Chemical Tools for Studying the Impact of cis/trans Prolyl Isomerization on Signaling: A Case Study on RNA Polymerase II Phosphatase Activity and Specificity

Abstract

Proline isomerization is ubiquitous in proteins and is important for regulating important processes such as folding, recognition, and enzymatic activity. In humans, peptidyl-prolyl isomerase cis–trans isomerase NIMA interacting 1 (Pin1) is responsible for mediating fast conversion between cis- and trans-conformations of serine/threonine–proline (S/T–P) motifs in a large number of cellular pathways, many of which are involved in normal development as well as progression of several cancers and diseases. One of the major processes that Pin1 regulates is phosphatase activity against the RNA polymerase II C-terminal domain (RNAPII CTD). However, molecular tools capable of distinguishing the effects of proline conformation on phosphatase function have been lacking. A key tool that allows us to understand isomeric specificity of proteins toward their substrates is the usage of proline mimicking isosteres that are locked to prevent cis/trans-proline conversion. These locked isosteres can be incorporated into standard peptide synthesis and then used in replacement of native substrates in various experimental techniques such as kinetic and thermodynamic assays as well as X-ray crystallography. We will describe the application of these chemical tools in detail using CTD phosphatases as an example. We will also discuss alternative methods for analyzing the effect of proline conformation such as ¹³C NMR and the biological implications of proline isomeric specificity of proteins. The chemical and analytical tools presented in this chapter are widely applicable and should help elucidate many questions on the role of proline isomerization in biology.

1. INTRODUCTION

Among the 20 natural amino acids, proline is the only residue with a significant faction (5%–20% depending on context) found in isomeric cis-conformations (Fig. 1A). The capacity for interconversion between cis- and trans-conformations at peptidyl-prolyl bonds can promote energetically favorable folding patterns in proteins or even act as regulatory switches for protein recognition (Lu, Liou, & Zhou, 2002). However, the presence of cis-proline is generally low and isomerization of peptidyl-prolyl bonds is inherently slow (Wedemeyer, Welker, & Scheraga, 2002). In order to regulate cis/trans-proline isomerization, organisms have evolved a large family of peptidyl-prolyl cis/trans isomerases (PPIases) that increase the rate of interconversion between cis- and trans-proline residues to mediate the equilibration of the population of cis/trans-prolines (Shaw, 2002). An important PPIase in humans is called PPIase NIMA-interacting 1 (Pin1), which targets phosphorylated Ser/Thr–Pro (S/T–P) motifs that are enriched in phosphorylation-mediated signaling pathways (Yaffe et al., 1997). Regulation of S/T–P motifs by Pin1 can alter the enzymatic activity, protein stability, and cellular localization of its substrates (Liou, Zhou, & Lu, 2011; Lu, Finn, Lee, & Nicholson, 2007; Wulf, Finn, Suizu, & Lu, 2005; Zhou & Lu, 2016). Studying the roles and mechanisms of cis/trans-proline-mediated regulation is critical for understanding key signaling pathways as well as elucidating the underlying mechanisms of certain cancers and diseases (Liou, Zhou, & Lu, 2011; Lu & Zhou, 2007). However, it is methodologically difficult to distinguish cis- and trans-isomers due to their identical chemical composition and molecular weight. Thus, innovative strategies are needed to detect proline conformation in various biological contexts.

Fig. 1.

Isomers of dipeptide motifs. (A) General structures of cis- and trans-proline peptides. Isomerization occurs around the peptidyl-prolyl bond between proline and the preceding residue. The trans-proline conformation is more abundant that the cis-conformation (~5%–20%) in nature. (B) Core structures of the phosphorylated SP cis- and trans-isosteres. Installing a double bond between the serine and proline residues locks the isostere in the cis- or trans-conformation.

