Physiological functions of protein kinase D in vivo

Authors


Abstract

The cellular functions of the serine/threonine protein kinase D (PKD) have been extensively studied within the last decade and distinct roles such as fission of vesicles at the Golgi compartment, coordination of cell migration and invasion, and regulation of gene transcription have been correlated with this kinase family. Here, we highlight the current state of in vivo studies on PKD function with a focus on animal models and discuss the molecular basis of the observed phenotypic characteristics associated with this kinase family. © 2013 IUBMB Life, 65(2)98–107, 2013

Introduction

The protein kinase D (PKD) family of serine/threonine kinases belongs to the Calmodulin-dependent protein kinases (CaMKs) superfamily and comprises three isoforms in mammals, PKD1, PKD2, and PKD3. PKD1 was the first member identified in human and mouse in 1994 (1, 2). In 1999 and 2001, PKD3 (3) and PKD2 (4), respectively, complemented the family. PKDs are activated downstream of novel PKCs by phosphorylation of serines within the activation loop of the kinase domain [reviewed in (5, 6)]. PKDs are receptors for diacylglycerol (DAG), which binds to the cysteine-rich domain of the kinases and is crucial for kinase activation. Thus, the local amount of DAG determines the localization and activation state of PKD, thereby restricting PKD responses to specific organelles such as the Golgi complex, the nucleus, and the plasma membrane (PM). It is thus not surprising that PKD is implicated in several intracellular processes and signaling pathways such as vesicle trafficking, survival responses, and cell motility [reviewed in (5)]. In the past, a number of PKD-specific substrates mediating these processes have been identified: At the trans-Golgi, PKD signals to a set of substrates including PI4KIIIβ (7), the lipid transfer proteins CERT (8), and OSBP (9) to locally control membrane lipid composition and thus fission of vesicles containing cargo destined for the PM. Besides the regulation of constitutive secretion, recent data also provided evidence for a role of PKD and its substrate Arfaptin-1 in regulated secretion of insulin from the trans-Golgi network (TGN) (10). At the PM, PKD suppresses cell motility by controlling the function of regulators of the actin cytoskeleton such as SSH1L, PAK4, cortactin, E-Cadherin, SNAIL, and RIN1 (11–18). Finally, PKD is involved in the regulation of transcription factors such as MEF2 and RUNX2 by phosphorylation of class IIa HDACs (19, 20).

Most of the presently known PKD substrates and functions have been identified and characterized in tissue cultures and assessed using siRNA techniques, expression of a dominant-negative PKD, or pharmacological inhibition. Whether these assigned functions are of any relevance in vivo, however, is still poorly addressed and little is known about PKD functions and effectors in normal tissues of whole organisms. Thus, elucidation of regulation of PKDs, their effectors, and physiological roles in vivo are major challenges to be tackled in ongoing and future studies. Here, we review the status quo of studies at the organism level with a focus on transgenic models and discuss to what extent PKD functions identified in vitro translate to a physiologic role in vivo.

Conserved Structural Properties of PKD

The three mammalian PKD isoforms share a conserved C-terminal kinase domain and N-terminal regulatory domains. Of note, PKD1 and PKD2 contain an N-terminal hydrophobic stretch of amino acids that could potentially insert into the membranes (21). In addition, the C-terminal autophosphorylation sites in PKD1 and PKD2 are part of a postsynaptic density-95/discs large/zonula occludens-1 (PDZ)-binding motif, which is not present in PKD3 and thus contributes to isoform-specific functions (22). In contrast to vertebrates, D. melanogaster contains only a single PKD gene (23). In the nematode C. elegans two orthologous genes are present, termed dkf-1 and dkf-2 (24–26). The isoforms share conserved structural features that are consistent with their classification in the PKD family of kinases (Fig. 1). The functional domains (cysteine-rich domains (CRDs), pleckstrin homology (PH) domain, and catalytic domain) show a substantial sequence similarity even for evolutionarily distant species like mouse and fly. Only the activation loop sequence of DKF-1 exhibits an exceptional single threonine phosphorylation motif, which is in line with the discovery that DKF-1 is activated by PKC-independent pathways (25).

