AANAT Kinetics of CoASH-Targeted Electrophiles of Tryptamine and Related Analogs

Abstract

Arylalkylamine N-acetyltransferase (AANAT) catalyzes the rate-limiting step in melatonin synthesis and is a potential target for disorders involving melatonin overproduction, such as seasonal affective disorder. Previously described AANAT inhibitor bromoacetyltryptamine (BAT) and benzothiophenes analogs were reported to react with CoASH to form potent bisubstrate inhibitors through AANAT’s alkyltransferase function, which is secondary to its role as an acetyltransferase. We replaced the bromoacetyl group in BAT with various Michael acceptors to mitigate possible off-target activity of its bromoacetyl group. Additionally, we modified the length of the carbon linker between the Michael acceptor and indole bicycle of tryptamine to determine its effect on potency. An AANAT enzymatic assay showed a two-carbon linker present in BAT was optimal, while none of the new warheads had activity. Kinetic analysis indicated that these Michael acceptors reacted with CoASH much slower than BAT and not within the timeframe of our enzymatic assay. Additionally, we confirmed earlier reports that the acetyltransferase function of AANAT follows an ordered bi bi mechanism in which AcCoA binds before serotonin. In contrast, BAT’s alkyltransferase kinetics revealed a bi uni mechanism in which BAT binds to AANAT before CoASH. Our model combines both functions of AANAT into one kinetic mechanism.

Graphical Abstract

Considerable research has focused on using melatonin as a sleep aid. However, the effects of melatonin overproduction have been much less studied. Abnormally high levels of melatonin may cause disruptions in circadian rhythm (CR) and potentially contribute to mental health disorders, including seasonal affective disorder (SAD).¹ The enzyme arylalkylamine-N-acetyltransferase (AANAT)^{2, 3} catalyzes the rate-limiting step in the synthesis of melatonin and has been implicated as a potential target for treating sleep disorders. AANAT has a strong association with CR, but its role in mental health disorders such as SAD is not well understood. Better inhibitors are needed to study AANAT’s function, as the most potent inhibitors have poor cell permeability (Figure 1, compound 2)⁴ or lack selectivity against melatonin receptors (Figure 1, compounds 1 and 6).⁵

Figure 1.

Previously published AANAT inhibitors relevant to the current work: 1,⁶ 2,⁴ 3a,⁶ 3b,⁶ 4,⁸ and 5–7.⁵ IC₅₀ values reported for sheep AANAT enzyme, unless indicated otherwise. AANAT was shown to catalyze the reaction between 1 and CoASH to form 2 in situ.⁶

Despite poor selectivity and possible off-target reactivity of its bromoacetyl group we chose previously published “pro-inhibitor” bromoacetyltryptamine (BAT) 1⁶ along with analogous benzothiophenes 5 and 6⁵ as starting points for optimization due to their unique mechanism of action. Specifically, these compounds coupled with CoASH in situ via AANAT’s novel, but understudied alkyltransferase mechanism (Figure 2 shows one possible mechanism) to form highly potent bisubstrate inhibitors (e.g. 2) that span both the serotonin and CoA binding sites.⁷ Though it was previously shown that alkyltransferase activity is distinct from its acetyltransferase activity and likely occurs at two different sites,⁶ here we account for both activities in one collective mechanism to help establish the kinetic feasibility of bisubstrate synthesis and inhibition during a sub-minute time frame. In addition to developing a kinetic mechanism that addresses this question, we sought alternative pro-inhibitors to determine whether the alkyltransferase substrate activity could be maintained with potentially less reactive electrophiles. To encompass a range of reactivities and different trajectories for the putative addition of CoASH, we selected the following electrophiles: acrylamides, sulfonamides, propiolamides and allenamides for comparison with previously described bromo- and chloro-acetyl moieties.

Figure 2.

One possible mechanism for AANAT’s alkyltransferase activity to form potent bisubstrate inhibitor 2. The order of substrate addition to the two sites has been determined by kinetic simulation in the current work.

