Selective Inhibitor of Protein Kinase PKN3 Generated by Conjugation of a Structurally Optimized Bumped N-(2-Aminoethyl)-8-anilinoisoquinoline-5-sulfonamide (H-9) with d-Arginine-Rich Chain

Abstract

The protein kinase N family belongs to the AGC kinase group and contains three isozymes: PKN1, PKN2, and PKN3. Catalytic domains of PKNs share high sequence similarity, yet the proteins differ in tissue distribution, functions, and involvement in pathological processes. In particular, PKN3 has been implicated in tumor growth and metastatic progression, highlighting the need for isozyme-selective inhibitors as both research tools and therapeutic leads. Here, we report the rational design of selective PKN3 inhibitors based on distinctive structural features of this kinase. Two strategies were applied. First, the smaller threonine gatekeeper residue unique to PKN3 within the AGC group was exploited by derivatization of N-(2-aminoethyl)isoquinoline-5-sulfonamide (H-9) at position C8. Among the resulting compounds, a phenylamino-substituted derivative displayed the highest affinity, with a dissociation constant (K_D) of 23 nM and more than 1000-fold selectivity over protein kinase A. Second, bisubstrate-analog design was employed to enhance binding to basophilic AGC kinases through covalent attachment of a (d-Arg)₃-containing chain to H-9 derivatives. This approach yielded ARC-2603, which bound PKN3 with a K_D value of 0.2 nM and showed 5500-fold selectivity over PKAcα. The selectivity of ARC-2603 was further evaluated in a commercial panel of 397 protein kinases, which supported its utility as a highly selective PKN3 inhibitor.

1. Introduction

The protein kinase N (PKN) family belongs to the AGC group of serine/threonine PKs. It comprises three isozymes—PKN1 (or PRK1), PKN2 (PRK2), and PKN3 (PRK3). The catalytic domains of PKN isozymes share high sequence similarity and are homologous to the novel PKC isozymes [1,2]. PKNs are basophilic PKs with high resemblance of their phosphorylation consensus sequences, characterized by their preference for positively charged amino acid residues on the N-terminal side of the phosphorylatable serine/threonine residue of the substrate protein, most importantly at position -3 [3,4]. However, PKN isozymes have distinct physiological functions, expression levels, tissue distribution, and different roles in pathological processes [5,6]. Unlike the other PKN isozymes that are widely distributed, the high expression level of PKN3 is more restricted to specific cell types, including cancer cells [7,8]. The association of PKN3 activity with tumor growth and metastatic progression has pointed to this enzyme as a potential target for cancer treatment [9]. This necessitates the development of isozyme-selective inhibitors that could be used as research tools and potential pharmaceutical agents. A liposomal small interfering RNA (siRNA) formulation against PKN3 Atu027 has reached clinical trials for treating advanced cancer (accession numbers NCT00938574 [10] and NCT01808638 [11]).

PKN3 is an understudied PK. Its crystal structure has not been disclosed. Until recently, selectivity testing of PKN3 inhibitors was not possible in commercially available selectivity-screening panels [9]. Several potent ATP-competitive inhibitors of PKN3 have been identified; however, most of these were discovered through screening libraries of inhibitors originally developed to target other protein kinases. The respective examples are staurosporine derivative PKC412 (Midostaurin), a broad-spectrum PK inhibitor that has been approved for the treatment of acute myeloid leukemia (subnanomolar K_i value for all PKN isozymes [6]); Y-27632, a known Rho-kinase inhibitor, which shows even higher potency for PKN3 (18 nM K_i for PKN3 and submicromolar K_i values for PKN1, PKN2, ROCK1, ROCK2 [6]); PP1 and SB-202190, inhibitors of Src and p38 MAPK, respectively (both show single-digit nanomolar K_i values for PKN3 and high PKN isozyme selectivity [6]); isoquinoline-5-sulfonamide derivatives H-8 and Fasudil that are known inhibitors of other AGC kinases (ROCK, PKA, PKG) [6,12]; covalent inhibitors THZ1 and JZ128, whicht were discovered among previously developed inhibitors of cyclin-dependent PKs (CDKs) and react with Cys840 on PKN3 [13]; Cdk1/2 inhibitor III, whose targets include Cdk1/cyclin B and Cdk2/Cyclin A [14] (34 nM K_i value for PKN3 [6]); an inhibitor of CaMKII KN62 (isoenzyme-selective inhibitor of PKN3 with 46 nM K_i value [6]); and GSK902056A (K_D^app = 10 nM for PKN3), an inhibitor discovered by analyzing large compound libraries by KinoBead assay and which is currently the only PKN3-related entry in the Chemical Probe database [15,16] (Figure A1).

Only a limited number of reports focus on the systematic development of PKN3 inhibitors. Derivatives of H-8 were tested for binding to PKN3 in the NanoBRET cellular target engagement assay, but no inhibitors with desirable activity were found within the series [17]. Asquith et al. developed inhibitors based on 4-anilinoquin(az)oline scaffolds [18,19]. This series included several narrow-spectrum inhibitors, a notable example being UNC-CA94 (Figure A1), which showed micromolar in-cell inhibitory potency and two-digit nanomolar IC₅₀ for PKN3 in the biochemical assay. In addition to ATP-competitive inhibitors, a protein substrate-competitive 15-amino acid inhibitor peptide, PRL, was derived from the autoinhibitory domain of PKN1. The peptide inhibited the PKN isozymes with micromolar K_i values [6,20].

Previously, we have developed bisubstrate inhibitors of protein kinases, termed ARCs, that incorporated two moieties, an adenosine analog and a peptide chain, targeting the ATP-binding site and the protein substrate binding site of PKs, respectively. These two moieties were connected by a flexible linker whose structure was optimized in structure-affinity studies. ARCs comprising nucleoside analogs and (d-arginine)-rich peptides are group-selective inhibitors that bind with high affinity (K_D values in the picomolar range) to several basophilic PKs, mostly belonging to the AGC group [21,22].

A hydrophobic back-cavity in the ATP-binding pocket is located inside the catalytic domains of most PKs. The shape and size of this cavity are controlled by the nature of a single amino-acid residue adjacent to the hinge region, called the gatekeeper [23]. There are 63 serine/threonine PKs in the AGC group of human kinases, 58 of which contain relatively large gatekeeper residues—Met, Leu, or Ile (Figure A2). A distinctive feature of the ATP-binding site of PKN3 is the presence of a relatively small gatekeeper residue, Thr639. A unique Thr gatekeeper affects the topology of the ATP-binding site of PKN3, potentially enabling selective targeting of this enzyme with inhibitors.

In this study, we utilize an integrative approach that combines the bisubstrate inhibitor design and introduction of a bulky group (“bump”) to the ATP-binding pocket-targeting fragment to facilitate the formation of highly effective and selective PKN3 inhibitors. We postulate that the introduction of a guanidine-functionalized chain molecule to an inhibitor targeting the ATP-binding pocket of PKs would enhance the selectivity and affinity toward basophilic protein kinases. Concurrently, the steric repulsion of the bump at a presumable position of the inhibitor to big methionine gatekeepers present at ATP-binding pockets of the majority of AGC kinases would diminish the affinity of the inhibitor and thus would increase the selectivity toward PKN3. This hypothesis is supported by the results of our previously published study describing the development of a photocaged bisubstrate inhibitor of PKAcα, in which we showed that the incorporation of a bulky substituent into the ATP-binding site-targeting fragment of a bisubstrate inhibitor drastically reduced binding affinity to the target PK [24].

2. Results

The first step of this project was to establish the structure of a bisubstrate inhibitor that would bind with high affinity to PKN3, but that would not be excessively selective toward other PKs. As an additional factor, knowledge of the inhibitor’s binding mode in the ATP-binding site would facilitate prediction of the optimal position for attachment of the bump. Unfortunately, the commercial PK selectivity panels where different ARC-inhibitors had been tested previously did not include the PKN3 protein. However, other members of the PKN family were inhibited by d-arginine-rich ARCs: ARC-669 [25], ARC-1411 [26], and ARC-3125 [27]. Therefore, we anticipated that the general structure of positively charged d-arginine-comprising ARC-inhibitors could also be compatible with PKN3 and would serve as a suitable lead structure of an inhibitor for this kinase.

