Stereospecific Cross-Coupling Reactions Provide Conformationally-Biased Arylalkanes with Anti-Leukemia Activity

Abstract

A focused small library of carbamates and alcohols was prepared employing stereospecific Kumada-ring opening reactions of tetrahydropyrans. The core framework of the library members is acyclic and incorporates 1,3-substituents, to provide a conformational bias in avoiding syn-pentane interactions. A new compound with micromolar activity against MOLT-4, CCRF-CEM, and HL-60(TB) leukemia cell lines was identified from this series.

1. Introduction

Drug discovery efforts often show bias towards flat, achiral molecules, however candidates containing multiple sp³ atoms and stereogenic centers have been associated with increased clinical success.^[1][2] Biological targets are inherently three-dimensional structures, therefore ligands that extend in all three dimensions may increase interactions to improve potency, while reducing off-target binding.^[3] For example, an increase in dimensionality of the core skeleton can be important for assisting appendage π systems to more effectively interact with binding sites.^[1]

In contrast to the flat and rigid structures of sp²-rich molecules, compounds with numerous sp³ centers exhibit potentially large conformational profiles. Limitation of this profile can be an important attribute for activity by reducing conformational entropic costs upon binding to the biological target.^[4] Biologically active natural products, including polyketides, typically inhabit a preferred conformation, while still retaining a degree of flexibility.^[5],[6] For example, polypropionates often adopt a low-energy conformer to avoid syn-pentane interactions.^[5]

The polyketide discodermolide is one example where molecular structure reduces the number of populated conformers (Figure 1).^[6] Extracted from sea sponge Discodermia dissoluta,^[7] discodermolide has been investigated for its nanomolar ability to inhibit growth of paclitaxel-resistant cancer cells.^[8] Discodermolide contains two polypropionate motifs that avoid syn-pentane interactions, causing two hairpin-like turns in the linear backbone.^[9] Analogues of discodermolide conserve these motifs, as the conformation is critical to its activity.^[6] Partly due to this restrictive effect on conformation, methods to access polypropionates and similar structures are of great value.^{[10],[11],[12]}

Figure 1.

Polyketide discodermolide.

In 2014, our laboratory reported an C_sp³-C_sp³ Kumada cross-coupling reaction of aryl tetrahydropyrans (THPs) and tetrahydrofurans (Scheme 1a).^{[13],[14],[15]} This ring-opening reaction utilizes an achiral nickel catalyst to couple a benzylic ether with Grignard reagents in a stereospecific manner. Importantly, the THP starting material is generated in a single step from the commercially available aldehyde by a diastereoselective Prins cyclization.^{[16],[17],[18]} Upon ring-opening of the THP, a compound rich in stereochemical information is generated, including 1,3-substitutients on the linear backbone.

Scheme 1.

a) Stereospecific ring-opening Kumada reaction. b) Library synthesis.

We set out to utilize this ring-opening reaction to synthesize a small library of compounds and test for anticancer activity (Scheme 1b). We hypothesized that the sp³ core skeleton, 1,3-substituent motif, ability to diversify at the alcohol position, and low molecular weight (<500 da) made this scaffold an appropriate candidate for library generation.^[19],[20] Additionally, this substructure shares similarities to our previously synthesized compounds that demonstrated activity towards breast cancer cell lines, further directing our testing efforts.^[21][22] The aryl group was modified to include various heterocycles, and the alcohol was unaltered or further modified to carbamates or hydroxymethyl pyridines. These modifications were chosen due to their prevalence in successful pharmaceutical agents and to increase logP of the series.^{[23],[24],[20]}

Our four target alcohols were the first compounds in our library to be synthesized (Scheme 2). The THP substrates cis-(±)-1 to cis-(±)-4 were synthesized via a clay-mediated Prins cyclization, employing four different aromatic aldehydes.^[16] With THPs in hand, the ring-opening Kumada cross-coupling reactions were performed to yield desired alcohols, syn-(±)-5, syn-(±)-6, syn-(±)-7, and syn-(±)-8. This reaction proceeds reliably with inversion at the benzylic carbon, therefore cis-disubstituted tetrahydropyrans produce alcohols with a syn configuration.^[13] Each alcohol was tested for anti-cancer activity.

Scheme 2.

a) Alcohols synthesized via Kumada ring-opening cross-coupling reaction.

Next, the alcohols were acylated to yield carbamate derivatives (Scheme 3). A series of dimethyl carbamates and morpholine carbamates (syn-(±)-9 to syn-(±)-16) were synthesized. Either carbamate could be synthesized using sodium hydride and corresponding carbamoyl chloride,^[25] however, a more reliable method utilized carbonyldiimidazole (CDI) and the corresponding amine.^[26] All carbamate derivatives were subjected to biological testing. Lastly, our synthetic efforts concluded with transformation of the pendant alcohols to hydroxymethyl pyridines (Scheme 4). Compounds syn-(±)-17 to syn-(±)-20 were synthesized using sodium hydride and 2-bromo(methyl)pyridine. All pyridines were isolated in >20:1 dr and subjected to biological testing.

