [15,16]. Conversion of PIP2 to PIP3 promotes the activation of downstream proteins AKT and mTOR [17]. PI3K activity is antagonized by the PTEN phosphatase, which hydrolyzes PIP3 to PIP2 [18]. As PIP3 plays a critical role in cell growth and replication, aberrant lipid kinase activity can promote oncogenic signaling. Class-I proteins have been recognized as promising drug targets due to their involvement in cancer and immune disease pathogenesis via the PI3K/mTOR pathway. The structural conservation of the ATP-binding pocket has complicated the development of isoform-specific inhibitors. Pan-PI3K inhibitors are active against all class-I isoforms and result in off-target toxicity and side effects. To date, there are five FDA-approved PI3K inhibitors (alpelisib, copanlisib, duvelisib, idelalisib, and umbralisib). While PI3K kinase inhibitors have been successfully applied to certain malignancies, their use and development has been hindered by poor drug tolerance and toxicity [19].

PIK3CA encodes the phosphatidylinositol 3-kinase catalytic subunit alpha (PI3Kα), also known as p110α. PI3Kα is the predominant catalytic isoform for regulation of glucose homeostasis and is commonly mutated and amplified in a variety of cancers [20]. The catalytic p110α subunit is composed of five domains, including an adaptor binding domain (ABD) that binds to the Class-IA regulatory domains, a Ras binding domain (RBD), a C2 domain that binds to cell membranes, a helical domain with unknown function, and a catalytic kinase domain. PI3Kα activation is mediated by receptor tyrosine kinases (RTKs), Ras proteins, and additional small molecules including calmodulin [21,22].

The catalytic p110α subunit can couple to regulatory subunit p85α, the SH2 domain of which contains high-affinity binding sites to the phosphorylated tyrosine motif (pYXXM) found in the C-terminus of RTKs [23]. The RTK pYXXM motif disrupts interactions between the catalytic p110α subunit and regulatory p85α subunit and activates PI3Kα by releasing the inhibitory p85α SH2 domains from the catalytic p110α subunit. This release triggers a conformational change in PI3Kα that exposes the kinase domain and permits membrane binding [24]. It has been shown that p110α phosphorylates Ser608 of the p85α regulatory subunit and decreases catalytic activity in vitro [25]. Mutation of Ser608 to Ala or Glu reduced lipid kinase activity, as well as the interactions between the p110α catalytic subunit and p85α regulatory subunit. Thus, phosphorylation of Ser608 may reveal a mechanism for regulating activity by stabilizing the autoinhibited state.

Mutations in PI3Kα are frequently found in brain, breast, head and neck, endometrial, cervical, and gastric cancers [17,26,27]. Most mutations cause gain-of-function (GOF) and are found within the helical or kinase domains. It has been reported that GOF mutations in the helical domain require interactions with Ras-GTP, whereas kinase domain mutations require interactions with the p85 regulatory subunit [28]. Co-existing mutations in both domains have been found to synergistically increase the catalytic function and tumorigenic activity [29]. In addition, deletions in the C2 domain have been found to activate PI3K signaling while also increasing the sensitivity to PI3Kα inhibitors, suggesting that residues within the C2 domain are critical for for PI3Kα function [30]. There are several hotspot mutations in p110α that confer a gain of function, including helical domain mutations E545K and E542K and the catalytic domain mutation H1047R. Helical domain mutations E545K and E542K have been shown to suppress inhibition of p110α by the p85 regulatory subunit, which results in mutation-driven signaling that promotes glucose metabolism and cervical cancer cell proliferation [31]. Catalytic domain mutation H1047R has been shown to enhance interactions between p110α and the lipid membrane, thereby enhancing its lipid kinase activity and downstream signaling.

The importance of PI3Kα in disease has been well-established; in May 2019, alpelisib (Piqray™, Novartis, Morris County, NJ, USA) was the first PI3Kα inhibitor approved by the FDA for the treatment of breast cancer [32]. Alpelisib has specific activity for PI3Kα over other isoforms, despite their nearly identical active sites, and potently inhibits common mutations such as E545K and H1047R.

