Small Molecule Inhibitors of Protein Kinase D: Early Development, Current Approaches, and Future Directions

Abstract

Now entering its fourth decade, research on the biological function, small molecule inhibition, and disease relevance of the three known isoforms of protein kinase D, PKD1, PKD2, and PKD3, has entered a mature development stage. This miniperspective focuses on the medicinal chemistry that provided a structurally diverse set of mainly active site inhibitors, which, for a brief time period, moved through preclinical development stages but have yet to be tested in clinical trials. In particular, between 2006 and 2012, a rapid expansion of synthetic efforts led to several moderately to highly PKD-selective chemotypes but did not yet achieve PKD subtype selectivity or resolve general toxicity and pharmacokinetic challenges. In addition to cancer, other unresolved medical needs in cardiovascular, inflammatory, and metabolic diseases would, however, benefit from a renewed focus on potent and selective PKD modulators.

Graphical Abstract

INTRODUCTION

Since the first report of Protein Kinase D1 (PKD1, also known as PKCμ) in 1994,¹ two additional isoforms (PKD2 and PKD3) have been characterized, and extensive progress has been made on the disease-related biological processes that engage this family of diacylglycerol (DAG)-regulated serine-threonine protein kinases. PKDs belong to the Ca²⁺/calmodulin-dependent kinase (CaMK) superfamily based on homology in the catalytic domain. In addition to cancer and cardiovascular diseases, the methionine gatekeeper kinases PKD1–3 are implicated in inflammatory, metabolic, CNS, and angiogenesis-related disorders as well as immune dysregulation and even viral infections. Several drug discovery programs were launched with the goal to generate selective PKD inhibitors for clinical trials. PKD biology, function, and signaling mechanisms have been reviewed in considerable depth,² and this perspective will therefore mainly focus on the structural classifications, syntheses, and potential for further development of small molecule PKD inhibitors, augmenting other recent medicinal chemistry reviews.³

Chart 1 provides an overview of the most important classes of PKD inhibitors that have exhibited nanomolar potencies in in vitro enzyme assays. Chemotypes I–X are introduced to classify the scaffolds, and specific lead examples 1–10 allow for a structure-guided comparison between chemotype classes that in many cases are representative for general kinase active site inhibitor designs. Chart 2 shows the timeline of significant discoveries in this now firmly established field, highlighting the most active development period from 2006 to 2012.

Chart 1. Overview of Major Chemotypes and Specific Exemplified Structures for Potent PKD1–3 Inhibitors and Representative In Vitro IC₅₀ or K_D Data for PKD1 Inhibition^a.

^aThe chemotype scaffolds are highlighted in blue in the specific examples. (*) K_D measurement.

Chart 2.

Timeline of the PKD Inhibitor Discovery and Development Wave, Peaking between 2006 and 2012

Shortly after the discovery of PKD1, Gschwendt et al. screened a series of indolocarbazoles including staurosporine (1), K252a, Gö6976, and Gö6983 (Figure 1).⁴ Remarkably, these four bisindoles spanned a 3000-fold potency range, with the IC₅₀ of the unselective pan-kinase inhibitor K252a being in the single-digit nanomolar and the maleimide Gö6983 in the double-digit micromolar range. The latter compound proved to be useful as a negative control in kinase assays. Furthermore, in contrast to the lack of kinase isoform selectivity observed for K252a and 1, the structurally closely related otherwise classical DAG- and calcium-dependent PKCα,β,γ-selective Gö6976 inhibited the calcium-independent, DAG-regulated PKD1 also very effectively at 20 nM. These and other SAR comparisons suggested that the central benzene ring in 1, K252a, and Gö6976 was necessary for PKD1 inhibition but that indole nitrogen substituents could be readily exchanged for achieving kinase selectivity for PKD. Accordingly, the discovery of the PKD-inhibitory chemotype I validated this kinase as a suitable target for medicinal chemistry studies.

Figure 1.

Structures and PKD1 potencies for selected chemotype I inhibitors; data suggest that the fused aromatic system (shown in red) is necessary for high potency but that indole nitrogen substituents (shown in blue) can convey increased selectivity for PKD over other kinases.

Synthesis and Representative SAR Analyses of Chemotypes I–X.

A concise synthesis of (+)-K252a, a representative example for inhibitors of chemotype I, is shown in Scheme 1, but many other strategies have culminated in the successful preparation of these natural product scaffolds and their analogs.¹⁶ In a versatile regioselective approach in 1999, Kobayashi et al. reported the conversion of indole 11 to the target indolocarbazole alkaloid.¹⁷ The allyl ester derived from 11 was first brominated with NBS and then N-glycosylated to generate the β-furanosylated intermediate 12. The allyl ester protective group was removed with Pd(0) catalysis, and the acid was coupled to tryptamine with the water-soluble reagent 1-ethyl-3-(3-(dimethylamino)propyl)carbodiimide (EDCI) in high overall yield from 11. Amide 13 was then oxidized regioselectively with DDQ to introduce a ketone group at the more reactive indole side chain, followed by acetylation to generate imide 14. An intramolecular aldol reaction/dehydration utilized DBU in the presence of 4 Å molecular sieves, and the resulting lactam 15 was subjected to a photochemical 6π-electrocyclization to generate the benzene ring after elimination of HBr from the intermediate cyclohexadiene. Potassium hydroxide cleavage of the tolyl esters and selective conversion of the primary alcohol to the iodide led to key intermediate 16. Cyclization of this compound to introduce the seven-membered ring was problematic but ultimately accomplished by a multistep sequence starting with a conversion of the iodide to the selenide, acetylation of the secondary alcohol, oxidation to the selenoxide, and elimination with triethylamine to give alkene 17. Dihydropyran (DHP) was used to prevent an undesired addition of phenylselenous acid to the reactive enol ether. Treatment of 17 with iodine, potassium iodide, and DBU effected a smooth iodoglycosylation to close the desired seven-membered ring, and a radical-mediated deiodination, methanolysis of the acetate, and Swern oxidation provided ketone 18 in excellent overall yield.

Scheme 1.

Regioselective Synthesis of (+)-K252a¹⁷

The α-hydroxy ester in (+)-K252a was introduced via formation of the cyanohydrin of 18, O-acetylation, and conversion of the nitrile to the amide 19 with gaseous HCl in formic acid. Interestingly, a single, kinetically favored cyanohydrin acetate was formed when the cyanohydrin formation was performed with a solution of hydrogen cyanide in pyridine. Finally, 19 was hydrolyzed under basic conditions, and the acid was converted to the methyl ester (+)-K252a by treatment with diazomethane. This linear but high-yielding synthetic route is also potentially useful for the preparation of a variety of interesting analogs of (+)-K252a.

For the preparation of simplified (+)-K252a analogs related to the potent checkpoint kinase inhibitor Gö6976, Roy et al. used the N-alkylated glyoxylyl chloride 20 and indole-3-acetic acid 21 to form the anhydride 22 in 31% yield (Scheme 2).¹⁸ Treatment of 22 with hexamethyldisilazane (HDMS) and methanol gave the corresponding imide in 96% yield, and oxidation with palladium(II) trifluoroacetate led to analog 23 in 30% yield.

Scheme 2.

Preparation of Gö6976 Analogs¹⁸

Subsequent to the discovery of PKD and the characterization of staurosporine-type inhibitors of this PKC variant, a trypanocidal drug, suramin (Figure 2), was shown to differentially affect the activity of several PKC isoforms¹⁹ and to activate PKD.²⁰ However, the pharmacological properties of this hexasulfonated rather promiscuous compound are complex,²¹ and its properties or those of its analogs as modulators of PKD have not been further evaluated.

Figure 2.

Structure of the PKC activity modulator suramin.