Figure 1.

Conserved regulatory and catalytic domains and phosphorylation sites of mouse PKD isoforms and orthologous proteins in C. elegans and D. melanogaster. Conserved Lys residues within the catalytic domain are involved in ATP binding. Labeled Ser (and Thr in the case of DKF-1) residues are conserved sites of phosphorylation within the activation loop. C-terminal autophosphorylation sites (Ser residues) are present in mammalian PKD1 and PKD2 and are part of a PDZ-binding motif. CRD1/CRD2 denotes cysteine-rich, zinc finger-like domains; PH, pleckstrin homology domain. Sequence homology of the functional domains is specified as % identity to corresponding PKD1 sequences of M. musculus. Domains are not drawn to scale.

PKD Animal Models

In the past years, several PKD animal models have been established (Table 1). The D. melanogaster model made use of the Gal4/UAS system to address tissue specificity of PKD function. Similarly, the C. elegans PKD proteins DKF-1 and -2A/B were expressed under the control of tissue-specific promoters to study specific functions. Some of the established PKD mouse models allow a conditional knock-out or transgene expression, either by the use of the Cre/LoxP system or the Tet expression system. The expression patterns of the distinct PKD isoforms were shown to be partially overlapping (38), suggesting that, at least in part, they are functionally redundant. For example, the PKD isoforms are functionally redundant with respect to class II HDAC phosphorylation. Here, the loss of HDAC regulation was only visible on deletion of the complete cellular PKD protein pool (39). Moreover, all three PKD isoforms are involved in the transport of basolateral cargo from the TGN to the PM. However, it was postulated that the isoforms may exert a cargo selective function in vesicular transport (40, 41). In line with this, PKD1 and PKD2, but not PKD3, phosphorylate PI4KIIIβ at the TGN to promote vesicular transport (7). These findings clearly support the existence of nonredundant functions of PKD isoforms, too. Thus, a knockout of one PKD family member may or may not be compensated by the presence of a redundantly acting isoform, depending on the particular physiological function studied. In the former case, it would be necessary to target all three PKD isoforms and create double- or triple-knockout animals. Alternatively, the transgenic overexpression of a kinase-dead PKD isoform could exert a dominant-negative effect on all PKD isoforms. For example, in HeLa cells TGN-to-PM trafficking was equally inhibited with dominant-negative, kinase-dead mutants of PKD1, PKD2, or PKD3 (41, 42). However, substrate competition with other PKD-related kinases such as CaMK could blur the interpretation of observed phenotypes. Thus, the phenotype of a mouse overexpressing dominant-negative PKD is not necessarily expected to be equivalent to the phenotype of a PKD knockout mouse.

