The Role of CDKs and CDKIs in Murine Development

Cyclin-dependent kinases (CDKs) and their inhibitors (CDKIs) play pivotal roles in the regulation of the cell cycle. As a result of these functions, it may be extrapolated that they are essential for appropriate embryonic development. The twenty known mouse CDKs and eight CDKIs have been studied to varying degrees in the developing mouse, but only a handful of CDKs and a single CDKI have been shown to be absolutely required for murine embryonic development. What has become apparent, as more studies have shone light on these family members, is that in addition to their primary functional role in regulating the cell cycle, many of these genes are also controlling specific cell fates by directing differentiation in various tissues. Here we review the extensive mouse models that have been generated to study the functions of CDKs and CDKIs, and discuss their varying roles in murine embryonic development, with a particular focus on the brain, pancreas and fertility.


Introduction
Cyclin-dependent kinases (CDKs) are proteins that, by definition, require the binding of partner cyclin proteins in order to phosphorylate a series of target proteins. These complexes are required for appropriate progression of the cell cycle and, as such, are essential for development. CDK inhibitors (CDKIs) are a class of proteins that preferentially bind to and inhibit the activity of CDK/cyclin complexes and are thus involved in stopping or delaying the progression of the cell cycle. The roles and functional redundancies of CDKs and CDKIs in regulating the cell cycle have been extensively reviewed previously [1][2][3]. The structures of these proteins, including potential sites for interactions with inhibitors, have also been extensively studied and reviewed [4][5][6][7][8].
Twenty CDKs have been described in the mammalian genome, derived from a single ancestral gene in yeast, and many of these have been shown to be expressed during embryonic development [9,10]. Likewise, there are two major families of CDKIs: the CIP/KIP family, encoded by the Cdkn1 gene family comprising p21 CIP1 , p27 KIP1 and p57 KIP2 ; and the INK4 family, encoded by the Cdkn2 gene family comprising p16 INK4a , p15 INK4b , p18 INK4c and p19 INK4d . Many CDKIs have been shown to play key roles in embryonic development [11][12][13]. In addition to their shared ability to modulate the cell cycle, several of these CDKs and CDKIs have been shown to direct differentiation of a number of different cell types with a particular convergence in the brain, pancreas and gonadal ridges, which will be discussed in greater detail later on.
For many of these proteins, their functions have been studied primarily through knock-out (KO) mouse models and cell lines. These are mice or cells with one or both alleles of the gene in question constitutively deleted from the genome with any phenotypic abnormalities ostensibly due to the absence of the specified protein. In the case of CDKs and CDKIs, family members often display redundant functions, and as a result, many of the expected phenotypes that might be caused by the Figure 1. Schematic of principal areas of CDK and CDKI involvement during murine development, as determined by KO and cKO mouse models. The ears, brain, pituitary gland, eyes, spine, lungs, heart, liver, pancreas, kidneys, spleen, intestine and testes/ovaries are colour-coded based on the Figure 1. Schematic of principal areas of CDK and CDKI involvement during murine development, as determined by KO and cKO mouse models. The ears, brain, pituitary gland, eyes, spine, lungs, heart, liver, pancreas, kidneys, spleen, intestine and testes/ovaries are colour-coded based on the presence of an altered phenotype upon KO of a CDK or CDKI, depicted here in an E17.5 mouse embryo. The area of each colour is not reflective of the magnitude of the effect on the mouse phenotype. KOs of Cdk1, Cdk4, Cdk6, Cdk11, Cdk13, Cdkn1b, Cdkn1c and Cdkn2c affect overall embryonic development. Table 1. Phenotypes of CDK (cyclin-dependent kinase) KO (knock-out) and cKO (conditional KO) in mice.