Scheme 3.

Dimethyl and morpholine carbamate derivatives.

Scheme 4.

Pyridine derivatives.

With the synthesis of the library complete, we began our evaluation of biological activity. We collaborated with the National Institute of Health Developmental Therapeutics Program (NIH DTP) to evaluate our experimental compounds against 60 human cancer cell lines (at 10 μM).^[27],[28] If substantial growth inhibition or cell death is detected, the compound is selected by the NIH for further concentration dependent testing.

Results of the initial evaluation are shown in Table 1. Each compound is presented with the specific cell line the compound was most active against and associated growth percentages.¹ Untreated cell lines have a growth percentage of 100%, therefore a positive value >100% indicates accelerated growth, a positive value <100% indicates inhibition of growth, and a negative growth percentage indicates cell death.

Table 1.

Biological data obtained from NIH60 one-dose study. Cell line most susceptible to each tested compound is reported.

Compound	Cancer Type	Cell Line	Percent Growth[a]
syn-(±)-5	Leukemia	MOLT-4	−54
syn-(±)-6	Renal	CAKI-1	76
syn-(±)-7	Renal	A498	82
syn-(±)-8	Renal	A498	67
syn-(±)-9	Breast	T-47D	75
syn-(±)-10	Renal	CAKI-1	73
syn-(±)-11	Melanoma	UACC-62	56
syn-(±)-12	Leukemia	CCRF-CEM	37
syn-(±)-13	Melanoma	UACC-62	75
syn-(±)-14	Renal	UO-31	67
syn-(±)-15	Renal	HL-60(TB)	78
syn-(±)-16	Renal	UO-31	75
syn-(±)-17	Ovarian	OVCAR-8	48
syn-(±)-18	Leukemia	RPMI-8226	63
syn-(±)-19	Leukemia	RPMI-8226	75
syn-(±)-20	Leukemia	HL-60(TB)	70

^[a]Percentages are compared to a no-drug control. Positive 0–100% growth indicates growth inhibition, and negative growth percentage indicates lethality.

Alcohol syn-(±)-5 showed greatest potency, with activity against a range of cell lines, including all leukemia cell lines. For example, it provided –54% growth of leukemia cell line MOLT-4 (acute lymphoblastic leukemia). Alcohol syn-(±)-5 was then subjected to concentration dependent testing by the NIH DTP (Table 2).² Alcohol syn-(±)-5 demonstrated micromolar activity towards the range of cell lines, and was most potent towards MOLT-4 with an LC₅₀ value of 6.1 μM. Additionally, its activity towards HL-60(TB) at 8.3 μM is intriguing. Cell line HL-60(TB) is an acute promyelocytic leukemia (APL) line, a subtype of acute myeloid leukemia (AML) with a 5-year survival rate for AML of only 24%.^[29] Alcohol syn-(±)-5 also showed activity towards colon cancer cell lines HCT-116 and KM12.

Table 2.

LC₅₀ values for syn-(±)-5.

Cancer Type	Cell Line	LC50 (μM)
Leukemia	MOLT-4	6.1
Leukemia	CCRF-CEM	5.1
Leukemia	HL-60(TB)	8.3
Colon	HCT-116	5.2
Colon	KM12	5.7

To further investigate the properties of syn-(±)-5, we sought to determine its preferred conformation. We calculated the energy of a series of conformers in a density functional theory (DFT) study at the B3LYP level employing the 6–31G(d) basis set.^{[30],[31],[32]} The lowest energy conformer (Figure 1) confirms our expectations based on hand-held models. The sterically bulky naphthyl and phenyl rings align syn to hydrogen atoms, to adopt a conformation that minimizes syn-pentane interactions.

In summary, we have synthesized a small library of molecules utilizing our previously developed Kumada ring-opening cross-coupling reaction. The molecules in our library contained an sp³ core scaffold, 1,3-substituents to induce conformational bias, aromatic and heterocyclic rings, along with various alcohol, carbamate, and hydroxymethyl pyridine appendages to provide potential hydrogen bond acceptors and donors. The library compounds were tested for anti-cancer activity in collaboration with the NIH DTP and one alcohol, syn-(±)-5, demonstrated micromolar activity against MOLT-4, CCRF-CEM, and HL-60(TB) leukemia cell lines. Based on the observed structure activity relationships, the naphthyl aromatic ring and primary alcohol were both critical functional groups for anti-leukemia activity of this compound.

Experimental Section

In a glovebox, a flame-dried 7 mL vial equipped with a stir bar was charged with substrate (1.0 equiv), Ni(cod)₂ (0.10 equiv), rac-BINAP or DPEphos (0.10 equiv), and PhMe (2.4 mL). Methylmagnesium iodide (2.5 equiv) was then added dropwise. After 24 h, the reaction was removed from the glovebox, quenched with methanol (2 mL), filtered through a plug of silica gel (neat Et₂O), and concentrated in vacuo.^[13]

Figure 2.

Lowest energy conformer of syn-(±)-5 obtained via DFT calculations at the B3LYP/6–31G(d) level.