Table 1. PKD animal models
inline image

PKD in D. melanogaster and C. elegans

In 2007, the first studies on PKD function in D. melanogaster were published. In transgenic flies ubiquitously overexpressing wild type (wt) or kinase-dead (kd) versions of Drosophila PKD, severe effects on wing vein development were observed. Specifically, cross vein formation was affected in all cases (27). The cross vein formation is especially sensitive to reductions in local BMP-like signaling (43). In mammals, BMPs are known to play an important role in vasculature formation during embryonal development through, among others, action on vascular smooth muscle cells [reviewed in (44)]. PKD is positioned downstream of BMP-2 and BMP-7 signaling in osteogenic differentiation of human osteogenic progenitor cells (45, 46). Thus, PKD activity could be controlled by a local BMP-like signal during Drosophila wing cross vein formation, too. On ubiquitous expression of a constitutively active PKD (PKD-SE), all larvae died at first larval instar. In addition, expression in the posterior compartment was fully lethal at late pupal stage (27). Thus, a tight control of PKD activity is crucial to overall fly development. Intriguingly, within the developing retina of larvae overexpressing PKD-SE, the nuclei of the photoreceptor cells that are normally localized at the apical side of the cell were mislocalized to a more basal position and showed an accumulation of F-actin and phosphorylated cofilin (27, 47). In addition, splintered hair bristles were observed in PKD-SE flies (47). Of note, similar phenotypes occur in flies lacking the cofilin phosphatase Slingshot (SSH) (48, 49). Coexpression of a wt SSH protein together with PKD-SE reduced the accumulation of F-actin in the retina of pupae, whereas coexpression of a phosphatase-inactive SSH potentiated the PKD-SE-induced effects (47), corroborating data from mammalian cell systems showing a PKD-mediated negative regulation of SSH activity (13). In mammalian cells, PKD also regulates the cofilin kinase LIMK1 through direct phosphorylation and activation of its upstream kinase p21-activated kinase 4 (PAK4) (14). PAK4 shows high similarity to the Drosophila group II PAK protein mushroom bodies tiny (Mbt) (50) and the PKD phosphorylation motif VPRRKpSLV in human PAK4 is highly conserved in Drosophila Mbt (LPRRKS521LV). Mutations in Mbt result in the frequent absence of photoreceptor cells in the eye (51, 52). Of note, expression of an activated MbtS492N,S521E protein, which contains a phosphomimetic mutation at the proposed PKD phosphorylation site at serine 521, caused a severe rough eye phenotype characterized by ommatidia of different size and shape and reduced overall eye size (51). Furthermore, ommatidia contained a variable number of photoreceptor cells and the rhabdomeres of these cells were misshaped. This phenotype resembles at least in part the phenotype observed in flies expressing PKD-SE (47) and, thus, it is intriguing to speculate that similarly to SSH, Mbt is a target of Drosophila PKD during eye development. In sum, although several studies on Drosophila PKD function have been conducted already, the availability of a (tissue selective) Drosophila PKD knock is still pending and will be crucial to elucidate development and tissue-specific functions of PKD in the fly because overexpression studies are prone to artifacts and need to be controlled by loss-of-function studies. Nevertheless, initial studies with RNAi-based knock down of endogenous PKD support the results from dominant-negative overexpression studies by demonstrating severe tissue loss in the wing and degeneration of the adult retina (27).

A C. elegans specific D kinase family isoform, DKF-1, was discovered and characterized in 2006 (25, 26). Detailed analysis of the kinase properties in vitro and in cell culture revealed that DKF-1 is a phorbolester-activated, but PKC-independent D kinase displaying similar substrate specificities as human PKDs. RNAi-mediated depletion of DKF-1 in vivo strongly impaired the mobility of adult worms. Loss of muscle contraction near the anus pointed toward a neuromuscular defect in the tail region (26). Dkf-1 null mutants (dkf-1(−/−)) develop normally; however, adult dkf-1(-/-) mutants also displayed an uncoordinated tail movement confirming the RNAi based results (26). The molecular basis for this motility phenotype is not known, nevertheless, the dkf-1 null model provides a helpful tool to investigate physiological PKD functions and downstream targets. Moreover, rescue experiments using human PKD isoforms would also aid to further understand redundant and individual functions of the three family members. Besides DKF-1, a second C. elegans specific D kinase, DKF-2, was identified in 2007 (24). The dkf-2 gene encodes two isoforms, DKF-2A and DKF-2B differing especially in their N-terminal part (Fig. 1) (28). Although DKF-2A is highly expressed in intestinal cells, a dkf-2 null mutant lacking both isoforms does not show effects on intestinal physiology. Secretion of proteins from the intestine as well as intestinal cell differentiation and development of the gut were unaffected. Surprisingly, the life span of adult dkf-2 null C. elegans increased by 40% when compared with wild type animals (24). This is attributed to the negative regulation of the transcription factor DAF-16, which is positioned downstream of insulin/IGF-1-like peptides and the receptor tyrosine kinase DAF-2 (53). DAF-16 is implicated in blocking transcription of genes encoding proteins that protect C. elegans against various stresses resulting in an extended life span (53). A genetic relationship between DKF-2 and DAF-16 was demonstrated in dkf-2 null postlarval animals, in which stress-induced nuclear translocation of DAF-16 in intestinal cells was blocked, identifying DKF-2 as an upstream negative regulator of DAF-16 (24). Still, the direct link between DKF-2 and DAF-16 could not be resolved and upstream signals activating DKF-2 remain unknown.

