Benzisothiazolones as modulators of macrophage migration inhibitory factor

Abstract

Substituted N-phenylbenzisothiazolones have been investigated as inhibitors of the tautomerase activity of the proinflammatory cytokine MIF (macrophage migration inhibitory factor). Numerous compounds were found to possess antagonist activity in the low micromolar range with the most potent being the 6-hydroxy analog 1w. 1w and the p-cyano analog 1c were also shown to exhibit significant inhibition of the binding of MIF to its transmembrane receptor CD74. Consistently, both compounds were also found to retard the MIF-dependent phosphorylation of ERK1/2 in human synovial fibroblasts.

The cytokine MIF (macrophage migration inhibitory factor) is emerging as an important therapeutic target for inflammatory, autoimmune, and hyperproliferative diseases.¹ MIF possesses multiple biological functions and it participates in many cellular processes. In addition to activating mitogen-activated protein kinases (MAPK),² MIF directly down-regulates p53 expression and function,³ and it counter-regulates the expression of glucocorticoids.⁴ Via binding to its cellular receptor CD74, MIF can promote cell survival through activation of the PI3K/Akt pathway. The roles of MIF in cardiovascular diseases, inflammation, and innate immune responses are well documented.^1,5,6 MIF is implicated in multiple disorders including rheumatoid arthritis,⁷ glomerulonephritis,⁸ diabetes,⁹ atherosclerosis,¹⁰ sepsis,¹¹ asthma,¹² and acute respiratory distress syndrome.¹³ Moreover, it has been shown that inhibition of MIF or CD74 attenuates prostate cancer cell growth.¹⁴ In addition to its activities as a cytokine, MIF is an enzyme that catalyzes ketoenol tautomerization. It appears that the tautomerase active site is located in proximity to the MIF-CD74 binding site.¹⁵ Thus, although evidence has not emerged that the enzymatic activity of MIF is significant in mammals, substantial effort has been directed at the discovery of small-molecule inhibitors of both MIF’s enzymatic and signaling activities.¹⁶ Our joint efforts began with virtual screening that yielded 11 structurally diverse inhibitors of MIF-CD74 binding with micromolar activities.¹⁷ One series, with a benzoxazolone core, was further pursued and yielded antagonists with IC₅₀ values as low as 7.5 nM in the tautomerase assay and 80 nM in the receptor binding assay.¹⁸ In parallel de novo design efforts, triazole derivatives emerged as viable alternatives. This led to the report of additional antagonists, and to the first agonists of MIF-CD74 signaling.¹⁹ Another promising hit from the virtual screening was 2-(4-chlorophenyl)-6-fluoro-benzo[d]isothiazol-3(2H)-one 1a with an IC₅₀ of 4.2 µM in the tautomerase assay.¹⁷ Results of further explorations of benzisothiazolones are reported here.

Our computational studies indicated that 1a should bind to MIF in the manner illustrated in Figure 1. Notably, results of Monte Carlo free-energy perturbation calculations²⁰ found that (a) the orientation with the carbonyl group of 1a pointing towards Ile64 is preferred by ca. 6 kcal/mol over the 180°-rotated alternative with the carbonyl group pointed towards Tyr95, and (b) substitutions at C5 or C6 with F, Cl, OH, OCH₃, and CN should be comparably viable and preferred over CH₃. This model received subsequent support from crystallographic studies of complexes of 2 and 3 with MIF.²¹ 1a, 2, and 3 align well with the bicyclic ring systems inserted into the active site and with the fluorine of 1a and bicyclic OH groups of 2 and 3 proximal to Asn97, and the C=O groups of 1a and 2 hydrogen-bonded to the NH of Ile64. The synthetic efforts were focused accordingly with particular interest in preparing the 6-OH analog in view of precedents indicating potential value of hydrogen-bonding with Asn97.¹⁷ The structures of the synthesized analogs, 1b – 1y, are recorded in Table 1.

Figure 1.

Computed structure of the complex of 1a bound in the MIF tautomerase site after energy-minimization with MCPRO²⁰ starting with protein coordinates from the 1GCZ crystal structure.^16a

Table 1.

Inhibitory activities for in the MIF tautomerase assay

Compd	R5	R6	R7	Rm	Rp	IC50 (µM)
1a	H	F	H	H	Cl	4.2
1b	H	F	H	H	OMe	6.2
1c	H	F	H	H	CN	6.4
1d	H	F	H	OMe	H	25
1e	H	F	H	Cl	H	5.0
1f	H	F	H	CH2OAc	H	2.3
1g	H	F	H	CH2OMe	H	2.4
1h	H	OMe	H	H	Cl	1.9
1i	H	H	F	OMe	H	8.0
1j	H	H	F	H	Cl	2.8
1k	H	H	Cl	H	Cl	6.6
1l	H	F	F	H	Cl	5.6
1m	F	H	H	H	Cl	5.6
1n	F	H	H	H	OMe	6.5
1o	F	F	H	H	Cl	4.9
1p	F	H	H	OMe	H	2.8
1q	NO2	H	H	H	Cl	6.2
1r	CF3	H	H	H	Cl	8.5
1s	H	Br	H	H	Cl	11
1t	H	CN	H	H	Cl	3.1
1u	Br	H	H	H	Cl	7.9
1v	CN	H	H	H	Cl	19
1w	H	OH	H	H	Cl	1.0
1x	H	H	F	OH	H	5.9
1y	H	F	H	CH2OH	H	4.8