Cdk1
Whole Body Mutant line with truncated protein-homozygous early embryonic lethal [21] Whole Body KO-lethal at the 1-cell stage due to inability to complete cell division [22] Testes cKO in spermatocytes-cells arrest at prometaphase resulting in male infertility [23] Ovaries cKO in oocytes-females infertile due to failure to resume meiosis. Rescued with microinjection of Cdk1 [24] Brain cKO in Nex (postmitotic neurons of cortex and hippocampus)-normal brain development [25] Liver cKO in liver-failure to complete cell division but normal phenotype due to compensatory cell growth [22] Cdk2 Whole Body/ Testes/Ovaries KO-develop normally but both sexes sterile. Spermatocytes and oocytes arrest at prophase I, oocytes slightly later than spermatocytes [26] Whole Body/ Testes/Ovaries KO-normal development but both sexes infertile [27] Ovaries cKO in oocytes-develop normally [24] Cdk3 Whole Body Null point mutation-normal development. Possible functional redundancy [28] Cdk4 Whole Body/ Testes/Ovaries/ Endocrine System/ Pancreas KO-reduced mutant numbers in heterozygous cross, suggesting some embryonic lethality. Decreased body size. Defective spermatogenesis reducing fertility. Females infertile due to deficit in hypothalamic-pituitary axis. Mice are insulin-deficient diabetic due to affected β-cells. [29] Whole Body/ Pancreas/ Testes/Ovaries KO-mice are undersized and develop diabetes as above. Males show testicular atrophy due to defective proliferation of gonocytes and hypoplastic seminiferous tubules. Female infertility due to perturbed corpus luteum formation [30] Whole Body/ Endocrine System KO-decreased body size with small pituitaries, with particularly reduced lactotrophs resulting in female infertility [31]

Cdk8
Whole Body/ Intestine KO-normal growth, but increased size and growth rate of intestinal tumours [42] Pancreas cKO in β-cells-improved glucose tolerance due to increased insulin secretion. Increased cell death following stress conditions [43] Cdk9 Whole Body KO-complete embryonic lethality from day E9.5 *

Cdk13
Whole Body/ Lung/Liver/ Kidney/Heart Null mutant-developmental delay, growth retardation, underdeveloped lungs, liver and kidneys. Embryo lethal due to heart defect [45]  Blank rows indicate a lack of experimental data for the indicated gene, to the best of our knowledge. * Mouse model is available from the International Mouse Phenotyping Consortium, but a referenced embryonic study could not be found. Brain KO-increased neural migration and differentiation in response to brain injury [48] Intestine KO-increased susceptibility to intestinal tumours [49] Immune System KO-no altered susceptibility to retrovirally induced lymphatic tumours [50] Immune System cKO in T cells-hyposensitive to programmed cell death in response to radiation-induced DNA damage [51]  Whole Body/Endocrine System/Ovaries KO-oversized with larger internal organs, especially the pituitary, and female sterility due to impairment of the formation of corpora lutea [52] Whole Body/ Endocrine System/Ovaries KO-oversized due to increased cell number and proliferation. Females infertile due to impaired luteal cell differentiation reflecting a disturbance in the hypothalamic-pituitary-ovarian axis [53] Whole Body/ Immune System/ Endocrine System/Eyes/ Ovaries KO-oversized, especially in thymus, pituitary and adrenal glands and gonadal organs. Disturbed organization of retinas. Female sterility due to impaired development of ovarian follicles [54] Endocrine System/Intestine/ Lung KO-increased susceptibility to pituitary and intestinal tumours, lesions of the female reproductive tract and lung adenomas following exposure to carcinogens [55] Heart/Ovaries KO-protective delay of ageing and apoptosis of heart due to ovariectomy by upregulating antioxidant enzymes [56] Testes KO-decreased testosterone production, despite increased proliferation of testosterone-producing cells [57] Brain KO-altered neuron migration and differentiation [58] Brain KO-increased neural progenitor cell proliferation under basal and injury conditions [59] Brain KO-increased cell proliferation of lower layer neurons [60] Immune System KO-increased susceptibility to retrovirally induced lymphatic tumours. Increased spleen proliferation [50] Immune System KO-developed spontaneous T cell lymphomas [61] Eyes KO-increased cell proliferation and poor repair response after photoreceptor damage [62]  Whole Body/ Brain KO-oversized embryo and macrocephaly, with increased cell proliferation, particularly of neuronal precursors and lower layer neurons. Decreased cell cycle length and increased cell cycle exit at days E14.5 and E16.5 [60] Kidney KO-normal kidney differentiation and development [66] Whole Body/ Spine cKO maternally-skeletal deformities, reduced body size and perinatal lethality [67] Whole Body cKO maternally-perinatal lethality, overgrowth during early gestation and weaning, but constrained growth in late gestation [68] Brain cKO paternally-decreased brain size and reduced number of neural progenitor cells during development [69]  Brain KO-increased susceptibility to CNS tumours and hydrocephaly following carcinogenic exposure [71] Immune System KO-normal development, but with thymic hyperplasia [72] Heart KO-increased cardiomyocyte proliferation and cardiac repair [73] Kidney KO-increased kidney cell proliferation and decreased apoptosis in response to ischemia-reperfusion injury [74] Lung KO-reduced inflammation in response to cigarette smoke, but no change to proinflammatory cytokine response [75] Lung cKO in lung epithelia-reduction in inflammation response to cigarette smoke and suppression of proinflammatory cytokines, but no protective effect [75] Lung cKO in pleura-no development of malignant mesotheliomas [76] Spine cKO in intervertebral disc-no change in senescence, reduced apoptosis and altered matrix homeostasis [77] Cdkn2b p15 INK4b Whole Body/ Testes/Ovaries/ Immune System KO-viable and fertile with no morphological or behavioural abnormalities at birth.