Previously, we have developed a fluorescence anisotropy (FA)-based binding assay for several PKs as a high-throughput method for the determination of K_D-values of PK/inhibitor complexes. We have shown that the obtained K_D-values well correlate with the inhibitory potencies of the compounds obtained in enzymatic assays [28].

We started the study by testing several previously characterized fluorescence-labeled ARC-inhibitors that have revealed high affinity toward some other basophilic kinases of the AGC-group as potential tracers in an FA assay for PKN1 and PKN3. ARC-1059 emerged as a suitable tracer for these PKs with sub-nanomolar and low-nanomolar affinity toward PKN3 and PKN1, respectively (

Table 1. K_D-values (±S.D., n = 2) of ARC-1059/PK complexes as measured with the FA-based binding assay. The K_D-values were determined in the present work or taken from the literature as referenced.

PK	KD/nM	Gatekeeper Residue
PKGIα	<1 [30]	Met
Akt3	16 ± 1.2 [22]	Met
MSK1	5.3 ± 0.85 [22]	Leu
p70S6K	11 ± 4.2 [22]	Leu
PKCη	5.2 ± 2.2 [22]	Met
PKAcα	0.52 ± 0.17 [22]	Met
ROCK2	<0.2 [22]	Met
PKN1	2.3 ± 0.3	Met
PKN3	0.20 ± 0.04	Thr

, Figure 1A). ARC-1059 is a 5-TAMRA-labeled conjugate of Hidaka’s inhibitor H-9 [29] and hexa-d-arginine that was previously characterized as a high-affinity fluorescent tracer for other AGC kinases (Table 1). The data presented herein is the first proof that d-arginine-comprising ARCs bind to PKN3 with high affinity. Considering its high PKN3 affinity and agreeable AGC kinase selectivity profile (Table 1), ARC-1059 was a promising lead compound for PKN3-selective inhibitor development.

Tracer ARC-1059 was used in titration experiments to establish concentrations of the active form of PKs, as described previously [28]. The commercial PKs used in this study were stored at −80 °C and used on the day the solutions were thawed. In several batches of PKN kinases from different companies, the content of active kinase was as low as 10–20% for PKN3 and 10% for PKN1 of the amount reported by the provider for the kinase protein. It needs further investigation whether such low activity originates from insufficient activation or low stability of the enzymes.

With a suitable fluorescent tracer found, the FA-based assay was used in a competitive binding/displacement format to determine K_D values for PKN3/inhibitor complexes for a set of previously developed compounds. First, reference compounds H-8, H-9, and Y-27632 were used to demonstrate the feasibility of the application of an FA-based displacement assay for PKN1 and PKN3 (Figure 1B,C,

Table 2. Comparison of K_D-values of inhibitor/PK complexes as determined by FA-based competitive displacement assay (±S.D., n = 2) with the previously published K_i values.

Compound Code	Structure	PK	KD/nM	Ki/nM
Y-27632		PKAcα	14,000 ± 2000	25,000 [31]
		ROCK2	250 ± 17	300 [32]
		PKN1	360 ± 50	700 ± 200 [6]
		PKN3	16 ± 9	13 ± 1 [6]
H-8		PKAcα	1600 ± 500	1200 [33]
		PKN1	2200 ± 300	1800 ± 300 [6]
		PKN3	270 ± 50	10 ± 2 [6]
H-9		PKAcα	1100 ± 100	1900 [33]
		ROCK2	1060 ± 20
		PKN1	1600 ± 100
		PKN3	260 ± 9

). An identical assay was used in parallel for PKs PKAcα and ROCK2, two other reference basophilic PKs of the AGC-group in this study. The determined K_D-values were in good agreement with the previously reported K_i values, with the exception of the inhibitor H-8, which, according to our data, was substantially less selective towards PKN3 than previously reported [6]. The affinities of H-8 and its structurally close analog H-9 were similar according to the results of our study with both PKN isozymes (Table 2). Thereafter, a small set of previously developed d-arginine-comprising ARCs was screened for binding to PKN3 (

Table 3. Affinities of d-arginine-comprising ARCs towards PKAcα and PKN3, as determined by FA-based competitive displacement assay, presented as K_D-values (±S.D., n = 2) and ratios of the obtained K_D values.

Code	Structure	KD (PKAcα)/nM	KD (PKN3)/nM	PKAcα/PKN3
ARC-1034 [34]	Adc-Ahx-[d-Arg]2-NH2	319 ± 57 [34]	>1000	<0.3
ARC-902 (compound 5 in [21])	Adc-Ahx-[d-Arg]6-NH2	0.33 [28]	5.2 ± 3.0	0.06
ARC-1408 [26]	7DP-Pip-Nda-d-Arg-Ahx-d-Arg-NH2	0.18 [26]	>70	<0.003
ARC-2127	7DP-Pip-Nda-d-Arg-Ahx-[d-Arg]3-NH2	0.004 ± 0.001	1.7 ± 0.3	0.002
ARC-2128	7DP-Pip-Nda-d-Arg-Ahx-[d-Arg]4-NH2	0.0010 ± 0.0001	0.5 ± 0.3	0.002
ARC-1411 [26]	7DP-Pip-Nda-[d-Arg]6-d-Lys-NH2	0.003 [26]	0.63 ± 0.45	0.005
ARC-3002 [22]	Fas-Ahx-d-Lys-Ahx-[d-Arg]6-NH2	3.3 [22]	<0.2	>16
ARC-903 (compound 26 in [21])	(H-9)-[CH2]5-CO-[d-Arg]6-NH2	0.53 [22]	0.9 ± 0.2	0.6
ARC-3010 [22]	(H-9)-CO-CH2NHCH2CO-[d-Arg]6-NH2	120 [22]	45 ± 4	2.6

). If grouped by the structure of the moiety targeting the ATP-binding site of PKs, this set of compounds included three series of compounds: conjugates of adenosine-5′-carboxylic acid (Adc), conjugates of 7-deazapurine-piperazine adduct (7DP-Pip), and conjugates of isoquinoline-5-sulfonamide (derivatives of Fasudil (Fas) and H-9). The number of d-arginine residues in the examined structures varied from 2 to 6. The ATP-binding site-targeting fragment and the C-terminal oligo-d-arginine fragment were separated by a linker comprising hydrophobic 6-aminohexanoic acid (Ahx) residue or an extended linker composed of two hydrophobic chains (Ahx or nonanedioic acid (Nda)), which were separated by a chiral spacer (d-Lys or d-Arg). All screened ARCs were C-terminal amides. The established K_D-values of PKN3-complexes of these ARCs were compared with their previously determined K_D-values of complexes with PKAcα, which, like PKN3, is a basophilic PK in the AGC group, but differs from PKN3 by the presence of a large gatekeeper residue (Met120). Although the screened set of compounds was small, the K_D-values (Table 3) still revealed general structure/affinity trends. First, the affinities toward both PKAcα and PKN3 increased with the number of d-Arg residues in the structure. Second, PKN3 showed a preference for structures with the extended linker. Third, the Adc-conjugates showed slight and 7DP-Pip-conjugates high PKAcα vs. PKN3 selectivity. The 7DP-Pip series of compounds in this set had the lowest K_D-values toward PKAcα, which made these compounds not suitable as lead structures for the development of the PKN3-selective bumped inhibitors. On the contrary, some of the compounds in the isoquinoline-5-sulfonamide series already exhibited very high affinity and moderate selectivity toward PKN3, as was also shown with the fluorescent tracer ARC-1059, and with the small-molecule inhibitors H-8 and H-9 (Table 2). Therefore, H-9 was selected as the ATPanalog lead for constructing bumped bisubstrate inhibitors for PKN3 targeting.

