Chemical Modification of the Multi-Target Neuroprotective Compound Fisetin

Abstract

Many factors are implicated in age-related CNS disorders making it unlikely that modulating only a single factor will provide effective treatment. Perhaps a better approach is to identify small molecules that have multiple biological activities relevant to the maintenance of brain function. Recently, we identified an orally active, neuroprotective and cognition-enhancing molecule, the flavonoid fisetin, that is effective in several animal models of CNS disorders. Fisetin has direct antioxidant activity and can also increase the intracellular levels of glutathione (GSH), the major endogenous antioxidant. In addition, fisetin has both neurotrophic and anti-inflammatory activity. However, its relatively high EC₅₀ in cell based assays, low lipophilicity, high tPSA and poor bioavailability suggest that there is room for medicinal chemical improvement. Here we describe a multi-tiered approach to screening that has allowed us to identify fisetin derivatives with significantly enhanced activity in an in vitro neuroprotection model while at the same time maintaining other key activities.

INTRODUCTION

There are currently no drugs available that prevent the nerve cell death associated with the majority of age-related disorders of the CNS. A prime example of this problem is ischemic stroke which is the leading cause of adult disability and the third leading cause of death in the US¹. Worldwide, approximately 5 million people die each year of stroke and the mortality rates are estimated to double by the year 2020². The nerve cell death associated with cerebral ischemia is due to multiple factors resulting from the lack of oxygen to support respiration and ATP synthesis, acidosis due to the buildup of the glycolytic product lactic acid, the loss of neurotrophic support, multiple metabolic stresses and inflammation^3a,b. While the focus of current drug discovery paradigms is primarily on the development of high affinity, single target ligands, a drug directed against a single molecular target may not be effective in treating the nerve cell death associated with conditions such as stroke because of the multitude of insults that contribute to the cell’s demise. This conclusion is supported by the lack of drugs for the treatment of stroke. Indeed, the only FDA-approved treatment to date is recombinant tissue-type plasminogen activator (rt-PA)⁴, which is a vascular agent. An alternative approach is to identify small molecules that have multiple biological activities relevant to the maintenance of neurological function.

Over the last few years, we have identified an orally active, neuroprotective and cognition-enhancing molecule, the flavonol fisetin⁵. Fisetin not only has direct antioxidant activity but it can also increase the intracellular levels of glutathione (GSH), the major intracellular antioxidant, via the activation of transcription factors such as Nrf2⁵. Fisetin can also maintain mitochondrial function in the presence of oxidative stress. In addition, it has anti-inflammatory activity against immune cells and inhibits the activity of 5-lipoxygenase, thereby reducing the production of lipid peroxides and their pro-inflammatory by-products⁵. This wide range of actions suggests that fisetin has the ability to reduce the loss of neurological function associated with multiple disorders, including stroke.

Although fisetin has been shown to be effective in the rabbit small clot embolism model of stroke⁶, its relatively high EC₅₀ in cell based assays (2–5 µM) as well as its low lipophilicity (CLogP 1.24), high tPSA (107Å), high number of hydrogen bond donors (HBD = 5) and poor bioavailability²⁸ suggest that there is room for medicinal chemical improvement if fisetin is to be used therapeutically for treating neurological disorders such as stroke. However, given its ability to activate multiple target pathways related to neuroprotection, screening for improvements is significantly more complicated than with the current classical approach to the development of a single target drug. In this paper, we describe a multi-tiered approach to screening that has allowed us to identify fisetin derivatives with significantly enhanced neuroprotective activity in an in vitro ischemia model while at the same time maintaining other key actions including anti-inflammatory and neurotrophic activity as well as the ability to maintain GSH under conditions of oxidative stress.

CHEMISTRY

The synthesis of substituted chalcones 013, 032, 033, 057, 063, 085, 086, 086A, 105 –108 and 137 was carried out by condensation of 2'-hydroxy acetophenones with appropriately substituted aldehydes using Ba(OH)₂ in methanol⁷ (Scheme 1). The tri-hydroxy chalcones 011, 034 and 087 were prepared from the corresponding chalcones by treatment with BBr₃ in dicloromethane⁸ and the di-hydroxy chalcone 088 was synthesized by THP deprotection using pTSA in methanol⁷ from the corresponding chalcone. The suitably substituted flavones 018, 038, 058, 068, 089, 115, 116, 119 and 120 were synthesized from the corresponding chalcones using I₂ in DMSO⁹ (Scheme 2). The hydroxy flavones 002, 028, 064, 072 and 094 were obtained from the corresponding chalcones by demethylation/deethylation or debenzylation using BBr₃ in dicloromethane⁸ or H₂, Pd/C in EtOAc/methanol¹⁰ respectively.

Scheme 1.

Synthesis of chalcone derivatives.

Scheme 2.

Synthesis of flavone derivatives.

Substituted flavonols 025, 036, 037, 059, 065, 090, 091, 114, 117, 118, 122 and 139 were prepared (Scheme 3) using 5.4% NaOH, 30% H₂O₂ in methanol¹¹ from the corresponding aldehydes. The known compounds fisetin, 002 and 04P were purchased from Indofine chemicals and the other hydroxy flavonols 027, 040, 041, 069, 070, 092, 093 and 140 were obtained from their corresponding flavonols (Scheme 3) by demethylation/deethylation (BBr₃ in dicloromethane)⁸ or debenzylation (H₂, Pd/C in EtOAc/methanol)¹⁰ methods. Finally the substituted quinolines 001, 004, 007, 017, 021–024, 083, 084, 109–113 and 121 were synthesized (Scheme 4) by condensation of 2'-amino acetophenones with appropriately substituted aldehydes using H₂SO₄ in methanol¹². Experimental procedures and data for all of the compounds are reported below or in the Supporting Information.

Scheme 3.

Synthesis of flavonol derivatives

Scheme 4.

Synthesis of quinoline derivatives.