Both isoforms, DKF-2A and B, are essential for Na+-induced learning, but not required for salt detection (28). Genetic relationship studies identified EGL-8, a PLC-β4 homolog, and the PKCβ homolog TPA-1 to be upstream of DKF-2A and 2B activation in Na+-induced learning, supporting a critical role for DAG in this process (28). As both DKF-2 isoforms are crucial to Na+-induced learning, signals emerging from kinase activation need to be integrated by a still unknown crosstalk between neuronal and intestinal cells. The phenotype of dkf-2 null animals also implies a role for PKD isoforms in plasticity. This is of special interest as mammalian PKDs are highly expressed in neuronal cells (38) and shown to act downstream of metabotropic glutamate receptors (54). Thus, exploring the roles of mammalian PKDs in long-term potentiation will advance the understanding of plasticity in the central nervous system.

DKF-2-deficient animals also displayed a hypersensitivity toward pathogenic bacteria, which was rescued by reexpression of DKF-2A-GFP, pointing to a role of the kinase in innate immunity (29). In addition, also elevated DKF-2A-GFP expression in wild-type animals conferred pathogen resistance. Taken together, the studies on DKF-1 and DKF-2A and -2B functions in C. elegans support the hypothesis for differential functions of PKD isoforms, which might be due to slight differences in structure and thus localization and activation (Fig. 1). Nevertheless, DKF-1 and -2-specific substrates mediating these functions still remain to be identified.

PKD Mouse Models

The first study that made use of mouse transgenics to unravel PKD function in vivo was published in 2003 (30). Specifically, PKD function in T-cell maturation was addressed. The major isoform expressed in primary mouse thymocytes and peripheral T cells is PKD2 (36). PKD2 is strongly activated by antigen receptor stimulation in T, B, and mast cells in nPKC-dependent manner (55). On activation, PKD2 rapidly translocated to the PM. However, during sustained responses to antigen receptor engagement, PKD2 relocated to the cytosol where it remained active for several hours (56). To address the role of PKD translocation during T-cell receptor (TCR) stimulation, the catalytic core of PKD1 was fused either to GFP or to the extracellular and transmembrane domain of the cell surface receptor rCD2 targeting PKD exclusively to the PM and the cytosol, respectively. Transgenic overexpression of these targeted active PKD1 mutants in mouse thymi differentially affected thymocyte development, indicating that certain PKD functions appear context dependent (30). Although both PKD1 mutants, membrane and cytosol targeted, induced T-cell differentiation, two striking differences were observed: Membrane but not cytosolic PKD1-induced expression of CD8 and CD4 in recombinase null mice pointing to a bypass of the β-selection checkpoint, whereas cytosolic but not membrane PKD1 suppressed Vβ to DJβ rearrangements of the TCRβ chain locus in wild-type T cells (30). These results are in accordance with a TCR-specific activation of PKD in premature and mature T cells and underline the importance of location-dependent PKD function. In line with this, membrane-localized PKD1 regulation of pre-T cell differentiation is dependent on Rho, whereas the cytosol-localized PKD1 acts in a Rho-independent manner (57). Studies in tissue culture revealed that TCR signaling causes nuclear export and thus inactivation of HDAC7 via PKD1-dependent phosphorylation resulting in enhanced