In 2008, the first potent and selective cell-active small molecule inhibitor of PKD, benzoxoloazepinolone CID755673, was reported as part of a high-throughput assay of 196 173 compounds from the National Institutes of Health small molecule repository.²² In part due to its ready availability, CID755673 has become a probe for PKD and a potential therapeutic molecule for inflammation- and metabolic dysregulation-driven pathologies.^23,24

CID755673 introduced chemotype II as a novel class of selective PKD inhibitors and inspired the discovery of numerous potent analogs, including 9-hydroxy-3,4-dihydrobenzo[4,5]-thieno[2,3-f][1,4]thiazepin-5(2H)-one (2, kb-NB142-70) (Scheme 3).⁶ Several approaches to construct the fused tricyclic core system of these analogs have been reported. Benzylation of commercially available 3-hydroxycinnamic acid 24 and a Higa cyclization²⁵ with thionyl chloride provided the benzo-[b]-thiophene acid chloride intermediate, which was subsequently converted to methyl ester 25 with triethylamine and methanol. Condensation of 25 with cysteamine hydrochloride (26) in the presence of DBU resulted in the formation of the desired benzothienothiazepinone 27, and debenzylation with boron tribromide led to kb-NB142-70 in good yield. Interestingly, the replacement of the furan and methylene group in CID755673 with a thiophene and a sulfur atom, respectively, in 2 increased the potency more than 6-fold: CID755673 showed an IC₅₀ of 182 nM against PKD1, while 2 had an IC₅₀ of 28 nM.

Scheme 3.

Structure of HTS Hit CID755673, and Preparation of the Potent and Selective PKD Inhibitor kb-NB142-70⁶

To improve the plasma stability of 2, the pyrimidine kmg-NB4–23 was also prepared (Scheme 4).⁶ In this analog, an electron-deficient pyrimidine methyl ether was used in place of the phenol in 2 to reduce the rate of phase I and II metabolism of the latter moiety. Starting with commercially available 28, annulation of the pyrimidine was accomplished with potassium cyanate in acetic acid and base treatment to generate diol 29. The two hydroxy groups were converted to chlorides with POCl₃, and a palladium-catalyzed reductive dechlorination occurred exclusively at the C-4 position to yield 30. The methanolysis of the C-2 chloride generated a methyl ether intermediate that was brominated to obtain the desired 31. Selective deprotonation at C-6 with Knochel’s magnesium tetramethylpiperidite base and treatment with Mander’s reagent led to methyl ester 32, which was cyclized with 26 and completed the formation of kmg-NB4-23. In spite of the significant structural modifications compared to kb-NB142–70, this compound retained an IC₅₀ of 124 nM in the PKD1 assay.

Scheme 4.

Preparation of the Pyrimidine Analog kmg-NB4-23⁶

The SAR of these benzofuroazepine, benzothienothiazepine, and pyrimidothienothiazepine PKD inhibitors covers several orders of magnitude. Chart 3 illustrates the strong dependence of the PKD1 inhibitory activity on the nature of the heterocycle, the ring size of the fused lactam, and the phenolic substituent. Exchanging the furan with a pyrrole slightly improves the activity from 0.18 μM in CID755673 to 0.13 μM in kb-NB123-57. The thiophene analog kb-NB142-70 that also contains a thioether in the lactam tether is significantly more potent with an IC₅₀ of 28 nM. In contrast, enlarging the furan to a pyran core decreases the activity ca. 10-fold (CID755673 vs CID797718) and ca. 5-fold when the seven-membered lactam is enlarged to an eight-membered analog (kb-NB96-53). The methyl ether analog of CID755673, kb-NB77-56, also drops ca. 10-fold in activity, but the inhibitory potency is partly regained in the benzothiophene kb-NB184-02 and the pyrimidothiophene kmg-NB4-23.⁵ The latter modification is significant because it prevents O-glucoronidation and rapid elimination of the active component in in vivo experiments.⁷

Chart 3.

Chemical Structures and PKD1 Inhibitory Activities (IC₅₀’s) of Representative Chemotype II Analogs

Several other lead compounds that directly inhibit PKD or influence its activity were also reported in the patent literature between 2007 and 2009, and some of these leads were further developed or inspired the design of related structures (see, for example, chemotypes III and IV) (Figure 3).^26-29

Figure 3.

Representative examples of PKC/PKD inhibitors claimed in the early patent literature.

Also, during this time period of vivid interest in the therapeutic potential of PKD in cardiovascular diseases, Novartis and Myogen (Gilead) reported several heterocyclic scaffolds with strong PKD-inhibitory effects and claims covering a broad range of indications,^30-32 including cancer and heart failure³³ and, more recently, neurodegenerative disorders such as Friedreich’s Ataxia (FRDA).³⁴ For example, the bipyridyl PKD inhibitor 3 (BPKDi, 2′-(cyclohexylamino)-6-(1-piperazinyl)[2,4′-bipyridine]-4-carboxamide) was based on a hit from a time-resolved fluorescence resonance transfer (TR-FRET) library screen of >650 000 compounds from full-length, human PKD1 overexpressed in insect cells (Scheme 5).⁸ An earlier 2,6-naphthyridine scaffold analog proved potent and orally bioavailable but showed moderate kinase selectivity and comparative nanomolar off-target activity vs GSK3β.³⁵ In contrast, 3 inhibited PKD1, PKD2, and PKD3 with IC₅₀ values of 1, 9, and 1 nM, respectively, but did not inhibit other related kinases.⁸ Typical for SAR analyses of this chemotype, citrazinic acid (33) was converted to dibromide 34 followed by selective aminolysis and Suzuki coupling to generate the bipyridine backbone 38 (Scheme 5). Ester aminolysis, an S_NAr reaction with cyclohexylamine, and deprotection of the Boc group completed the synthesis of the lead compound, BPKDi (3).

Scheme 5.

Preparation of the Bipyridine BPKDi (3)⁸

Cancer Research Technology’s compound CRT0066101 (47, CRT101, (S)-2-(4-((2-aminobutyl)amino)pyrimidin-2-yl)-4-(1-methyl-1H-pyrazol-4-yl)phenol) represents another example of a chemotype III lead structure.^26,36 A large number of analogs of this lead structure were prepared using variations of the synthetic pathway shown in Scheme 6. For the modular assembly of the parent compound, aldehyde 39 was converted to nitrile 40 followed by aminolysis with lithium hexamethyldisilazide (LiHMDS) to give amidine intermediate 41. Condensation with ethyl propiolate and saponification provided hydroxypyrimidine 42, which was converted to the chloride, demethylated, and combined with Boc-protected 44 to attach the diamine side chain. Suzuki coupling of 45 with boronic acid 46 and Boc cleavage then led to 47.

Scheme 6.

Preparation of the 2-Arylpyrimidine 47²⁶

Meta-substituted aromatics are a popular pharmacophore in medicinal chemistry, and several of the PKD-inhibitory scaffolds take advantage of this motif. The orally bioavailable 3,5-diarylisoxazole 4 represents chemotype IV and was shown to inhibit PKD selectively with a low nanomolar EC₅₀, demonstrating ca. 9- and 3-fold selectivity for PKD1 versus PKD2 and PKD3.⁹ For the preparation of 4 and related 3,5-diarylazoles, a nitrile oxide cycloaddition of chloroimidate 48 and alkyne 49 was used to access the isoxazole moiety in 50 (Scheme 7). Radical bromination of the methyl group and substitution of the benzylic bromide with amine 51 provided intermediate 52, which was saponified and coupled with alanine amide 53. Treatment of amide 54 with trifluoroacetic anhydride (TFAA) led to a nitrile containing a trifluoroacetyl group at the benzylic amine. The trifluoroacetamide was then removed chemoselectively by treatment with sodium borohydride to yield inhibitor 4.

Scheme 7.