RESULTS

The goals of this study were to improve the potency of fisetin based upon the activation of multiple neuroprotective pathways while at the same time altering its physicochemical properties to be more consistent with those of successful CNS drugs (molecular weight ≤ 400, CLogP ≤ 5, tPSA ≤ 90, HBD ≤ 3, HBA ≤ 7)^26,33 in order to increase the possibility of efficient brain penetration, and to better understand its SAR. We took two different approaches to the improvement of fisetin. In the first, we removed/modified/replaced the different hydroxyl groups in a systematic manner. In the second approach, we modified the flavone scaffold by changing it to a quinoline while at the same time maintaining key structural elements.

We chose to use for our primary screen an in vitro ischemia model in combination with the HT22 hippocampal nerve cell line⁶. For this screen, we set a cut-off for the EC₅₀ of 1 µM. To induce ischemia in the HT22 cells we used iodoacetic acid (IAA), a well known, irreversible inhibitor of the glycolytic enzyme glyceraldehyde 3-phosphate dehydrogenase (G3PDH)¹⁴. IAA has been used in a number of other studies to induce ischemia in nerve cells^15a–e and we have used it in several recent screens for neuroprotective molecules^16a,b. The changes following IAA treatment of neural cells are very similar to those seen in animal models of ischemic stroke¹⁷ and include alterations in membrane potential¹⁸, breakdown of phospholipids¹⁹, loss of ATP^20a,b and an increase in reactive oxygen species (ROS)^21,19.

We used three secondary screens that allowed us to assess three key activities of fisetin that are highly relevant to stroke, as well as other neurological disorders: maintenance of GSH, the major endogenous cellular antioxidant, inhibition of bacterial lipopolysaccharide (LPS)-induced microglial activation, an indicator of anti-inflammatory activity and PC12 cell differentiation, a measure of neurotrophic activity. All of these activities are relevant to the nerve cell loss seen in stroke^22a–c. Previous and ongoing studies suggest that these activities of fisetin are mediated via distinct pathways but that all three may be important for the neuroprotective effects of fisetin in vivo⁵. To assess GSH maintenance we looked at total intracellular GSH levels after a 24 hr treatment with the compound both in the absence and presence of glutamate, an inducer of GSH loss and oxidative stress^23a,b. Inhibition of LPS-induced microglial activation was determined by treating N9 mouse microglial cells with LPS alone and in the presence of the compounds and assaying nitrite (spontaneously produced by air oxidation of nitric oxide) levels in the medium 24 hr later²⁴. PC12 cell differentiation was determined by treating PC12 cells with the compounds and looking at neurite outgrowth after 24 hr. In all cases, fisetin was used as a positive control²⁵.

STRUCTURE ACTIVITY RELATIONSHIP

We began by assessing the roles of the four different hydroxyl groups in the activity of fisetin. We found that removal of the 7 hydroxyl (04P) not only improved the neuroprotective activity ~6-fold over fisetin in our primary screen of in vitro ischemia without loss of either the GSH maintaining activity or PC12 cell differentiation but also enhanced lipophilicity increasing the CLogP from 1.24 to 1.82 (Table 1). Furthermore, this modification did not alter the anti-inflammatory activity of fisetin (Table 1). This finding allowed us to replace the 7 hydroxyl with hydrophobic groups in order to further improve the lipophilicity and tPSA to values more consistent with typical CNS drugs^26,33. The addition of a benzene ring (040) to the A ring further enhanced neuroprotective activity ~5.5-fold with a much more pronounced effect seen with the α-naphtha derivative (040) as opposed to the β-naphtha (041) derivative (Table 1). However, this modification eliminated the ability of the derivative to maintain GSH under conditions of oxidative stress. For this derivative, the 3 hydroxyl was not important for neuroprotective activity (040 vs 002) but did enhance anti-inflammatory activity.

Table 1.

Half maximal effective concentrations (EC₅₀s) for protection in the in vitro ischemia assay were determined by exposing HT22 cells to different doses of each derivative in the presence of 20 µM IAA for 2 hr (HT22/IAA). Cell viability was determined after 24 hr by the MTT assay. The ability to maintain GSH (GSH) was determined by treating HT22 cells with different doses of each derivative (1–10 µM) in the presence of 5 mM glutamate. After 24 hr cell extracts were prepared and analyzed for total GSH. Fisetin (10 µM) was used as a positive control. The ability to induce PC12 cell differentiation (PC12 diff’n) was determined by treating PC12 cells in N₂ medium with different doses of each derivative (1–10 µM) for 24 hr. Differentiation was assessed by visual inspection with fisetin (10 µM) as a positive control. Anti-inflammatory activity (microglia) was assessed in N9 microglial cells treated with bacterial lipopolysaccharide alone or in the presence different doses of each derivative (1–10 µM) for 24 hr. Fisetin was used as a positive control. TEAC values, a measure of direct antioxidant activity, were determined using the ABTS⁺ decolorization assay.

Compound	M.Wt	tPSA	CLogP	Structure	EC50 in vitro ischemia (µM)	GSH	PC12 diff'n	microglia	TEAC
Fisetin	286	107	1.24		3	yes	yes	80%	3
002	304	67	3.52		0.08	no	yes	55%	0.18
04P	270	87	1.82		0.5	yes	yes	80%	2
018	338	45	5.16		no	no	no	2%	0.15
025	354	65	4.71		0.5	no	no	11%	0.84
027	298	87	2.77		0.5	no	no	82%	1.89
028	282	67	3.30		0.25	no	no	93%	0.27
036	376	65	4.94		0.3	no	no	13%	0.15
038	360	45	5.39		no	no	no	2%	0.2
040	320	87	2.99		0.09	no	yes	91%	2.4
041	320	87	2.99		0.25	no	yes	87%	1.26
058	386	45	5.87		0.5	no	no	83%	0.09
059	402	65	5.42		0.17	no	no	14%	0.27
064	296	56	3.66		0.03	no	no	41%	0.12
065	424	65	5.65		0.08	no	yes	88%	0
069	312	76	3.19		0.04	no	no	77%	1.89
070	334	76	3.41		>0.5	no	yes	78%	0.63
072	318	56	3.88		0.04	no	no	19%	0.2
092	312	76.00	3.19		0.02	no	no	72%	1.74
093	298	87.00	2.40		>0.5	no	no	0%	1.56
094	282	66.76	2.93		>0.5	no	no	5%	0.15
114	357	49.17	4.51		0.07	no	yes	26%	1.44
115	341	29.54	4.99		0.2	no	no	23%	0.03
116	319	29.54	4.70		>0.5	no	no	2%	0.12
117	309	49.77	4.17		0.02	yes	no	11%	3
118	331	49.77	4.40		0.04	yes	no	0%	0.93
119	293	29.54	4.65		>0.5	no	no	8%	0
120	315	29.54	4.88		0.25	no	no	35%	0.24
122	335	49.77	4.28		0.09	yes	yes	0%	2.4
140	372	70.00	4.09		0.005	yes	yes	50%	2.1
142	351	70.00	3.64		0.015	yes	yes	21%	2.79