Next, we focused on the development of bumped analogs of H-9. Since the crystal structure of PKN3 is not yet available, we overlaid the AlfaFold-simulated 3D-structure of this PK with the co-crystal structure of the inhibitor H-8 (the methylated derivative of H-9) and PKAcα, which was published in 1996 [35]. From this overlay, we identified positions 7 and 8 (Figure 2) as the sites oriented towards the gatekeeper residues of these PKs. For reasons of availability of the starting material and synthetic simplicity, we chose to modify the position C8 of isoquinoline in this study. C8 substituents were chosen to span differences in size, flexibility, and hydrogen-bonding capacity. Derivatives of H-9 were synthesized by sulfonation of 8-chloroisoquinoline at C5, followed by the attachment of Boc-protected ethylenediamine to the activated sulfonyl moiety. C8 functionalization was then achieved through nucleophilic aromatic substitution (compounds 3 and 4), Suzuki–Miyaura coupling (5), or Buchwald–Hartwig coupling (6). Final Boc deprotection yielded bumped H-9 derivatives 2′–6′ (Scheme 1).

In the small-molecule series of compounds (

Table 4. Affinities of unsubstituted and 8-substituted derivatives of H-9 toward PKAcα and PKN3, as determined by FA-based displacement assay, presented as K_D-values (±S.D., n = 2) and ratios of the obtained K_D values. * n.d.—not determined.

Code of Compound Number	R- in Structure: 8-R-(H-9)	KD (PKAcα)/nM	KD (PKN3)/nM	PKAcα/PKN3
H-9	H-	1100 ± 100	260 ± 9	4
2′	Cl-	1800 ± 400	3500 ± 400	0.5
5′	Phe-	150,000 ± 40,000	7300 ± 1700	21
6′	Phe-NH-	>300,000	23 ± 2	>13,000
4′	Bn-NH-	64,000 ± 4000	220 ± 75	290
3′	Morph-	>300,000	>300,000	n.d. *

), the chloro-substitution did not remarkably affect the K_D-value for PKAcα. All other substitutions resulted in substantial loss of affinity towards PKAcα. Phenylamino and morpholino derivatives of H-9 were not able to displace the fluorescent tracer ARC-583 [28] from the complex with PKAcα at the highest concentration of 300 µM. In the case of PKN3, the chloro-, phenyl-, benzylamino, and morpholino-substituents reduced the affinity. The morpholino derivative 3′ did not displace the tracer ARC-1059 from the PKN3 complex at the highest concentration of 300 µM. Chloro-substitution had a stronger effect on the binding to PKN3 than to PKAcα, which reversed the PKN3 vs. PKAcα selectivity. The phenylamino-substituted compound 6′ bound to PKN3 with a smaller K_D-value than H-9, which points to specific steric constraints in the ATP-binding site and possible hydrogen bonding between PKN3 and the exocyclic NH-group in the C8 position of isoquinoline.

Next, ARC-type conjugates were constructed from all synthesized derivatives of H-9. The R-(H-9)-fragments were interlinked by the generation of a secondary amine between the ethylene diamine of the H-9 moiety and the hexanoic acid linker, as outlined in the previously published structures of ARC-903 (Table 3) and ARC-1059 [30]. If compared to the latter structures, the number of d-arginines was reduced. The C-terminal tri-d-arginine fragment was conjugated to the isoquinoline fragment via an extended linker with d-arginine spacer between the alkyl chains. The d-arginine spacer has been shown to give a strong interaction with the glycine-rich loop of PKAcα, according to the results of crystal structure analysis of ARC/PKAcα co-crystals [34]. All new ARC-type compounds thus possessed the general sequence 8-R-(H-9)-[CH₂]₅-CO-d-Arg-Ahx-[d-Arg]₃-d-Lys-NH₂, where R is the bump group. Although peptides containing 4–6 d-Arg residues in the C-terminus of the sequence typically yielded higher affinities (Table 3), a reduction in the number of d-Arg residues was motivated by the need to keep the inhibitor potency in a range that is reliably characterizable in the FA-based displacement assay. A C-terminal D-Lys residue was incorporated to allow attachment of fluorescent labels in the future. Optimization of the molecular chain -[CH₂]₅-CO-d-Arg-Ahx-[d-Arg]₃-d-Lys-NH₂ is planned for a future study.

ARCs were synthesized on Rink amide MBHA resin using solid-phase peptide synthesis technology to obtain C-terminal amides. The C-terminal free carboxylic acid group was avoided because of its possible unfavorable interactions with positively charged residues [21,36]. For conjugate assembly, resin-bound sequences were first acylated at the N-terminus with 6-bromohexanoic acid, followed by nucleophilic substitution with the corresponding 8-substituted H-9 derivatives.

K_D-values of the synthesized ARC-s are presented in

Table 5. Affinities of constructed ARC-inhibitors toward PKAcα and PKN3, as determined by FA-based displacement assay, presented as K_D-values (±S.D., n = 2) and ratios of the obtained K_D values.

Code	R- in Structure: 8-R-(H-9)-[CH2]5-CO-d-Arg-Ahx-[d-Arg]3-d-Lys-NH2	KD (PKAcα)/nM	KD (PKN3)/nM	PKAcα/PKN3
ARC-2600	H-	4.1 ± 0.1	1.7 ± 0.2	2.4
ARC-2601	Cl-	3.1 ± 0.2	8.5 ± 0.8	0.4
ARC-2602	Phe-	1150 ± 180	21 ± 2	55
ARC-2603	Phe-NH-	1100 ± 200	0.20 ± 0.03	5500
ARC-2604	Bn-NH-	440 ± 30	1.0 ± 0.1	440
ARC-2605	Morph-	3420 ± 460	221 ± 21	15.5

. Conjugation with the chain molecule incorporating (d-Arg)₃ fragment increased the affinities toward PKAcα and PKN3 with all H-9 derivatives by over 2 orders of magnitude. The order of affinities of the conjugates for both studied PKs was quite similar to that of the corresponding small-molecule derivatives of H-9. The un-bumped compound ARC-2600 was slightly PKN3-selective, although less than H-9. All newly developed ARCs had higher affinities toward PKN3 than PKAcα, except the chloro-substituted derivative. The same was true for their small-molecule counterparts. The morpholino derivative ARC-2605 had the lowest affinity to both tested PKs, with low-micromolar affinity toward PKAcα and submicromolar affinity toward PKN3. In contrast, ARC-2603 showed remarkable PKN3 vs. PKAcα selectivity; upon attachment of the aniline bump to ARC-2600, the K_D-value for PKN3 increased, as it was also observed with the small-molecule counterparts, and selectivity as expressed by PKAcα/PKN3 K_D-ratios changed from 2.4 (ARC-2600) to 5500 (ARC-2603).

Ultimately, this study yielded two highly potent PKN3 inhibitors: the small-molecule phenylamino-derivative of H-9 (MW = 342) and the corresponding bisubstrate inhibitor ARC-2603 (MW = 1321). Thereafter, these two compounds were tested against a broader set of PKs for a more thorough assessment of selectivity.

First, an in-house FA-based displacement assay was conducted with AGC-PKs ROCK2 and PKN1. PKN1 is a close homolog of PKN3. ROCK2 is an established target for isoquinoline-5-sulfonamide-type inhibitors (Table 2). Both PKs possess big methionine gatekeepers. The K_D-values of complexes of ARC-2600 with these PKs were in a two-digit nanomolar range (

Table 6. Affinities of H-9, 6’ (8-phenylamino-(H-9)) and the corresponding ARC-inhibitors toward PKN1 and ROCK2, as determined by FA-based displacement assay, presented as K_D-values (±S.D., n = 2) and ratios of the obtained K_D values.

Compound	KD (ROCK2)/nM	KD (PKN1)/nM
H-9	1060 ± 20	1600 ± 100
6′ (8-Phe-NH-(H-9))	>100,000	175,000 ± 15,000
ARC-2600	23 ± 1	84 ± 2
ARC-2603	430 ± 30	4150 ± 150

), indicating that the initial PKN3-selectivity of the H-9 scaffold is retained in the case of the bisubstrate inhibitor. The bumped small molecule 6′ was not able to displace the fluorescent tracer ARC-1042 from the ROCK2 complex at the highest concentration of 100 µM, and the same concentration of 6′ only partially displaced the tracer ARC-1059 from the PKN1 complex. The bumped bisubstrate inhibitor ARC-2603 bound to ROCK2 with a higher affinity (sub-micromolar K_D-value) than to PKAcα. The K_D-value of the complex of ARC-2603 with PKN1 was in a one-digit micromolar range, as it was also observed with PKAcα. In conclusion, the attachment of a phenylamino bump increased the affinity toward PKN3 of both the small-molecule and the bisubstrate inhibitor and increased the PKN3 selectivity by more than 3 orders of magnitude if the K_D-values for PKN3 were compared with those of PKN1, PKAcα, and ROCK2.