Another project goal was to determine whether the chain separating the warhead from the indole bicycle could be optimized for enzymatic potency. It was previously shown that acrylamide 7⁵ was considerably less potent than bromoacetyl analog 5 (Figure 1). However, because of the extra carbon present in the acryloyl group of 7 in relation to BAT’s bromoacetyl group, we speculated that 7 might have lost potency due to an ineffective trajectory for reacting with CoASH in the alkyltransferase site. This idea was further supported by the fact that having an extra carbon had little effect on potency of the extended bisubstrate inhibitor 3b⁶ (Figure 1), so poor reactivity with CoASH is the more likely reason for the diminished activity of 7. To address the importance of the linker for pro-inhibitor potency, we tested both one- and three-carbon homologs of BAT 1, as well as one- and two-carbon chain linkers for all the electrophiles of the current study.

AANAT inhibitors 1 and 2 were synthesized as previously reported.⁴ In an analogous manner to compound 1’s preparation, chloroacetyl-containing compound 4⁹ was synthesized by acylation of tryptamine with chloroacetyl chloride (Scheme 1). Benzothiophene analogs 5 and 6 were prepared by a multistep synthesis as in the literature^{5, 10} with procedural modifications described in supporting information. All compounds were purified by flash chromatography and/or recrystallized as needed to obtain >95% purity by HPLC.

Scheme 1.

Synthesis of bromoacetyl (Y = -CH₂Br) and chloroacetyl (Y = -CH₂Cl) aminoalkylindoles 1, 4, 8, 9.

Homologs of 1 with one less or one more carbon atom in the chain separating indole from the warhead were prepared by acylation of commercially available (1H-indol-3-yl)methanamine and 3-(1H-indol-3-yl)propan-1-amine to give 8 and 9, respectively (Scheme 1).

Because acryloyl chloride is expensive and has poor stability, we prepared acrylamide analogs 10 and 11 by a different route than compounds of Scheme 1. Instead, we used acrylic acid and coupling reagent 1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU, Scheme 2) to obtain 10 and 11 in moderate yields.

Scheme 2.

Synthesis of acryloyl aminoalkylindoles 10 and 11.

Vinyl sulfonamides 12 and 13 were prepared via sulfonylation of aminoalkylindoles with 2-chloroethane-1-sulfonyl chloride followed by in situ elimination (Scheme 3) according to the procedure provided by Ji.¹¹

Scheme 3.

Synthesis of vinyl sulfonamides 12 and 13.

Due to the commercial inaccessibility of propioloyl chloride, which can be challenging to work with¹² and is more conveniently prepared in situ,¹³ we coupled 3-(trimethylsilyl)propiolic acid with aminoalkylindoles using DCC and N-hydroxysuccinimide.¹⁴ The TMS group was subsequently removed in situ using TBAF to provide propiolamides 14 and 15, albeit in poor yields (Scheme 4).

Scheme 4.

Synthesis of propiolamides 14 and 15.

Allenamides 16 and 17 were prepared using a method described by Abbas,¹⁵ in which 3-butynoic acid was activated with 2-chloro-1-methylpyridinium iodide (Mukaiyama reagent) and then coupled with aminoalkylindoles. A base-induced isomerization, which likely involved deprotonation of the more acidic alpha proton of the 2-pyridinyl ester intermediate (not shown), resulted in the allenes 16 and 17 (Scheme 5).

Scheme 5.

Synthesis of allenamides 16 and 17.

Enzymatic potency of final products was assessed (Figure 1 and Tables 1) with an assay we previously described¹⁶ using purified recombinant N-terminally GST-tagged form of the protein from Ovis aries (sheep).¹⁷ We chose the sheep isoform, because it is easier to prepare than the human version and also allows for direct comparisons with published inhibitors (Figure 1) and crystal structure of bound inhibitor 2 (PDB: 1IB1).⁷

Table 1.

Effects of various electrophile-substituted tryptamines and homologues on AANAT enzymatic activity.

Compd	n	Z	AANAT IC50 [μM][a]
8	1	-COCH2Br	23 ± 22
1	2	-COCH2Br	3.8 ± 2.2
4	2	-COCH2O	18 ± 3
9	3	-COCH2Br	22 ± 5
10, 11	1, 2	-COCH=CH2	>20
12, 13	1, 2	-SO2CH=CH2	>20
14, 15	1, 2	-COC=CH	>20
16, 17	1, 2	-COCH=C=CH2	>20

^[a]Values for compounds 1, 4, 8 and 9 were calculated from two independent experiments. For compounds designated as “>20 μM”, no inhibition was observed at 10 μM test compound concentration, so IC₅₀ was estimated to be at least 2-fold greater.