We also examined the role of the B ring hydroxyls in neuroprotection as well as the other key activities of a-naphtha derivative. Changing both hydroxyls to ethoxy groups (036, 038) not only greatly reduced neuroprotective activity but also eliminated both the anti-inflammatory activity and the ability to induce PC12 cell differentiation. Changing only one of the hydroxyls to a methoxy group enhanced neuroprotective activity ~2-fold over 040 in the absence of the 3 hydroxyl group (072) but greatly reduced neuroprotective activity relative to 040 in the presence of the 3 hydroxyl (070). Furthermore, this modification did not restore the ability to maintain GSH under conditions of oxidative stress and the derivative without the 3 hydroxyl (072) also lacked anti-inflammatory activity and the ability to induce PC12 cell differentiation. Surprisingly, changing the 4'-hydroxyl to a benzyloxy group (065) restored neuroprotective activity in the presence of the 3 hydroxyl. Compounds possessing tertiary nitrogen, a feature of many CNS drugs, show a higher degree of brain permeation^26,33,34. With this observation in mind, both hydroxyls on the B ring were replaced with a single dimethyl amino group at the 4' position resulting in a highly neuroprotective compound in the presence of the 3 hydroxyl group (118) and a somewhat less effective compound in its absence (120). This modification also eliminated 2 hydrogen bond donors. Although 118 regained the ability to maintain GSH levels, it lacked both anti-inflammatory and neurotrophic activity. Modification of the dimethyl amine to a pyrrolidine group at the 4' position gave a compound that had excellent neuroprotective activity in the presence of the 3 hydroxyl (114) and could also induce PC12 cell differentiation but had poor anti-inflammatory activity and did not maintain GSH levels. However, addition of a 3' hydroxyl to this derivative resulted in a compound with outstanding neuroprotective activity (EC₅₀ = 5 nM) (140) that could also maintain GSH under conditions of oxidative stress, induce PC12 differentiation and had reasonably good anti-inflammatory activity.

As a second approach, we replaced the benzene ring with two methyl groups (027) in order to generate a derivative with a similar CLogP and tPSA as 040 but with a less bulky addition to the A ring (Table 1). Surprisingly, this derivative not only showed significantly decreased neuroprotective activity as compared with 040 but also lost the ability to induce PC12 cell differentiation along with the continued failure to maintain GSH levels. Removal of the 3 hydroxyl enhanced neuroprotective activity 2-fold (028) but did not restore the induction of PC12 cell differentiation or the maintenance of GSH. Modification of the B ring hydroxyls produced mixed results. Modification of one hydroxyl to a methoxy (064, 069, 092) improved neuroprotective activity ~10–20-fold but slightly reduced anti-inflammatory activity. Similar to the results with the derivatives of 040, modification of both the B ring hydroxyl groups to ethoxy groups (018, 025) did not improve neuroprotective activity. Furthermore, none of these derivatives regained the ability to maintain GSH or induce PC12 cell differentiation and they also showed reduced anti-inflammatory activity. While the methoxy, benzyloxy dimethyl derivative did have somewhat enhanced neuroprotective activity relative to 027 in the presence of the 3 hydroxyl (059), it was deficient in anti-inflammatory activity. Furthermore, separation of the B ring hydroxyls (093, 094) not only eliminated neuroprotective activity but the other key activities as well. However, similar to the results with the derivatives of 040, replacement of the hydroxyls with a single dimethyl amino group at the 4' position produced a compound with excellent neuroprotective activity but only in the presence of the 3 hydroxyl (117 vs 119). This compound also regained the ability to maintain GSH but lacked both neurotrophic and anti-inflammatory activity. Addition of a single pyrrolidine group to the 4' position instead gave a compound that had excellent neuroprotective activity only in the presence of the 3 hydroxyl (122 vs 116) and could also maintain GSH levels and induce PC12 cell differentiation but still had poor anti-inflammatory activity. Addition of a 3' hydroxyl to this derivative resulted in a compound with outstanding neuroprotective activity (142) that could also maintain GSH under conditions of oxidative stress and induce PC12 differentiation but had lower anti-inflammatory activity than 140.

Chalcones are intermediates in the synthesis of flavonoids and were used to determine the effect of opening up the C-ring on activity (Table 2). Surprisingly, the chalcones of both the naphtha (034) and dimethyl derivatives (011) had similar (034) or enhanced (011) neuroprotective activity compared to their flavone counterparts and also regained all of the key activities including the ability to maintain GSH under conditions of oxidative stress. In contrast, the chalcones where both the B ring hydroxyls were modified had either no (032, 013, 086) or greatly reduced (063) neuroprotective activity. Furthermore, splitting the B ring hydroxyls of 011 (087) eliminated the ability to maintain GSH under conditions of oxidative stress. The conversion of a hydroxyl to a methoxy (088) also eliminated the ability to promote PC12 cell differentiation. Thus, of the chalcones tested 011 and 034 are superior to fisetin by both the selection criteria and medicinal chemistry properties.

Table 2.

