CDK5 is not known to be mutated in cancer. However, unlike neurodegenerative diseases, where CDK5 deregulation is predominantly believed to be due to p25 formation, in cancers, it is mainly due to CDK5 and/or p35 (and sometimes p39) overexpression/amplification, although in medullary thyroid carcinoma (MTC), high p25 levels have been reported [34] (Table 1). Accordingly, in breast cancer, prostate cancer, lung cancer, colorectal cancer, melanoma, pituitary adenoma, leukemia, hepatocellular carcinoma (HCC), gliomas and MTC, both CDK5 and p35 levels are increased (Table 1). P35 is transcriptionally upregulated in response to stimuli such as bFGF, IFN-γ, 1,25-dihydroxyvitamin D3, tumor necrosis factor (TNF)-α and TGFβ. In pancreatic tumors, CDK5, p35 and p39 are overexpressed in more than 90%, 94%, and 75%, respectively, predominantly due to genomic amplification [35]. CDK5 and p35 are also highly expressed in pancreatic neuroendocrine tumors [36]. In ovarian cancer, multiple myeloma (MM), and head and neck cancer, only CDK5 upregulation has been reported. In most of the aforementioned cancers, increased CDK5 and/or activator expression matches with poor prognosis, lymph node metastasis, and overall poor survival, while low CDK5 levels correlate with metastasis-free disease [37]. A single-nucleotide polymorphism (SNP) in the CDK5 promoter region is linked to aggressive prostate cancer in the African-American population [38] and to lung cancer in the Korean population [39].

Table 1 Expression of CDK5 and its activators in human cancers

Importantly, in gastric cancer, CDK5 levels are much lower than those in normal gastric tissues, which correlates with decreased patient survival and metastases [40, 41]. This unique function of CDK5 as a tumor suppressor has been attributed to its nuclear localization in these tissues.

CDK5 acts as an oncogene in breast cancer

CDK5 and p35 are highly expressed in breast cancer tissues and positively correlate with tumor progression and poor prognosis [42, 43] (Table 1). The mechanism of their upregulation is largely unclear, except for one study demonstrating that transforming growth factor β1 (TGFβ1) stimulation increases their mRNA levels [42]. At the molecular level, active CDK5 facilitates cell migration, metastasis and epithelial to mesenchymal transition (EMT) via actin remodeling, and promotes tumorigenesis by inhibiting apoptotic pathways. TGFβ1 treatment in breast cancer cells triggers active CDK5/p35 to phosphorylate Focal Adhesion Kinase (FAK) at S732, causing metastatic invasion and EMT by potentiating F-actin bundle formation [42] (Fig. 1A). Similarly, Su et al. showed that EGF treatment of MDA-MB-231 cells triggered CDK5-mediated phosphorylation of Adducin-1 (ADD1) at T724, which reduced its ability to bind F-actin (Table 2). ADD1 bundles and caps F-actin barbed ends. Thus, CDK5-mediated phosphorylation exposed the barbed ends, causing elongation and reorganization of the actin cytoskeleton, resulting in cell migration and invasion (Fig. 1A) [44]. Interestingly, the exact mechanism by which EGF activates CDK5 was not explored in this study.

Fig. 1

(A) CDK5 promotes aggressive oncogenic phenotypes in breast cancer by several mechanisms. The addition of TGF-β1 activates CDK5, which phosphorylates the FAK protein at S732, resulting in changes in the actin cytoskeleton and subsequently leading to EMT and breast cancer progression. CDK5 also promotes the migration of breast cancer cells by directly phosphorylating talin (S425) and PIPKIγ90 (S453). CDK5 is also activated upon EGF stimulation, leading to adducin 1 (ADD1) phosphorylation at T724. This was followed by dynamic remodeling of the actin cytoskeleton, promoting cell migration and invasion in breast cancer. CDK5 phosphorylates PPARγ at S273, which releases ESRP1. ESRP1 is stabilized, and PPARγ is self-degraded. ESRP1 stabilization switches CD44s to the CD44v isoform, which promotes metastasis and stemness. Green and red circles show activating and inactivating phosphorylation events, respectively. (B) CDK5 promotes aggressive oncogenic phenotypes in breast cancer by inhibiting apoptosis. CDK5 depletion in breast cancer cells triggers mitochondrial permeability transition pore (mPTP) opening followed by mitochondrial depolarization and an increase in ROS production, which leads to the activation of caspases and cell death. Opening of the mPTP and mitochondrial depolarization also cause calcium release and activation of calcineurin, which causes dephosphorylation of dynamin-related protein 1 (DRP1) at S637, resulting in mitochondrial fragmentation and ultimately cell death. CDK5 knockdown also induces apoptosis in breast tumorospheres by increasing the proapoptotic protein Bim. CDK5 inhibits FOXO1 by phosphorylating it at the S249 site, favoring its nuclear export and inhibiting its transcriptional activity. In the absence of CDK5, nuclear FOXO1 thereafter induces the expression of downstream proapoptotic genes such as Bim, leading to apoptosis