After two months of age, mice had atypical haematopoiesis and increased lymphocytes in the spleen [78] Cdkn2c p18 INK4c Whole Body/ Endocrine System/ Immune System KO-gigantism and organomegaly, particularly of the pituitary, spleen and thymus, due to altered cell cycle control. Altered cell cycle entry of resting B cells [79] Whole Body/ Testes/Ovaries/ Immune System/ Endocrine System KO-viable and fertile with no morphological or behavioural abnormalities at birth. From six months of age mice developed splenomegaly, enlarged lymph nodes and increased incidence of pituitary tumours [78] Endocrine System/Immune System KO-spontaneous pituitary and lymphatic tumours. Also observed in heterozygous mice, implying haploinsufficiency [80]  Whole Body/ Testes KO-normal growth with no obvious malformations. Decreased, but not total loss of, male fertility due to testicular atrophy and increased apoptosis of germ cells [81] Immune System KO-increase in mean ploidy level in bone marrow [82] Ears KO-aberrant cell cycle entry and subsequent apoptosis in sensory hairs leading to hearing loss [83] Cdkn3 Brain Exon deletion-decreased grip strength * * Mouse model is available from the International Mouse Phenotyping Consortium, but a referenced study could not be found.

Absolute Requirement of Cdk1
Cdk1 is essential for successful completion of the cell cycle [18,21], preventing the creation of constitutive Cdk1 −/− cell lines or mouse models. Attempts to create Cdk1 −/− offspring result in immediate failure following fertilisation, due to the inability to successfully complete cell division. This requirement cannot be compensated for by other members of the CDK family [84] and is even observed in a mutant line with cassette insertion resulting in heterozygous cells expressing only 50% of wild-type levels of Cdk1, which also results in early embryonic lethality [22]. It is therefore perhaps surprising that such a requirement for successful completion of the cell cycle is not observed in any of the other CDK family members.
Cdk1 is also indispensable for cell cycle completion during gamete development [85]. It is essential for the formation of sperm, as Cdk1 cKO spermatocytes result in cell arrest at prometaphase and thus male infertility [23]. Additionally, Cdk1 cKO oocytes lead to female infertility as the oocytes are unable to resume meiosis, with a microinjection of Cdk1 rescuing the phenotype, thus confirming the necessity of Cdk1 in mouse female fertility [24]. Cdk1 maintains phosphorylation of protein phosphatase 1 and lamin A/C for meiosis resumption in the oocytes [24], and is necessary as one of the factors required to phosphorylate eIF4E-BP1, a protein involved in facilitating spindle formation and stability during meiosis in oocytes [86]. Cdk1 phosphorylation is believed to be particularly required for resumption of meiosis in oocytes, with a natural increase in Cdk1 levels in later life resulting in speeding up meiosis I, thus potentially leading to errors such as aneuploidy in later-life pregnancies [87]. This can be ameliorated by inducing excess cyclin B1, as exploitation of preferential destruction of unbound B1 results in sustained Cdk1 levels, prolonging the cell cycle, and thus reducing aneuploidy errors [88]. Other members of the CDK family, in particular Cdk2, are also required for fertility, as discussed below.