Half maximal effective concentrations (EC₅₀s) for protection in the in vitro ischemia assay were determined by exposing HT22 cells to different doses of each derivative in the presence of 20 µM IAA for 2 hr (HT22/IAA). Cell viability was determined after 24 hr by the MTT assay. The ability to maintain GSH (GSH) was determined by treating HT22 cells with different doses of each derivative (1–10 µM) in the presence of 5 mM glutamate. After 24 hr cell extracts were prepared and analyzed for total GSH. Fisetin (10 µM) was used as a positive control. The ability to induce PC12 cell differentiation (PC12 diff’n) was determined by treating PC12 cells in N₂ medium with different doses of each derivative (1–10 µM) for 24 hr. Differentiation was assessed by visual inspection with fisetin (10 µM) as a positive control. Anti-inflammatory activity (microglia) was assessed in N9 microglial cells treated with bacterial lipopolysaccharide alone or in the presence different doses of each derivative (1–10 µM) for 24 hr. Fisetin was used as a positive control. TEAC values, a measure of direct antioxidant activity, were determined using the ABTS⁺ decolorization assay.

Compound	M.Wt	tPSA	CLogP	Structure	EC50 in vitro ischemia (µM)	GSH	PC12 diff'n	microglia	TEAC
Fisetin	286	107	1.24		3	yes	yes	80%	3
011	284	78	3.64		0.05	yes	yes	94%	2.7
013	340	56	5.62		no	no	no	56%	0.09
032	362	56	5.84		no	no	no	5%	0.12
034	306	78	3.86		0.08	yes	yes	75%	2.8
063	410	56	6.55		0.5	no	no	9%	0.15
086	388	55.76	6.33		no	yes	yes	70%	0.12
087	284	77.76	3.57		0.05	no	yes	53%	0.93
088	298	66.76	4.08		0.2	no	no	56%	0.12

As an alternative approach to improving fisetin, we modified the flavone scaffold changing it to a quinoline scaffold (Table 3) in an attempt to further improve potency and physiochemical properties while retaining the key structural elements of the flavone in the quinoline scaffold. The simplest version, 007, showed a ~75-fold increase in neuroprotective activity relative to fisetin, maintained GSH under conditions of oxidative stress and had strong anti-inflammatory activity. However, it did not induce PC12 cell differentiation. We explored a number of modifications to see if we could enhance neuroprotective activity and/or restore the PC12 cell differentiating activity. Interestingly, the substitution of an ethoxy (023) or an iso-propoxy (024) for the methoxy group on the C ring did restore the differentiating activity while also slightly improving (~2-fold) the neuroprotective activity relative to 007. Importantly, replacement of the O-methyl group with an O-cyclopentyl ring resulted in a compound with a >400-fold decrease in EC₅₀ relative to fisetin for neuroprotective activity (121) and maintenance of all of the key activities. For all forms of the quinoline-based derivative, removal of one (022) or both (021) of the B ring hydroxyls or conversion of one or both of these hydroxyls to methoxy (001, 017), ethoxy (004), nitro (111) or chlorine or fluorine (not shown) greatly reduced or eliminated neuroprotective activity. All of these changes also reduced or eliminated all of the other key activities. Splitting the two ring hydroxyls (083, 084) also reduced neuroprotective activity and eliminated the ability to maintain GSH and induce PC12 cell differentiation but did not impact anti-inflammatory activity. In contrast to the derivatives based on the flavone scaffold, the addition of a single dimethyl amino (109, 112) or pyrrolidine group (110, 113) to the 4' position of the B ring did not enhance neuroprotective activity relative to the 3', 4' dihydroxy derivative and generally resulted in a reduction or elimination of the other key activities. Thus, in the presence of the quinoline scaffold the catechol group on the B ring is essential for activity.

Table 3.

Half maximal effective concentrations (EC₅₀s) for protection in the in vitro ischemia assay were determined by exposing HT22 cells to different doses of each derivative in the presence of 20 µM IAA for 2 hr (HT22/IAA). Cell viability was determined after 24 hr by the MTT assay. The ability to maintain GSH (GSH) was determined by treating HT22 cells with different doses of each derivative (1–10 µM) in the presence of 5 mM glutamate. After 24 hr cell extracts were prepared and analyzed for total GSH. Fisetin (10 µM) was used as a positive control. The ability to induce PC12 cell differentiation (PC12 diff’n) was determined by treating PC12 cells in N₂ medium with different doses of each derivative (1–10 µM) for 24 hr. Differentiation was assessed by visual inspection with fisetin (10 µM) as a positive control. Anti-inflammatory activity (microglia) was assessed in N9 microglial cells treated with bacterial lipopolysaccharide alone or in the presence different doses of each derivative (1–10 µM) for 24 hr. Fisetin was used as a positive control. TEAC values, a measure of direct antioxidant activity, were determined using the ABTS⁺ decolorization assay.

Compound	M.Wt	tPSA	CLogP	Structure	EC50 in vitro ischemia (µM)	GSH	PC12 diff'n	microglia	TEAC
Fisetin	286	107	1.24		3	yes	yes	80%	3
001	281	51	3.89		no	no	no	69%	0.24
004	323	40	5.33		no	no	no	76%	0.12
007	267	62	3.66		0.04	yes	no	85%	0.36
017	281	51	3.89		0.5	no	no	90%	0.15
021	235	21	4.55		0.75	no	no	6%	0.12
022	251	42	4.07		no	no	no	4%	0.81
023	281	62	4.20		0.02	yes	yes	90%	0.90
024	295	62	4.50		0.02	yes	yes	80%	0.18
083	267	62.06	3.30		>0.5	no	no	84%	0.05
084	295	62.05	4.13		0.21	no	no	64%	0.27
109	278	24.83	4.82		0.06	no	no	62%	0.06
110	304	24.83	4.93		no	no	no	67%	0.27
111	296	93.63	4.33		no	no	no	25%	0.18
112	306	24.83	5.66		0.05	no	no	71%	0.15
113	332	24.83	5.77		0.5	no	no	67%	0.27
121	321	62.05	5.14		0.007	yes	yes	82%	0.40

The transcription factor Nrf2 plays a key role in regulating GSH metabolism in many different cell types²⁷. We have shown that fisetin can induce Nrf2 and that this correlates with its ability to enhance GSH levels⁵. To determine if the derivatives which can maintain GSH levels do so by increasing Nrf2 we looked at Nrf2 levels in the nuclei of derivative-treated cells using fisetin as a positive control (Table 4). Surprisingly, not all of the derivatives that maintain GSH levels induce Nrf2. This was particularly true for the derivatives based on the quinoline scaffold where none of them increased Nrf2 despite being very effective at maintaining GSH levels.

