Design and Interpretation of Human Sulfotransferase 1A1 Assays

Abstract

The human sulfotransferases (SULTs) regulate the activities of hundreds, if not thousands, of small molecule metabolites via transfer of the sulfuryl-moiety (-SO₃) from the nucleotide donor, 3′-phosphoadenosine 5′-phosphosulfate (PAPS) to the hydroxyls and amines of the recipients. Our understanding of the molecular basis of SULT catalysis has expanded considerably in recent years. The basic kinetic mechanism of these enzymes, previously thought to be ordered, has been redefined as random for SULT2A1, a representative member of the superfamily. An active-site cap whose structure and dynamics are highly responsive to nucleotides was discovered and shown to be critical in determining SULT selectivity, a topic of longstanding interest to the field. We now realize that a given SULT can operate in two specificity modes—broad and narrow—depending on the disposition of the cap. More recent work has revealed that the caps of the SULT1A1 are controlled by homotropic allosteric interactions between PAPS molecules bound at the dimer’s active sites. These interactions cause the catalytic efficiency of SULT1A1 to vary in a substrate-dependent fashion by as much as two orders of magnitude over a range of PAPS concentrations that spans those found in human tissues. SULT catalysis is further complicated by the fact that these enzymes are frequently inhibited by their substrates. This review provides an overview of the mechanistic features of SULT1A1 that are important for the design and interpretation of SULT1A1 assays.

Introduction

In humans, the activities of the cytosolic sulfotransferases (SULTs) account for approximately one third of phase 2 metabolism (Blanchard et al., 2004; Nowell and Falany, 2006). This 13-member superfamily of enzymes regulates the activities of thousands of small-molecule metabolites (Blanchard et al., 2004; Cook et al., 2013c). The SULT1A1 isoform is present in numerous tissues but is most abundant in the liver, where it is present in near-gram quantities (Riches et al., 2009). The extremely broad specificity of SULT1A1 enables it to sulfonate the scores of small molecules—both endogenous metabolites and xenobiotics—needed to perform its homoeostatic and detoxicating functions. Imbalances in these functions have been linked to numerous diseases, including breast (Falany et al., 2002) and endometrial (Falany and Falany, 1996) cancer, Parkinson disease (Steventon et al., 1989), cystic fibrosis (Li and Falany, 2007), diabetes (Yalcin et al., 2013), and hemophilia (Moore, 2003).

Our understanding of the molecular basis of SULT catalytic behavior has changed substantially in recent years (Cook et al., 2013a–c; Wang et al., 2014a,b). Studies of SULT1A1 and 2A1 have revealed that isomerization of a highly conserved ∼30-residue stretch of amino acids (the active-site cap) controls substrate selectivity by sterically controlling access to the active site of the enzyme (Cook et al., 2013a). Further, a recent discovery revealed that SULT1A1 selectivity and its catalytic efficiency are regulated by homotropic allosteric interactions between PAPS molecules bound at the active site of the dimer (Wang et al., 2014b). Binding of the first PAPS molecule closes the cap of the subunit to which it has bound, whereas addition of the second causes the caps of both subunits to open. Finally, the well known selectivity of SULT1A1 toward small phenolic substrates has been explained at the molecular level. The enzyme uses the phenyl moieties of two phenylalanines to “sandwich” the phenolic moieties of the acceptors, bind them tightly, and orient them for nucleophilic attack (Cook et al., 2015b).

Recent work reveals that SULT1A1 turnover and selectivity are highly responsive to PAPS at concentrations that are rarely used experimentally. These concentrations are known to exist in cells and should be considered when designing experiments. The combination of the PAPS sensitivity and acceptor-specific partial substrate inhibition of SULT1A1 results in a complex mechanism. With the goal of providing a practical framework for the design and interpretation of SULT1A1 assays, this review provides an overview of mechanistic behaviors of SULT1A1 that are relevant to assay design and an in-depth discussion of SULT partial substrate inhibition.

Discussion

Mechanisms of SULT1A1 Selectivity.