Embryo
We would predict that loss of expression of any of the CDK family members would result in impaired embryonic growth, due to disruption in the normal pathways of the cell cycle. Indeed, this trend is largely followed, with Cdk4 and Cdk6 null animals displaying reduced embryonic growth [27,29,31,41,89], and Cdk11 −/− embryos failing to progress beyond day E4.5 [44]. While Cdk4 is not absolutely required for embryonic viability, Cdk4 −/− animals are born at a reduced frequency from heterozygous matings, suggesting partial embryonic lethality [29]. Cdk3, however, appears to be functionally redundant, as a null point mutation was not found to affect embryonic development [28].
While single KOs of Cdk2, Cdk4 and Cdk6 are viable, this may be due to functional redundancy as tKO mice without Cdk2, Cdk4 and Cdk6 begin dying at day E13.5 of gestation, and have malformed livers and hearts exhibiting thinner ventricular walls than wild-type mice [21]. It is thought that loss of Cdk4 in particular is key for the embryonic lethality of the tKO, with Cdk2 and Cdk6 dKO mice being viable but exhibiting a combination of the phenotypes seen in Cdk2 and Cdk6 single KOs, such as reduced body size and haematopoietic defects [41].
A functional relationship between Cdk2 and Cdk4 remains unclear, with one study showing that a Cdk2 and Cdk4 dKO also results in embryonic death at around day E15, due to heart defects [90]. In contrast, another study found that Cdk2 and Cdk4 dKO mice complete embryonic development and are born, slightly underweight, but die shortly after birth, potentially also due to heart failure [91]. Cdk4 KO mice with a Cdk2 ERT cKO however, show similar developmental phenotypes to Cdk4 KO mice [91].
Cdk4 and Cdk6 dKO mice have altered haematopoiesis resulting in severe anaemia, with embryos being undersized and beginning to die from day E14.5, with pups that survive to birth dying soon afterwards [41]. A partial dKO where one allele of Cdk4 is expressed but Cdk6 is knocked-out results in viable offspring, while expression of one allele of Cdk6 and complete Cdk4 KO results in partial embryonic lethality, again suggesting that Cdk4 has the more important role in embryonic viability [41].
In contrast to Cdk2, Cdk4 and Cdk6, which all display some degree of functional redundancy, deletions of Cdk5, Cdk11 and Cdk13 display variable levels of embryonic lethality in isolation [32,44,45].
Cdk5 expression is largely restricted to the brain, with constitutive and neuron-specific deletion of the gene resulting in overlapping phenotypes and perinatal lethality (~60%) [32]. Cdk11 −/− blastocysts are terminated around day E3.5 due to a massive failure in mitosis [44], while Cdk13 −/− embryos fail by day E16.5 [45]. For the remaining CDKs, a number remain untargeted, restricting our knowledge of their potential impact on embryonic growth parameters.
The data generated from CDK mouse models largely corresponds with the presumed function of promoting embryonic growth. Conversely, we would expect that expression of CDKIs will consequently restrict embryonic growth, due to negative regulation of the cell cycle. It is therefore surprising that although control of the cell cycle at G1/S phase was impaired in the absence of Cdkn1a (p21 CIP1 ) [47], most likely through a general reduction in inhibition of CDK activity, the mouse model did not present with altered embryonic growth parameters. This finding is in stark contrast to the other members of the CIP/KIP family of CDKIs. Indeed, Cdkn1b −/− mice, which lack p27 KIP1 , are oversized due to increased cell proliferation, and exhibit proportionally larger internal organs [52][53][54]. This phenotype was not accompanied by reduced embryonic survival. However, the loss of control of the cell cycle did present with postnatal tumour formation [54]. It has been noted that many of the functions of Cdkn1b and Cdkn1c are functionally redundant, with a dKO line displaying greatly enhanced phenotypes of the single KOs [92]. tKO mice lacking all Cdkn1 family members develop normally until mid-gestation, dying around day E13.5, marginally earlier than Cdkn1b/Cdkn1c dKO mice at day E15.5, with both Cdkn1a/Cdkn1b and Cdkn1a/Cdkn1c dKO mice surviving past this point [93]. Knock-in of p27 KIP1 to replace p57 KIP2 results in viable and healthy mice, with only a few developmental defects, suggesting p27 KIP1 and p57 KIP2 have overlapping but not identical developmental roles [94]. Both p27 KIP1 and p57 KIP2 appear to be expressed frequently in postmitotic cells, with the role of p57 KIP2 restricted to organogenesis, as expression is not maintained into adulthood, while p27 KIP1 is expressed both during development and in adult mice [95].