The potency of ARC-2603 to inhibit the phosphorylation reaction catalyzed by PKN3 and PKN1 was characterized by real-time PhosphoSens enzymatic assay (available from AssayQuant Technologies Inc, Marlborough, MA, USA) at 1 mM ATP, which is close to the intracellular ATP concentration. An IC₅₀-value of 3.1 ± 0.6 nM for PKN3 was obtained, which is in accordance with the sub-nanomolar K_D-value of the compound, assuming the ATP-competitive inhibition mechanism (Figure 3). Y-27632 was used as a reference compound in this assay (observed IC₅₀ = 740 ± 20 nM). PKN1, on the contrary, was not inhibited in the tested concentration range (up to 1 µM). Thus, the results of the enzymatic assay confirmed the PKN3 vs. PKN1 selectivity observed in the FA-based displacement assay.

Finally, the most potent compound, ARC-2603, was analyzed in a panel of 397 PKs. An enzymatic activity panel of AssayQuant Technologies that utilizes the same type of PhosphoSens assay was used. Based on our inhibition curve, 300 nM ARC-2603 was chosen as the end concentration in the panel because it showed near complete inhibition of PKN3 in our conditions (Figure 3). The panel was run at a fixed 1 mM concentration of ATP. The data for PKs inhibited by more than 50% and for some other selected PKs are given in

Table 7. Inhibitory percentages of selected PKs in the presence of ARC-2603 (300 nM final concentration). The selectivity panel was conducted on a commercial basis by AssayQuant Technologies using the PhosphoSens assay. The ATP concentration used was 1 mM. Km values are given as reported by the company. The full panel is given in the Supporting Information (Table S1).

PK	Group	Gatekeeper	ATP Km/µM	Inhibition Percentage
PKN3	AGC	Thr	37	96.3 ± 0.4
PKN1	AGC	Met	24	90.2 ± 0.7
DDR1	TK	Thr	120	87.5 ± 1.4
HER4	TK	Thr	25	68.1 ± 2.4
TEC	TK	Thr	50	67.9 ± 0.6
EGFR	TK	Thr	44	67.8 ± 3.1
PIM1	CAMK	Leu	29	65.4 ± 1.5
HASPIN	Other	Phe	47	63.8 ± 6.3
FMS/CSF1R	TK	Thr	140	62.4 ± 7.9
AURC	Other	Leu	60	58.7 ± 1.5
SLK	STE	Ile	16	52.8 ± 1.3
PKCι	AGC	Ile	23	51.2 ± 4.5
EPHB4	TK	Thr	43	50.3 ± 2.3
PKN2	AGC	Met	16	37.1 ± 0.2
PKAcα	AGC	Met	15	17.8 ± 0.5
ROCK2	AGC	Met	11	25.9 ± 0.9

and the full panel is given in Table S1.

A total of 13 PKs in this panel were inhibited by over 50%. Seven of these PKs possess Thr-gatekeeper residues and six PKs are tyrosine kinases. The three most inhibited PKs stand out, with more than 70% inhibition: PKN3, which was the most inhibited PK (residual activity 3.7%); PKN1; and tyrosine kinase DDR1. The high inhibition percentage of PKN1 contradicts the previous data obtained by the displacement and enzymatic assays, and it needs additional investigation. The third member of the PKN family, PKN2, which also has a Met gatekeeper, showed only 37.1% inhibition. The low inhibition of PKAcα and ROCK2 was in better agreement with the K_D-values obtained by the displacement assay. PKCι was the only other AGC kinase in the panel that was inhibited by over 50% (residual activity: 48.8%), in addition to PKN1 and PKN3, which shows that the addition of bump successfully reduced the inhibitory potency toward PKs with large gatekeeper residues in the AGC group. All Ser/Thr PKs that were inhibited by over 50% have positively charged amino acids in their consensus sequences [4]. Tyrosine kinase DDR1 also has a preference to phosphorylate residues adjacent to arginine-rich sequences [37]. In addition, DDR1 possesses a Thr gatekeeper and a high ATP Km-value, which may explain the high inhibition percentage of this PK at 1 mM ATP. Other tyrosine PKs in the >50% inhibition range are rather acidophilic. This may point to the different binding modes of d-arginine-comprising bisubstrate inhibitors to the catalytic subunits of these Ser/Thr and Tyr PKs; however, to draw more reliable conclusions, each of these kinases must be examined individually in greater detail.

3. Discussion

PKs have been recognized as drug targets since the 1980s. The isoquinoline-5-sulfonamide-based inhibitors, first introduced in 1984 by Hidaka et al. [29] were among the earliest small molecules to demonstrate the potential of protein kinase inhibitors as a viable class of therapeutic agents [38]. A representative substance of this class is Fasudil, which was the first systematically developed small-molecule PK inhibitor that was approved in Japan in 1995 for the treatment of cerebral vasospasm [39]. The ATP-binding pocket of PKs is well druggable. The challenges associated with the development of ATP-competitive inhibitors of PKs include achieving selectivity, since the ATP-binding pockets of PKs are highly conserved, and the fact that ATP-competitive inhibitors must compete with ATP under conditions of its high intracellular concentration. Nevertheless, the development of effective and selective PK-targeting drugs has proven to be productive; by the end of 2025, the number of small-molecule kinase inhibitors approved by the FDA as drugs reached one hundred, the majority of which target PKs [40]. From a drug development perspective, absolute inhibitory specificity is not always required or may even be undesirable [40]. At the same time, the search continues for novel strategies for the development of highly selective and potent inhibitors of PKs that could serve as drug candidates and chemical probes.

In this study, we constructed a selective subnanomolar inhibitor of basophilic PKN3 by the conjugation of a structurally optimized 8-substituted isoquinoline heterocycle with a guanidine-rich positively charged chain. Modifications at this position of isoquinoline have been performed in a limited number of developments of PK inhibitors. Here we demonstrated that conjugation of the ATP-competitive small-molecule H-9 to a flexible molecular chain comprising a (d-arginine)₃ fragment remarkably (by over two orders of magnitude) increased the affinity toward PKN3. The attachment of an aniline moiety to the C8-position of the isoquinoline ring effectively (more than 1000-fold in case of some PKs) diminished the affinity of the inhibitors toward PKs with large gatekeeper residues. These effects together lead to a remarkable enhancement in selectivity toward PKN3. Interestingly, the inhibitor ARC-2603 co-targeted DDR1, a tyrosine PK whose dysregulation is also linked to cancer [41]. Thus, from the structure–activity studies, two perspective inhibitors of PKN3 emerged: small-molecule 8-anilinoderivative of H-9 (K_D = 23 nM) and the corresponding d-arginine-rich inhibitor ARC-2603 (K_D = 0.2 nM). The fact that the 8-anilino group was effective at diminishing the affinity of both the small-molecule H-9 and the corresponding bisubstrate inhibitor towards PKs possessing methionine gatekeeper residues may point to the similar positioning of the isoquinoline ring of these inhibitors in the ATP-binding pockets of PKs, which could be verified by structural analysis in the future. Results of our systematic study point to an important role of the bump group in achieving the PKN3 selectivity of inhibitors. Previously, we have demonstrated that structures structurally similar to ARC-2603 penetrate the cell plasma membrane and are stable in the intracellular milieu [27]. The applicability of the developed inhibitors for the regulation of PKN3 activity in the cell will be determined in future studies.

4. Materials and Methods

4.1. Equipment and Software

¹H NMR (700 MHz), ¹³C NMR (176 MHz), and 2D HMBC NMR spectra were acquired using a Bruker (Billerica, MA, U.S.) Avance III 700 MHz spectrometer (16.4 T). Tetramethylsilane (TMS) was used as the internal standard. HIQ denotes isoquinoline protons; numbering corresponds to the isoquinoline ring shown in Scheme 1.

