Structure-Activity Relationship Studies of Functionalized Aromatic Peptidomimetics as Neurolysin Activators

Abstract

Modulating peptidase neurolysin (Nln) has been identified as a potential cerebroprotective target for the development of therapeutics for ischemic stroke. Continued structure-activity relationship studies on peptidomimetic small molecule activators of Nln bearing electron-donating and electron-withdrawing functionalized phenyls are explored. Incorporation of fluorine or trifluoromethyl groups produces Nln activators with enhanced A₅₀, while methoxy substitution produces derivatives with enhanced A_max. Selected activators containing methoxy or trifluoromethyl substitution are selective for Nln over related peptidases and possess increased blood-brain barrier penetrability than initial hits.

Stroke is a cerebrovascular disease that is caused by interruption of the blood flow to the brain, with resultant ischemia can result in mild to severe disability and/or eventually death. Stroke is classified as either hemorrhagic stroke or ischemic stroke, with the latter being the most common form (87% of presentations).^1,2 Globally stroke is a leading cause of death and disability accounting for 11% of all deaths.^3,4 Recent data on mortality in the U.S. place stroke as the fifth leading cause of death.⁵ Evidence is emerging that COVID-19 infection increases both morbidity and mortality of ischemic stroke.⁶ Only one pharmacological treatment option is currently available to treat ischemic stroke which is tissue plasminogen activator (tPA). This agent suffers from the major drawback of a limited window of opportunity with a requirement to administer tPA within three to four hours of suffering a stroke. In many cases of ischemic stroke, the patient, or relatives, are not aware they have suffered a stroke until hours later.^7,8,9 Thus, the search for new therapeutic modalities to treat ischemic stroke remains an unmet clinical need.

Recent in vitro and in vivo experimental studies have identified peptidase neurolysin (Nln) as a compensatory cerebroprotective mechanism in the post-ischemic brain.¹⁰ This function of Nln is primarily explained by its ability to inactivate several cerebrotoxic and generate a few other cerebroprotective neuropeptides and by that reduce excitotoxicity, oxidative stress, edema formation, blood brain barrier (BBB) hyper-permeability, and neuroinflammation.^11,12

Previously we reported the screening of a small molecule library that identified several dipeptide hit compounds, including His-Phe (1) that possessed the ability to activate Nln.¹³ Subsequent hit-to-lead optimization led to the discovery of the first lead compounds with enhanced activation activity, selectivity, brain penetration and stability (Figure 1).¹⁴ Several compounds, represented by 2, 3 and 4 (Figure 1) provided significant increase in potency, as measured by A₅₀, the concentration required to activate Nln by 50%, over 1. However, maximum percent activation (A_max) was limited to approximately half that of the hit compound. Unfunctionalized phenyl derivative 5 was observed to have one of the highest A_max values of compounds synthesized (467%), above that of the original hit compound 1 (442%), but with limiting A₅₀ (252 μM). Our prior structure-activity relationship (SAR) studies showed that aromatic functionalization with electron-donating and -withdrawing groups successfully modulated both the A₅₀ and A_max values of Nln activators. Herein, we report expansion of SAR studies focused on the Eastern aromatic moiety fragment of histidine-based peptidomimetics to enhance A₅₀ and A_max activity towards Nln. Two compounds were identified that possess enhanced activation profiles while maintaining peptidase selectivity and BBB penetration.

Figure 1.

Structures, A₅₀ and A_max activity of previously identified Nln activators.

Target compounds were synthesized in moderate to good yields (40–70%) by amide coupling reaction of commercially available Boc-L or Boc-D histidine with an appropriately substituted aniline using the benzotriazol-1-yloxytris(dimethylamino)phosphonium hexafluorophosphate (BOP) coupling agent. Subsequent trifluoroacetic acid (TFA)-mediated deprotection of the tert-butoxycarbonyl protecting group afforded the desired peptidomimetics without racemization of the chiral starting material as previously reported (Scheme 1).¹⁴

Scheme 1:

Reagents and conditions: a) BOP, DIPEA, DMF, 50 °C, 1h b) Substituted aniline, DIPEA, DMF, r.t., 30 min (compounds 5–24) or NaH, 0 °C, 30 min (compounds 25–28); then 0–50 °C, overnight; c) 20% TFA in DCM, r.t., 3h.