At the heart of the molecular machinery that determines SULT selectivity and turnover is a conserved 30-residue active site “cap” that sits directly over the nucleotide- and acceptor-binding pockets (Fig. 1) (Cook et al., 2012, 2013a,b; Leyh et al., 2013). The structure and dynamics of the cap respond to nucleotide in ways that determine the enzyme’s specificity. With nucleotide bound at only one subunit of the SULT1A1 dimer, the cap of that subunit will spend most of its time in the “closed” conformation (Fig. 2). In this configuration, the cap encapsulates nucleotide, preventing its escape, and forms a pore through which acceptors must pass to enter the active site. The pore acts as a molecular “sieve” whose dimensions determine the size and geometry of acceptors that can be sulfonated. When nucleotide is bound, the acceptor “half” of the cap can occasionally “peel” away from the base of the active site and thus isomerize into the open state (Cook et al., 2013b). In this half-open configuration, the enzyme can sulfonate far larger and more complex acceptors. In effect, SULT1A1 has two specificity settings—narrow and broad —that are determined by the open/closed status of the cap, and the balance between these forms is determined by the isomerization equilibrium constant (K_iso) that governs their interconversion. For singly nucleotide-bound SULT1A1 in the absence of acceptor, K_iso = 26 in favor of the closed configuration (Cook et al., 2013a).

Fig. 1.

Nucleotide-linked cap closure determines substrate selection. The open and closed forms of SULT1A1 were aligned by root-mean-square-deviation minimization (PyMol) of the C_α positions of the four β-sheets that form part of core of the structure (residues 41–45, 103–108, 124–129, and 188–192). The C_α positions of these sheets are independent of the cap position. The open and closed caps are rendered in white and gold, respectively. Estradiol (E2), a small substrate, can enter the active site regardless of the cap conformation. Raloxifene (RAL), a large substrate, is sterically prevented from entering the active site when the cap is closed, and thus it binds only to the open conformation. The open and closed structures were generated using Groningen Machine for Chemical Simulation (University of Groningen, Groningen, Netherlands) as described previously (Gamage et al., 2005). The initial coordinates were taken from the SULT1A1∙E2∙PAP crystal structure (2D06). PAPS and raloxifene parameter files were generated using PQS (Parallel Quantum Solutions, Fayetteville, AR).

Fig. 2.

The coupling of PAPS binding and cap closure in SULT1A1.The substrate-binding sites of the ligand-free enzyme (left-most panel) are open and able to bind both small and large acceptors. The addition of the first PAPS molecule closes the cap over the PAPS- and acceptor-binding sites of the subunit to which it binds. In this configuration, PAPS cannot escape, and only small acceptors can enter the active site. Occasionally, the enzyme isomerizes to an open state (not shown) to which large and small acceptors can bind (K_iso = 26 in favor of the closed state). Upon binding of the second PAPS molecule, the caps of all of the sites open, which alleviates the bias toward small substrates and causes a 4-fold increases in the turnover of each subunit (Wang et al., 2014b).

The segments of the endogenous-metabolite acceptors that are sulfonated by SULT1A1 are often phenolic—hence, the enzyme’s common name, phenol sulfotransferase (Whittemore et al., 1986). Their planarity allows these acceptors to pass unimpeded through the pore. The rate constants for their entry and exit from the active site are independent of pore status, and their affinities for the open and closed forms of the enzyme are indistinguishable. In contrast, many xenobiotics (an enormous, structurally complex class of compounds) are too large to pass through the closed pore and can bind only the open form (Cook et al., 2013c). As they bind, the enzyme is drawn out of the closed conformation by mass action. Typical initial-rate and equilibrium binding experiments measure the affinity of ligand for the ensemble of enzyme forms to which it can bind (directly or indirectly) rather than the affinity for any particular species. For such measurements, the affinity of a large acceptor (K_app) is a convolution of its affinity for the open form (K_d) and the isomerization equilibrium constant (K_iso) and is given by the following equation: K_app = K_d (1 + K_iso). Thus, for SULT1A1, where K_iso = 26, the affinity for the open form is reduced relative to the ensemble by a factor that is nearly equal to K_iso (Cook et al., 2013a).