Individually, Cdkn1c must be viewed in a slightly different light to its family members, as this gene undergoes genomic imprinting. The paternally inherited allele of this gene is epigenetically silenced throughout the mouse's life-course, with functionally monoallelic expression from the maternally inherited allele [96]. To the best of our knowledge, none of the other CDKs or CDKIs have been shown to undergo this process. As a result of this, a heterozygous deletion of just the paternal allele (silenced) results in wild type expression levels of the gene, while a deletion of the maternal allele (active) presents as functionally homozygous-null. Independent deletions of this gene have shown near total late perinatal or very early neonatal lethality [64,65,67,97]. The reasons for this lethality remain unclear, as although mice present with a cleft palate, preventing feeding and causing breathing difficulties [63][64][65], this would not explain the large degree of perinatal lethality that is observed. Likely the cleft palate represents one of a number of defects that combine to cause lethality, alongside abdominal wall and heart defects. Due to the role of Cdkn1c in blocking the cell cycle, it would be predicted that null animals would be overgrown. However, no such phenotype is observed at birth [63][64][65]68,98]. Instead, researchers have found a complex picture, where early embryonic overgrowth is lost in the late stages of a pregnancy, due to a restriction in growth potential in part attributable to the size of mouse litters [68]. Partly mirroring these results, increased dosage of Cdkn1c through the use of bacterial artificial chromosomes (BAC) has found an exquisitely dosage-sensitive embryonic growth restriction phenotype [99].
While a plethora of work has examined the roles of the CIP/KIP family in embryonic growth and development, the function of the INK4 family in the embryo remains comparatively untested, with studies predominantly focussed on postnatal consequences and tumour formation following deletion of a family member. Loss of Cdkn2c (p18 INK4c ) results in giantism and organomegaly, with the organomegaly phenotype increased upon dKO with Cdkn1b (p27 KIP1 ) [79]. Both Cdkn2b (p15 INK4a )/Cdkn2c (p18 INK4c ) and Cdkn2c (p18 INK4c )/Cdkn2d (p19 INK4d ) dKO mice are viable with no morphological or behavioural abnormalities at birth [78,100]. While we cannot assume that this is the extent of the roles for the INK4 family in the regulation of embryonic growth, it must be remembered that this family of proteins is specific in its activity towards Cdk4/6 complexes, while the CIP/KIP family displays a more general inhibition of CDK family activity [101], potentially differentiating the families with regards to embryonic requirements.

Brain Development and Function
Functional roles for CDKs in brain development have so far been restricted to just a handful. While it is essential for embryonic development and assumedly in brain development, Cdk1 does not appear to have a postmitotic role in neurons, with a cKO in such cells resulting in normal development [25]. Further testing of a Cdk1 role in different regions of the brain remains unfeasible outside of postmitotic cells.
Cdk5 has been shown to be essential in mouse brain development, with cKOs indicating a variety of functions in various regions of the brain, overcoming the embryonic lethality observed in constitutive KOs. A constitutive KO of Cdk5 results in altered brain development, with more than 60% of developing embryos dying in utero [32]. cKOs of Cdk5 do not result in the same embryonic lethality, but still cause behavioural issues [34][35][36]. The dosage and activity of Cdk5 is tightly regulated in the brain [102]; in particular, p35 and p39 are both essential for Cdk5 activity and may be the principal activators of Cdk5 [103]. Cdk5 is involved in dendritic spine formation, is important in Purkinje cell migration and dendritic growth, and is required for formation of ventral striation; loss of Cdk5 causes GABAergic neuronal death in the developing mouse foetus, as shown in p35 cKO and p39 KO mice [104][105][106]. Cdk5 phosphorylation of the neural protein Synapsin III is needed to regulate neurite outgrowth and axonal elongation [107]. Cdk5 phosphorylation is also required for the membrane targeting of, and promotion of proteasome-dependent degradation of Cx43, a protein involved in neuron migration and embryonic brain development, which is associated with adult anxiety-related behaviour [108].
Loss of Cdk5 specifically in CNP cells affects myelination and results in impaired learning and memory [36,37]. CaMKII Cdk5 cKO, which results in deletion in the mouse forebrain, causes many behavioural abnormalities, seizures, and, intriguingly, increased susceptibility to cocaine [34,35], while absence of Cdk5 in the mid and hind-brain or in Emx1-expressing cells, which are involved in corticogenesis, results only in morphological defects with no reported behavioural changes [38][39][40]. It is interesting to note that despite none of the CDKIs having a specific affinity for Cdk5, it is the only CDK to have overlapping expression patterns in the brain with CDKI family members.