HPLC analyses and purification of the novel compounds were performed using a Shimadzu (Kyoto, Japan) Prominence LC Solution system equipped with an SPD-M20A PDA detector and an LCMS-2020 ESI-MS detector. Purification was carried out on a Gemini C18 reverse-phase column (250 mm × 4.6 mm, 5 μm) using a MeCN/H₂O gradient containing 0.1% TFA at a flow rate of 1 mL/min. The column was thermostated at 40 °C. The purity of compounds 2–6 and 2′–6′ was assessed with the same column, whereas the purity of the novel ARC compounds was assessed with a Luna C18 reverse-phase column (250 mm × 4.6 mm, 5 μm) under the same conditions (see Sections S3 and S5 in the Supplementary Materials for gradient programs). Chromatograms were processed using LabSolutions CS software (Shimadzu, version 5.117).

High-resolution mass spectra (HRMS) were acquired on a Thermo Electron LTQ Orbitrap instrument (ESI-HRMS; Thermo Scientific Inc, Waltham, MA, USA) in positive ion mode.

A NanoDrop 2000c spectrophotometer (Thermo Scientific Inc) was used for recording UV-Vis absorption spectra and product quantification. Concentrations of the compounds were calculated according to the Beer–Lambert law using experimentally determined molar extinction coefficients (Figure S1). Molar extinction coefficients were determined from DMSO stock solutions prepared from accurately weighed samples of compounds 2–6 that were serially diluted in the assay buffer (pH 7.5).

Solution-phase reactions were monitored by TLC on Polygram^® Sil G/UV254 plates (Macherey-Nagel, Düren, Germany), with visualization under UV light at 254 nm and 365 nm. A PRO 30 ultrasonic cleaner (ASonic, Ljubljana, Solvenia) was used for sonication. Column chromatography was performed on silica gel 60 (0.04–0.063 mm) purchased from Fluorochem EU (Cork, Ireland). A dual-wavelength (250 and 280 nm) UV-Vis absorbance detector, model RD2 (Reach Devices LLC, Boulder, CO, USA), was used as the detector for monitoring the chromatography. The solid-phase synthesis was performed in polyproylene (PP) SPE tubes equipped with PP filters. Compounds were dried using an Alpha 2-4 LD plus lyophilizer (Martin Christ, Osterode, Germany) for intermediate compounds or an RVC 2-25 CDplus vacuum concentrator (Martin Christ, Osterode, Germany) for final compounds, connected to an RC6 vacuum pump (Vacuubrand, Wertheim, Germany).

Buffered aqueous solutions of biologically active compounds were prepared in low-binding centrifuge tubes (Protein LoBind, Eppendorf AG, Hamburg, Germany). The biochemical binding assays and measurements of the catalytic activity of kinases were performed on black non-binding-surface 384-well polystyrene microplates (catalogue number #4514, Corning, NY, USA). Eppendorf Research (Eppendorf, Hamburg, Germany) pipettes with PP pipet tips (premium surface, Nerbe Plus, Winsen/Luhe, Germany) and 12-channel electronic pipettes with original tips (E1 ClipTip, Thermo Scientific, Waltham, MA, USA) were used for liquid handling in biochemical assays. The fluorescence intensity (FI) and fluorescence anisotropy (FA) measurements were performed with a PHERAstar (BMG Labtech, Ortenberg, Germany) microplate reader. The FI and FA measurements of the photoluminescent tracers ARC-583, ARC-1059, and ARC-1042 were performed with an FA optical module [EX 540(50) nm, EM 590(50) nm]. The detector was adjusted with the solution of the free tracer (in the absence of kinase). Data were processed with Graphpad Prism software (v 6.04, GraphPad Software, La Jolla, CA, USA) and Microsoft Excel (v 2203, Redmond, WA, USA).

4.2. Reagents and Enzymes

Chemicals were purchased from the following commercial sources: Acros Organics (Geel, Belgium), Aldrich (Burlington, MA, USA), Amresco (Framingham, MA, USA), BLDPharm (Shanghai, China), Calbiochem (San Diego, CA, USA), Carl Roth (Karlsruhe, Germany), Chempur (Piekary Śląskie, Poland), Fluorochem EU (Cork, Ireland), Fluka (Buchs, Switzerland), Fisher Scientific (Waltham, MA, USA), Iris Biotech GmbH (Marktredwitz, Germany), Lach-Ner (Neratovice, Czech Republic), Merck (Darmstadt, Germany), NovaBiochem (London, UK), Peaxum (Moscow, Russia), Riedel-de Haën (Seelze, Germany), Sigma-Aldrich (Saint Louis, MO, USA), and Thermo Fisher Scientific (Waltham, MA, USA). Y-27632 and H-8 were from Cayman (Ann Arbor, MI, USA) and H-9 was from BLDPharm (Shanghai, China). The Sox kinase sensor substrate (code AQT05099) was purchased from AssayQuant Technologies Inc (Marlborough, MA, USA). Fluorescence tracers ARC-583, ARC-1042, and ARC-1059, as well as all ARC conjugates listed in Table 3, were previously synthesized in the same laboratory.

Benzylamine was freshly distilled prior to use. All other chemicals were used as supplied by the manufacturers. Enzymes PKN1, PKN3, and ROCK2 were purchased from Carna Biosciences (Kobe, Japan). PKAcα was produced by Dr Tanel Sõrmus in the laboratory of Prof. Richard Engh (University of Tromsø, Tromsø, Norway).

4.3. Chemical Synthesis

4.3.1. 8-Chloroisoquinoline-5-sulfonic Acid (1)

The synthetic procedure was adapted from Koelsch and Albertson [42]. 8-Chloroisoquinoline (2500 mg, 15.28 mmol, 1 equiv.) was dissolved in methanol (10 mL). The solution was cooled in an ice–water bath. Concentrated H₂SO₄ (96%, 0.853 mL, 1 equiv) was added dropwise under vigorous stirring and the resulting suspension was stirred for additional 10 min at room temperature. The mixture was then concentrated and dried in vacuo. Fuming sulfuric acid (20% SO₃, 20 mL) was added to the crude 8-chloroisoquinoline sulfate salt, and the reaction mixture was allowed to stir at room temperature for two days. The reaction mixture was poured onto crushed ice (100 g) and the precipitated solid was collected by filtration, washed once with water, and dried in vacuo to give the title compound as white needle-like crystals (2557 mg, 69% yield). The structure of the obtained compound corresponded to the desired 5-sulfonated regioisomer, as confirmed by 2D HMBC NMR spectroscopy (see S2.1 in Supplementary Material).

¹H NMR (700 MHz, DMSO-d₆, TMS) δ 9.88 (s, 1H, HIQ-1), 9.13 (d, J = 6.3 Hz, 1H, HIQ-4), 8.80 (d, J = 6.3 Hz, 1H, HIQ-3), 8.38 (d, J = 7.7 Hz, 1H, HIQ-6), 8.06 (d, J = 7.7 Hz, 1H, HIQ-7). ¹³C NMR (176 MHz, DMSO-d₆, TMS) δ 145.4, 143.2, 135.9, 135.2, 133.5, 132.8, 129.5, 125.3, 123.9.

4.3.2. N-(2-Boc-aminoethyl)-8-chloroisoquinoline-5-sulfonamide (2)

Compound 1 (1449 mg, 5.95 mmol, 1 equiv.) was suspended in SOCl₂ (21 mL), and a catalytic amount of DMF (0.1 mL, 0.2 equiv.) was added. The reaction mixture was refluxed on a hot plate set to 120 °C until complete dissolution of the material, followed by an additional 1.5 h of reflux under the same conditions. The mixture was then cooled to room temperature and concentrated. The residue was taken up in CHCl₃ and concentrated repeatedly to remove traces of SOCl₂, then dried in vacuo. The obtained sulfonyl chloride was dissolved in dry THF (30 mL), and the solution was cooled in an ice–water bath. TEA (2.5 mL, 3 equiv. relative to compound 1) was added, followed by N-Boc-ethylenediamine (1904 mg, 2 equiv. relative to compound 1) dissolved in dry THF (5 mL). The reaction mixture was allowed to stand at room temperature for three days under a N₂ atmosphere (full conversion is typically reached within a few hours). The mixture was subsequently filtered, and the filtrate was concentrated and dried in vacuo.