Table 4.

The ability of the derivatives that maintain GSH levels to induce the transcription factor Nrf2 was assayed by SDS-PAGE and Western blotting of nuclear extracts of untreated and derivative-treated cells. Fisetin treatment was used as a positive control.

Compound	Nrf2
Fisetin	yes
04P	yes
007	no
011	yes
034	yes
086	yes
117	no
118	no
121	no
122	no
140	no
142	no

DISCUSSION and CONCLUSIONS

Several important findings emerge from this study. First, within the flavone scaffold we were able to demonstrate SARs with respect to four distinct biological activities and to improve neuroprotective activity up to 600-fold (140). We also show that while it is possible to maintain all of the biological activities that are likely to be important for in vivo efficacy, each of these activities has specific and unique structural requirements. Thus, it is possible to balance enhanced neuroprotective activity with the other key activities as well as the physical characteristics of the compounds in order to arrive at compounds that have the best chance for efficacy in vivo. An additional key finding is that neither the neuroprotective activity nor any of the other three key activities of the fisetin derivatives show any correlation with antioxidant activity as defined by the TEAC value (Table 1).

Each of the key activities of the fisetin derivatives shows distinct structural requirements. For example, within the flavone structure (Table 1), the maintenance of GSH poses the strictest structural requirements. It is highly sensitive to modification of the A ring (040, 027). However, substitution of the B ring hydroxyls with a single tertiary-amino group is compatible with the maintenance of GSH even in the presence of A ring modifications (117, 118) as long as a 3 hydroxyl group is present. In contrast, the anti-inflammatory activity of the flavone-based derivatives is not particularly sensitive to modification of the A ring, especially in the presence of a 3 hydroxyl group (e.g. 040 vs. 04P). The anti-inflammatory activity of the flavone-based derivatives, however, is not very tolerant of modification of the B ring hydroxyls (e.g. 036, 072) and is also not tolerant of the substitution of the tertiary-amino groups regardless of the presence of a 3 hydroxyl group (e.g. 117, 119). However, the anti-inflammatory dampening effect of the tertiary amino groups can be reduced by the re-addition of a hydroxyl group to the 3' position (140). The PC12 differentiation promoting activity of the flavone-based derivatives shows a similar but less demanding set of structural requirements as the GSH maintaining activity for it is somewhat more tolerant of modifications to the A ring (e.g. 040 but not 027). In addition, while this activity is sensitive to modifications of the B ring hydroxyls, it tolerates limited modifications that eliminate the GSH maintaining activity (e.g. 065).

Once the flavone structure is opened up to give the chalcone (Table 2), only modification of the B ring hydroxyls affects the GSH maintaining activity of the fisetin derivatives. The one exception is 086 which has a methoxy and a benzyloxy group on the B ring. The PC12 differentiation promoting activity of the chalcone-based derivatives shows similar structural requirements as the GSH maintaining activity. Interestingly, while the anti-inflammatory activity of the α-naphtha chalcone-based derivatives is eliminated by modification of the B ring hydroxyls, the anti-inflammatory activity of the dimethyl chalcone based-derivatives is much more tolerant of this type of modification.

We have also identified a new quinoline scaffold that reserves the key structural elements of the flavone but results in enhanced neuroprotective activity (up to >400×) while maintaining the other key activities. Although these derivatives have greatly reduced free radical scavenging activity relative to fisetin based on TEAC values (Table 3), several are highly neuroprotective in our in vitro assay. In addition, while the most neuroprotective compounds with this scaffold do have hydroxyl groups, they are not polyphenols. Interestingly, within the context of this scaffold, the structural requirements for each key activity are somewhat sharper. For the maintenance of GSH, a catechol group on the B ring is essential. PC12 differentiation promoting activity requires both a catechol group on the B ring and a hydrophobic group on the 4-position of the C ring. The requirements for anti-inflammatory activity are somewhat less stringent but are sensitive to modifications of the B ring hydroxyls in a manner similar to the flavone-based derivatives.

We have also found that it is possible to separate neuroprotective activity from the three other key activities of fisetin. This result suggests that none of the three key activities play a role in neuroprotection in our in vitro ischemia assay. Both the differentiation-promoting and anti-inflammatory activities could have critical roles in maintaining CNS function in vivo but are less likely to be relevant in an in vitro assay with a single cell type. What is more surprising is that the ability to maintain GSH is not essential for neuroprotection in the in vitro ischemia assay as GSH loss is a component of this cell death paradigm⁶. However, the compounds with the lowest EC₅₀s for neuroprotection are all effective at maintaining GSH levels. Furthermore, many of the effective neuroprotective compounds that do not maintain GSH are also not good antioxidants as defined by the TEAC assay, an in vitro assay for antioxidant activity. Together, these results suggest that the neuroprotection by these compounds is mediated by some other, as yet undefined, action that is independent of GSH. This is currently under investigation.

In addition, we also found that the ability to maintain GSH levels did not correlate with the induction of Nrf2 by the compounds. There are a number of other mechanisms for maintaining GSH levels that could be modulated by these compounds including reduction of GSH utilization or inhibition of GSH export²⁷. These possibilities will be explored in future studies.

Many of our most effective fisetin derivatives also have improved medicinal chemical properties in terms of HBD, CLogP and tPSA, falling within the criteria for CNS drugs^26,33. Hydrogen bonding properties of drugs can significantly influence their CNS uptake profiles. Polar molecules are generally poor CNS agents and low lipophilicity (CLogP) and high hydrogen bonding decreases BBB penetration³³. Another important aspect of our fisetin derivatives is that they all lack A ring hydroxyl groups which are known to be subject to modification following oral administration²⁸. Thus, they are less likely to be metabolized in this way, leading to enhanced bioavailability and brain penetration.