Table 2 Direct CDK5 substrates in various cancer types

CDK5 also promotes chemotaxis in breast cancer cells by phosphorylating Talin’s head domain. Calpain cleaves talin, creating a globular head domain and a C-terminal rod domain. The talin head domain binds to Smurf1, an E3 ubiquitin ligase, and is degraded (Fig. 1A). CDK5-mediated phosphorylation of talin at S425 prevents its binding to Smurf1, inhibiting its degradation (Table 2). The talin head domain limits focal adhesion turnover, stabilizing lamellipodia and leading to cell migration [45]. CDK5 phosphorylates phosphatidylinositol 4-phosphate 5-kinase type I γ90 (PIPKIγ90) at S453 (Table 2), which regulates PIPKIγ90 activity to control fibronectin secretion and, consequently, cell invasion [46]. CDK5 mediates stemness in triple-negative breast cancer (TNBC) by phosphorylating Peroxisome proliferator-activated receptor gamma (PPARγ) at S273 (Table 2), which curbs its E3 ligase activity and releases epithelial splicing regulatory protein 1 (ESRP1). As a result, ESRP1 is stabilized, and PPARγ is degraded. ESRP1 promotes CD44s isoform switching to CD44v, which causes stemness transformation and metastasis [43] (Fig. 1A).

CDK5 likewise promotes tumorigenesis by inhibiting cell death pathways. Loss of CDK5 in breast cancer cells opens up the mitochondrial permeability transition pore (mPTP), initiating increased mitochondrial depolarization and higher ROS levels, with subsequent activation of caspases and cell death (Fig. 1B). This is in contrast with AD, where CDK5 hyperactivation results in mitochondrial dysfunction resulting in neuronal death [47]. In tandem, mitochondrial depolarization also promotes calcium release, causing calcineurin activation, which dephosphorylates dynamin-related protein 1 (DRP1) at S637. DRP1 dephosphorylation triggers mitochondrial fragmentation, resulting in cell death [48]. Similarly, CDK5 inhibition or knockdown induces apoptosis in tumor spheres by stabilizing the transcription factor FOXO1. This results in increased levels of the proapoptotic protein Bim1 and decreased in vivo tumor volume [49] (Fig. 1B).

Role of CDK5 in prostate cancer (PCa)

Similar to breast cancer cells, prostate cancer cells express high levels of CDK5 and p35 (Table 1) [50, 51]. Thus, CDK5 is frequently deregulated in prostate cancer, and its levels strongly correlate with poor clinical prognosis. CDK5 promotes cell proliferation and invasion in PCa.

Androgen receptor (AR) is the key driver in PCa pathogenesis. In healthy prostatic epithelial tissues, AR plays an essential role in regulating terminal differentiation, apoptosis suppression and hormone secretion [52]. However, AR signaling is highly deregulated in PCa and facilitates cell proliferation, survival, and invasion during PCa development [53]. CDK5 promotes prostate cancer growth by both activating and stabilizing AR either directly or indirectly [50, 54, 55]. Hsu et al. demonstrated that CDK5 phosphorylates AR at S81 in vitro and in vivo [50] (Table 2). This stabilizes AR, causing enhanced nuclear translocation and activation. Furthermore, AR protein levels showed a significant correlation with CDK5 or p35 in 177 AR-positive PCa patients, underscoring the clinical significance of the CDK5-AR axis in PCa progression [50]. CDK5 also indirectly stabilizes and activates AR by directly phosphorylating STAT3 at S727 in cells and xenograft tumors [55]. Phosphorylated STAT3 interacts with AR, which upregulates AR protein stability and transactivation, enhancing cell proliferation. The authors further provided clinical evidence that the level of p-Ser727-STAT3 significantly correlated with Gleason score and CDK5, p35 and AR levels [55] (Fig. 2A).