Addition of the second molecule of PAPS to the SULT1A1 dimer causes the caps of both subunits to open and profoundly affects the selectivity of the enzyme (Fig. 2). The affinities of large substrates increase ∼26-fold, making them comparable to those of small acceptors, and the turnover of the enzyme increases 8-fold (Wang et al., 2014b). Consequently, the catalytic efficiency (k_cat/K_m) of small and large acceptors increases 8-fold and ∼160-fold, respectively. These PAPS‑concentration dependent changes in specificity need to be considered in designing and interpreting SULT1A1 assays. The affinity of the second PAPS to add (30 μM) is 81‑fold less than that of the first (0.37 μM); hence, saturation of the second site requires a PAPS concentration of ∼0.50 mM (∼17 K_m), which is higher than what has often been used. Publications that preceded the discovery of these allosteric interactions typically use PAPS concentrations from 2 to 30 μM. Over this range, the enzyme transitions from predominantly singly to substantially doubly occupied, causing initial-rate parameters to differ considerably, particularly for large acceptors. The sensitivity of the enzyme to PAPS concentration in this range is almost certainly a major source of the high variability seen in published SULT1A1 initial‑rate constants (E2 0.6–2.4 seconds⁻¹ (Gamage et al., 2005; Rohn et al., 2012) and for 4-nitrophenol from 11 to 66 minutes⁻¹ (Duffel and Jakoby, 1981; Tyapochkin et al., 2009; Cook et al., 2015a) and may explain the apparent discrepancy in catalytic efficiency of fulvestrant toward SULT1A1 (Edavana et al., 2011; Cook et al., 2013a).

The PAPS concentrations in human tissues are such that the specificity of SULT1A1 is expected to be highly tissue dependent. In tissues that experience a high xenobiotic load (e.g., liver) the enzyme is expected to be doubly occupied, whereas in those where the load is slight (e.g., heart), the enzyme will be singly occupied. A correlation of PAPS tissue concentration and PAPS occupancy is shown in Fig. 3 (Brzeznicka et al., 1987; Alnouti and Klaassen, 2006). Given the 81-fold difference in the affinities of the first and second nucleotide, it is possible to adjust PAPS concentrations to study the singly and doubly occupied enzyme in isolation or in combination. The “cleanest” measurements, whether they are to determine initial-rate parameters or SULT1A1 levels, are done at 0.50 mM PAPS. Under double-occupancy conditions, one need not be concerned with whether a substrate can bind open-only forms or whether the probe acceptor will bias total activity measurements. If the goal is to assess catalytic behavior under conditions likely to prevail in a particular tissue, assays should be performed at the in vivo concentration of PAPS, which will likely yield a hybrid of open and closed forms. For many tissues, PAPS concentrations are not yet well defined; where they are, one should attempt to use cytosolic volume rather that cellular volume, in estimating concentrations (Goresky, 1963).

Fig. 3.

The fraction of doubly nucleotide-bound SULT1A1 dimer in human tissues. Occupancy was calculated using published PAPS concentrations (Klaassen and Boles, 1997; Alnouti and Klaassen, 2006; Wang et al., 2014b).

Partial Substrate Inhibition.

The SULTs typically exhibit partial substrate inhibition, a type of inhibition in which turnover decreases to a nonzero value at saturating substrate. Several mechanisms have been proposed to explain this inhibition in SULTs, including an allosteric-binding pocket (Zhang et al., 1998), gating (Lu et al., 2008; Cook et al., 2010), the binding of multiple acceptors in the active site (Gamage et al., 2003), and the formation of a dead-end complex (Gamage et al., 2005; Tyapochkin et al., 2009; Gulcan and Duffel, 2011; Wang et al., 2014a). In a recent study of SULT2A1, each of the 22 microscopic rate constants associated with interconversion of the 11 complexes in the mechanism were determined (Wang et al., 2014a). These studies provide an in-depth view of the inhibition mechanism. During initial-rate turnover under conditions where inhibition is negligible, PAP release from the binary complex (E∙PAP) is largely rate-determining; thus, E∙PAP accumulates in the presteady-state phase of the reaction, contributing to a burst of product, and it reaches a fixed concentration in the steady-state. As the concentration of substrate increases, it adds more quickly to E∙PAP, increasing the concentration of the dead-end complex (E∙PAP∙S), from which PAP escapes slowly relative to the binary complex (Wang et al., 2014a). Studies with oxidized rat enzyme implicate PAP release as the rate-liming step during uninhibited turnover of SULT1A1 (Marshall et al., 2000).