Recent studies in our laboratory have shown that fisetin is effective in multiple animal models of neurological disorders including stroke⁶ and Huntington’s disease²⁹. Furthermore, fisetin can reduce both the kidney and CNS complications of diabetes in the Akita model of type 1 diabetes³⁰. Thus, it shows promise for the treatment of multiple diseases for which there are currently no good treatments. The identification and characterization of more efficacious derivatives is the first step in moving this lead compound towards the clinic.

In summary, starting with the multi-target polyphenol fisetin, we have generated a number of derivatives with greatly enhanced neuroprotective activity (e.g. 011, 50 nM; 121, 7 nM and 140, 5 nM) in a cell culture-based model of ischemia. Many of the more potent fisetin derivatives also have good CNS medicinal chemical properties. In addition, some of these derivatives maintain the other three key activities of fisetin including anti-inflammatory, neurotrophic and GSH-maintaining activities making them good candidates for further testing in animal models of stroke as well as other neurological diseases. In creating these derivatives, we have shown that it is possible to enhance a primary activity of a polyphenol such as fisetin while at the same time maintaining other key activities which are not necessarily directly related to this primary activity. Thus, we are able to maintain the multi-target qualities while improving both the physiochemical and pharmacological properties of the compound.

EXPERIMENTAL SECTION

Biology: Cell culture

Fetal calf serum (FCS) and dialyzed FCS (DFCS) were from Hyclone (Logan, UT). Dulbecco’s Modified Eagle’s Medium (DMEM) was purchased from Invitrogen (Carlsbad, CA). HT22 cells⁶ were grown in DMEM supplemented with 10% FCS and antibiotics. PC12 cells were grown in DMEM supplemented with 10% FCS, 5% horse serum and antibiotics. N9 microglial cells were grown in DMEM supplemented with 10% FCS, 1× non-essential amino acids, 1× essential amino acids and antibiotics.

Cytotoxicity assay

Cell viability was determined by a modified version of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay based on the standard procedure⁶. Cells were seeded onto 96-well microtiter plates at a density of 5 × 10³ cells per well. For the in vitro ischemia assay, the next day, the medium was replaced with DMEM supplemented with 7.5% DFCS and the cells were treated with 20 µM iodoacetic acid (IAA) alone or in the presence of the different derivatives. After 2 hr the medium in each well was aspirated and replaced with fresh medium without IAA but containing the derivatives. 20 hr later, the medium in each well was aspirated and replaced with fresh medium containing 2.5 µg/ml MTT. After 4 hr of incubation at 37 °C, the cells were solubilized with 100 µl of a solution containing 50% dimethylformamide and 20% SDS (pH 4.7). The absorbance at 570 nm was measured on the following day with a microplate reader (Molecular Devices). Results were confirmed by visual inspection of the wells. Controls included compound alone to test for toxicity and compound with no cells to test for interference with the assay chemistry. All of the derivatives were tested twice in this assay and those that showed a strongly positive response (EC₅₀ <1 µM) were tested a third time for confirmation.

Differentiation assay

PC12 cells in N₂ medium were treated with the derivatives (1–10 µM) or fisetin (10 µM) as a positive control for 24 hr at which time the cells were scored for the presence of neurites. PC12 cells produce neurites much more rapidly when treated in N₂ medium than when treated in regular growth medium. For each treatment, 100 cells in each of three separate fields were counted. Cells were scored positive if one or more neurites >1 cell body diameter in length were observed. All of the derivatives were tested twice in this assay and those that showed a positive response were tested a third time for confirmation.

Anti-inflammatory assay

Mouse N9 microglial cells plated in DME with 7.5% DFCS were treated with 10 µg/ml bacterial lipopolysaccharide (Sigma) alone or in the presence of the fisetin derivatives (1–10 µM) or fisetin (10 µM) as a positive control. After 24 hr the medium was removed, spun briefly to remove floating cells and 100 µl assayed for nitrite using 100 µl of the Griess Reagent (Sigma) in a 96 well plate. After incubation for 10 min at room temperature the absorbance at 550 nm was read on a microplate reader. All of the derivatives were tested twice in this assay and those that showed a positive response were tested a third time for confirmation.

Total glutathione

Total intracellular glutathione was determined by a chemical assay as described³¹. All of the derivatives were tested twice in this assay and those that showed a positive response were tested a third time for confirmation.

SDS-PAGE and immunoblotting

For immunoblotting of Nrf2, nuclear extracts were prepared as described³² from untreated cells and cells treated with the fisetin derivatives for 1, 2 and 4 hr. Fisetin was used as a positive control. For each derivative, the concentration which was most effective at preventing cell death was used. Protein concentrations were determined using the BCA protein assay (Pierce). Equal amounts of protein were solubilized in 2.5X SDS-sample buffer, separated on 10% SDS-polyacrylamide gels and transferred to nitrocellulose. Equal loading and transfer of the samples was confirmed by staining the nitrocellulose with Ponceau-S. Transfers were blocked for 1 hr at room temperature with 5% nonfat milk in TBS/0.1% Tween 20 and then incubated overnight at 4 °C in the primary antibody diluted in 5% BSA in TBS/0.05% Tween 20. The primary antibodies used were: anti-Nrf2 (#SC13032; 1/1000) from Santa Cruz Biotechnology (Santa Cruz, CA) and anti-β-actin (#5125; 1/20,000) from Cell Signaling (Beverly, MA). The transfers were rinsed with TBS/0.05% Tween 20 and incubated for 1 hr at room temperature in horseradish peroxidase-goat anti-rabbit or goat anti-mouse (Biorad, Hercules, CA) diluted 1/5000 in 5% nonfat milk in TBS/0.1% Tween 20. The immunoblots were developed with the Super Signal reagent (Pierce, Rockford, IL). All of the derivatives were tested twice in this assay and those that showed a positive response were tested a third time for confirmation.