PIK3CG encodes for the phosphatidylinositol 3-kinase catalytic subunit gamma (PI3Kγ), also known as p110γ. PI3Kγ belongs to class-IB PI3Ks and has both lipid and Ser/Thr protein kinase activity. It regulates immune stimulation and suppression in inflammation and cancer [33]. The PI3Kγ catalytic subunit has five domains, consisting of a putative uncharacterized adaptor binding domain (ABD), a Ras-binding domain (RBD), a C2 domain for binding cell membranes, an α-helical domain, and a catalytic kinase domain [34,35]. PI3Kγ activation is primarily regulated through interactions with GPCRs, which occur through association of the PI3Kγ regulatory subunit with the G-protein βγ subunits [36,37]. It can also be activated through interaction between the PI3Kγ RBD domain and Ras GTPases [38]. In the absence of lipid membrane binding, PI3Kγ maintains an inactive conformation [39].

In contrast to PI3Kα, PI3Kγ is commonly upregulated or overexpressed in cancer rather than mutated; however, mutations have been identified in cancer patients [40,41]. Loss-of-function mutations in PI3Kγ cause severe immunodeficiency, highlighting PI3Kγ’s critical role in promoting appropriate adaptive immune responses [42]. PI3Kγ is also a driver of inflammatory and metabolic disorders including rheumatoid arthritis, atherosclerosis, lupus, obesity, and pulmonary fibrosis [43].

PIKfyve is a bifunctional lipid kinase with Ser/Thr protein kinase activity. PIKfyve and its enzymatic products regulate cellular processes including membrane trafficking, ion channel activity, cytoskeletal dynamics, nuclear transport, stress- and hormone-induced signaling, transcription, and cell cycle progression [44,45].

As a lipid kinase, it synthesizes phosphoinositides (PIs) PtdIns(3,5)P2 from PtdIns3P and PtdIns5P from PtdIns. Synthesis of PtdIns(3,5)P2 by PIKfyve is negatively regulated by the formation of a multi-protein complex with scaffolding regulator ArPIKfyve and phosphatase Sac3 [46]. The cryo-EM structure revealed that formation of this complex sterically hinders PIKfyve from accessing membrane-associated PIs [45]. Serine residue autophosphorylation of PIKfyve represses its own lipid kinase activity and simultaneously activates Sac3 lipid phosphatase activity to downregulate lipid product synthesis [45]. Sac3 is also a serine phosphatase that acts on PIKfyve to increase the lipid kinase activity [45]. PIKfyve kinase activity is required for Sac3 lipid phosphatase activity [45], meaning that its dual function is critical for maintaining the delicate balance of lipid homeostasis. However, given the presence of multiple phospho-sites on PIKfyve and its role as a target in multiple signaling pathways, how the delicate balance between the dual activities of PIKfyve and Sac3 is maintained remains unknown.

PIKfyve has four structured domains including a FYVE finger domain, which targets the protein to PtdIns3P-enriched endosome membranes, and a DEP domain with unknown function. The middle of the protein interacts with several binding partners and contains two domains, a Cpn60-TCP1 domain, which has sequence similarity to chaperonins, and a CHK homology region containing the conserved Cys, His, and Lys residues found in all PIKfyve orthologs. The PIKfyve catalytic domain controls all three of its catalytic functions (PtdIns(3,5)P2/PtdIns5P synthesis and protein phosphorylation) and contains a single ATP-binding site. Mutation of the catalytic Lys-1831 abrogates all three kinase activities [47]. Mutations in PIKfyve cause functional defects in endosomal sorting, leading to Francois-Neetens fleck corneal dystrophy [48].

Ire1α is a ubiquitously expressed transmembrane ER stress sensor that functions as a Ser/Thr protein kinase and endoribonuclease. It is the best-studied branch of the unfolded protein response (UPR), an integrated intracellular signal transduction pathway that is activated in response to the accumulation of unfolded proteins in the ER [49]. The UPR initially responds to ER stress by blocking translation and upregulating molecular chaperones and folding enzymes. Prolonged cell stress causes a switch to pro-apoptotic and pro-inflammatory signaling cascades. The UPR has become an attractive pathway for drug discovery efforts due to its involvement in cancer, inflammation, neurological disorders, diabetes, and ischemia-reperfusion injury [50].

Ire1α contains an N-terminal ER luminal domain bound to cytoplasmic Ser/Thr kinase and endoribonuclease domains through a transmembrane linker [51]. The accumulation of misfolded protein causes Ire1α to dimerize in a “back-to-back” orientation [52], resulting in the formation of high-order oligomers and release of the ER Hsp70 chaperone BiP from the inactive Ire1α monomer. Thus, Ire1α activation is facilitated by luminal domain dimerization and trans-autophosphorylation of the cytosolic kinase domains [53,54]. Phosphorylation of the Ire1α activation segment stabilizes the RNase domain, which excises a 26-nucleotide intron from XBP1 [52,55]. The two spliced exons are then ligated by RNA ligase RtcB to form XBP1s [56], which encode an essential transcription factor for downstream activation of UPR response genes [57,58].