The crude product was purified by normal-phase column chromatography (CHCl₃/MeOH, 97:3, v/v) using dry loading on Celite. The obtained material was recrystallized from DCE, affording a white crystalline solid (1203 mg, 52% yield over two steps).

TLC (DCM/MeOH, 10:1) R_f 0.58. HPLC purity 99.4% (220 nm). ¹H NMR (700 MHz, CDCl₃, TMS) δ 9.80 (s, 1H, HIQ), 8.77 (d, J = 6.0 Hz, 1H, HIQ), 8.44 (d, J = 6.0 Hz, 1H, HIQ), 8.33 (d, J = 8.4 Hz, 1H, HIQ), 7.72 (d, J = 8.4 Hz, 1H, HIQ), 5.96 (br s, 1H, NH), 4.79 (br t, J = 5.6 Hz, 1H, NH), 3.20 (q, J = 5.6 Hz, 2H, CH₂), 3.06 (q, J = 5.6 Hz, 2H, CH₂), 1.39 (s, 9H, C(CH₃)₃). ¹³C NMR (176 MHz, CDCl₃, TMS) δ 156.9, 150.4, 146.1, 138.7, 133.5, 133.1, 132.6, 126.3, 126.0, 117.0, 80.3, 44.3, 40.3, 28.3.

4.3.3. N-(2-Boc-aminoethyl)-8-morpholinoisoquinoline-5-sulfonamide (3)

The synthetic procedure was adapted from [43]. Compound 2 (150 mg, 0.39 mmol, 1 equiv.) was dissolved in morpholine (1.5 mL), and the mixture was heated on a hot plate set to 100 °C for 24 h under N₂ atmosphere. After cooling to room temperature, the reaction mixture was diluted with water and the pH was adjusted to 7 using 1 M HCl. The aqueous layer was extracted with EtOAc (3 × 15 mL). The combined organic layers were washed with brine, dried over Na₂SO₄, and concentrated under reduced pressure.

The crude product was purified by normal-phase column chromatography (DCM/MeOH, 10:1, v/v). The obtained material was triturated with hexane and recrystallized from Hex/DCE (1:2, v/v) to afford the title compound as pale yellow crystals (122 mg, 67% yield).

TLC (DCM/MeOH, 10:1) R_f 0.54. HPLC purity 99.8% (220 nm). ¹H NMR (700 MHz, CDCl₃, TMS) δ 9.61 (s, 1H, HIQ), 8.64 (d, J = 6.0 Hz, 1H, HIQ), 8.36 (d, J = 6.0 Hz, 1H, HIQ), 8.32 (d, J = 8.0 Hz, 1H, HIQ), 7.09 (d, J = 8.0 Hz, 1H, HIQ), 5.68 (br s, 1H, NH), 4.83 (br s, 1H, NH), 4.02 (m, 4H), 3.27 (4H), 3.20 (q, J = 5.6 Hz, 2H, CH₂), 3.03 (q, J = 5.6 Hz, 2H, CH₂), 1.39 (s, 9H, C(CH₃)₃). ¹³C NMR (176 MHz, CDCl₃, TMS) δ 156.6, 155.4, 149.7, 145.1, 134.7, 133.3, 127.5, 123.1, 117.4, 113.2, 79.9, 66.8, 53.6, 43.7, 40.3, 28.3.

4.3.4. N-(2-Boc-aminoethyl)-8-benzylaminoisoquinoline-5-sulfonamide (4)

The compound was prepared following the procedure described for compound 3, except that benzylamine (1.5 mL) was used in place of morpholine, and the heating period was extended to 48 h. Workup and chromatographic purification were performed as described for Compound 3. The product was recrystallized from Hex/EtOAc (1:3, v/v) to afford the title compound as light yellow crystals (102 mg, 57% yield).

TLC (DCM/MeOH, 10:1) R_f 0.49. HPLC purity 99.5% (220 nm). ¹H NMR (700 MHz, CDCl₃, TMS) δ 9.39 (s, 1H, HIQ), 8.60 (d, J = 5.6 Hz, 1H, HIQ), 8.34 (d, J = 5.6 Hz, 1H, HIQ), 8.20 (d, J = 9.1 Hz, 1H, HIQ), 7.42–7.39 (m, 4H, PhH), 7.35 (t, J = 7.0 Hz, 1H, PhH), 6.60 (d, J = 9.1 Hz, 1H, HIQ), 5.96 (br s, 1H, NH), 5.47 (br s, 1H, NH), 4.85 (br s, 1H, NH), 4.59 (d, J = 4.9 Hz, 2H, Ph–CH₂), 3.17 (q, J = 5.6 Hz, 2H, CH₂), 2.98 (q, J = 5.6 Hz, 2H, CH₂), 1.38 (s, 9H, C(CH₃)₃). ¹³C NMR (176 MHz, CDCl₃, TMS) δ 156.5, 148.8, 145.5, 145.3, 136.8, 136.7, 133.2, 129.1, 128.1, 127.6, 119.8, 118.0, 117.6, 103.3, 79.9, 48.0, 43.6, 40.3, 28.3.

4.3.5. N-(2-Boc-aminoethyl)-8-phenylisoquinoline-5-sulfonamide (5)

Compound 2 (150 mg, 0.39 mmol, 1 equiv.) was dissolved in THF (7 mL). Phenylboronic acid pinacol ester (110 mg, 0.54 mmol, 1.4 equiv.) was added to the solution along with THF (1 mL), followed by the addition of degassed 2M Na₂CO₃ (2 mL). The mixture was degassed in an ultrasonic bath and subsequently purged with N₂ under stirring; this procedure was repeated twice. Pd(PPh₃)₄ (48 mg, 0.1 equiv.) was added under an N₂ flow and the mixture was refluxed on a hot plate set to 100 °C for 4 h under a N₂ atmosphere. The mixture was then cooled to room temperature and THF was removed under reduced pressure. The residue was diluted with 30 mL of water and extracted two times with DCM (1 × 30 mL and 1 × 15 mL). The combined organic layers were washed with brine, dried over Na₂SO₄, and concentrated under reduced pressure.

The crude product was purified by normal-phase column chromatography (CHCl₃/iPrOH, 94:6, v/v). The obtained material was triturated with hexane and recrystallized from Hex/DCE (2:1, v/v) to afford the product as white crystals (76 mg, 46% yield).

TLC (DCM/MeOH, 10:1) R_f 0.55. HPLC purity 98.5% (220 nm). ¹H NMR (700 MHz, CDCl₃ + TMS) δ 9.37 (s, 1H, HIQ), 8.69 (d, J = 6.0 Hz, 1H, HIQ), 8.48 (d, J = 6.0 Hz, 1H, HIQ), 8.45 (d, J = 7.7 Hz, 1H, HIQ), 7.59–7.50 (m, 6H, PhH + HIQ), 5.92 (br s, 1H, S–NH), 4.90 (br s, 1H, C(O)–NH), 3.25 (q, J = 5.6 Hz, 2H, CH₂), 3.11 (q, J = 5.6 Hz, 2H, CH₂), 1.40 (s, 9H, C(CH₃)₃). ¹³C NMR (176 MHz, CDCl₃, TMS) δ 156.8, 152.2, 147.0, 145.0, 137.4, 133.2, 132.8, 131.8, 129.9, 128.8, 128.7, 127.2, 126.7, 117.0, 80.2, 44.1, 40.4, 28.3.

4.3.6. N-(2-Boc-aminoethyl)-8-phenylaminoisoquinoline-5-sulfonamide (6)

The synthetic procedure was adapted from [44]. Compound 2 (150 mg, 0.39 mmol, 1 equiv.) was weighed into an oven-dried round-bottom flask. Anhydrous THF (2 mL), Cs₂CO₃ (375 mg, 3 equiv), and aniline (0.043 mL, 0.47 mmol, 1.2 equiv.) were added under an N₂ flow. The mixture was degassed in an ultrasonic bath and subsequently purged with N₂ under stirring; this procedure was repeated twice. Pd(PPh₃)₄ (39.8 mg, 0.09 equiv.) was then added and the mixture was refluxed on a hot plate at 100 °C for 43 h. The mixture was then cooled to room temperature and THF was removed under reduced pressure. The residue was diluted with 30 mL of water and extracted three times with DCM (1 × 30 mL and 2 × 15 mL). The combined organic layers were washed with brine, dried over Na₂SO₄, and concentrated under reduced pressure.