Determination of the Trolox Equivalent Activity Concentration (TEAC)

TEAC values for the flavonoids were determined as described³¹. Briefly, 250 µl of 2,2'-azinobis(3-ethylbenzothiazoline 6-sulfonate) (ABTS) treated overnight with potassium persulfate and diluted to an OD of ~0.7 at 734 nm was added to 2.5 µl of a derivative solution in ethanol. The change in absorbance due to the reduction of the ABTS radical cation was measured at 734 nm for 4 min. To calculate the TEAC, the gradient of the plot of the percentage inhibition of absorbance vs. concentration for the derivative in question was divided by the gradient of the plot for Trolox.

Chemistry: General Methods

All reagents and anhydrous solvents were obtained from commercial sources and used as received. ¹H NMR and ¹³C NMR were recorded at 500 and 125 MHz, respectively, on a Varian VNMRS-500 spectrometer using the indicated solvents. Chemical shift (δ) is given in parts per million (ppm) relative to tetramethylsilane (TMS) as an internal standard. Coupling constants (J) are expressed in hertz (Hz), and conventional abbreviations used for signal shape are: s = singlet; d = doublet; t = triplet; m = multiplet; dd, doublet of doublets; brs = broad singlet. Mass spectrometry (LCMS) was carried out using a Shimadzu LC-20AD spectrometer and electro spray ionization (ESI) mass analysis with a Thermo Scientific LTQ Orbitrap-XL spectrometer. Melting points were determined with a Thomas Hoover capillary melting point apparatus and are uncorrected. All final compounds were characterized by LCMS and ¹H NMR and gave satisfactory results in agreement with the proposed structure. All of the tested compounds have a purity of at least 95% which was determined by analysis on a C18 reverse phase HPLC column [PHENOMENEX–LUNA (50 × 4.60 mm, 3µ)] using 10–90% CH₃CN/H₂O containing 0.02% AcOH with a flow rate 1 mL/min (5 min gradient) and monitoring by a UV detector operating at 254 nm. Mass spectra were acquired in the positive mode scanning over the mass range of 50–1000. LC/MS M+H signals were consistent with the expected molecular weights for all of the reported compounds. Thin layer chromatography (TLC) used EMD silica gel F-254 plates (thickness 0.25 mm). Flash chromatography used EMD silica gel 60, 230–400 mesh.

General Procedure A for the Synthesis of Chalcone Derivatives 013, 032, 033, 057, 063, 085, 086, 105–108 137 and 138

A mixture of 2'-hydroxy acetophenone (1eq), aryl aldehyde (1eq) and Ba(OH)₂ (1eq) in MeOH (3 mL/mmol) was stirred for 12 h at 40 °C. Methanol was evaporated and the residue was diluted with water, neutralized with 1N HCl and extracted with ethyl acetate, the organic layer was washed with brine solution, dried (Na₂SO₄) and evaporated. Solid residues were recrystallized from CH₂Cl₂/Hexane, liquid residues were purified by flash chromatography using silica gel (230–400 mesh) with 10–30% EtOAc/Hexane gave chalcones with 30–90% yield.

General Procedure B (methyl/ethyl deprotection) for the Synthesis of Compounds 002, 011, 027, 028, 034, 041, 087, 093, 094 140 and 142

To a stirred and cooled 0 °C solution of suitably protected starting material (1eq) in CH₂Cl₂ (5 mL/mmol) was added BBr₃ (2 eq/alkoxy group) and the mixture was stirred for overnight at room temperature under nitrogen atmosphere. The reaction mixture was quenched by adding 5% Na₂HPO₄ solution, extracted with CH₂Cl₂, combined organic extracts were washed with brine, dried (Na₂SO₄) and evaporated. The resulting solids were recrystallized from methanol.

Method C for the Synthesis of Chalcone 088

To a stirred solution of chalcone 086A (74.7 mg, 0.203 mmol) in MeOH (2 ml) was added p-toluenesulfonic acid (77.3 mg, 0.407 mmol). The reaction mixture was stirred for 3 h at room temperature, after completing the reaction solvent was evaporated, the residue was diluted with water (20 mL), then neutralized with saturated NaHCO₃, and extracted with EtOAc. Combined extracts were washed with brine, dried (Na₂SO₄), and evaporated. The residue was purified by flash chromatography using silica gel (230–400 mesh) with 20% EtOAc/Hexane gave 088 94% yield, as a yellow solid.

General Procedure D (debenzylation) for the Synthesis of Compounds 64, 069, 070, 072 and 092

The benzyl protected flavones and flavonols were dissolved in 1:1 EtOAc/Ethanol (10 mL/mmol) then treated with 5% palladium on charcoal (5% w/w) and the mixture was stirred under hydrogen atmosphere (balloon pressure) for overnight. The reaction mixture was filtered and the solvent was evaporated, the resulting solids were recrystallized from dichloromethane/methanol.

General Procedure E for the Synthesis of Flavone Derivatives 018, 038, 058, 068, 089, 115, 116, 119 and 120

A solution of chalcone (1eq) and Iodine (0.01eq) in DMSO (1 mL/mmol) was heated at 130 °C for 3–6 h. Reaction mixture was cooled and diluted with water, extracted with CH₂Cl₂, washed with aqueous saturated Na₂S₂O₃, dried (Na₂SO₄) and evaporated. Solid residues were recrystallized from CH₂Cl₂/Hexane liquid residues were purified by flash chromatography using silica gel (230–400 mesh) with 30–80% EtOAc/Hexane gave flavones with 50–95% yield.

General Procedure F for the Synthesis of Flavonol Derivatives 025, 036, 037, 059, 065, 090, 091, 114, 117, 118, 122 139 and 141

To a stirred and cooled 0 °C solution of chalcone in MeOH (5 mL/mmol) was added 5.4% NaOH (3.2 mL/mmol) followed by 30% H₂O₂ (0.37 mL/mmol) drop wise, and the mixture was stirred for 3h at 0 °C, then the ice bath was left in place but not recharged, and stirring was continued overnight. The reaction mixture was acidified with 2M HCl, and the resulting precipitate was collected by filtration and washed with water and recrystallized from dichloromethane gave flavonols with 40–90% yield.