Fig. 2

(A) CDK5 facilitates highly oncogenic phenotypes in prostate cancer. CDK5 can simultaneously affect numerous targets and promote PCa progression. CDK5/p35 overexpression in prostate tumor cells increases the phosphorylation of talin1 at S425, which induces a conformational change resulting in talin1 activation. Phosphorylated talin1 then binds to the cytoplasmic tail of β1 integrin, leading to β1 integrin activation and downstream integrin signaling. This leads to increased cell survival, adhesion, motility and metastatic potential of PCa cells. CDK5 also activates STAT3 and AR proteins by phosphorylation at specific sites. Activation of STAT3 or AR causes prostate cancer proliferation. CDK5 indirectly activates AKT resulting in p53 and p21CIP downregulation and cell cycle progression. CDK5 directly phosphorylates p21CIP at S130, which degrades it, promoting oncogenesis. Enzalutamide treatment upregulates GR signaling, which transcriptionally increases MYCN levels. MYCN increases p35, p39 and E2F1 transcription. p35 and p39 binding activates CDK5, which phosphorylates Rb at S807 and S811, releasing E2F1, which in turn transcriptionally upregulates the NEPC genes, leading to neuroendocrine differentiation. Green and red circles show activating and inactivating phosphorylation events, respectively. Green and red arrows represent activating and inactivating pathways, respectively. (B) CDK5 signaling in pancreatic cancer. K-Ras activates CDK5 by promoting the cleavage of p35 to p25, although the exact mechanism is not known. CDK5 induces cell migration and invasion through activation of RalA and RalB in pancreatic cancer. CDK5 directly phosphorylates EZH2 at T261, causing its ubiquitylation by FBW7, which in turn inhibits pancreatic cancer cell migration and invasion

In contrast, Lindqvist et al. demonstrated that CDK5 phosphorylates AR at S308; however, the consequences of this phosphorylation event on AR stability or transactivation were not analyzed [54]. Nevertheless, this study confirmed that CDK5 knockdown indeed decreases AR stability. The authors further showed that CDK5’s major growth-promoting activity stemmed from its activation of the AKT pathway, although the exact mechanism leading to AKT phosphorylation at S473 remained unclear [54]. AKT activation downregulated p21CIP1 and p53 proteins, allowing cell cycle progression. Huang et al. further explored the link between CDK5 and p21CIP1 and showed that CDK5 directly phosphorylates p21CIP1 at S130, which degrades it, thereby releasing CDK2, which promotes oncogenesis [56] (Fig. 2A).

Similar to breast cancer, CDK5 activity is required to control cell motility and the metastatic potential of prostate cancer cells [51]. CDK5 knockdown or pharmacological inhibition resulted in cytoskeletal remodeling and loss of motility and invasiveness in the highly metastatic AT6.3 prostate cancer cell line. Similarly, dominant-negative CDK5-expressing xenografts revealed significantly fewer metastases than controls [51]. Talin1 was also shown to be phosphorylated by CDK5 in prostate cancer cells, which activates β1 integrin, promoting invasion [57] (Fig. 2A).

A recent study revealed CDK5 as a key player that promotes neuroendocrine phenotypes in a prostate cancer patient–derived xenograft (PDX) treatment model upon ADT therapy [58]. Accordingly, enzalutamide treatment in the PDX133-4 model upregulated glucocorticoid receptor (GR) signaling, which transcriptionally increased MYCN levels. MYCN upregulated p35, p39 and E2F1 transcription. P35 and p39 activated CDK5, which directly phosphorylated Rb at S807 and S811, releasing E2F1, which in turn transcriptionally upregulated the NEPC genes synaptophysin (SYP), chromogranin A (CHGA) and neuron-specific enolase (NSE), leading to neuroendocrine differentiation (Fig. 2A) [58]. This study revealed CDK5 as a potential target for NEPC.

CDK5 acts as an oncogene in pancreatic cancer

CDK5, p35 and p39 are overexpressed in more than 90%, 94%, and 75% of pancreatic ductal adenocarcinoma (PDAC) tumors, respectively, predominantly due to genomic amplification (Table 1) [35]. As noted before, CDK5 and p35 are also abundantly expressed in pancreatic neuroendocrine tumors [36]. Furthermore, CDK5 is also hyperactivated in pancreatic cancer due to mutant K-Ras. K-Ras activating mutations are hallmarks of pancreatic cancer. K-Ras hyperactivates CDK5 by cleaving p35 into p25, promoting increased cell migration and invasion [35]. However, the exact mechanism by which K-Ras promotes p25 formation remains unknown. CDK5 ablation inhibited orthotopic tumor formation and systemic metastasis in vivo [59] (Fig. 2B).