Ire1β is a close paralog of Ire1α. It has dual Ser/Thr protein kinase and endoribonuclease activities and induces the unfolded protein response (UPR) through transcription factor XBP1 [55,58], albeit less effectively than Ire1α [59]. Like Ire1α, it contains separate kinase and endoribonuclease domains. Ire1β serves as a direct and dominant-negative suppressor of Ire1α, dampening the UPR and ER stress response in epithelial cells of the intestine and other mucosal surfaces [53,60].

The eukaryotic translation initiation factor 2α kinase 2 (EIF2AK2), also known as protein kinase R (PKR) and interferon-induced double-stranded RNA-activated protein kinase, is a dual-specificity Ser/Thr and Tyr protein kinase. It contains two dsRNA binding motifs (dsRBD) and a kinase domain. PKR has been extensively studied as a regulator of the integrated stress response to RNA and DNA viruses, working as both a translational repressor through phosphorylation of Ser-51 on EIF2α and as a transcriptional upregulator of innate immune pathway proteins including Iκβ and NF-κβ [61,62,63]. On DNA damage, PKR halts the G2-M cell cycle transition by phosphorylating Tyr-4 of CDK1, causing its ubiquitination and proteasomal degradation [64]. PKR is a critical signaling mediator of cell proliferation and apoptosis through its interactions with p38 MAP kinase, NF-κβ, and the insulin signaling pathways [63,65,66,67]. It is ubiquitously expressed in vertebrates [68] and is implicated in cancer, neurodegeneration, inflammation, aging, and metabolic disorders [69,70].

The most widely used PKR inhibitor is C16, an imidazo-oxindole inhibitor with an in vitro IC₅₀ of 210 nM; however, less potent inhibitors including 2-aminopurine have also been developed [70]. As an ATP-competitive inhibitor, C16 inhibits both the Ser/Thr and Tyr kinase functions of PKR and has been investigated for use in memory enhancement [71], neurodegeneration [72], inflammation [73], and metabolic disorders [74].

RIOK1 is a Ser/Thr protein kinase and ATPase in the Rio (right open reading frame) family of atypical protein kinases. In humans, there are three Rio subfamilies: Rio1 and Rio2 are conserved across all kingdoms of life, while Rio3 is only found in multicellular eukaryotes [75]. Specifically, RIOK1 functions in processes essential for cell proliferation, cell cycle progression, and chromosome maintenance. Yeast and human RIOK1 kinases are essential for the biogenesis of small ribosomal subunits, and depletion of this enzyme causes cell cycle arrest in yeast. Although RIOK1 is classified as a Ser/Thr kinase [76], it predominantly acts as an ATPase and regulates pre-40S ribosomal subunit association [77].

The RIO kinase domain adopts the conserved bilobal kinase domain structure but lacks an activation loop and substrate recognition sites. It instead contains a C-terminal RIO-kinase specific αR helix and a flexible loop between β3 and helix αC, which together allow it to accommodate a phospho-Asp substrate and serve as an ATPase [78]. At this time, there are no specific RIOK1 inhibitors. However, RNAi-mediated knockdown of RIOK1 reduced cell proliferation in RAS-driven cancer cell lines [79], making it an alluring target for drug discovery efforts.

TATA-box binding protein associated factor 1 (TAF1) is a Ser/Thr protein kinase and histone acetyltransferase with ubiquitin-activation and -conjugation activities [80]. TAF1 is the largest subunit of TFIID and binds core promoter sequences and transcriptional regulators. TFIID, which initiates RNA-polymerase II-dependent transcription, is the core scaffold for the TFIID basal transcription factor complex [81,82]. TFIID is comprised of the evolutionarily conserved TATA binding protein (TBP) and multiple TBP-associated factors (TAFs).