expression of Nur77 (58, 59). However, pre-T cells recombinase null animals lacking HDAC7 do not have altered Nur77 expression levels during thymocyte selection and bypass β-selection (60). Moreover, PKD2 null mice normally upregulate Nur77 expression on TCR triggering (37). Thus, the mechanism(s) underlying PKD-dependent T-cell differentiation in vivo still remains unclear. Of note, a gain-of-function mutant does not answer the question whether the protein is relevant for this specific biological process. To overcome this problem, a targeted knock-in approach was used to generate kinase-dead Prkd1S744/748A and Prkd2S707/711A alleles, respectively (36). The majority of mice with homozygous Prkd1S744/748A alleles died at day 9.5 of embryonic development, similar to mice with early embryonic deletion of PKD1 (32), confirming this stage as an essential phase in mouse embryogenesis, which requires PKD1 catalytic activity. This is in line with the observation that PKD1 is already expressed in early embryonic stages in mice (38). In contrast, mice with homozygous expression of PKD2S707/711A are born at normal mendelian frequencies and are phenotypically indistinguishable from wild-type littermates (36). Similar results were obtained by the use of mice derived from ES cells with a gene-trap cassette insertion into the Prkd2 locus (Prkd2Gt/Gt) (36). Consequently, proper embryonic development requires PKD1 but is independent of PKD2. A detailed analysis of the immunological properties of Prkd2Gt/Gt mice as well as Prkd2S707/711A mice revealed that PKD2 kinase activity is dispensable for the normal development of mature, peripheral T- and B-lymphocytes. However, mice expressing only a catalytically inactive PKD2 display reduced antigen receptor induced cytokine production in mature peripheral T lymphocytes and an impaired antigen-specific humoral immune response. Thus, a unique function for PKD2 in adaptive immunity is suggested (36). To analyze PKD2 function in TCR signaling in more detail, mice expressing defined α/β TCRs were backcrossed to either Prkd2Gt/Gt mice or mice deficient in PKD2 catalytic activity (37). The loss of PKD2 catalytic activity caused a tremendous increase in total thymic cellularity as well as peripheral T cells resulting in enlarged lymph nodes and spleens. PKD2-null TCR transgenic T-cells that exit the thymus have the transcriptional profile of normal naive T cells; however, they lack the ability to fully reprogram their transcriptome in response to TCR engagement. This is evident from decreased levels of key cytokines such as IL-2 and chemokines known to be essential for T-cell effector function (37). Taken together, these studies point to a crucial role of PKD2 in mediating α/β TCR-dependent limitation of thymic output and imply a negative role for PKD2 in cell proliferation. Of note, PKD3 catalytic activity has been shown to be normal in lymphocytes isolated from Prkd2S707/711A mice (36), suggesting that PKD2S707/711A does not act in a dominant-negative manner toward PKD3 in this case. In contrast to mutations of the ATP-binding site (K/W or K/N), inactivating mutations of the activation loop serines do not impact on PKD1 autophosphorylation at serine 916 (61, 62). It is known that autophosphorylation at this C-terminal residue reverses the binding of PKD1 (and PKD2) to PDZ-domain proteins, thereby regulating the duration and amplitude of localized PKD activity (22). This raises the question whether a PKD2S707/711A protein can act in a dominant-negative manner at all.