In our hands, previously published bromo- and chloro-acetylated tryptamines (Figure 1: 1 and 4, respectively) and benzothiophenes (Figure 1: 5 and 6) were 3–16X less potent than reported, with the biggest discrepancy for hydroxy-substituted benzothiophene 6 (2.9 μM vs. 0.18 μM). It should be noted that compounds 5 and 6 were previously tested using human enzyme. However, compound 1 was shown to be more potent in sheep enzyme than human,¹⁸ so one would have expected benzothiophenes 5 and 6 to be even more potent in our assay than what had been published using human enzyme. Likely a more important difference between the two assays was that a thirty-minute incubation period was previously applied for 5 and 6, giving more time to form bisubstrate inhibitor. For compounds 1 and 4, a 2–4 minute incubation time had been applied previously.⁸ In our assay, we used a stopped-flow instrument which simultaneously added all reagents together and began UV absorbance measurements on the resulting reaction immediately, giving less time to form bisubstrate inhibitor. Despite these differences, there was some consistency with earlier results. For example, chloroacetyltryptamine 4 was less potent than 1 (18 μM vs. 3.8 μM) as reported, likely because it forms the bisubstrate inhibitor more slowly as chloride ion is a poorer leaving group than bromide ion.⁸ Also, good agreement was obtained between our IC₅₀ value for bisubstrate inhibitor 2 and the reported one (0.14 μM vs. 0.15 μM).

Homologs of 1 containing either one or three carbons between the indole bicycle and warhead, 8 and 9, respectively, were 6X less potent than 1. It should be noted that Khalil⁶ previously showed that adding more than two carbon atoms between the amide carbonyl and the CoA sulfur for bisubstrate inhibitors led to substantially greater losses in potency, so for that reason we did not prepare homologs with longer chains. Given that adding an extra carbon atom in going from bisubstrate inhibitor 2 to 3b had only a modest effect on potency (reported K_i of 0.048 μM vs. 0.067 μM),⁶ the more substantial drop in potency for pro-inhibitor 9 was likely due to poorer reactivity with CoASH, consistent with the weaker electrophilicity of an acrylamide versus and alpha-bromoacetamide group. On the other hand, truncated bisubstrate inhibitor 3a (Figure 1) was 4X less potent than 2, which is comparable to the 6X decrease we observe in going from pro-inhibitor 1 to 8. Taken together, these findings indicate that the distance between the indole bicycle and warhead in 1 is optimal for this mechanism of inhibition.

Unfortunately, none of the new warheads (Table 1: 10–17) were active at 10 μM, so IC₅₀ values were not determined. Though 10–17 have not previously been tested against AANAT, benzothiophene acrylamide 7 (Figure 1) was reported to be 40X less potent than its bromoacetyl analog 5 (IC₅₀ of 25 μM vs. 0.61 μM). Assuming there was a similar 40X shift in potency going from 1’s activity in our assay to acrylamide 11, it is not surprising that the latter was inactive at 10 μM. We had originally hypothesized that a shorter chain length (n = 1) such as in 10 would more closely mimic the distance between the warhead and indole bicycle as in 1, resulting in improved alkyltransferase activity. However, shortening the chain (e.g. 10 vs. 11) offered no apparent benefit in potency for the various warheads tested. Also, altering of the putative trajectory of nucleophilic attack by CoASH on the warhead by changing from alkene to alkyne (14 and 15) or allene (16 and 17) was not helpful. Likewise, more reactive vinyl sulfonamides 12 and 13, were inactive. To better understand why these new warheads were inactive in our enzyme assay, we studied the kinetics of their reaction with CoASH (details below).