All three members of the CIP/KIP family have described roles in neural development and regulation. Interestingly, while the role of Cdkn1a appears to exclusively revolve around its ability to regulate cell cycle progression [48], Cdkn1b and Cdkn1c both perform this role, and in addition to this, are responsible for directing differentiation of a number of neural cell types [58][59][60]. Cdkn1b and Cdkn1c are expressed transiently in the brain during development [95], with Cdkn1c being expressed in the brain at day E11.5-E13.5 [109]. Cdkn1c has been shown to suppress growth with a constitutive KO resulting in macrocephaly due to an increased number of cortical precursors entering the cell cycle, which combines with a shortened cell cycle length and leads to an increase in cortical surface, thickness and cell numbers [60]. However, this impact has recently been shown to only occur non-cell-autonomously, with neuronal cell-specific KOs identifying a cell-autonomous role of Cdkn1c that promotes growth and cell survival [110]. Mosaic analysis with double markers technology was used to show that this cell-autonomous role is dependent on the existence of the locus, regardless of parental origin or transcription [110]. This role is further solidified by the observation that overexpression of Cdkn1c in embryos is associated with a reduction in the number of cortical progenitor cells leading to a decrease in upper layer neurons in the developing brain [111]. Finally, a recent study has hinted at a possible role for the paternal allele of Cdkn1c in neocortical development, despite this gene being paternally imprinted and therefore not transcribed [69].
An oligodendrocyte disorder resulting in loss of white brain matter links increased progenitor cell proliferation and increased differentiation with increased Cdk2 and decreased p21 CIP1 and p27 KIP1 activity [112]. Additionally, impairment of potassium channels in the brain related to seizure disorders was reversed by decreased Cdk2 or increased p27 KIP1 activity [113]. In wild-type embryos, p27 KIP1 expression is found to occur along a gradient in developing neural tubes, with the highest expression in postmitotic cells of the neuroepithelium, with the deletion of a component of the nuclear pore complex showing a correlation between altered p27 KIP1 expression patterns and abnormalities in cell proliferation during development [114].
The INK4 CDKI family does not appear to have an important role in the developing mouse brain, with no known studies showing a requirement for any of the Cdkn2 genes. Indeed, even a dKO of both Cdkn2c and Cdkn2d appears to result in no neurological defects [100].

Pancreas
Both CDKs and CDKIs appear to have a role in insulin regulation, as changes in their expression affect pancreatic development, with a particular functional convergence in islets. Cdk4 in particular is vital to pancreatic function as a Cdk4 −/− model results in mice with greatly impaired insulin secretion [29,30]. In contrast, a Cdk8 cKO specific to β-cells increases insulin secretion and thus improves glucose tolerance [43]. However, these cells have lost their ability to deal with stress, leading to increased apoptosis under oxidative stress [43]. Phosphorylation of Ngn3 by CDKs, including Cdk1 and Cdk2, controls differentiation of endocrine cells in the embryonic pancreas and is required to maintain adult β-cell function [115]. The importance of CDKs in controlling the pancreatic tissue cell cycle is attracting pharmaceutical interest, with CDK inhibitors being shown as potentially viable novel approaches to treat pancreatic adenocarcinomas [116].
The specific roles of CDKIs in pancreatic development are less clear. A dKO of Cdkn2c and Cdkn2d causes hyperplasia of pancreatic β-cells [100], but no other role for the INK4 family in the pancreas has yet been identified. KOs of members of the CIP/KIP family do not appear to greatly affect the developing pancreas, but their expression is still evident. For instance, upregulation of Cdkn1a in the pancreas is associated with a decrease in islet cell proliferation [117]. While p57 KIP2 expression has been observed in the pancreas, most notably at day E13.5 and then rapidly declining [109], p27 KIP1 does not appear to be expressed [95]. Rat p57 KIP2 is highly homologous to mouse p57 KIP2 , while both differ from human p57 KIP2 [118]. Similar to mice, rat p57 KIP2 expression is tightly controlled in the pancreas, decreasing at birth and increasing again two weeks after birth, but is not evident in β-cells [118]. A complex picture has emerged for Cdkn1c whereby the gene indirectly regulates islet mass and β-cell populations [119]. Overexpression models have found newborn mice to have significantly altered blood glucose profiles, which are then retained through to adulthood [120]. Some of the exact nature of this role remains unclear, in part due to the difficulty of dissecting the functions of Cdkn1c from those of surrounding genes in the Kcnq1 imprinting cluster, with further studies needed to elucidate these specific roles.