Studies with primary cells from PKD2-deficient mice revealed that PKD2 is dispensable for integrin-mediated lymphocyte adhesion and homing to lymphoid tissues in vivo (63). This was surprising, as several studies with transformed cell lines proposed a role of PKD in integrin-recycling and integrin-mediated cell adhesion (64–66). In addition, studies with DT40 PKD1/3 knock out B cells showed that PKD enzymes do neither regulate basic cellular processes such as proliferation or survival responses, nor NFκB transcriptional activity downstream of the B-cell antigen receptor (67). These findings are contrary to earlier studies in established cell lines (68–70). The discrepancy between results obtained with transformed cell lines and primary cells from genetically engineered mice thus emphasizes the necessity for appropriate animal models to draw conclusions on in vivo functions. This is also evident from studies with isolated platelets from Prkd2S707/711A mice. These platelets show impaired thrombus formation in in vitro assays, whereas the transgenic mice retain a normal haemostatic response (71).

Based on findings that PKD is a class II HDAC kinase in vitro (72), further studies with genetically engineered PKD mice focused on a role of PKD in heart and skeletal muscle tissue. Class II HDACs play a key role as transcriptional regulators of pathological cardiac remodeling. Specifically, HDACs 5 and 9 are negative regulators of cardiac hypertrophy by interacting with and blocking of the transcription factor myocyte enhancer factor 2 (MEF2) [reviewed in (73)]. In vitro, each PKD isoform is capable of phosphorylating HDAC5 as well as other members of the class II HDACs such as HDAC4, 7, and 9, mediating interaction with 14-3-3 proteins and thus controlling nuclear localization and function of HDAC proteins (74). Accordingly, only the combined knock out of PKD1 and PKD3 in chicken DT40 B lymphocytes effectively blocked HDAC5 and 7 phosphorylation on antigen receptor signaling (39). In transgenic mice, the cardiac-specific overexpression of constitutively active (ca) PKD1 (MHC-PKD1ca) caused a brief phase of cardiac hypertrophy, followed by chamber dilation and impaired systolic function and death (31). In line with this, expression of fetal cardiac genes was strongly increased in hearts of PKD1ca expressing animals. Endogenous PKD1 is activated in ventricular tissue during pathological cardiac hypertrophy in vivo, however, all three PKD isoforms are expressed in cardiac myocytes (31) making it likely that PKD2 and PKD3 act redundantly or together with PKD1 to control HDAC5 phosphorylation and cardiac hypertrophy. To address this in more detail, mice with floxed Prkd1 alleles (Prkd1flox) were generated to allow a conditional PKD1 knockout (32). Cardiomyocyte specific deletion of PKD1 using α-MHC-Cre transgenic mice confirmed the crucial role of this isoform in cardiac hypertrophy: On pressure overload, chronic angiotensin II or isoprotenerol stimulation, PKD1-deficient mice were resistant to hypertrophy and fibrosis. In addition, fetal gene activation was partially blocked (32). The blunted hypertrophy in mice with heart selective PKD1 KO suggests that neither PKD2 and PKD3 nor other stress-responsive kinases such as CaMK, potentially acting on common downstream targets, can fully compensate for the loss of PKD1. Although these studies suggest that PKD exerts its essential function in cardiac remodeling through class II HDACs, a direct link in vivo was missing. In addition, PKD might also have other cardiac-specific functions: For example, PKD phosphorylated cardiac troponin I and reduced myofilament Ca(2+) sensitivity in skinned adult rat ventricular myocytes (75). Furthermore, recent studies with PKD1 deficient cardiomyocytes suggest a role for PKD in contraction-induced glucose but not fatty acid uptake (76). Finally, PKD was activated in an acute model of diabetes in cardiomyocytes. Here, PKD activation resulted in enhanced lipoprotein lipase trafficking in a PKCδ-dependent manner (77), which is in line with the well-characterized role of PKD in vesicle fission at the TGN (42).

A direct link between PKD1 and class II HDACs was provided in 2008 using mice with a skeletal-muscle specific expression of PKD1ca (MCK-PKD1ca) (33). On exercise, the skeletal muscle remodels its composition forcing the formation of slow-twitch fatigue-resistant type I muscle fibers in a class II HDAC-MEF2 dependent manner. Forced expression of a ca PKD1 protein in skeletal muscle induced a fiber type switch resulting in an increase in type I fibers (33). In line with this, transgenic mice had a lean appearance with unchanged metabolic activity and displayed fatigue resistance. Accordingly, deletion of PKD1 in skeletal muscle enhanced susceptibility to fatigue (33). Expression of PKD1ca enhanced HDAC4 and HDAC5 phosphorylation pointing to increased MEF2 activation in skeletal muscle in vivo. Indeed, crossing of MCK-PKD1ca mice with 3xMEF2-LacZ reporter mice revealed a high MEF2 activity in skeletal muscle (33). The essential role of PKD in skeletal muscle remodeling was confirmed by conditional expression of a kinase-dead (kd), dominant-negative PKD1K612W-GFP protein, which inhibited exercise-induced fiber type switch and thus skeletal muscle remodeling in a MEF2-dependent manner (34). Thus, mice expressing PKD1kd exhibited altered muscle fiber composition and decreased running performance compared to control mice. Of note, skeletal muscle-specific deletion of PKD1 did not result in a fiber type switch (33) suggesting that in this tissue, PKD2 and PKD3 or other class II HDAC kinases such as CaMK could compensate for the loss of PKD1. Interestingly, mice expressing PKD1kd as well as mice with deletion of PKD1 in skeletal muscle were indistinguishable from their wild-type littermates (33, 34). This demonstrates that an external signal, such as exercise, is required to activate PKD1 to exert its function in skeletal muscle.