Synthetic routes used for the preparation of N-arylbenzisothiazolones 1b – 1v are summarized in Scheme 1. 1b – 1l were prepared starting from the easily available secondary arylamides²² 4 via method A (Scheme 1). Ortho-lithiation of 4 with sec-butyllithium in the presence of tetramethylethylenediamine (TMEDA),²³ followed by quenching of the dianion with dimethyl disulfide afforded the corresponding arylmethylthioethers 5 in good yields. Sulfides 5 were heated with NaIO₄ in methanol to yield the sulfoxides 6, which were cyclized with thionyl chloride to afford the desired N-arylbenzisothiazolones 1b – 1l in moderate to good yields.²⁴ Benzisothiazolones 1m – 1p were prepared via Method B in Scheme 1. Introduction of sulfide functionality into arylamide 7 en route to 9 required protection of the bisactivated ortho-site of arylamides 7 with a trimethylsilyl group.²⁵ Thus, 7 was subjected to standard directed lithiation conditions followed by trapping of the generated dianion by TMS chloride to provide amides 8. Then 8 was subjected to the second ortho-lithiation, dimethyl disulfide quench, and desilylation with tetrabutylamonium fluoride (TBAF) to yield the sulfides 9. These products were subsequently oxidized and cyclized as in Method A to afford the desired 1m – 1p. Analogs 1q – 1t, which possess functional groups that are incompatible with ortho-lithiation methodology, were prepared starting from the corresponding 2-fluorobenzoic acids 10 via Method C. Benzoyl chlorides, formed by refluxing 10 in thionyl chloride, were reacted with anilines to afford amides 11 in good to excellent yields. For 1t, the cyano group was introduced via Pd-catalyzed cyanation of the aryl bromide.²⁶ Amides 11 were subjected to S_nAr reaction with sodium thiomethoxide in DMSO to afford thioethers 12 in good yields. Finally, 12 was cyclized to 1q – 1t upon treatment with sulfuryl chloride.²⁷ Benzisothiazolones 1u – 1v were prepared as depicted in Method D starting from readily available 5-bromo-2-fluorobenzamide 13.²⁸ Cu-catalyzed N-arylation of 13 with 1-choro-4-iodobenzene (14) allowed preparation of arylamides 15 en route to 1u and 1v via the protocol in Method C. Finally, preparation of hydroxyl-substituted analogs 1w – 1y required use of protecting groups, as detailed in Scheme 2. Precursors 1f, 1h and 17 were prepared by the methods in Scheme 1 and deprotected in a final step.

Scheme 1.

Synthesis of N-arylbenzisothiazolones 1b–1v.

Scheme 2.

Final deprotection step in the syntheses of 1w–1y.

The ability of the N-arylbenzisothiazolones 1 to inhibit the tautomerase activity of MIF was evaluated with 4-hydroxyphenylpyruvate (4-HPP) as the substrate, as previously presented.^17,29 Human MIF, prepared recombinantly, was used throughout. All compounds show good activities ranging from 1 to 25 µM (Table 1). As expected from the structural model, there was little effect on the activities from varying the solvent-exposed substituents at the para (R^p) position (1a, 1b, 1c). Some substitutions at the meta position provided improvement, e.g., 1f and 1g, while variation at the 6-position did lead to the most potent compounds, 1h and 1w. Indeed, the hydroxyl analog 1w is the most active at 1 µM. We have found few compounds more active than this in the 4-HPP tautomerase assay including the covalent inhibitor 4-IPP (4-iodo-6-phenylpyrimidine), which yields an IC₅₀ of 4.5 µM.^17–19

Many of the compounds were also tested in the MIF-CD74 binding assay. As previously detailed, this assay features biotinylated MIF and immobilized CD74 ectodomain (CD74^73–232) with streptavidin conjugated alkaline phosphatase processing p-nitrophenylphosphate as the reporter.^17,30 Most of the present benzisothiazolones showed weak inhibitory activity that did not reach the 50% level in the micromolar range. The two strongest antagonists are 1c and 1w, which showed 49% inhibition of binding at 78 nM and 23% inhibition at 4 nM, respectively. A few of the compounds emerged as agonists;¹⁹ the strongest is 1i, which enhances the MIF-CD74 binding by 2.7-fold at 4 µM.