TAF1 is a bipartite kinase composed of N- and C-terminal kinase domains (NTK and CTK) and a centrally located histone acetyltransferase domain (HAT) [83]. Between the HAT and CTK domains are two bromodomains that selectively bind to multiply acetylated histone H4 peptides [84]. TAF1 acetylates histones H3 and H4 [85], and TAF1-mediated acetylation of H3 is required for Sp1 activation of cyclin D1 transcription and G1 to S-phase cell cycle progression [86]. Both TAF1 kinase domains are necessary for Ser phosphorylation of the TFIIF subunit transcription factor RAP74 and initiation of the RNA polymerase II transcription complex assembly [83,87]. TAF1 also phosphorylates tumor suppressor p53 on Thr-55, resulting in Mdm2-mediated p53 degradation and progression through G1 of the cell cycle [88]. Interaction between retinoblastoma protein Rb and TAF1 inhibits its kinase activity, which suggests that TAF1 may play a role in tumor suppression [89]. Mutations to the TAF1 gene cause X-linked dystonia parkinsonism and X-linked syndromic intellectual development disorder-33 [90,91]; however, the specific mechanisms by which TAF1 mutations drive these diseases are unknown.

MHC class-II transactivator (CIITA) is a Ser/Thr protein kinase and histone acetyltransferase that functions as a transcriptional coactivator and as the master regulator of major histone compatibility complex (MHC) class-II gene expression [92]. CIITA is a functional homolog of TAF1 and uses multiple mechanisms to activate transcription initiation and elongation [93,94]. CIITA phosphorylates the C-terminal of TAF7, a component of the TFIID promoter complex, and Ser-36 of histone H2B [95]. Its acetyltransferase domain is required for de novo transcription of MHC class-II genes and enhanced transcription of MHC class-I genes [96]. Autophosphorylation of CIITA markedly increases its acetyltransferase activity, which suggests that the acetyltransferase domain is regulated by CIITA’s intrinsic kinase activity [95].

CIITA is the only documented transcription factor in the nucleotide-binding oligomerization (NOD) family and contains a conserved tripartite structure consisting of a variable N-terminal effector-binding domain (EBD), a central NOD domain, and a C-terminal region with a variable number of leucine-rich repeats [93]. Its N-terminal acidic domain for transcriptional transactivation contains its histone acetyltransferase domain (HAT), followed by a proline-, serine-, and threonine-rich domain (P/S/T domain) with multiple phosphorylation sites [97]. The central nucleotide-binding domain (NACHT domain) acts as a kinase domain, and in conjunction with the C-terminal leucine-rich repeats, it affects self-association and nuclear import [98,99]. Mutations in the CIITA gene cause severe immunodeficiency syndrome, also known as bare lymphocyte syndrome [100].

The p53-related protein kinase (PRPK) and its homologs Bud32/Kae1 are Ser/Thr protein kinases and ATPases present in eukaryotes including yeast, humans, and pathogenic fungi including Candida albicans, Coccidioides immitis, and Cryptococcus neoformans. PRPK is one component of the highly conserved EKC/KEOPS complex (endopeptidase-like and kinase associated to transcribed chromatin/kinase, endopeptidase, and other proteins of small size). The EKC/KEOPS complex is required for the universal threonyl carbamoyl adenosine (t6a) tRNA modification, which is found in all tRNAs that pair with ANN codons where N is any of the four bases [101]. The EKC/KEOPS complex and t6a modification are critical for life, with remarkable sequence conservation from archaea to mammals [102,103].

PRPK maintains the conserved human kinase domain architecture including the regulatory and catalytic spines, a DFG motif, and a salt bridge between the invariant lysine in β3 and the invariant glutamate in helix αC [104]. There is high functional conservation between PRPK and its yeast homolog Bud32 [105]. Bud32′s regulatory partner Kae1 (OSGEP in humans) switches its intrinsic kinase activity to ATPase activity, which is required for EKC/KEOPS complex function [106]. Kae1 binds to the C-terminal lobe of Bud32 while Cgi121 (TPRKB in humans) binds to the N-terminal lobe, with the catalytic residues Lys-52, Asp-161, and Asp-182 interacting between the two lobes [107]. Based on the similarity between Bud32 and the Rio2 atypical kinase, it is hypothesized that its ATPase activity is required for the dissociation of tRNA from the complex.

Mutations in PRPK and the EKC/KEOPS complex cause Galloway-Mowat syndrome, an autosomal recessive disease characterized by a combination of early-onset nephrotic syndrome and microcephaly with brain anomalies [108]. PRPK is phosphorylated at Ser-250 by AKT [109] and is a promising therapeutic target in colon adenocarcinomas, cutaneous squamous carcinomas [108], and multiple myeloma [110].