A few studies have also addressed PKD function in neurons as well as in pancreatic beta-cells in vivo. In the central nervous system, all three PKD isoforms were found to be expressed in the embryo (38, 78). Transgenic mice with neuron-specific, inducible expression of a dominant-negative PKD1kd showed localization of the PKD1 transgene to the TGN in hippocampal neurons (35). Expression of PKD1kd-GFP in this tissue changed the morphology of the Golgi compartment from a thread-like structure to a dispersed phenotype. This is in line with results obtained from cultured primary hippocampal neurons (35) and points to a crucial role of PKD in Golgi structure and function in the brain. In addition, PKD-mediated Hsp27 phosphorylation (67) is essential for mediating neuroprotection against ischemic neuronal injury in vivo (79).

In pancreatic beta-cells, PKD localized to the Golgi compartment stimulating insulin secretion at this site (80). In this case, PKD activity was negatively controlled by p38δ-mediated phosphorylation. Accordingly, p38δ-deficient mice showed high constitutive PKD activity accompanied by enhanced glucose-mediated insulin secretion and thus protection against high-fat-feeding induced insulin resistance (80). The essential role of PKD1 in glucose-stimulated insulin secretion has been confirmed using islets from PKD1 KO mice (81). More recently, Arfaptin-1 was identified as a PKD-specific substrate mediating biogenesis of transport carriers containing insulin at the TGN in rat INS-1 cells (10). However, the identification of in vivo PKD downstream targets at the Golgi compartment remains a challenge for the future.

Comparison of Animal Models Phenotypes

D. melanogaster and C. elegans physiology and behavior are regulated by signaling molecules, mechanisms, and pathways that are often conserved in mammals. However, concerning PKD, there is a lack of similarity between the phenotypes of the genetically altered animals, which could be explained by the existence of three differentially regulated PKD isoforms in mammals compared to two and only one PKD gene in C. elegans and D. melanogaster, respectively. In addition, there are differences not only between the vertebrate and invertebrate system with respect to physiological responses such as innate immunity but also between D. melanogaster and C. elegans. Future work should, therefore, aim at the key question of to what extent findings on PKD function in D. melanogaster and C. elegans can be translated to the mammalian system.

Therapeutic Aspects of PKD Inhibition

In the past, PKD-specific inhibitors have been developed and were successfully tested for their ability to treat prostate and pancreatic cancer in mice, making PKD an attractive therapeutic target [extensively reviewed in (82)]. However, systemic inhibition of PKD in vivo can be a double-edged sword: For example, inhibition of PKD activity in pancreatic beta-cells will result in diminished insulin secretion and could thus aid in development of diabetes. By contrast, PKD inhibition in the heart could reduce cardiac hypertrophy and failure. Therefore, future strategies should focus on a local, organelle and tissue-specific PKD inhibition.

Concluding Remarks

The studies reviewed herein on PKD function using D. melanogaster, C. elegans, and mouse as model organisms already revealed a crucial role of the kinase family in diverse biological processes such as innate immunity, plasticity, cardiac, and skeletal muscle remodeling and T lymphocyte function. This diversity of PKD functions in vivo might be explained by a dynamic change of PKD localization and activation together with a specific combination of substrates. So far, only class II HDACs and SSH1 have been validated as PKD substrates in vivo. The identification of PKD-specific substrates is thus crucial to fully understand the functions of this kinase family in vivo. Because of embryonic lethality and in part contradictory results between studies in tissue culture and transgenic animals emphasize the need for further work to unravel the full spectrum of the three PKD isoforms activities in vivo, in particular their specific and overlapping physiological functions. To address these open questions, more refined transgenic animal models, allowing tissue-specific and conditional interference with PKD activity, either in total or isoform selective, are required.

Acknowledgements

The authors thank Monilola Olayioye and Klaus Pfizenmaier for critical reading of the manuscript. Work in the lab of Angelika Hausser is supported by grants from the German Cancer Aid and the Heidelberger Akademie der Wissenschaften. Kornelia Ellwanger is supported by the German Research Foundation (DFG PF247/13-1).

Ancillary