Interestingly, in pancreatic cancer, Ral guanine nucleotide-exchange factors (Ral GEFs) have emerged as the key effectors of the K-Ras pathway instead of the B-Raf, extracellular signal-regulated kinase (ERK) or MAPK pathways. As Ral GEFs activate RalA and RalB small G proteins, Lim et al. demonstrated that RalA is critical for Ras-driven tumor initiation and that RalB is crucial for Ras-driven tumor metastasis [60]. As CDK5 is also critical for pancreatic cancer progression, Feldmann et al. investigated a potential link between CDK5 and RalA/B GTPases [59]. Using KRAS2 mutants, they reported that the loss of RalA function inhibits the tumorigenicity of pancreatic cancer cells, which is dependent on CDK5. Accordingly, dominant-negative CDK5-expressing cells exhibited decreased levels of active RalA-GTP and RalB-GTP. Furthermore, constitutively active forms of RalA or RalB could overcome the effect of dominant-negative CDK5, suggesting that Ral proteins are downstream of CDK5. Although the exact molecular mechanism was not delineated, CDK5 may regulate the activation of RalA and RalB by controlling the function of either RalGEFs or RalGAP, suggesting CDK5 as a possible target for pancreatic cancer [59] (Fig. 2B).

In pancreatic cancer, active CDK5 phosphorylates enhancer of zeste homolog 2 (EZH2), which triggers its degradation via FBW7, a component of ubiquitin ligase (Table 2). As EZH2 degradation inhibits tumor migration and invasion [61], CDK5 acts as a tumor suppressor in pancreatic cancer (Fig. 2B). Nevertheless, as many studies have emphatically shown that CDK5 is critical for pancreatic cancer progression, the clinical significance of CDK5-mediated EZH2 degradation needs to be explored further.

CDK5 promotes medullary thyroid cancer (MTC)

MTC, although slow-growing, is a highly deadly cancer, partly because it is often diagnosed at an advanced metastatic stage. While thyroid cancers originate from thyroid follicular cells, which produce thyroid hormone, MTC stems from neuroendocrine parafollicular C cells (aka C cells), which synthesize calcitonin. MTC accounts for ~ 1–2% of all thyroid cancers and can be sporadic (75% MTC cases) or hereditary (25%). CDK5 and its activators are present in normal human thyroid tissues but overexpressed in hereditary and sporadic MTC clinical specimens (Table 1) [34]. Pozo et al. showed that inducible expression of p25 in thyroid C cells in vivo leads to MTC, while preventing p25 overexpression inhibits tumor growth [34]. Similarly, targeting CDK5 inhibits sporadic MTC patient-derived cell proliferation. Mechanistically, they showed that active CDK5 phosphorylates retinoblastoma (Rb) protein at S807/811, leading to the expression of CDK2 and cyclin A, which results in cell proliferation (Table 2). Thus, MTC cells overexpressing dominant negative, kinase-dead CDK5 or subjected to p35 knockdown prevented Rb phosphorylation and decreased CDK2 and Cyclin A expression, causing cell cycle arrest (Fig. 3A) [34].

Fig. 3

(A) CDK5 signaling in MTC. Upon Rb phosphorylation by CDK5, E2F is released and activates the transcription of target genes, including CDK2 and Cyclin A, that mediate cell proliferation. CDK2-Cyclin A further phosphorylates Rb in a positive feedback loop. HER2 signaling activates CDK5 by an unknown mechanism, which in turn phosphorylates STAT3 at S727, promoting cell proliferation and tumorigenesis through cFos-JunB signaling. GDNF activates RET signaling leading to ERK-EGR1 activation, which increases p35 transcription, increasing CDK5 activation. Active CDK5 phosphorylates STAT3 at S727, promoting tumorigenesis. Green circles show activating phosphorylation events. (B) CDK5-mediated signaling pathways in gastric and colorectal cancer. In gastric cancer, nuclear CDK5 overexpression inhibits the proliferation and metastasis of gastric cancer cells. Nuclear CDK5 upregulates the CDK inhibitor p16INK4a, which inhibits the S-G2 phase transition, leading to cell cycle arrest. CDK5 binds with p27, which results in its nuclear translocation. CDK5 also binds PP2A, which inhibits metastasis; however, it is unknown whether this event occurs in the cytoplasm and/or nucleus. (C) CDK5 modulates the ERK5–AP-1 axis to regulate the oncogenic pathway in colorectal cancer. CDK5 directly interacts with ERK5 and phosphorylates it at T732, thus upregulating the expression of AP-1 and some of its target genes (VEGFA, MMP1 and c-myc). This event results in the malignant development of human CRC both in vitro and in vivo