In view of these results, inhibition of MIF signal transduction by 1c and 1w was further investigated in human target cells, specifically, primary synovial fibroblasts. The expectation is that MIF-CD74 inhibition should reduce ERK (extracellular-signal-regulated kinase) phosphorylation, which is relevant to the control of hyperproliferative and autoimmune diseases.³¹ Thus, the test compounds were added together with MIF for 30 minutes to 1×10⁶ synovial fibroblasts cultured in DMEM/0.1% fetal bovine serum. The cells then were lysed and the intracellular contents of phospho-ERK1/2 and total ERK1/2 were quantified by specific antibodies and western blotting. The results in Figure 2 show that addition of MIF enhances ERK phosphorylation, while co-administration with 1c and 1w results in substantial reduction of phosphorylation. The most profound effects are obtained with the 6-hydoxy analog 1w, which reduces the phosphorylation by 35% at 1 µM and still by 20% at 100 nM. For comparison, results are included for the well-studied MIF tautomerase inhibitor ISO-1;³² the racemate yields a 19% reduction of ERK phosphorylation at 2 µM.

Figure 2.

Relative potency of MIF antagonists in reducing MIF-dependent ERK1/2 phosphorylation in human synovial fibroblasts. The bottom numerals show the ratio of phosphorylated to total ERK protein relative to MIF alone from densitometric scanning. Values are expressed in relative units from two experiments. *P<0.05, **P<0.01 by t-test.

A final issue that was addressed was the possible covalent modification of MIF by the benzisothiazolones. Many MIF inhibitors have been reported to function in this manner by bonding to the nucleophilic Pro1 or Cys residues.^15b,33 Indeed, Ouertatani-Sakouhi et al. recently reported 12 covalent modifiers of Pro1 from high-throughput screening and mass spectrometric analysis;³³ these compounds feature reactive carbonyl functionality that is normally avoided when seeking reversible inhibitors.³⁴ However, they also demonstrated that the anti-inflammatory agent ebselen inhibits MIF by covalent attachment specifically to Cys80 among the three Cys options (56, 59, and 80).³³ The modification causes disruption of MIF trimer assembly and, therefore, negates formation of the tautomerase active sites.³³ Besides the structural similarity of 1a and ebselen, it has also been shown that some N-arylbenzisothiazolones can react with cysteines yielding opening of the benzisothiazolone ring and disulfide formation.³⁵ The latter process should be facilitated by electron-withdrawing substituents in the para position of the N-phenyl ring as in the reported case with a p-sulfamoyl group.

Thus, mass spectrometric studies were pursued to assess possible covalent attachment of 1a to MIF in a similar manner to the prior investigation of 4-IPP and a benzoxazolone MIF inhibitor.¹⁸ Equimolar quantities of human MIF and 1a were mixed (30 µM of each reactant in 0.01 mL) for one hour at 25 °C and the reaction mixture was analyzed (Bruker 9.4T FT-ICR). A small peak appeared at the mass position expected for the covalent attachment of one molecule of 1a with a MIF monomer (12,616 Da), as shown in Figure 3. Analysis of the four highest peaks in the MIF alone and MIF + 1a isotopic distributions indicates that ca. 3% of the MIF is modified by 1a. With 4-IPP, which has a similar IC₅₀ (4.5 µM) in the tautomerase assay, ca. 20% of the protein was modified.¹⁸ Further FT-ICR MS analysis of a trypsin-digested sample revealed a peptide of M = 6083 Da that corresponds to the covalent adduct of the MIF^12–66 fragment and 1a (Figure 3). These data are consistent with covalent attachment of MIF and 1a at Cys56 or Cys59. The fact that Cys59 is well-buried in the structure of the MIF trimer shifts the likelihood of the modified residue towards being Cys56. While the catalytic Pro1 is not modified by 1a, cysteine modification could influence MIF conformational integrity and account for alterations in tautomerization activity and receptor binding functions. However, the extent of covalent modification appears small, and it is unlikely that it can account for the absolute activities in Table 1 or of the relative values. For example, electron-withdrawing groups should make the benzisothiazolones more reactive electrophiles; however, this is not apparent in comparisons of the IC₅₀ values such as for 1f and 1g versus 1a and 1c, and in comparing the isomeric 1t and 1v, where 1v would be expected to be more reactive towards nucleophilic attack at the conjugating para-sulfur site.

Figure 3.

Mass spectrometric results demonstrating minor covalent modification of MIF by 1a. The inset shows the product after trypsin digestion of the modified protein.

In summary, N-phenylbenzisothiazolones have been examined for their ability to inhibit the tautomerase and signaling activities of the cytokine MIF. Numerous analogs have been reported with activities in the low micromolar range for inhibition of the tautomerase activity. Though the expectation is that they are reversible inhibitors that bind near the tautomerase active site, mass spectroscopic investigation of the original virtual screening hit 1a demonstrated that it yields a small amount of covalent modification of Cys56 or Cys59 of MIF. The most potent MIF tautomerase inhibitor reported here is the 6-hydroxy analog 1w at 1 µM. This compound was also shown to cause striking retardation of ERK phosphorylation in synovial fibroblasts. Investigations continue in several veins including evaluation of the anti-inflammatory and anti-proliferative activities of the compounds.

Research Highlights.

> Benzisothiazolones as inhibitors of macrophage migration inhibitory factor (MIF). > Inhibition of MIF-CD74 binding and ERK phosphorylation. > Mass spectral analysis of covalent modification of cysteines. > Potential anti-inflammatory and anti-proliferative agents.