Fertility
Many CDK KO mouse models result in reduced fertility or even complete infertility, highlighting essential roles for CDKs in mouse fertility. Cdk2, Cdk4, Cdk16 and Cdk18 have all been found to be expressed in developing ovaries [121]. Cdk2 is isolated to premeiotic spermatocytes in the testes, while Cdk16 and Cdk18 are expressed in postmeiotic spermatids, suggesting diverse roles in male fertility [121]. In particular, Cdk2 is dispensable for growth but not fertility [122]. A Cdk2 KO causes both male and female infertility, due to both spermatocytes and oocytes arresting at prophase I [26,27,85]. However, an oocyte-specific Cdk2 cKO resulted in normal oocyte development, with the authors concluding that Cdk1 was the only family member required for resumption of meiosis [24]. Cdk2 is also required for differentiation of spermatocytes, with a loss-of-function allele resulting in increased apoptosis of gonocyte precursors following depressed differentiation into spermatogonial stem cells [123]. A Cdk2 and Cdk6 dKO also results in infertility in both sexes, while a Cdk4 and Cdk6 dKO displays the same infertility seen in a Cdk4 single KO, with total female infertility and partial male infertility [29][30][31], signifying that Cdk6 does not have a dominant phenotype in mouse fertility [41]. The impaired spermatogenesis in Cdk4 KOs appears to be due to defective proliferation of gonocytes and hypoplastic seminiferous tubules resulting in testicular atrophy [30]. A lack of Cdk4 causes female sterility through suppressing development of the pituitary, resulting in reduced lactotrophs and a deficit in the hypothalamic-pituitary axis, leading to perturbed corpus luteum formation [29][30][31]. Cdk16 interacts with, and is phosphorylated by, cyclin Y-like1 to regulate spermatogenesis [89]. A Cdk16 cKO resulted in structurally malformed and consequently hypomotile spermatozoa, leading to reduced male fertility, placing Cdk16 function in the terminal stages of spermatogenesis [46]. Cdk16 does not appear to be required for female fertility [46]. A more in-depth review of the role of CDKs in spermatogenesis has been compiled by Palmer et al. [124].
A lack of Cdkn1b (p27 KIP1 ) results in female sterility due to impairment of ovarian follicle development and corpora lutea formation, believed to be related to the oversized pituitary and thus alteration of the hypothalamic-pituitary-ovarian axis which is seen in these KO mice [52][53][54]. Despite having overly large testes, male fertility was not affected in Cdkn1b KO mice [54]. Loss of Cdkn1c (p57 KIP2 ) has been shown to result in sexual immaturity in both sexes [65]. Decreased Cdkn1a and Cdkn1c expression in the testes is associated with mitotic arrest in the male germ cells [125]. Cdkn1c expression is observable in rat testes at day E17, declines at day E19.5 to day E21.5, but is again evident in spermatogonia from the second week after birth and in adult spermatids [118]. In mice, while Cdkn1b expression has been observed in the ovaries and testes, Cdkn1c does not appear to be expressed [95].
Contrastingly, loss of Cdkn2a (p16 INK4a ), Cdkn2b (p16 INK4b ) or Cdkn2c (p18 INK4c ) does not appear to affect fertility [70,78]. Additionally, mice without both Cdkn1b and Cdkn1c remain fertile [78]. Loss of Cdkn2d (p19 INK4d ), however, decreases but does not abolish male fertility, due to testicular atrophy and increased apoptosis of male germ cells [81]. Similarly, a dKO of both Cdkn2c and Cdkn2d produces fertile females but sterile males [100]. These dKO males exhibit small testes with delayed progression through meiosis leading to increased germ cell apoptosis and decreased sperm cell count with malformed existing spermatozoa, suggesting both p18 INK4c and p19 INK4d are required to regulate differentiation of mature sperm [100]. Alternative splicing of Cdkn2c leads to two distinct mRNAs, the shorter of which is found in various differentiated cell types such as muscle, lung and liver, while the longer mRNA is isolated to the testes in undifferentiated, mitotically active cells, suggesting that the two protein isotypes have separate roles, particularly for differentiation [126]. p15 INK4a , p16 INK4b , p21 CIP1 and p27 KIP1 are all expressed in the ovary, but p16 INK4b is expressed more highly than the other CDKIs and seems to decrease during oocyte growth [127].