As observed in prostate cancer, CDK5 also phosphorylates STAT3 at S727 in MTC cells. However, contrary to PCa, in which phospho-STAT stabilized AR causing proliferation, in MTC, phospho-STAT3 led to increased cell proliferation and tumor formation via the downstream genes c-FOS and JUNB. Accordingly, phospho-dead STAT3 (S727A) prevented p35-induced human TT and mouse MTC-M cell proliferation (Fig. 3A) [62]. Subsequent studies revealed the mechanism of CDK5 activation in MTC cells. Upon GDNF stimulation, CDK5 physically interacts with Rearranged-during-transfection (RET) kinase, which activates CDK5 by increasing p35 expression via the ERK1/2-EGR1 pathway [63]. Active CDK5 phosphorylates STAT3 at S727 to promote human medullary cancer cell growth. As activating germline mutations of RET are observed in 100% hereditary and ~ 40% sporadic MTC, CDK5 activation by RET could have profound clinical consequences (Fig. 3A).

Role of CDK5 in gastric cancer- nuclear CDK5 acts as a Tumor suppressor

Unlike most other cancers, CDK5 acts as a tumor suppressor in gastric cancer. Consequently, CDK5 expression levels are significantly lower in the majority of gastric tumors than in normal gastric tissues, which correlates with decreased patient survival and metastases (Table 1) [40, 41, 64]. Equally importantly, CDK5’s tumor-suppressive functions are mainly attributed to its nuclear localization in these tissues. Cao et al. demonstrated that CDK5 is both nuclear and cytoplasmic in non-tumor tissues. However, it is completely excluded from the nucleus of tumor cells, indicating that nuclear CDK5 acts as anti-oncogene [40]. CDK5 has no intrinsic nuclear localization signal but relies on its binding to p27 using its N-terminal residues (N-17) to translocate to the nucleus [41]. Accordingly, patients with lower levels of CDK5 and p27 display poorer survival compared to patients with either high CDK5 or high p27 or both [41].

Interestingly, CDK5 possesses two nuclear export signals (NES) that bind with the nuclear export mediator CRM-1 and result in cytoplasmic residence of CDK5 [64]. Cao et al. showed that the small molecule NS-0011 increases CDK5 accumulation in the nucleus by disrupting its binding with CRM-1, which in turn suppresses both cancer cell proliferation and xenograft tumorigenesis [40]. Similarly, exogenous expression of nuclear CDK5 inhibits the proliferation of tumor cells and xenografts in nude mice. Overall, these results underscore that CDK5’s nuclear residence drives its function as a tumor suppressor. However, the molecular mechanisms by which nuclear CDK5 acts as a tumor suppressor are largely unknown in gastric cancer, except for one study that showed that nuclear CDK5 upregulates the levels of the CDK inhibitor p16INK4a, which contributes to cell cycle arrest [40] (Fig. 3B).

Lu et al. further revealed that downregulation of CDK5 promotes metastasis in gastric cancer cells, which depends on its interaction with PP2A, a serine/threonine phosphatase [65]; however, this study did not analyze whether PP2A binds cytoplasmic and/or nuclear CDK5. Similarly, the impact of PP2A binding on CDK5 activity or levels was not examined. Nevertheless, similar to CDK5, PP2A is also downregulated in gastric cancer, which is associated with poorer overall survival [65]. Together, these studies demonstrate that CDK5 acts as a tumor suppressor in gastric cancer, likely due to its nuclear localization.

Role of CDK5 in Colorectal cancer (CRC)

CDK5 and p35 are broadly expressed in human colorectal cancer (CRC) cell lines and human CRC tissues compared to paired normal tissues and cell lines (Table 1) [66, 67]. Higher levels of CDK5 in CRC correlate with advanced disease stage, poor differentiation, increased tumor size and poor prognosis [67]. Likewise, CDK5 kinase activity is also higher in aggressive cell lines such as HCT116 and SW480 compared to less aggressive cell lines such as Caco-2 and Lovo [66]. Similar to other cancers, CDK5 enhances both proliferation and metastasis in CRC [66, 67]. At the molecular level, CDK5 upr