The crude product was purified by normal-phase column chromatography (CHCl₃/MeOH, 10:1, v/v). The obtained material was dissolved in hot DCE and precipitated by the addition of hexane, affording the title compound as yellow crystals (32 mg, 19% yield).

TLC (DCM/MeOH, 10:1) R_f 0.38. HPLC purity 98.6% (220 nm). ¹H NMR (700 MHz, DMSO-d₆, TMS) δ 9.80 (s, 1H, HIQ), 9.35 (s, 1H, NH), 8.65 (d, J = 6.0 Hz, 1H, HIQ), 8.33 (d, J = 6.0 Hz, 1H, HIQ), 8.08 (d, J = 8.4 Hz, 1H, HIQ), 7.77 (br t, J = 5.6 Hz, 1H, NH), 7.43 (t, J = 7.7 Hz, 2H, PhH), 7.38 (d, J = 7.7 Hz, 2H, PhH), 7.19–7.15 (m, 2H, PhH and HIQ), 6.66 (br t, J = 5.6 Hz, 1H, NH), 2.89 (q, J = 6.3 Hz, 2H, CH₂), 2.72 (q, J = 6.3 Hz, 2H, CH₂), 1.30 (s, 9H, C(CH₃)₃). ¹³C NMR (176 MHz, DMSO-d₆, TMS) δ 155.4, 148.0, 146.8, 144.8, 140.7, 135.0, 132.6, 129.5, 123.9, 122.3, 122.2, 119.1, 117.1, 106.6, 77.7, 45.1, 41.9, 28.1.

4.3.7. Deprotection of Compounds 2–6

General deprotection procedure. Boc deprotection was performed on a 50 mg scale for compounds 2–5 and on a 30 mg scale for compound 6. The corresponding N-(2-Boc-aminoethyl)-8-substituted isoquinoline-5-sulfonamide was dissolved in TFA/DCM (1:1, 1 mL) and stirred at room temperature overnight. The reaction mixture was concentrated under reduced pressure and the residue was neutralized with 5 mL of 5% aqueous NaHCO₃.

Workup. For compounds 2′–5′, the aqueous layer was extracted 6–8 times with DCM; the combined organic layers were dried over Na₂SO₄, concentrated, and dried in vacuo. For compounds 3′ and 6′, the aqueous layer was extracted 6–8 times with MTBE; in later experiments, DCM proved more effective for extraction. The combined organic layers were dried over Na₂SO₄, concentrated, and dried in vacuo. For compound 4′, the suspension was filtered, and the solid was washed with water (2×) and THF (2×) on filter. The solid was then dissolved in methanol, concentrated, and dried in vacuo.

A portion of compounds 2′–6′ was purified by reverse-phase HPLC as described in Section 4.1, while the remaining material was used for the synthesis of the corresponding ARC conjugates without further purification. NMR spectra were recorded from crude material, while only HPLC-purified portions of compounds 2′–6′ were used for biochemical assays.

N-(2-Aminoethyl)-8-chloroisoquinoline-5-sulfonamide (compound 2′, 8-Cl-(H-9)). HPLC purity 98.8% (220 nm). White solid. HRMS (ESI) for C₁₁H₁₂ClN₃O₂S: [M + H]⁺ calcd m/z 286.04115, found 286.0417; [M + 2H]²⁺ calcd m/z 143.52421; found 143.5239. ¹H NMR (700 MHz, DMSO-d₆, TMS) δ 9.71 (s, 1H, HIQ), 8.84 (d, J = 6.0 Hz, 1H, HIQ), 8.54 (d, J = 6.0 Hz, 1H, HIQ), 8.30 (d, J = 7.8 Hz, 1H, HIQ), 7.97 (br signal, 1H, HIQ), 2.79 (t, J = 6.4 Hz, 2H, CH₂), 2.47 (t, J = 6.4 Hz, 2H, CH₂). ¹³C NMR (176 MHz, DMSO-d₆, TMS) δ 149.3, 145.4, 136.2, 134.7, 132.6, 131.9, 126.9, 125.0, 117.3, 45.8, 41.3.

N-(2-Aminoethyl)-8-morpholinoisoquinoline-5-sulfonamide (compound 3′, 8-morpholino-(H-9)). Off-yellow solid. HPLC purity 92.5% (220 nm). HRMS (ESI) for C₁₅H₂₀N₄O₃S: [M + H]⁺ calcd m/z 337.13289, found 337.1339; [M + 2H]²⁺ calcd m/z 169.07008; found 169.0700. ¹H NMR (700 MHz, DMSO-d₆, TMS) δ 9.55 (s, 1H, HIQ), 8.68 (d, J = 5.8 Hz, 1H, HIQ), 8.37 (d, J = 5.8 Hz, 1H, HIQ), 8.24 (d, J = 8.1 Hz, 1H, HIQ), 7.29 (d, J = 8.1 Hz, 1H, HIQ), 3.92 (m, 4H), 3.22 (m, 4H), 2.87 (t, J = 6.6 Hz, 2H, CH₂), 2.71 (t, J = 6.5 Hz, 2H, CH₂). ¹³C NMR (176 MHz, DMSO-d₆, TMS) δ 154.5, 149.2, 144.5, 134.0, 132.3, 127.1, 122.2, 117.2, 113.7, 66.0, 53.2, 41.5, 39.9.

N-(2-Aminoethyl)-8-benzylaminoisoquinoline-5-sulfonamide (compound 4′, 8-BnNH-(H-9)). Yellow solid. HPLC purity 97.2% (220 nm). HRMS (ESI) for C₁₈H₂₀N₄O₂S: [M + H]⁺ calcd m/z 357.13797, found 357.1390; [M + 2H]²⁺ calcd m/z 179.07262; found 179.0725. ¹H NMR (700 MHz, DMSO-d₆, TMS) δ 9.85–9.84 (m, 1H, HIQ), 8.69–8.66 (br m, 1H, NH), 8.58 (d, J = 6.0 Hz, 1H, HIQ), 8.27 (d, J = 6.0 Hz, 1H, HIQ), 7.96 (d, J = 8.6 Hz, 1H, HIQ), 7.42 (d, J = 7.6 Hz, 2H, PhH), 7.33 (t, J = 7.6 Hz, 2H, PhH), 7.23 (t, J = 7.2 Hz, 1H, PhH), 6.53 (d, J = 8.6 Hz, 1H, HIQ), 4.60 (d, J = 5.7 Hz, 2H, CH₂), 2.76 (t, J = 6.5 Hz, 2H, CH₂), 2.58 (t, J = 6.5 Hz, 2H, CH₂). ¹³C NMR (176 MHz, DMSO-d₆, TMS) δ 149.3, 147.6, 144.7, 138.6, 135.6, 132.6, 128.5, 127.0, 126.9, 118.2, 117.7, 117.1, 102.6, 45.8, 42.6, 40.0.

N-(2-Aminoethyl)-8-phenylisoquinoline-5-sulfonamide (compound 5′, 8-Ph-(H-9)). HPLC purity 98.1% (220 nm). Off-white solid. HRMS (ESI) for C₁₇H₁₇N₃O₂S: [M + H]⁺ calcd m/z 328.11142, found 328.1121; [M + 2H]²⁺ calcd m/z 164.55935; found 164.5591. ¹H NMR (700 MHz, DMSO-d₆, TMS) δ 9.23 (s, 1H, HIQ), 8.73 (m, 1H, HIQ), 8.57 (m, 1H, HIQ), 8.40 (m, 1H, HIQ), 7.73 (m, 1H, PhH), 7.60–7.57 (m, 5H, PhH + HIQ), 2.84 (m, 2H, CH₂), 2.54 (m, 2H, CH₂). ¹³C NMR (176 MHz, DMSO-d₆, TMS) δ 151.0, 145.2, 144.2, 137.0, 134.1, 131.9, 131.0, 129.8, 128.7, 128.6, 127.1, 126.3, 117.3, 45.7, 41.3.