Preparation of the 3,5-Diarylisoxazole 4⁹

The SAR of 4 reveals a strong dependence of the inhibitory activity on the end groups of the meta-linked arene triad (Chart 4). The enantiomeric aminonitrile in 4′ increases the IC₅₀ ca. 15-fold, establishing a strong stereoselectivity in the PKD1 binding mode. The polar nitrile function is also required for low nanomolar inhibition since the isopropylamide 4a is ca. 35 times less active than 4. In contrast, exchanging the 4-aminopyran in 4 with the isopropylamine in 4b has only a minor effect on the IC₅₀. Isoxazole 4b and pyrazole 4c also have single-digit nanomolar activity, but similar to the isoxazole series, removing the nitrile from the terminal amide in isopropylamide 4d diminishes inhibition ca. 25-fold. Pyrazoles 4e and 4f have even lower activities, with IC₅₀’s of 1.0 and 2.5 μM, respectively.⁹ The structurally related furan CID2011756 has similarly moderate activity for PKD1 but ca. 5-fold increased activity for the other PKD isoforms with IC₅₀ values of 3.2, 0.6, and 0.7 μM, respectively, for PKD1, PKD2, and PKD3.³⁷ CID2011756 might therefore be an interesting lead structure for the design of PKD2/3-selective inhibitors.

Chart 4.

Chemical Structures and PKD1 Inhibitory Activities (IC₅₀’s) of Representative Chemotype IV Analogs

The pyridine benzamide CRT0066051 (61, CRT51, CRT5, 3-(6-amino-5-(6-ethoxynaphthalen-2-yl)pyridin-3-yl)-N-(2-(dimethylamino)ethyl)benzamide) is another example of a structural variation of chemotype IV (Scheme 8).³⁸ Coupling of acid 55 with diamine 56 in the presence of EDCI provided amide 57 in 63% yield. The meta-substituted pharmacophore was then readily assembled by two Suzuki couplings first with bromide 58 and then boronate 60 to give the target inhibitor 61. Both CRT0066051 (61) and CRT0066101 (47) have low-nanomolar pan-PKD inhibitory activities and have been used to characterize the role of PKD in various biological processes and disease models. A potential concern for in vivo applications of CRT0066051 is the low aqueous solubility at pH 7 for this compound, calculated at log S = −5.9, versus log S = −3.4 for CRT0066101,³⁹ which might also limit dosing regimens and bioavailability.

Scheme 8.

Preparation of the Pyridine Benzamide 61²⁷

In 2017, CRT0066101 and CRT0066051 were reported to inhibit the replication of HRV, PV, and FMDV, suggesting that PKD may also present opportunities in antiviral drug discovery.⁴⁰ Furthermore, in 2019, an industrial collaboration disclosed the structure-based design of potent, conformationally restricted imidazole inhibitors that broadly fit into this chemotype (Chart 5).⁴¹ While Im-1 inhibited PKD1 in an ADP FP assay with high stereoselectivity over its enantiomer Im-1′, it was also a potent inhibitor of NF-κB-inducing kinase (NIK). Similarly, oxazole Im-2 in spite of its increased PKD1 activity was still very potent in the NIK assay. High selectivity for PKD1 over NIK was observed with the cyclopropyl-substituted Im-3. While the ether-bridged Im-4 proved to be selective for NIK over PKD1, the desired preference for PKD1 was regained by introducing the cyclopropyl group in Im-5. In fact, among 220 kinases tested, Im-5 only inhibited PKD1 >30% at 0.1 μM concentration. Interestingly, Im-5 has the (S)-configuration, which was the less-preferred stereochemistry for the PKD1 inhibitory potency of Im-1 but provided a 10-fold increased selectivity over NIK. Unfortunately, PKD2 and PKD3 inhibitory data are not available for these compounds.

Chart 5. Chemical Structures and PKD1 Inhibitory Activities (K_i’s) of Conformationally Rigidified Chemotype IV Analogs^a.

^aThe ratio of NIK/PKD1 K_i’s is shown in parentheses. The higher the ratio, the greater the selectivity for PKD1.⁴¹

A screen of a diverse collection of kinase-biased inhibitors in 2012 revealed additional scaffolds for PKD inhibition (Figure 4).¹⁰ In particular, the 4-azaindole chemotype V exemplified by compound 139 (5) distinguished itself as a solid ATP-competitive lead structure, responsive to structural modifications, with a low double-digit nanomolar potency and robust cell activity. Furthermore, this chemotype offers the hitherto untapped opportunity for dual PKD-p38 MAP kinase inhibitors.⁴² In addition to 5, 4-azaindoles 68 and 69 demonstrated a moderate to high level of inhibition of PKD1 with the indole NH being a prerequisite for the highest degree of inhibition. Lead compound 5 also lacked activity for undesired kinase targets (PKCα, PKCδ, and CAMKIIα).¹⁰

Figure 4.

Chemical structures of novel PKD1 small molecule inhibitors found as primary hits in a radiometric PKD1 assay. Hits were selected based on their ability to inhibit PKD1 at or above 50% at 1 μM (percent inhibition in parentheses).¹⁰

The preparation of 4-azaindole inhibitor 5 used dimethylacetal 70 for a chain extension with 4-fluorobenzyl bromide to generate N-oxide 71 after oxidation with methyltrioxorhenium (Scheme 9). Condensation with aminopyridine 72 and Pd-catalyzed Heck cyclization of the enamine 73 provided the azaindole 74, which was chlorinated and treated with amine 76 to complete the synthesis of 5.⁴²

Scheme 9.

Preparation of 4-Azaindole 5⁴²

In 2013, the generally well-known kinase-inhibitory pyrazolopyrimidine chemotype VI was also found to act as a strong inhibitor of PKD1–3.¹¹ In particular 1-naphthyl-PP1 (6, 1-NA-PP1, 1-(tert-butyl)-3-(naphthalen-1-yl)-1H-pyrazolo[3,4-d]-pyrimidin-4-amine) was found to inhibit PKD isoforms at about 100 nM as an ATP-competitive inhibitor. However, among 17 synthetic analogs of 6, only 83 provided a minor modulation of PKD1 activity, confirming that the sterically demanding 6 already provides a tight fit in the binding pocket. This property could potentially be used in the future for the development of analogs of 6 and 83 as PKD subtype-specific inhibitors. The synthesis of 83 and related compounds was accomplished starting with a Vilsmeier–Haack reaction of barbituric acid 77 to give the trichloropyrimidine 78 in 57% yield (Scheme 10). The pyrazole ring was closed by treatment of 78 with methyl hydrazine in the presence of triethylamine to yield 59% of the pyrazolo[3,4-d]pyrimidine 79. Selective aminolysis at C-6 with ammonia at room temperature led in 81% yield to 80, which was brominated to give the pyrazolopyrimidine 81 in 40% yield. A selective Suzuki coupling at C-3 with 1-naphthyl boronic acid provided intermediate 82, and the chlorine group was reduced by a catalytic hydrogen transfer process to give 83 in 68% yield over two steps.

Scheme 10.

Preparation of the Pyrazolopyrimidine 83¹¹

More recently, an analog of 6, 3-indolylmethyl-pyrazolo[3,4-d]pyrimidine (3-IN-PP1), was found to be significantly more potent with an IC₅₀ of 33 nM against PKD2.⁴³

Shortly after the discovery of the PKD-inhibitory effects of the pyrazolopyrimidine chemotype, a targeted kinase inhibitor library screen revealed the pteridine chemotype VII 2-(5-chloro-2-fluorophenyl)-N-(pyridin-4-yl)pteridin-4-amine (7, SD-208) as an ATP-competitive, cell-active pan-PKD inhibitor.¹² The SAR profile of 7 proved to be rather narrow, and among diverse substitutions, only analogs with the 4-aminopyridine moiety maintained a low nanomolar potency. For the preparation of the lead compound as well as most analogs, methyl 3-aminopyrazine-2-carboxylate (84) was converted to the imide 86 with an excess of acid chloride 85 (Scheme 11). Cyclization of 86 in the presence of ammonium hydroxide provided the pteridine 87 in high yield. Anilines such as 88 were then fused to the heterocyclic core in the presence of the coupling reagent PyBOP to give target compound 7.

Scheme 11.

Preparation of Pteridine 7¹²

Chemotype VIII has not yet been rendered specific for PKD but was found to be a potent inhibitor of this kinase in an off-target panel assay of the aurora kinase inhibitor AT9283 (8).¹³ This pyrazol-4-yl-urea is best described as a multitargeted kinase inhibitor, and interestingly, several benzimidazole fragments also showed submicromolar PKD activity.³ Therefore, it is quite feasible that more PKD-selective analogs of 8 could be developed in future SAR campaigns.