A focused library of substituted phenyl peptidomimetics bearing electron-donating (EDG) or electron-withdrawing groups (EWG) were obtained to determine effect to activate Nln. Our prior studies indicated that addition of a methoxy EDG resulted in increased A_max, and to this end we synthesized L and D ortho, meta and para-methoxy derivatives 7–12 (Table 1). Gratifyingly, this modification resulted in both increased potency and maximum activation activity over the parent (5; A₅₀ = 252 μM, A_max = 467%). Activity was found to be dependent on substituent position for A₅₀ with a pattern of increasing activity of ortho > meta > para wherein the A₅₀ of compounds 7, 9, and 11 were 45.1, 105.4 and 241.4 μM respectively. Compounds with unnatural D amino acid stereochemistry followed the same pattern of activity wherein A₅₀ values of 8, 10 and 12 were 53.5, 120.8 and 214.2 μM respectively. When A_max is considered, the same activity pattern is preserved with compounds 7, 9, and 11 possessing 426%, 454% and 582% respectively for the natural L amino acid and compounds 8, 10 and 12 possessing 428%, 478% and 540% respectively, for the D amino acid. Overall derivatives with ortho-OMe substitution possessed an activity profile greater than the unsubstituted parent compound, with up to 5.5-fold lower A₅₀ and retained or increased A_max.

Table 1:

Structure, A₅₀, A_max, MPO score and LLE profile of functionalized phenyl histidine-containing peptidomimetics.

Compd #	Structure		A501 (μM; 95% CI)	Amax2 (%; 95% CI)	MPO3 Score	LLE4
	L/D-His	R
His-Phe (1)	L	N/A	74.5 (56.2 to 98.8)	442.3 (418.8 to 467.6)	4.1	6.78
5	L	H	252 (170.6 to 406.6)	466.6 (402.3 to 579.7)	5	4.02
6	D	H	113.6 (92.8 to 140.5)	440.3 (414.9 to 470.4)	5	4.37
7	L	2-OMe	45.1 (34.5 to 59.2)	426 (401.5 to 453.5)	5	5.28
8	D	2-OMe	53.5 (42.5 to 67.8)	427.6 (405.7 to 452.1)	5	5.21
9	L	3-OMe	105.4 (82.5 to 136.3)	453.5 (423.2 to 490.5)	5	4.91
10	D	3-OMe	120.8 (95.5 to 154.9)	477.6 (445.6.6 to 517.2)	5	4.85
11	L	4-OMe	241.4 (182.4 to 332.5)	581.6 (520.8 to 670.3)	5	4.55
12	D	4-OMe	214.2 *(151.2 to 321.9)	540.2 (475.4 to 641.6)	5	4.61
13	L	3,5-OMe	52.1 (43.8 to 62.2)	462.8 (444.6 to 482.6)	4.73	4.60
14	L	3,4,5-OMe	117 (93 to 149)	436 (407.6 to 470.8)	4.4	4.99
15	L	2-F	37.5 (14.6 to 88.4)	268.2 (231.1 to 310.9)	5	4.45
16	D	2-F	78.2 (61.1 to 101.1)	330.9 (312.6 to 352.4)	5	4.13
17	L	3-F	24.8 (15.2 to 40.3)	382.1 (351.6 to 417.2)	5	4.63
18	D	3-F	30.9 (25.8 to 37.2)	396.5 (383 to 410.9)	5	4.53
19	L	4-F	136 (112.8 to 166.5)	487.9 (485.4 to 523.4)	5	3.89
20	D	4-F	36.1 (26.4 to 49.7)	403.3 (378.5 to 431.2)	5	4.46
21	L	3-CF3	28.6 (22.5 to 36.3)	311.8 (299.2 to 325.4)	5	3.63
22	D	3-CF3	22.8 (18.3 to 28.4)	362.4 (348.6 to 377)	5	3.73
23	L	4-CF3	32.2 (27 to 38.4)	4335.2 (325.1 to 346)	5	3.58
24	D	4-CF3	15.6 (10.7 to 22.7)	324.8 (306.1 to 345.1)	5	3.89
25	L	3,5-F	90.6 (66.9 to 125)	431.2 (397 to 474)	5	3.83
26	L	2,6-F,4-Br	26.7 (23.2 to 30.7)	370 (361 to 379)	5	4.67
27	L	3,5-CF3	279 (161 to 596)	633.2 (505.2 to 946.4)	4.96	3.05
28	L	2,4-CF3	25.2 (20.1 to 31.6)	368 (354 to 384)	4.96	4.09

¹Concentration required to activate Nln by 50%.