General Procedure G for the Synthesis of Quinoline Derivatives 001, 004, 007, 017, 021–024, 083, 084, 109–113 and 121

To a stirred solution of 2’-amino acetophenone (1eq) and aromatic aldehyde (1 to 3eq) in alcohol (3 mL/mmol) was added H₂SO₄ (0.75eq) and the mixture was refluxed for 12–24 h. Reaction mixture was cooled, solvent was evaporated and the residue was diluted with water, neutralized with 5% NaHCO₃ solution and extracted with ethyl acetate, the organic layer was washed with brine, dried (Na₂SO₄) and evaporated. Flash chromatography of the residue over silica gel using 10–50% EtOAc/Hexane gave 4-alkoxy 2-aryl quinolines with 15–50% yield.

Analytical data for selective compounds (011, 121, and 140): (E)-3-(3,4-dihydroxyphenyl)-1-(2-hydroxy-4,5-dimethylphenyl)prop-2-en-1-one (011)

Following general procedure B, 011 was obtained from chalcone 013 as an orange solid (95% yield); mp 174–177 °C; LCMS purity 99%; ¹H NMR (DMSO-d₆, 500 MHz) δ ppm 2.23 (s, 3H), 2.24 (s, 3H), 6.78 (s, 1H) 6.82 (d, J = 8.0 Hz, 1H), 7.23 (dd, J = 8.5, 2.0 Hz, 1H), 7.31 (d, J = 2.0 Hz, 1H), 7.72 (q, J = 15.5 Hz, 2H), 8.02 (s, 1H), 9.11 (br s, OH), 9.81 (br s, OH), 12.79 (s, OH); ¹³C NMR (DMSO-d₆, 125 MHz) δ ppm 18.72, 20.46, 116.18, 116.36, 117.94, 118.45, 118.69, 123.19, 126.65, 127.59, 130.91, 146.04, 146.12, 146.92, 149.59, 161.23, 193.36; LCMS: m/z 285 ([M + H]⁺), MS (ESI): m/z calcd for C₁₇H₁₆NO₄ ([M + H]⁺) 285.1082; found 285.1001 ([M + H]⁺).

4-(4-(cyclopentyloxy) quinolin-2-yl) benzene-1, 2-diol (121)

Following general procedure G, 121 was obtained as a dark yellow solid (16% yield); mp 199–201 °C; LCMS purity 98%; ¹H NMR (DMSO-d₆, 500 MHz) δ ppm 1.66 (m, 2H), 1.79 (m, 2H), 1.89 (m, 2H), 2.07 (m, 2H), 5.30 (m, 2H), 6.85 (d, J = 8.5 Hz, 1H), 7.31 (s, 1H), 7.44 (t, J = 8.0 Hz, 1H), 7.54 (dd, J = 8.5, 2.0 Hz, 1H), 7.67 (t, J = 8.0 Hz, 1H), 7.75 (d, J = 2.0 Hz, 1H), 7.88 (d, J = 8.0 Hz, 1H), 8.04 (d, J = 7.5 Hz, 1H), 9.21 (brs, OH); ¹³C NMR (DMSO-d₆, 125 MHz) δ ppm 24.19, 32.71, 80.12, 99.28, 115.02, 115.99, 119.41, 120.60, 121.97, 125.21, 128.93, 130.30, 131.11, 145.88, 147.71, 149.11, 157.92, 160.73; LCMS: m/z 322 ([M + H]⁺), MS (ESI): m/z calcd for C₂₀H₁₉NO₃ ([M + H]⁺) 322.1437; found 322.1412 ([M + H]⁺).

3-hydroxy-2-(3-hydroxy-4-(pyrrolidin-1-yl)phenyl)-4H-benzo[h]chromen-4-one (140)

Following general procedure B, 140 was obtained from compound 139 as an orange red solid (50% yield); mp 223–225 °C; LCMS purity 98%; ¹H NMR (DMSO-d₆, 500 MHz) δ ppm 1.88 (s, 4H), 3.45 (s, 4H), 6.74 (d, J = 8.5 Hz, 1H), 7.85 (m, 5H), 8.04 (d, J = 8.5 Hz, 1H), 8.11 (d, J = 8.5 Hz, 1H), 8.68 (d, J = 7.5 Hz, 1H), 9.37 (s, 1H); ¹³C NMR (DMSO-d₆, 125 MHz) δ ppm 25.16, 50.19, 114.38, 114.69, 117.94, 120.61, 120.74, 120.99, 122.54, 124.10, 124.87, 128.06, 128.87, 129.77, 135.37, 139.19, 140.29, 146.34, 146.44, 151.65, 172.19; LCMS: m/z 374 ([M + H]⁺), MS (ESI): m/z calcd for C₂₃H₁₉NO₄ ([M + H]⁺) 374.1386; found 374.1402 ([M + H]⁺).

Abbreviations

ABTS

2,2'-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid)

BCA

bicinchoninic acid assay

BSA

bovine serum albumin

CNS

central nervous system

DFCS

dialyzed fetal calf serum

DME

dulbecco’s modified eagle’s

DMEM

dulbecco’s modified eagle’s medium

DMSO

dimethyl sulfoxide

EC₅₀

half maximal effective concentration

EtOAc

ethyl acetate

ESI

electrospray ionization

FCS

fetal calf serum

FDA

food and drug administration

G3PDH

glycolytic enzyme glyceraldehyde 3-phosphate dehydrogenase

GSH

glutathione

HBA

hydrogen bond acceptor

HBD

hydrogen bond donor

IAA

iodoacetic acid

LPS

lipopolysaccharide

LCMS

liquid chromatography mass spectrometry

MS

mass spectrometry

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide

NMR

nuclear magnetic resonance

Nrf2

NF-E2 related factor 2

p-TSA

p-toluenesulfonic acid

rtPA

recombinant tissue-type plasminogen activator

ROS

reactive oxygen species

SAR

structure activity relationship

SDS

sodium dodecyl sulfate

SDS-PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis

TBS

tris buffered saline

TMS

tetramethylsilan

THP

tetrahydropyran

tPSA

topological polar surface area

TEAC

trolox equivalent activity concentration