Quite recently, a series of quinazolines such as 9 (NK-176, N-(2-(4-methylpyridin-2-yl)quinazolin-4-yl)benzo[d]oxazol-6-amine) was reported in a phenotypic assay for kinase regulators of lysosome function.¹⁴ Several of these quinazolines proved to have submicromolar PKD1 inhibitory activities. This chemotype IX is structurally related to the previously reported chemotypes III, VII, and VIII and has the potential to provide more potent analogs upon further optimization. Interestingly, 9 and its analogs have also been found as potent inhibitors of activin-like kinase (ALK5) with moderate activity on p38,⁴⁴ thus further highlighting the structural connection between PKD and these kinases and the need to perform kinase panel studies in order to determine particular selective PKD and multitargeted kinase activity profiles. The SAR of chemotype IX is exemplified in Chart 6. Moving from the benzoxazole 9 to the benzothiazole NK-140 did not change the measured PKD1 K_D value, but replacing the linker NH with an oxygen atom obliviated the kinase activity in NK-164, which therefore served as a negative control in the transcription factor E3 (TFE3) assays.¹⁴ Related benzofuran, benzoxazole, and indole analogs also showed decreased activity, suggesting that the position of the nitrogen atom as a hydrogen-bond acceptor in the benzazole and the position of the NH linker and hydrogen-bond donor are critical requirements for low nanomolar activity.

Chart 6.

Chemical Structures and PKD1 Inhibitory Activities (K_D’s) of Representative Chemotype IX Analogs¹⁴

Protein kinases can be inhibited reversibly and irreversibly by alkylation of key cysteine residues with suitable Michael acceptors, as exemplified by chemotype X.⁴⁵ This strategy has also been applied to PKD, which contains a conserved Cys residue in the ATP-binding site that is susceptible to Michael acceptors. The resorcyl lactone hypothemycin (10) demonstrated a 94% inhibition of PKD1 at 2 μM concentration in vitro;¹⁵ however, PKD1 was not readily inhibited by 10 in cells.⁴⁶

Structural Studies of PKD Inhibitors.

The crystal structure of PKD1 (or PKD2,3) is not yet available in the public domain, but several homology models have been built and used for docking studies with known inhibitors.⁴⁷ Another strategy for studying the binding pose of an inhibitor and applying structure-based design strategies is to solve inhibitor cocrystal structures with surrogate kinases, such as Map4K4.⁴¹ In a recent approach, a 3D architecture of PKD1 was generated from a close homologue, the calcium/calmodulin-dependent kinase CaMK-1G (PDB ID: 2JAM), subjected to molecular dynamics (MD) simulations for optimization, and then used to screen drug-like molecules, resulting in a series of interesting new potential inhibitors that have yet to be experimentally verified.⁴⁸

Figure 5 shows a typical docking analysis for two structurally closely related inhibitors, NPKDi and BPKDi, conducted using AutoDock Vina.⁴⁹ The hydrophobic pyridines fit the adenosine binding region, replacing the adenosine of ATP in a binding mode typical for an ATP-competitive type I inhibitor. In addition to interactions with the gatekeeper residue M659, the hinge loop that forms the hydrophobic binding pocket II, K612, a highly conserved residue required for the binding of ATP in protein kinases, is also engaged. Modification of the 2,6-naphthyridine in NPKDi with functional groups that are able to hydrogen bond to K612 can result in a significant improvement in the specificity of BPKDi, as shown in the docking model Figure 5C and 5D.^35,61

Figure 5.

Docking analyses of PKD1 inhibitors. (A) Structures of inhibitors NPKDi and BPKDi. (B) Docking modes in the ATP binding region. (C and D) Key PKD1 residue interactions.

Biochemical and Cellular Properties of PKD Inhibitors.

PKD small molecular inhibitors have been evaluated both as tool compounds and as potential clinical candidates in various biological systems and disease models. In the following sections, an overview of the in vitro and in vivo applications of the I–X chemotypes of PKD inhibitors is presented, highlighting the most significant findings and their implications for future drug development targeting PKD. This discussion mainly focuses on the cross-comparison of PKD inhibitors from different studies in an effort to reinvigorate PKD-targeted drug development and facilitate a selection of proof-of-concept inhibitors for future studies. Several inhibitor classes have also been reviewed in two previous reviews.^3,50

Potency.

Table 1 compares the in vitro properties of representative PKD inhibitors in each chemotype. It is clear that several scaffolds have been optimized to reach single-digit nanomolar potencies, particularly in chemotype III, i.e., NPKDi,³⁵ BPKDi,⁶¹ and CRT0066101.^36,62 At cellular levels, submicromolar IC₅₀ values have been demonstrated for the three Novartis PKD inhibitors (NPKDi, BPKDi, and 3,5-diarylisoxazole 4) in the range of 32–240 nM for inhibiting PKD1-mediated GFP-HDAC5 nuclear export in cardiac myocytes,^9,35,61 resulting from a direct phosphorylation of HDAC5 by PKD1.⁵¹ However, despite the remarkable potencies at the biochemical level, the on-target inhibition of PKD1 autophosphorylation at S916 in intact cells stalls in the micromolar range. For example, CRT0066101 inhibits p-S916-PKD1 autophosphorylation with an IC₅₀ of 6 μM in PANC-1 pancreatic cancer cells,³ displaying a remarkable >1000-fold difference between biochemical and cellular potencies. Note that this difference is not uncommon for ATP-competitive inhibitors, although it is yet to be determined whether CRT0066101 competes with ATP. In line with this pattern of activity shifts, the cellular biological activities of these inhibitors are also mostly in the micromolar range with only a few reported to reach submicromolar potency. For example, CRT0066101, the most potent PKD inhibitor in terms of anticancer activity in vitro, has been reported to inhibit tumor cell viability with an IC₅₀ of 0.6–1.9 μM.⁶² Interestingly, chemotype II compounds (CID755673, kb-NB142-70, kmg-NB4-23), despite being noncompetitive with ATP, also exhibit a significant shift (<100-fold) in IC₅₀ from biochemical to cellular environments.^5,60 Binding to serum proteins might partially account for a potency drop; however, kb-NB142-70 was 78% plasma protein bound at both 0.3 and 3 μg/mL, and kb-NB165-09 was 88% and 94% bound to plasma protein at 0.3 and 3 μg/mL, respectively.⁷ These relatively low levels of protein binding imply that other factors in addition to ATP concentration and plasma protein binding affect the activity of PKD inhibitors in cells. These factors might primarily influence the bioavailability of the inhibitors and include general compound stability, solubility, size, and cell permeability properties as well as effects of efflux pumps. For example, the TPSA of CRT0066101 is 102 Å^2,39 which suggests that the passive molecular transport of this molecule through cell membranes is likely a limiting factor. The impact of these ATP-independent properties could be evaluated by comparing the target binding affinity of the inhibitors in different experimental settings. Currently, there is scant information on the binding properties of PKD inhibitors; only two (compound 139 and NK-176) were evaluated in an active-site-directed competition binding assay (KINOMEscan) for kinase profiling. NK-176 was also examined using KiNativ, which measures small molecule binding to kinases in cell lysates, providing an opportunity to assess the impact of cellular factors other than ATP on the bioavailability of the inhibitor. Interestingly, despite the high binding affinity (K_D = 53 nM) at the biochemical level, NK-176’s cellular binding activity was much lower, ranging from 17% to 59% for PKD1–3 at 2 μM (correlating to a probable >40× drop in binding affinity) measured in the KiNativ assay, suggesting that ATP-independent factors could significantly impact the efficacy of the PKD inhibitors in cells. In addition, tumor cells are also known to have higher levels of ATP, which can further reduce the cellular efficacy of inhibitors.

Table 1.