N-(2-Aminoethyl)-8-phenylaminoisoquinoline-5-sulfonamide (compound 6′, 8-PhNH-(H-9)). Yellow solid. HPLC purity 99.0% (220 nm). HRMS (ESI) for C₁₇H₁₈N₄O₂S: [M + H]⁺ calcd m/z 343.12232, found 343.1232; [M + 2H]²⁺ calcd m/z 172.06480; found 172.0647. ¹H NMR (700 MHz, DMSO-d₆, TMS) δ 9.79 (s, 1H, HIQ), 9.34 (br s, 1H, NH), 8.65 (d, J = 6.0 Hz, 1H, HIQ), 8.36 (d, J = 6.0 Hz, 1H, HIQ), 8.08 (d, J = 8.5 Hz, 1H, HIQ), 7.43 (m, 2H, PhH), 7.38 (d, J = 7.6 Hz, 2H, PhH), 7.19–7.15 (m, 5H, PhH and HIQ), 2.70 (t, J = 6.4 Hz, 2H, CH₂), 2.45 (t, J = 6.4 Hz, 2H, CH₂). ¹³C NMR (176 MHz, DMSO-d₆, TMS) δ 147.9, 146.6, 144.7, 140.6, 134.9, 132.5, 129.4, 123.7, 122.4, 122.1, 119.1, 117.0, 106.5, 45.5, 41.1.

Neutralization of H-9·HCl. H-9·HCl (90 mg) was neutralized with 5% aqueous NaHCO₃ (5 mL) and extracted 6–8 times with DCM. The combined organic layers were dried over Na₂SO₄, concentrated, and dried in vacuo.

4.3.8. Solid-Phase Synthesis

The resin-bound peptide fragment Fmoc-d-Arg(Pbf)-Ahx-[d-Arg(Pbf)]₃-d-Lys(Boc)-NH-resin was prepared following standard Fmoc solid-phase peptide synthesis protocols on Rink amide MBHA resin (0.61 mmol/g). Protected amino acids (3 equiv.) were activated with HBTU (2.95 equiv.) and HOBt (3 equiv.) in the presence of NMM (6 equiv.) in DMF. After 3 min, the coupling solutions were added to the resin and the mixtures were shaken for 60 min. Fmoc removal was carried out with 20% piperidine in DMF in two treatments of 5 min and 15 min with fresh reagent. The resin was then washed with organic solvents (DMF (3×), iPrOH (1×), DCM (3×)) and dried. This resin-bound peptide fragment was used directly for the synthesis of the ARC conjugates.

Novel ARC compounds were prepared according to previously reported procedure [21], with minor modifications, on a 7–12 μmol scale, using compounds 2′–6′ and H-9 as the amine building blocks. For each conjugate, 6-bromohexanoic acid (10 equiv.) was activated with N,N′-diisopropylcarbodiimide (5 equiv.) in DMF and added to the resin-bound peptide fragment (0.27 mmol/g; 1 equiv.). After 45–60 min of agitation, the resin was washed. The resulting resin-bound alkyl bromide was then reacted with the corresponding amine building block (2′–6′ or H-9, 4 equiv.) in DMSO/DIPEA for 20–24 h.

After coupling, the resins were washed in sequence with DMF (3×), isopropanol (1×), and DCM (3×) and then dried. Treatment with TFA/H₂O/triisopropylsilane (95:2.5:2.5, v/v/v) for 3 h was used as the standard cleavage and deprotection procedure. The crude products were precipitated in diethyl ether, centrifuged, and washed twice more with diethyl ether. The solids were purified by reversed-phase HPLC, as described in Section 4.1, and lyophilized to yield the final ARC conjugates.

ARC-2600 (conjugate of H-9). HPLC purity 99.0% (220 nm). HRMS (ESI) for C₅₃H₉₅N₂₃O₉S: [M + 2H]²⁺ calcd m/z 615.87747, found 615.8793; [M + 3H]³⁺ calcd m/z 410.92074; found 410.9220.

ARC-2601 (conjugate of 2′). HPLC purity 96.5% (220 nm). HRMS (ESI) for C₅₃H₉₄ClN₂₃O₉S: [M + 2H]²⁺ calcd m/z 632.85798, found 632.8604; [M + 3H]³⁺ calcd m/z 422.24108; found 422.2429.

ARC-2602 (conjugate of 5′). HPLC purity 92.7% (220 nm). HRMS (ESI) for C₅₉H₉₉N₂₃O₉S: [M + 2H]²⁺ calcd m/z 653.89312, found 653.8958; [M + 3H]³⁺ calcd m/z 436.26450; found 436.2663.

ARC-2603 (conjugate of 6′). HPLC purity 97.7% (220 nm). HRMS (ESI) for C₅₉H₁₀₀N₂₄O₉S: [M + 2H]²⁺ calcd m/z 661.39857, found 661.4015; [M + 3H]³⁺ calcd m/z 441.26814; found 441.2699; [M + 4H]⁴⁺ calcd m/z 331.20292; found 331.2046.

ARC-2604 (conjugate of 4′). HPLC purity 97.3% (220 nm). HRMS (ESI) for C₆₀H₁₀₂N₂₄O₉S: [M + 2H]²⁺ calcd m/z 668.40639, found 668.4094; [M + 3H]³⁺ calcd m/z 445.94002; found 445.9418; [M + 4H]⁴⁺ calcd m/z 334.70683; found 334.7085.

ARC-2605 (conjugate of 3′). HPLC purity 98.2% (220 nm). HRMS (ESI) for C₅₇H₁₀₂N₂₄O₁₀S: [M + 2H]²⁺ calcd m/z 658.40385, found 658.4067; [M + 3H]³⁺ calcd m/z 439.27166; found 439.2733; [M + 4H]⁴⁺ calcd m/z 329.70556; found 329.7070.

4.3.9. FA-Based Binding/Displacement Assay

The assay was performed as described previously [28] in a final volume of 20 μL/well on a 384-well microtiter plate. The buffer solution contained 50 mM HEPES (pH = 7.5), 150 mM NaCl, 0.005% Tween-20 (P20) and 5 mM DTT. The measurements were carried out at 30 °C after 30 min incubation at the same temperature. The concentration of the active forms of PKs was determined on each day of experiment prior to performing FA-based displacement or enzyme kinetic assays by titration of a fixed concentration of the fluorescent tracer with the corresponding PK. The competitive binding (displacement) assays were carried out by mixing a dilution series of the competitor with a fixed concentration of the fluorescent tracer and PK. The following conditions were used in displacement assays with different PKs: 3 nM PKAcα and 2 nM tracer ARC-583 (K_D = 0.48 nM); 3 nM ROCK2 and 2 nM ARC-1042 (K_D = 0.1 nM), 2.5 or 4 nM PKN1 and 1 nM ARC-1059; and 0.9 or 1.3 nM PKN3 and 1 nM ARC-1059.

4.3.10. Kinase Activity Assay

The catalytic activity of PKN1 and PKN3 was measured using PhosphoSens^® technology (AssayQuant Technologies Inc, Marlborough, MA, USA) according to the manufacturer’s recommendations. The phosphorylation mixtures contained the inhibitors (three-fold dilutions), 0.75 nM PKN1 or 0.5 nM PKN3 (concentration of active kinase, determined by FA-assay prior to the assay); 1 mM ATP, 10 mM Mg(OAc)₂, 15 µM Sox Kinase Sensor substrate (code AQT050910); 50 mM HEPES (pH = 7.5); 0.012% Brij-35; 0.22 mg/mL BSA; 1.2 mM DTT; and 0.5 mM Na₂EDTA. The solutions of all other components were pre-incubated for 20 min at 30 °C before starting the phosphorylation by the addition of the peptide substrate. The progress of the phosphorylation at 30 °C was monitored by the change in fluorescence intensity, using a PHERAStar microplate reader (BMG Labtech, Ortenberg, Germany), for 60–90 min. The slopes of fluorescence intensity vs. time dependencies were calculated using the linear region of the data. The reference solutions without the inhibitors and without PK were used as positive and negative controls of the catalytic activity of the enzyme, respectively.

5. Conclusions

This study establishes a rational structure-guided strategy for generating highly potent and selective bumped inhibitors of PKN3. A C8-modified isoquinoline sulfonamide scaffold conjugated to d-arginine-rich peptide chain yielded an inhibitor with subnanomolar affinity and marked kinome selectivity.