In Vitro Studies
Cdk8 and Cdk19, as involved in the CDK-Mediator complex, are vital for regulating gene expression following differentiation of embryonic stem cells into neuronal precursors [128]. Cdk8 in particular is needed to bind to distal enhancers to promote gene activation upon differentiation [128]. In the absence of Jdp2, p21 CIP1 is required to bind to antioxidant response elements of the Slc7a11 promoter in order to provide redox control and block random oxygen species mediated apoptosis in granule cell progenitors, an essential component for normal development of the cerebellum [129].
Cdk16 is required for dendrite development and is regulated by Cdk5 phosphorylation [130]. Cdk16 binds to p35 and is phosphorylated by Cdk5 to enhance its own kinase activity [33]. Additionally, phosphorylation of Cdk16 by BRSK2 reduces glucose-stimulated insulin secretion in mouse β-cells [131].

Discussion and Future Perspectives
While a great deal of data has been generated on the roles of both CDKs and CDKIs in murine embryonic development, as has been presented in this review, there remains a number of as yet entirely untested gene family members. Much of the focus of research into these families will remain in cancer biology, given the oncogenic (CDKs) and tumour suppressing (CDKIs) nature of the groups, respectively. There is a wealth of data being generated on the use of CDK inhibitor compounds (CDKis) and their efficacy for tumour treatment [4,132]. While developmental biology and tumour therapy seem like divergent focuses, much of the data that has been generated in embryonic development can serve as a therapeutic guide. Thanks to multiple knockout lines and environmental studies, researchers have uncovered shared and redundant functions of both CDKs and CDKIs. This data will be invaluable in future determinations of appropriate CDKi therapies, based on the expression profiles of the CDK and CDKI families in tumours.
Further study of Cdk1 will remain challenging due to its critical role in cell cycle progression. However, researchers have demonstrated that this absolute requirement can be bypassed in certain scenarios using conditional deletions, which have also been utilized to dissect the multiple roles of Cdk5 in brain development. There remains the opportunity to target Cdk11 and Cdk13 in a similar manner, so as to bypass the total embryonic lethality seen in the respective constitutive knockout mouse lines and enable cell type specific functional role analyses. Beyond these, a number of the CDKs are awaiting their functional roles to be examined, as has been highlighted in Table 1.
The CDKIs present with a similar picture, where some family members have been studied extensively through both constitutive and conditional deletions, while others remain to be tested. Perhaps surprisingly, only Cdkn1c has been used to explore the impact of overexpression of a CDKI [99,120]. These lines have shown an exquisite dosage sensitivity of the gene on both embryonic growth and tissue differentiation. Given the known shared functional roles of the CIP/KIP family [92][93][94][95], Cdkn1a and Cdkn1b appear to be ideal candidates for a similar targeting strategy and further exploration of the shared and unique roles of these family members. Recently, researchers have begun applying a novel approach to the study of these genes, through the generation of bioluminescent reporter lines. Cdkn2a, Cdkn1b and Cdkn1c have all been independently targeted with a luciferase-reporter to monitor endogenous gene expression in vivo [133][134][135]. While these lines were initially largely developed to enable in vivo tumour formation, progressive technological development has allowed sensitive high-resolution reporting of embryonic Cdkn1c expression in utero. This approach uncovered a previously unseen sensitivity of the epigenetic markers that regulate this gene's expression to gestational dietary modification that resulted in lifelong misexpression of the gene. As a read-out for gene expression, these models offer unparalleled insights, due to the ability to image individual animals throughout their life-course, and therefore generate a complete picture of expression. It is easy to envisage this approach being applied to any of the CDKs or CDKIs to explore as yet unknown functions and provide a complementary addition to the catalogue of presently assembled mouse models, which will continue to expand and develop our knowledge of these gene families.

Conflicts of Interest:
The authors declare no conflict of interest.