Summary of Known PKD-Targeted SMIs^a

chemotype	compound name	biochemical IC50 (nM)	cellular IC50b (μM)	kinases inhibitedc (#/#,d %,e μMf)	ref
I	staurosporine	27 (PKD1)	n.d.	nonselective	58
		72 (PKD2)
		18 (PKD3)
		ATP competitive
	K252a	7 (PKD1)	n.d.	nonselective	4
		ATP competitive
	Gö6976	20 (PKD1)	n.d.	cPKC (no profiling data)	4, 59
		ATP competitive
	Gö6983	20 000 (PKD1)	n.d.	c/n/aPKC (no profiling data)	4, 59
		20 000 (PKD1)
II	CID755673	182 (PKD1)	11.8 (PKD1), PKD1 autophosphorylation at S916 in LNCaP cells	MK2, GSK-3/9, CK1δ, MK5/PRAK, CDK2, and ERK1 (6/48, ≥50%, 10 μM)	5, 22, 60
		280 (PKD2)
		227 (PKD3)
		non-ATP competitive
	kb-NB142-70	28 (PKD1)	2.2 (PKD1), PKD1 autophosphorylation at S916 in LNCaP cells	similar to CID755673	6, 60
		59 (PKD2)
		53 (PKD3)
		non-ATP competitive
	kmg-NB4-23	120 (PKD1)	6.8 (PKD1), PKD1 autophosphorylation at S916 in LNCaP cells	n.d.	5, 6
		non-ATP competitive
III	NPKDi (13c)	0.6 (PKD1)	0.032 (PKD1), GFP-HDAC5 nuclear export in cardiac myocytes	GSK-3α/β, hCaMKIIδ, hMARK1/2 (8/197, > 80%, 1 μM)	35, 61
	BPKDi (12a)	1 (PKD1)	0.077 (PKD1), GFP-HDAC5 nuclear export in cardiac myocytes	none (3/197, > 80%, 1 μM)	61
		9 (PKD2)
		1 (PKD3)
	CRT0066101	1 or 9 or 16 (PKD1)	6.0 (PKD1), PKD1 autophosphorylation at S916 in PANC-1 cells	CDK5, DYRK2, Pim-1 (5/127, > 90%, 1 μM)	36, 40, 62
		2.5 or 8.1 or 9 (PKD2)
		2 or 7.7 or 6 (PKD3)
IV	CRT0066051	1 or 9.2 (PKD1)	6.6 (PKD1), PKD1 autophosphorylation at S916 in PANC-1 cells	BRK (3/127, >90%, 1 μM)	38, 40
		2 or 8.4 (PKD2)
		1.5 or 7.6 (PKD3)
	3,5-diarylisoxazole 4	5.5 (PKD1)	0.24 (PKD1), GFP-HDAC5 nuclear export in cardiac myocytes	(4/230, ≥ 50%, 1 μM)	9
		48 (PKD2)
		17 (PKD3)
V	compound 139	16.8 (PKD1)	1.5 (PKD1), PKD1 autophosphorylation at S916 in LNCaP cells	(43/353, ≥50%, 10 μM); p38α/β, JNK1/2, STK36/21, CK1ε (8/353, 99–100%, 10 μM)g	10
VI	1-NA-PP1	155 (PKD1)	22.5 (PKD1), PKD1 autophosphorylation at S916 in LNCaP cells	p38α, CK1δ, Src, Lck, CSK, RIP2 (7–11/~70, >50%, 1 μM)	11, 63
		133 (PKD2)
		104 (PKD3)
	3-IN-PP1	108 (PKD1)	effective at 20 μM for blocking PDBu-stimulated cortactin phosphorylation at S298	n.d.	43, 62
		94 (PKD2)
		108 (PKD3)
	compound 17m	35 (PKD1)	effective at 20 μM for blocking PDBu-stimulated cortactin phosphorylation at S298	n.d.	62
		35 (PKD2)
		17 (PKD3)
VII	SD-208	107 (PKD1)	17.0 (PKD1), PKD1 autophosphorylation at S916 in LNCaP cells	TGFβR1 (no profiling data)	12
	94 (PKD2)
	105 (PKD3)
	ATP competitive
VIII	AT9283 (8)	10 (PKD1)	n.d.	aurora kinase A/B, JAK2/3, Abl (T315I), Flt-3, CDK2, etc. (28/144, IC50 > 30 nM)	13
IX	NK-176	53 (PKD1)	n.d.	none (3/392, >90%, 1 μM)	14
X	hypothemycin	1200 (PKD1)	not effective up to 4 μM on PKD1 autophosphorylation	n.d.	15, 46
	covalent inhibitor

^aThe in vitro and cellular IC₅₀ values of known PKD SMIs are listed.

^bCellular IC₅₀, inhibition of PKD1 activity (autophosphorylation at S916), or PKD1-regulated downstream event (HDAC5 nuclear export) in intact cells.

^cPKD is not listed but included when counting the number of hits.

^dNumber of hits (including PKD)/total number of kinases profiled.

^ePercent inhibition as cutoff in kinase profiling.

^fConcentration used in kinase profiling.

^gKinase profiling was performed on compound 140,¹⁰ which is similar in structure and activity to compound 139. n.d. not determined.

Selectivity.

Another key aspect of protein kinase inhibitor development is selectivity. Kinase profiling data at variable scale levels are available for several PKD inhibitors. Among them, NK-176 appears to be a remarkably selective PKD inhibitor. In a large-scale kinase profiling analysis encompassing 392 kinases, PKD1–3 emerged as the sole hits with >90% inhibition at 1 μM, demonstrating an almost exclusive specificity for the PKD family.¹⁴ Additionally, the selectivity profile of 3,5-diarylisoxazole 4 was also impressive.⁹ In a kinase profiling analysis of 230 kinases, only 4 hits with over 50% inhibition at 1 μM were identified.⁹ CRT0066101 and CRT0066051 also show high selectivity for PKD with only three (CDK5, DYRK2, Pim-1) and one (BRK) additional targets inhibited >90% at 1 μM besides PKD in a profiling involving 127 kinases. BPKDi exhibited an improved selectivity profile compared to NPKDi with no additional kinases inhibited >80% at 1 μM besides PKD1–3 in a panel of 197 kinases, as compared to five (GSK-3α/β, hCaMKIIδ, hMARK1/2) hits for NPKDi inhibited at >80% out of 197 kinases examined at 1 μM.^35,61 The 4-azaindole PKD inhibitors compounds 139 and 140, also reported as potent p38α/β inhibitors, similarly exhibited an attractive selectivity profile with eight (p38α/β, JNK1/2, STK36/21, CK1ε) additional kinases besides PKD inhibited at the 99–100% level at 10 μM in a large-scale profiling involving 353 kinases.¹⁰ Compared to these highly selective PKD inhibitors that belong to chemotypes III, IV, V, and IX, inhibitors classified in chemotypes II, VI, and VIII are relatively less selective based on the number of off-target kinases, the profiling scale, and the concentration of inhibitor used in the analysis. Structural studies of chemotypes with favorable selectivity profiles may therefore provide additional insights to aid in the design of more selective PKD inhibitors.

In summary, extensive in vitro analyses have been conducted on PKD inhibitors over several decades. It is clear from these studies that highly potent and selective PKD inhibitors of distinct structural classes have been successfully identified. Some of these inhibitors exhibit nanomolar potencies for PKD and on-target cellular PKD inhibition in cells in the low micromolar range. They also exert a wide variety of biological activities at cellular levels that are in agreement with the multitude of PKD biological functions in different biological systems. Future work should be directed at side-by-side analyses of these inhibitors at biochemical and cellular levels to validate the results reported in a diverse assay systems and performed by different research groups. In particular, kb-NB142-70, CID2011756, CRT0066101, CRT0066051, compound 139, Im-5, and NK-176 would form an attractive panel for such a cross-comparison.

In Vivo Evaluation of PKD Inhibitors.