As with our acetyltransferase enzymatic assay, we used the sheep isoform of AANAT to evaluate the alkyltransferase activity of selected inhibitors (Table 2). We monitored the formation of product between the warheads and CoASH by LC-MS, and we performed the reactions at pH 7.5 to match the pH used in our enzymatic assay. Surprisingly the rate of reaction for CoASH with pro-inhibitor 1 was the same in the presence and absence of AANAT at this pH. This result contrasts with the previously reported enzymatically-accelerated rate observed at pH 6.8.⁶ To compare, we also reacted 1 with CoASH in the absence of AANAT at pH 6.8 and obtained a second order rate constant of 1.31 M⁻¹ min⁻¹. Data were also collected with AANAT at pH 6.8 with varying concentrations of 1 and CoASH. These data were globally fitted to a simulated ordered substrate binding mechanism (Figure 3, alkyltransferase activity), where 1, or BAT, binds to AANAT before CoASH. Rate constants were used to calculate steady-state enzyme kinetic constants such as k_cat, K_m, and k_cat/K_m for substrates 1 and CoASH (Table S1). The k_cat/K_m for AANAT alkyltransferase activity (Table S1) is several orders of magnitude greater than the uncatalyzed second order rate constant (Table S2) and in good agreement with earlier reports at pH 6.8.⁶ Similar to 1, we also observed no effect of AANAT on rates of reaction with CoASH for electrophiles 10–13 at pH 7.5. Notably, their rates were 17–493X slower than 1. Therefore, it is unlikely that they would form sufficient bisubstrate inhibitors to show activity within the timeframe of our enzymatic assay.

Table 2.

Reaction rates of electrophile-substituted tryptamines with CoASH.

Electrophile Compd #	n	Z	Z’	kforward (M−1 min−1)[a]	kforward C.I.[b]
1	2	-COCH2Br	-COCH2-	4.30	(4.30, 4.82)
10	1	-COCH=CH2	-COCH2CH2-	0.0156	(0.0064, 0.024)
11	2	-COCH=CH2	-COCH2CH2-	0.00873	(0.00447, 0.0136)
12	1	-SO2CH=CH2	-SO2CH2CH2-	0.261	(0.209, 0.310)
13	2	-SO2CH=CH2	-SO2CH2CH2-	0.110	(0.103, 0.120)

^[a]Reactions were performed at 25 °C in pH 7.5 phosphate buffer with 2 mM each of CoASH and electrophile-substituted tryptamines (1, 10–13) and 1 mM tris(2-carboxyethyl)phosphine (TCEP). Reactions included 5% MeOH to allow for electrophile dissolution. Determined rate constants were the same for reactions with and without 1.2 μM AANAT. Reaction rates were obtained by monitoring formation of CoASH adducts by LC-MS. Rate constants were determined by fitting product accumulation using KinTek global kinetic explorer software.

^[b]Upper and lower bounds were determined within a 0.83 relative error threshold as previously described.¹⁹

Figure 3.

Global enzyme kinetic mechanism of AANAT acetyltransferase and alkyltransferase activity. AANAT enzyme (E) is shown in different binding states with various substrates and products such as acetyl-CoA (AcCoA), coenzyme A (CoASH), serotonin (sero), N-acetylserotonin (N-AcSero), bisubstrate inhibitor 2, and BAT 1.

In addition to evaluating various electrophiles as pro-inhibitors, we also wanted to fit AANAT acetyltransferase and alkyltransferase data globally to a single kinetic mechanism to demonstrate that the general bisubstrate inhibition mechanism described by Zheng⁸ could collectively and quantitatively fit the data. First, we collected AANAT enzyme kinetic data by varying serotonin and AcCoA to obtain AANAT acetyltransferase progress curves (Figures S1A and B). Second, to incorporate bisubstrate 2 inhibition effects on acetyltransferase activity, we collected progress data for the acetyltransferase reaction with various amounts of compound 1 and 2 (Figures S1C and D). Lastly, we incorporated progress curves following the generation of compound 2 using LC-MS based enzymatic and non-enzymatic assays (Figure S2). To model these data, we began by using the ordered bi bi mechanism described previously²⁰ to simulate acetyltransferase activity (Figures S1 and 3), which confirmed the reported bi-bi mechanism in which AcCoA binds to AANAT before serotonin (Figure 4).

Figure 4.

Proposed binding order of substrates based on kinetic modeling. The acetyltransferase and alkyltransferase sites may or may not be the same site in AANAT.