The v versus [S] plots for mechanisms involving partial substrate inhibition have the characteristic shape shown in Fig. 4A. Each of the four initial-rate constants (V_max, V_inh, K_m, and K_i) can be obtained from such data using graphical analysis to extract estimates of the constants that are then used to fit the v versus [S] data and obtain final constants. This discussion assumes that the PAPS concentration is fixed and saturating and that the acceptor concentration, [S], is varied. The algebra that describes the v versus [S] plot is given by eq. 1, which reveals that as [S] becomes small relative to K_i, the equation reduces to the noninhibited Michaelis-Menten equation (Cleland, 1979). Thus, when [S]/Ki ≤ ∼0.1, a double-reciprocal plot (i.e., 1/v versus 1/S) will approximate a straight line, from which estimates of K_m and V_max can be obtained from x- and y-axis intercepts in the normal way (see Fig. 4B). The algebra further reveals that as [S],

(1)

approaches infinity, descends to a horizontal line at, which is given by the plateau seen in Fig. 4C. Once approximate values for V_max, K_m, and V_inh have been determined, K_i can be estimated from the velocity at [S] = K_m using eq. 2 (see, Fig. 4D). To obtain final values for the constants and their associated errors, the estimates of the four constants are used as initial values for least-squares fitting of the data to eq. 1:

(2)

The protocol described in the preceding paragraph was used to extract constants from the representative v versus [S] plot seen in Fig. 4A. The velocities were generated computationally using eq. 1 and the constants listed in Table 1, which are representative of SULT mechanisms. To approximate experimental data, the velocities were allowed to vary randomly by ± 10%. Three such data sets were generated and the averaged velocities plotted. The resulting best-fit constants are listed in Table 1 and were used to draw the solid line seen passing through the data (Fig. 4A). It should be noted that the best-fit constants were independent of whether the fitting algorithm was initialized with the fixed constants or their graphically determined estimates. Thus, the lack of agreement between the fixed and fit values is due to the random scatter in the data.

Fig. 4.

Graphical analysis of partial substrate inhibition data. (A) A representative v versus [S] plot for a partial-inhibitor SULT acceptor. The solid lines through the data are the catalytic behavior predicted using the analysis described in the main text. (B) A double-reciprocal plot of the (A) data with indications for obtaining K_m and V_max. (C) The data in (A) extended to higher substrate concentrations to emphasize the plateau used to obtain V_inh. (D) An expanded view of the low-substrate concentration range of data in (A) that indicates how to obtain K_inh. In all cases, velocities were calculated using eq. 1 (see text) from the fixed constants listed in Table 1. Velocities were allowed to vary randomly by ± 10%. Three velocities were calculated at each substrate concentration; the averaged values are plotted.

TABLE 1.

Fixed and fit initial-rate parameters

Parameter	Fixed	Fit
Vmaxa	10	9.0 (0.3)c
Vinha	2.0	1.9 (0.2)
Kmb	1.0	0.82 (0.07)
Kinhb	10	13 (1)

Units are ^aarbitrary and ^bμM.

^cValues in parentheses indicate standard error.

Conclusions

The mechanism of SULT1A1 is complex. Mechanistic studies of the enzyme must consider both its allosteric regulation and partial-substrate inhibition. If the goal is to compare SULT activities across substrates or tissues, the most straightforward approach is to use high (∼0.50 mM) concentrations of PAPS. Doing so alleviates issues with respect to large versus small substrates or partial opening of the active-site cap. On the other hand, if the goal is to assess catalytic behavior in a certain tissue, assays should be performed at the PAPS concentrations in that tissue, ideally, in the cytosol of the cells in that tissue. Finally, partial substrate inhibition must be carefully considered in obtaining initial-rate parameters.

Abbreviations

PAP

3′, 5′-diphosphoadenosine

PAPS

3′-phosphoadenosine 5′-phosphosulfate

SULT

sulfotransferase

Authorship Contributions

Participated in research design: Wang, Cook, Leyh.

Performed data analysis: Wang, Cook, Leyh.

Wrote or contributed to the writing of the manuscript: Wang, Cook, Leyh.