PKD plays an important role in the pathogenesis of multiple diseases, including stress-induced cardiac hypertrophy, inflammatory diseases, metabolic disorders, and cancer. A total of six different PKD inhibitors from chemotypes I (Gö6976), II (CID755673), III (NPKDi, BPKDi, CRT0066101), and VII (SD-208) have been evaluated in vivo. These preclinical studies can be separated into two cohorts based on the disease models being (1) acute, which only requires short-term inhibitor application, or (2) chronic, which requires repeated long-term administration of inhibitors. In the following sections, inhibitors that have been tested in these two types of in vivo applications will be discussed. Table 2 provides a detailed comparison of the application of PKD inhibitors in these animal models, detailing their doses, routes, and schedules of administration, the animal species used, the outcome of the evaluation, and their underlying mechanisms of actions.

Table 2.

In Vivo Application of PKD Inhibitors^a

chemotype	compound	disease	dose, route, schedule and species	in vivo efficacy	mechanisms	ref
I	Gö6976	bone repair	2.35 mg/kg, ip, once daily for 4 days before growth plate tissue collection at day 10 in rats	reduction in bone formation and more cartilage repair in the growth plate injury site in rats	by decreasing bone-related genes (osterix, osteocalcin) and increasing cartilage-related genes (Sox9, collagen-2a, collagen-10a1)	64
		bacterial infection-induced inflammation, liver injury, sepsis	(1) 2.5 mg/kg in DMSO, ip, once 30 min before LPS/D-GalN challenge in mice	(1) inhibited LPS/D-GalN-induced acute liver injury in mice	(1) by inhibition of MAPKs activation to reduce TNF-α production in liver	65, 66
			(2) 2.3 mg/kg in 7.6% DMSO in PBS, ip, 1 and 4 h before or 2 h after the antibiotic killed the GBS challenge	(2) protected d-GalN-sensitized mice from shock-induced death caused by antibiotic-killed GBS	(2) by inhibiting MAPKs and NF-kB and expression of proinflammatory cytokines/chemokines
II	CID755673	pancreatitis	15 or 20 mg/kg in DMSO, ip, once 60 min before or after cerulein treatment in rats	reduction in necrosis, inflammatory responses, and severity of pancreatitis and similar effects in alcoholic pancreatitis	by inhibiting NF-κB-mediated inflammatory responses, trypsinogen activation, and several cell death mediators (ATP, Bcl-2, RIPK1)	67-69
		diabetes	10 mg/mL in 50% DMSO, osmotic pump, 2 weeks in mice	accelerated onset of diabetes by increasing fasting glucose in BTBRob/ob mice and decreased insulin secretion	by exacerbating stress-induced nascent granule degradation (SINGD) and inhibiting insulin secretion	70
			1 or 10 mg/kg, ip, once (acute) or once daily for 16 days (chronic) in mice	enhanced heart function and reduced heart weight in type 2 diabetes db/db mice	by improving both systolic and diastolic cardiac function with unknown mechanisms	71
		myocardial infarction	10 mg/kg in saline, iv, every other day for 4 weeks in rats	inhibited astragaloside IV-induced angiogenesis of myocardial tissue in rats with myocardial infarction	by blocking the PKD1-HDAC5-VEGF pathway	72
		learning and memory	182 nmol, infusion into the bilateral hippocampal CAI areas, before DHPG (a group I mGluRs) agonist in rats	prevented DHPG-induced learning and memory impairments in rats	possibly by affecting NMDAR internalization and activity	73
III	NPKDi (13c)	cardiac hypertrophy	(1) 5, 15, 50 mg/kg in vehicle (0.5% methylcellulose and 0.5% Tween-80), po, once daily for 14 days in a DSS rat model	reduced high salt-induced cardiac hypertrophy at the 50 mg/kg po dose, but no effect in pressure overload-induced cardiac hypertrophy at the 15 mg/kg sc dose	by blocking the phosphorylation and nuclear export of the MEF2 repressor HDAC in cardiac myocytes	35
			(2) 15 mg/kg in acidified captisol, sc, once daily for 12 days in a TAB rat model
	BPKDi (12a)	cardiac hypertrophy	0.5, 1.5, 5 mg/kg in acidified captisol, sc, once daily for 14 days in rat models	did not affect cardiac hypertrophy at 5 mg/kg sc dose in high salt- and pressure overload-induced cardiac hypertrophy rat models	by inhibiting phosphorylation and nuclear export of HDAC4/5 in cardiac myocytes in vitro	61
	CRT0066101	pancreatic cancer	80 mg/kg in 5% dextrose, po, once daily for 21 days in nude mice	inhibited PANC-1 tumor xenograft growth in a subcutaneous and an orthotopic pancreatic cancer models in nude mice	by blocking NF-κB activation and NF-κB-mediated tumor cell proliferation and survival	36
			80 mg/kg in 5% dextrose, po, every other day for 14 days before sacrifice at day 24 in immunocompetent MD4 mice	suppressed the growth of orthotopic KPC4662 pancreatic xenograft tumors in immunocompetent mice	by inhibiting cancer-cell autonomous tumor-promoting function of PKD in PDAC xenografts	74
		colon cancer	40, 80, 120 mg/kg in 5% dextrose, po, once daily for 21 days in mice	inhibited HCT116 colorectal tumor xenograft growth in nude mice	by inducing G2-M arrest and apoptosis via inhibiting Akt, ERK, and NF-κB in colorectal cancer cells	75
		breast cancer (ER− and TNBC)	80 mg/kg in 5% dextrose, po, every other day for 8 weeks in NOD scid mice	inhibited primary tumor growth, local invasion, and metastasis to distant organs in an orthotopic mouse tumor xenograft model of ER− breast cancer	by blocking the upregulated PKD3-mediated tumor cell proliferation, survival, and invasion in ER− cancer cells	76
			65 mg/kg, po, 5 times a week for 4 weeks in nude mice	inhibited subcutaneous MCF-7-ADR tumor xenograft growth in nude mice	by suppressing breast cancer stemness through GSK3/μ-catenin signaling	77
			50 or 80 mg/kg in 5% dextrose, po, once daily for 3 or 6 weeks in nude mice	inhibited TNBC tumor xenograft growth in nude mice	by causing G1 arrest and regulating phosphorylation of MYC, MAPK1/3, AKT, YAP, and CDC2 and by targeting a PKD3/CLU pathway in TNBC	78, 79
		bladder cancer	120 mg/kg in 5% dextrose, po, 3 days per week for 25 days in nude mice	inhibited bladder tumor xenograft growth in nude mice	by inducing cell cycle arrest via reducing CDK1/cyclin B complex activity	80
		liver metastasis	80 mg/kg in 5% dextrose, po, once daily for 28 days in nude mice	reduced lung metastasis of subcutaneously implanted hepatocellular carcinoma cells in nude mice	by inhibiting TNFα-induced EMT and HCC metastasis	81
		tumor immune response	50 mg/kg in PBS, ip, once at day 7 before sacrifice at day 21 alone or in combination with PD-1 in C57BL/6 mice	blocked the anticancer effect of ant-PD-1 therapy	by antagonizing T-cell reactivation by anti-PD-1 therapy via inhibiting Akt	82
		Obesity and diabetes	10 mg/kg in water, po, daily for 13 weeks in C57BL/6 mice	decreased intestinal fat absorption, reduced high-fat diet-induced obesity and diabetes and improved gut microbiota profile in mice	by inhibiting triglycerides absorption in the intestine by regulating APOA4	55
		anxiety in myocardial infarction (MI)	80 mg/kg, po, once daily for 14 days in rats	attenuated anxiety-like behaviors in MI rats	by blocking platelets granule release, thereby reducing serum IGF1 in MI rats	83
		pancreatitis	20 mg/kg in DMSO, ip, once 60 min before or after cerulein treatment in rats	reduction in inflammatory responses, cell necrotic death and the severity of disease in alcoholic pancreatitis	by inhibiting NF-κB, trypsinogen activation, and several cell death mediators (ATP, Bcl-2, RIPK1)	69
VII	SD-208	prostate cancer	60 mg/kg in 1% methylcellulose, po, twice daily for 21 days in nude mice	abrogated the growth of PC3 subcutaneous tumor xenografts in nude mice	by inducing G2/M arrest through induction of p21 and increased p-Cdc2 and downregulation of Cdc25c	12

^aThe list contains PKD inhibitors that have been examined in vivo, including their dosages, routes, and schedules of administration, animal species used, in vivo efficacy, and underlying mechanisms in various disease models.