To account for the generation of compound 2, or bisubstrate inhibitor in the mechanism, we added an ordered bi uni mechanism for the alkyltransferase reaction (Figure 3), described above, to arrive at a kinetic model that joins acetyltransferase and alkyltransferase activities of AANAT into one kinetic mechanism. We found that an alkyltransferase mechanism that allows BAT 1 to bind before CoASH (Figure 4) significantly improved the data fitting by reducing the error between the model and data by about 2-fold (Figures S1 and S2). Although fits to the data are generally very good to excellent, some significant deviation such as in Figure S2B between the model and data is observed. This can be rationalized based on the challenges of global fitting of progress curve data. Despite the apparent deviation, however, the relatively wide confidence intervals in rate constants reported here (Table S2) along with experimental variance may explain the discrepancies.

We have applied a two-pronged approach to study AANAT’s dual functions as an acetyltransferase and alkyltransferase. For the first piece of our work, we synthesized and enzymatically tested both existing and new electrophile-substituted tryptamines and related benzothiophenes. For the second piece, we developed a kinetic mechanism to link the two functions. Though we obtained similar potency as previously reported for bisubstrate inhibitor 2, we noted a significant difference in potency between bromo- and chloro-acetyl containing pro-inhibitors in our enzymatic assay relative to earlier results. We ascribed this difference to a shorter incubation time with CoASH in our assay and concomitant decreased formation of bisubstrate inhibitor.

A new SAR finding was that altering the length of the two-carbon linker between BAT’s bromoacetyl group and indole bicycle (i.e. 8 and 9, respectively) resulted in a significant drop in potency. Taking into account activities of previously reported bisubstrate inhibitors of different chain lengths, the drop in potency for the longer chain analog (9) is likely due to a negative effect on reaction rate with CoASH, whereas the shorter chain analog (8) might be losing activity due to forming a less potent bisubstrate inhibitor.

None of the new warheads worked in our enzymatic assay, which we conclude is due to slow nonenzymatic or AANAT-catalyzed reactions with CoASH. Notably, AANAT had no effect on their reactions, suggesting they are not substrates of its alkyltransferase function. However, one should keep in mind that at pH 7.5, AANAT had no effect on reaction rate between BAT 1 and CoASH, even though it did accelerate rate at pH 6.8. This can likely be explained in part by the relatively high concentrations of BAT and CoASH that we used to facilitate the analysis, which would not be pharmacologically relevant in cellular experiments. This difference in rate between the two pH levels can be understood based on the pK_a of the CoASH thiol (ca. 9) and the increased thiolate ion at higher pH. We also checked the reaction rate of acrylamide 10 with CoASH at pH 6.8 in the presence of enzyme and found it to be several times slower than at pH 7.5. Thus, we would not expect these new warheads to have enzyme activity at either pH, because they are forming very little bisubstrate inhibitor within the timeframe of our assay. We cannot be sure of their activity in a cellular environment, but one would not expect any rate enhancement for their reactions with CoASH due to AANAT. Overall, we would expect these Michael acceptor warheads to be inferior to the bromoacetyl group in vivo.

For the second piece of our work, involving a kinetic mechanism, we confirmed that the acetyltransferase function occurs through a bi-bi mechanism in which AcCoA binds to AANAT before serotonin. In contrast to an earlier report that the alkyltransferase mechanism involves random order of CoASH or BAT binding,⁸ our model predicts that BAT 1 binds to AANAT before CoASH via a bi uni mechanism. The consequences of this binding order and how it relates to AANAT’s physiological role remain to be determined. Taken together with our SAR, we feel that AANAT’s alkyltransferase function is relatively limited in scope of the warhead, which could make it less useful as a drug delivery approach for AANAT inhibitors.

New findings of the current work are summarized as follows: 1) AANAT pro-inhibitor BAT 1 was more potent than Michael acceptor analogs in an enzymatic assay, likely due to faster reaction with CoASH to form the bisubstrate inhibitor 2; 2) a two-carbon linker connecting the bromoacetyl group and indole bicycle of BAT was shown to be optimal for enzymatic potency; 3) a kinetic simulation of AANAT’s alkyltransferase and acetyltransferase functions revealed two distinct mechanisms for substrate binding.

Data availability

The authors declare that the data supporting the findings of this study is available within the paper and its Supplementary Information files, and is also available from the corresponding author. ¹H and ¹³C NMR, experimental procedures, and characterization for all newly synthesized final compounds are shown in the Supporting Information files.