²Maximum % activation achieved.

³Multi-Parameter Optimization.

⁴Ligand-lipophilicity efficiency.

We next investigated multiple substitution on the phenyl ring with 3,5-dimethoxy (13) and 3,4,5-trimethoxy (14) derivatives. The natural amino acid L stereocenter was retained since no eutomer was evident from the previous derivatives. Disubstituted compound 13 possessed an A₅₀ of 52.1 μM, approximately 2-fold greater than the respective mono-meta-methoxy substituted compound 9, but equipotent with mono-ortho-methoxy 7. Maximum activation activity (463%) was equipotent with both 7 and 9 within confidence levels. Trimethoxy 14 suffered an approximate 2-fold loss of A₅₀ (117 μM) compared with 13 and a slight suppression of A_max. This can be rationalized by the apparent deactivating effect of the 4-OMe substituent (as seen in 11) which is more than two-fold less active than other methoxy substituted compounds and its contribution to combined effects in the trimethoxy substituted compound would be anticipated to be deleterious to overall activity.

Along with activation activity in silico central nervous system (CNS) penetration by multiparameter optimization (MPO),¹⁵ and ligand-lipophilicity efficiency (LLE),¹⁶ were calculated to predict the ‘drug-likeness’ of compounds that would need to penetrate the BBB.¹⁷ All compounds were calculated to have MPO scores between 4.4–5, predictive of suitable BBB penetration. All compounds were found to reduce LLE compared with the His-Phe hit (1), however all EDG substituted compounds improved LLE above that of the unsubstituted parents 5 and 6, exemplified by compounds 7 and 8 with LLE values of 5.3 and 5.2 indicating greater ‘drug-likeness’.

We next investigated the effects of EWGs with fluorine and trifluoromethyl substituted derivatives. The substitution activity pattern was changed upon the addition of a fluorine with meta > ortho > para for the natural L amino acid-containing compounds 15, 17 and 19 with A₅₀ of 37.5, 24.8 and 136 μM, respectively. Interestingly for the unnatural D amino acid-containing compounds 16, 18 and 20 (A₅₀ of 78.2, 30.9 and 36.1 μM respectively) the substitution pattern to activity correlation was again changed to meta > para >> ortho. Further, differences in stereochemistry presented themselves within the ortho and para compounds with L-15 being twice as potent as D-16 yet L-19 was approximately 4-fold less potent than D-20. The meta derivatives L-17 and D-18 were equipotent within confidence levels. When maximum activation activity was analyzed, both sets of amino acids shared a common substitution pattern to activity correlation with para > meta > ortho (Table 1). In general A_max values for all fluorine substituted derivatives were substantially lower than both the non-substituted parent and the original hit compound His-Phe. The para-F compound 19 was the only derivative with an increased A_max (488%).

To explore the effect of greater electronegative groups we synthesized meta- and para-substituted trifluoromethyl derivatives (21–24) as ortho derivatives which proved less active and were not considered further. All four compounds proved highly active compared with both the unsubstituted parent and initial hit compound. In the L amino acid-based compounds both meta- and para-CF₃ substituted derivatives 21 and 23 were equipotent within confidence levels with A₅₀ of 28.6 and 32.2 μM and A_max of 312% and 335% respectively. Within the D amino acid-based compounds, derivative 24 with para-CF₃ was the most potent of all compounds synthesized with A₅₀ of 15.6 μM compared with meta-CF₃ 22 with A₅₀ of 22.8 μM. All trifluoromethyl substituted compounds possessed reduced A_max compared to both initial hit and unsubstituted parent. Due to the increased lipophilicity of the CF₃ group compounds 21–24 possessed much lower LLE than their monofluoride counterparts with values as low as 3.58.