Short-Term Application of PKD Inhibitors In Vivo.

The diseases evaluated in this category are primarily inflammatory diseases, including pancreatitis and bacterial infection-induced inflammatory conditions. Under these acute conditions, PKD inhibitors were mostly administered once before or after the initiation of the pathological conditions, and the animals were evaluated within a few hours to several days. Besides the studies discussed below, a short-term application of PKD inhibitors has also been also reported in a study on bone repair using Gö6976⁶⁴ and on learning and memory functions using CID755673,⁷³ as shown in Table 2.

Inflammatory Conditions Caused by Bacterial Infection.

The PKC/PKD inhibitor Gö6976 was evaluated in two studies involving bacterial infection-induced inflammation, liver injury, and septic shock. It was found that one-time intraperitoneal (ip) injection of Gö6976 at 2.5 mg/kg 30 min before the LPS/d-galactosamine (d-GalN) challenge that mimics bacterial infection blocked liver injury caused by LPS/D-GalN. Gö6976 acts by inhibiting MAPKs and subsequent TNF-α production in mouse liver.⁶⁵ In a similar study in d-GalN-sensitized mice, one-time ip injection of 2.3 mg/kg Gö6976 before or after GBS challenge significantly protected mice from shock-induced death by inhibiting MAPKs and NF-kB and decreasing expression of the proinflammatory cytokines/chemokines.⁶⁶ Since Gö6976 is a dual inhibitor of PKC and PKD, it is possible that its anti-inflammatory effects are due to the inhibition of both kinases.

Inflammatory Injuries in Pancreatitis.

Acute pancreatitis is an inflammatory condition of the pancreas that causes pancreatic acinar cell necrosis and in its severe form could lead to single or multiple organ damage and death. In three separate reports, the protective effects of two PKD inhibitors, CID755673 and CRT0066101, were demonstrated in a rat model of cerulein-induced pancreatitis. A single ip injection of CID755673 at 15 mg/kg before or after cerulein treatment significantly reduced inflammatory responses, necrosis, and severity of pancreatitis.^67,68 Both CID755673 and CRT0066101 given once at 20 mg/kg ip before or after cerulein treatment also effectively attenuated inflammation, necrotic cell death, and disease severity in an alcohol-aggravated pancreatitis.⁶⁹ Mechanistically, these inhibitors mainly act through inhibiting NF-κB-mediated inflammatory responses^67,68 and by modulating mitochondrial ATP production, Bcl-2 family proteins, and RIP1 kinase, as well as trypsinogen activation.⁶⁹ In these studies, the anti-inflammatory and protective effects of CID755673 are compelling in both alcoholic and nonalcoholic pancreatitis, indicating a broader applicability. Since CID755673 is less potent and selective compared to CRT0066101 whose effect was only apparent for alcoholic pancreatitis, it is possible that the beneficial effects of CID755673 may involve other off-target kinases. Nonetheless, the anti-inflammatory activities of CID755673 in vivo remain attractive for future drug development.

Long-Term Application of PKD Inhibitors In Vivo.

PKD inhibitors have been evaluated in several preclinical disease models that required repeated application over an extended period of several weeks to months. Below, we provide a review on the three main diseases in this category, i.e., cardiac hypertrophy, metabolic disorders (obesity and diabetes), and cancer. PKD inhibitors in chemotypes II (CID755673), III (NPKDi, BPKDi, CRT0066101), and VII (SD208) have been evaluated in this setting. Besides these diseases, two studies reported the use of PKD inhibitors in myocardial infarction (CID755673) and anxiety associated with myocardial infarction (CRT0066101), details of which are listed in Table 2.

Cardiac Hypertrophy.

Studies using genetically modified mouse models have firmly established PKD1 as a therapeutic target for stress-induced cardiac hypertrophy.^52,53 This has spurred significant interest in developing PKD inhibitors as potential therapeutic agents for cardiac hypertrophy. Novartis led this effort by contributing three highly potent and selective PKD inhibitors (NPKDi, BPKDi, and 3,5-diarylisoxazole 4). Two of these (NPKDi, BPKDi) have been evaluated in vivo in two distinct rat models of cardiac hypertrophy (high salt and pressure overload). The NPKDi 13c, despite being less selective, significantly reduced both PKD autophosphorylation and high salt-induced cardiac hypertrophy given orally at 50 mg/kg, but it had no effect in pressure overload-induced cardiac hypertrophy when administered subcutaneously at 15 mg/kg. However, BPKDi 12a, a more selective analog of NPKDi, was not effective at inhibiting PKD autophosphorylation and cardiac hypertrophy when administered subcutaneously at 5 mg/kg. It appears that blood pressure increase was one of the dose-limiting factors for BPKDi. Unfortunately, no further development in this area has been reported since 2010.

Diabetes and Obesity.

Although growing evidence has identified PKD as a critical metabolic regulator in multiple organs,^2,54 PKD inhibitors have not been examined extensively in metabolic diseases. PKD modulates insulin secretion from pancreatic β cells, and its dysregulation results in hyper-insulinemia and systemic insulin resistance that underlie metabolic disorders.^2,54 In Type 2 diabetes, metabolic stress causes elevated stress-induced nascent insulin granule degradation (SINGD), a process negatively regulated by PKD. It was found that inhibition of PKD by CID755673 applied by osmotic pump for 2 weeks accelerated diabetes by increasing SINGD and reducing insulin secretion.⁷⁰ On the other hand, CID755673 given ip once daily for 16 days at 1 or 10 mg/kg led to enhanced cardiac function in diabetic db/db mice,⁷ implying that inhibition of PKD may improve heart functions in obese and diabetic patients. The efficacy of CID755673 in chronic conditions that require long-term treatment is encouraging and somewhat surprising given the poor PK profile of its analog kb-NB142-70. Also, off-target effects of CID755673 could not be ruled out entirely. Another inhibitor that has been evaluated in metabolic disorders is CRT0066101. In a mouse model of high-fat-diet-induced obesity and diabetes, CRT0066101 given orally at 10 mg/kg daily for 13 weeks resulted in reduced intestinal fat absorption, decreased obesity and associated diabetes, and an improved gut microbiota profile in mice via inhibiting intestinal triglycerides absorption by regulating apolipoprotein A4 (APOA4).⁵⁵ Despite these promising early findings, more studies are needed to confirm the therapeutic potential of PKD inhibitors in metabolic disorders.

Cancer.

CRT0066101, developed by Cancer Research Technology (CRT), is the most intensively studied PKD inhibitor in preclinical mouse models of cancer. In 2010, a joint report by the Guha group from MD Anderson and CRT identified CRT0066101 as an anticancer agent for pancreatic cancer.³⁶ It was found to suppress the growth of subcutaneous and orthotopic PANC-1 pancreatic tumor xenografts in vivo when given orally at its maximum tolerated dose of 80 mg/kg/day for 3 weeks. No apparent signs of acute toxicity were observed during the treatment period as reflected by a stable body weight. The compound was found to have a half-life of 60 min and ~100% bioavailability. The optimal therapeutic concentration of 8 μmol/L was reached at 6 h after oral administration.³⁶ This favorable PK/PD profile set the stage for the subsequent evaluation of this inhibitor in multiple cancer models by different research groups. As detailed in Table 2, the anticancer efficacy of CRT0066101 has been demonstrated in colon cancer,⁷⁵ breast cancer including estrogen receptor negative (ER⁻),⁷⁶ triple-negative breast cancer (TNBC),⁷⁷ and bladder cancer.⁸⁰ Notably, it was also effective at blocking lung metastasis of subcutaneously implanted hepatocellular carcinoma cells in nude mice,⁸⁰ In these studies, CRT0066101 was given orally at a dose ranging from 40 to 120 mg/kg with 80 mg/kg/day being the most frequently used dosing regimen. Mechanistically, CRT0066101 inhibits tumor cell proliferation and survival by inducing G2/M arrest via inhibiting CDK1/cyclin B complex activity and other key pro-proliferative and pro-survival proteins, including NF-κB, ERK, and Akt.^36,75,80 CRT0066101 has also been shown to induce G1 arrest, possibly through regulating MYC, MAPK1/3, AKT, YAP, and CDC2.⁷⁷ In addition to CRT0066101, our group showed that SD-208, a dual inhibitor of PKD and TGFβR1, was also effective at inhibiting the growth of subcutaneous prostate cancer tumor xenografts in mice. This compound similarly induced G2/M arrest by modulating the activity of the CDK1/cyclin B complex.¹² It is interesting to note that a recent study demonstrated a concerning effect of CRT0066101 in antagonizing the anti-PD-1 immunotherapy by blocking AKT reactivation in T cells.⁵⁵ However, another study showed that inhibition of PKD could abrogate immunosuppressive properties of B cells and thereby promote effector T-cell function.⁷⁴ In the later study, CRT0066101 suppressed the growth of orthotopic pancreatic xenograft tumors in immunocompetent mice, demonstrating its viability as an anticancer agent despite its potential complications in the immune system. Overall, these studies have established CRT0066101 as an orally effective anticancer agent in multiple solid tumor models. However, the potentially dose-limiting solubility and long-term toxicity profile of CRT0066101 would need to be further elaborated and, if needed, improved by structural modifications.