Combinations of EWG substituents were investigated with 3,5-difluoro compound 25 (A₅₀ = 91 μM) >3-fold less active than the mono-meta-CF₃ derivative 17 but with slightly increased A_max, albeit lower than the unsubstituted parent. Compound 26 possessing a 2,6-difluoro, 4-bromo substitution pattern possessed an A₅₀ of 26.7 μM, equipotent with the monofluoro derivatives 21, 22 and 23 with only slightly increased A_max. The 3,5-CF₃ derivative 27 was the most inactive compound for A₅₀ (279 μM) of the entire series but possess the highest A_max of 633%. The 2,4-CF₃ derivative 28 (A₅₀ = 25.2 μM, A_max = 368%) was 10-fold more potent in A₅₀ than 27 but had overall lower activity than the para-CF₃ derivative 24 but with only marginally better A_max.

To determine the effects of fluorine and methoxy substitution on the ‘drug-like’ profile of Nln activators beyond potency the 3,5-dimethoxy derivative 13 and para-fluoro derivative 24 were evaluated for selectivity to related peptidases (Figure 2) and for BBB penetration (Figure 3).

Figure 2.

The effect of compounds 13 and 24 on activity of peptidases closely related to Nln. All panels document concentration-dependent effect of the indicated compounds on hydrolysis of a respective quenched fluorescent substrate (n = 4, mean ± SD): Mca-Pro-Leu-Gly-Pro-D-Lys(DNP)-OH at 15 μM for TOP, Mca-Arg-Pro-Pro-Gly-Phe-Ser-Ala-Phe-Lys(Dnp)-OH at 10 μM for NEP and ACE, and Mca-Ala-Pro-Lys- (Dnp)-OH at 10 μM for ACE2. V₀ %, represents the initial velocity of QFS hydrolysis by the peptidases. In all panels, the initial velocity of the hydrolysis in the absence of either compound corresponds to 100% on the vertical axis.

Figure 3.

Apical to basolateral transport of Nln activators at 10 μg/mL across an in vitro co-culture model of the BBB at 37 °C illustrates good brain permeability of peptidomimetic compounds 13 and 24. The permeability coefficient (Pe) was calculated from the cleared volume of each compound versus time. Values represent the mean ± SEM of 3–4 measurements. ns: non-statistically significant.

Both EDG and EWG-containing compounds 13 and 24 respectively, retained a high degree of selectivity for Nln over the closely related peptidases thimet oligopeptidase (TOP), neprilysin (NEP), angiotensin-converting enzyme 2 (ACE2), and ACE, which together with Nln, belong to the same family of enzymes (Figure 2). Interestingly the selectivity of derivative 24 for Nln represents the first such report for an unnatural D amino acid. Derivatives 2, 3, 4 and 13 which all show similar selectivity,¹⁴ are all L-histidine based, thus, stereochemistry does not impact selectivity.

Derivatives 13 and 24 were evaluated for their ability to penetrate the BBB using an in vitro co-culture model system comprised of primary astrocytes and bEnd3 cells to ensure that the additional substituents were not deleterious to this property (Figure 3). Both compounds which are in good correlation with MPO predictive data, trended to have improved BBB penetration compared to positive control compound 4, which itself showed significantly improved permeability over initial hits.¹⁴

In conclusion, a series of 22 EDG or EWG functionalized phenyl histidine-containing peptidomimetics have been synthesized and evaluated for their ability to activate Nln. In total, 21 derivatives possess increased A₅₀ over the unsubstituted parent, while 13 derivatives were more potent than the initial hit compound 1. Five compounds possessed greater maximum activation activity over the unsubstituted parent and six possessed greater A_max than the initial hit. Overall EWG incorporation produces more potent Nln activators based on A₅₀ than does EDG substitution. However, EDG incorporation produces derivatives with greater A_max. The greater electronegativity of CF₃ results in reduced A_max and LLE compared with the less electronegative fluorine substituent but greater A₅₀. Selectivity for related peptidases is retained in both methoxy and trifluoromethyl derivatives as is BBB penetration. Importantly the nature of the stereochemistry of the histidine amine is not important for peptidase selectivity, potentially leading to synthesis of more stable unnatural amino acid analogues. This study provides further information on the SAR required to modulate A₅₀ and A_max at the Eastern fragment of the peptidomimetic scaffold and will inform strategic modifications to enhance one or both metrics as needed in future studies directed towards the Western fragment which will be reported in due course.