PKD Inhibitors in Combination Therapy.

One area that deserves significant future attention is the use of PKD inhibitors in combination therapy. Previous studies have mainly focused on their use in combination therapy for cancer, and the outcomes have been mixed. PKD inhibitors have been shown to synergize with several chemotherapeutics, including paclitaxel⁵⁶ and cisplatin,⁵⁷ as well as targeted agents such as the multikinase inhibitor regorafenib.⁷⁵ Specifically, inhibition of PKD by CRT0066101 significantly enhanced the cytotoxic effect of regorafenib in multiple colon cancer cell lines by synergistically inducing apoptosis and downregulating ERK, AKT, and NF-κB signaling.⁷⁵ The combination of CRT0066101 and paclitaxel synergistically reduced oncosphere and colony formation in TNBC cells and, most importantly, reduced tumor recurrence in vivo in an orthotopic breast cancer mouse model.⁵⁶ Another PKD inhibitor, CID2011756,³⁷ enhanced the sensitivity of cancer cells to cisplatin or carboplatin. This PKD inhibitor increased proteasome-mediated degradation of ATPase copper-transporting α/β (ATP7A/B) by blocking phosphorylation.⁵⁷ Since the selectivity of CID2011756 is unknown, it remains to be determined if and how much inhibition of PKD1–3 accounts for its synergistic effects. It should be noted that the combined inhibition of PKD is not always desirable, and there is evidence that activating it could also be beneficial, such as the case of suramin (a PKD activator) and a DNA methyltransferase inhibitor combination for breast cancer treatment, where a role of PKD1 in tumor suppression is implicated.²⁰ In addition, it has been shown that CRT0066101, by inhibiting Akt activation and T-cell function, accelerated tumor growth and reversed the therapeutic effect of anti-PD-1 therapy in a syngeneic mouse tumor models.⁸² Overall, further in-depth studies are needed in this area to fully realize the utility of PKD inhibitors in combination therapy.

Perspectives and Future Directions.

After the first reports on the discovery and biochemical characterization of PKD, considerable enthusiasm for this new target drove a 15 year period of rapid growth, culminating in a vigorous preclinical development of a significant number of structurally distinct, PKD-selective chemotypes. However, these early lead structures still suffer from a lack of selectivity between the three PKD subtypes and decreased potency in cells and in vivo and did not thoroughly address potential on-target and off-target toxicity. Initial clinical entry strategies also diverged between cardiac diseases and cancer; PKD was found to regulate myofilament function⁸⁴ and influence blood pressure downstream of angiotensin II,^85,86 which complicates the design of human trials of small molecule pan-PKD modulators. While mechanistic and cell-based studies on PKD inhibitors are continuing, major therapeutic breakthroughs have yet to be realized. Different PKD isoforms have seemingly opposing functions in cancer; accordingly, future clinical studies in this therapeutic field will likely require isoform-selective inhibitors. In consideration of the very high level of structural homology between PKD1, PKD2, and PKD3, particularly in the vicinity of the active site, it would be very attractive to identify allosteric inhibition strategies in order to evaluate the therapeutic profile of isoform-selective agents. In particular, for noncancer-related applications, high subtype selectivity might be a key requirement in a clinical setting. Other current gaps in the field include the scarcity of in vivo toxicity profiles of inhibitor chemotypes as well as the lack of unique and reliable biomarkers that could be used to track PKD response in patients. A side-by-side, comprehensive evaluation of different classes of PKD inhibitors would support these efforts. Furthermore, alternative classes of PKD modulators (nucleotide based, peptide based, etc.) would also provide helpful new tool compounds. Finally, as with other experimental therapeutics, drug combinations are a particularly attractive opportunity for clinical studies. In light of the large body of published research on PKD, including its medicinal chemistry and wealth of biological functions (Figure 6), as well as the broad portfolio of modern targeting strategies and structural biology tools, another growth period in PKD-driven pharmaceutical research and development is likely to emerge in the near future.

Figure 6.

PKD as a potential therapeutic target in human diseases, illustrating key signaling targets of PKD and diverse PKD-regulated biological processes in four major types of human diseases.² Created with BioRender.com.

ABBREVIATIONS USED

APOA4

apolipoprotein A4

BTBR^ob/ob

black and tan brachyuric obese/obese mouse

CLU

clusterin

DBU

1,8-diazabicyclo[5.4.0]undec-7-ene

DDQ

2,3-dichloro-5,6-dicyano-1,4-benzoquinone

DHP

dihydropyran

DHPG

(S)-3,5-dihydroxyphenylglycine

d-GalN

d-galactosamine

DSS

Dahl salt sensitive

EDCI

1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide

ESC

embryonic stem cell

GBS

group B streptococci

FP

fluorescence polarization

ip

intraperitoneal injection

mGluRs

metabotropic glutamate receptors

MI

myocardial infarction

PDAC

pancreatic ductal adenocarcinoma

PKD

protein kinase D

PKD1

protein kinase D1

PKD2

protein kinase D2

PKD3

protein kinase D3

po

oral

PyBOP

benzotriazol-1-yloxytripyrrolidinophosphonium hexafluorophosphate

n.d.

not determined

sc

subcutaneous

SINGD

stress-induced nascent granule degradation

TAB

thoracic aortic banded

TFAA

trifluoroacetic anhydride

TNBC

triple-negative breast cancer

TPSA

topological polar surface area

Biographies

Qiming Jane Wang received her B.S. degree in Microbiology (1989) from Xiamen University in China. Thereafter she moved to the United States and obtained her Ph.D. degree in Biochemistry (1995) from Creighton University. After completing her postdoctoral training with Dr. Peter M. Blumberg at the National Cancer Institute, she joined the faculty of the University of Pittsburgh School of Medicine and is currently a Professor at the Department of Pharmacology and Chemical Biology at Pitt. Her current research mainly focuses on oncogenic protein kinases in the diacylglycerol signaling network and small molecule drug discovery and development targeting these kinases.

Peter Wipf received his Ph.D. degree in 1987 from the University of Zurich under the guidance of Professor Heinz Heimgartner. After 2 years as a Swiss NSF postdoctoral fellow with Professor Robert E. Ireland at the University of Virginia, he joined the University of Pittsburgh in 1990 and was promoted to Distinguished University Professor in 2004. He has secondary appointments in Pharmaceutical Sciences and in Bioengineering, and he is a coleader of the Cancer Therapeutics Program at the UPMC Hillman Cancer Center. His research expertise includes medicinal chemistry, total synthesis of natural products, heterocyclic chemistry, and study of strain-activated carbo- and heterocycles. He has coauthored more than 600 manuscripts, and he is a fellow of the AAAS, the RSC, and the ACS.