One of the essential processes that proline isomerization regulates is transcription mediated by RNA polymerase II (RNAPII), the main enzyme responsible for generation of messenger RNAs as well as small nuclear RNAs in eukaryotic cells (Hanes, 2014). In order to efficiently coordinate transcription of vastly complex genomes, RNAPII has evolved a highly repetitive C-terminal domain (CTD) consisting of consensus heptad repeats Y₁S₂P₃T₄S₅P₆S₇ that undergo a variety of posttranslational modifications (PTMs). Various PTMs on the CTD of RNAPII specifically and temporally recruit different transcriptional proteins to execute their appropriate functions at different stages of transcription (Eick & Geyer, 2013). The combination of writing, reading, and erasing of posttranslational marks on the RNAPII CTD is collectively referred to as the “CTD code,” which is essential for proper cycling through each stage of general transcription (Buratowski, 2003; Corden, 2013; Eick & Geyer, 2013; Jeronimo, Bataille, & Robert, 2013; Mayfield, Burkholder, & Zhang, 2016). Of the five residues within the heptad repeats that can be phosphorylated, the two SP motifs (YS₂P₃TS₅P₆S) are the major sites of dynamic phosphorylation/dephosphorylation required in each round of transcription (Eick & Geyer, 2013; Harlen & Churchman, 2017; Mayfield et al., 2016; Schüller et al., 2014). The presence of SP motifs in the CTD raises the possibility that cis/trans prolyl isomerization plays a role in CTD recognition. Indeed, temperature sensitive mutants of Ess1 (Pin1 homolog in Saccharomyces cerevisiae) have a slow growth or temperature sensitive phenotype along with transcriptional defects in elongation and termination (Wu et al., 2000). The termination defects may be in part related to an important regulator of termination called Nrd1 which prefers pS₅-cis-P₆ as its substrate (Kubicek et al., 2012). Studying how cis/trans prolyl isomerization impacts the specificity of CTD-binding proteins, as well as how altered recognition impacts transcriptional processing, is crucial for deciphering the CTD code.

In addition to altering protein–protein recognition, another mechanism for cis/trans-proline isomerization affecting the CTD code is through altering the enzymatic activity of CTD modification enzymes. A number of phosphatases have been found to dephosphorylate RNAPII CTD, some of which play critical roles in transcription. Ssu72 is one such CTD phosphatase which has a high specificity for phosphorylated S₅ of CTD heptad repeats (Krishnamurthy, He, Reyes-Reyes, Moore, & Hampsey, 2004) and has been implicated in both 3′ end processing and transcriptional termination (Steinmetz & Brow, 2003). Interestingly, the crystal structure of the complex between Ssu72 and phosphorylated S₅ (pS₅) CTD substrate revealed that the pS₅P₆ is selectively captured in the cis-conformation (Luo et al., 2013; Werner-Allen et al., 2011; Xiang et al., 2010; Fig. 2A) compared to another CTD phosphatase that captured P₆ in the trans-conformation (Zhang et al., 2006; Fig. 2B). For crystal structures showing substrate bound in the trans-conformation, this does not preclude the possibility of bound substrate being in the cis-conformation in some of the crystal units. However, since the trans-form substrate is more abundant in these structures the signal from bound cis-form substrates will be averaged out. The structural evidence for Ssu72 recognition of pS₅-cis-P₆ is especially convincing then, as the averaging effect resulted in the proline being observed only in the cis-conformation (Xiang et al., 2010). Furthermore, the availability of cis proline is critical for efficient dephosphorylation of pS₅ substrate by Ssu72 as shown through addition of Pin1 (Mayfield et al., 2015; Zhang et al., 2012). However, to fully elucidate the dynamics of Ssu72 recognition of cis and trans proline conformations, we and others have developed specialized chemical tools and methods to decipher the effects of one conformation vs the other.

Fig. 2.

Crystal structures of Ssu72 and Scp1 bound to native pS₅-P₆ CTD peptides. (A) Ssu72 (green) bound to the native pS₅ CTD peptide (gray) with P₆ in the cis-conformation (PDB: 3O2Q). (B) Scp1 (goldenrod) bound to native pS₅ CTD peptide with P₆ in the trans-conformation (PDB: 2GHT). The Van der Waals exterior of each protein is represented as a transparent gray surface. The 2Fo-Fc maps representing electron density are contoured to 1σ and illustrated as blue meshes around the CTD substrates.

While cis/trans-proline isomerization plays important biological roles in the function of many proteins, this process is difficult to study because tools such as DNA sequencing or mass spectrometry are not able to detect the differences in cis- and trans-proline within proteins. On the other hand, structural biology methods such as X-ray crystallography and nuclear magnetic resonance (NMR) can provide detailed information on the interaction between the target protein’s isomeric specificity and the ligand containing the proline residue. However, both of these techniques are highly dependent on the applicability of the methodology for each target protein and substrate. If the protein does not crystallize or does not incorporate the peptide ligand easily, then X-ray crystallography will be not suitable. NMR is a powerful method for detecting atomic detailed structure that can distinguish cis- and trans-prolines, but this technique is incompatible with larger proteins and complexes. In this chapter, we introduce the design of peptidomimetic isosteres that are locked in specific isomeric conformations that can be incorporated into any ligand peptide sequence for testing isomeric specificity of a target protein (Fig. 1B). As an example, we show how we generated peptidomimetic compounds with the sequence of the heptad repeats of RNAPII CTD and used them to understand how prolyl isomerization affects the activity and specificity of CTD phosphatases. In this chapter, we describe in detail how the locked proline peptidomimetics were designed and how these compounds were used as tools to determine the isomeric preference of CTD phosphatases. In essence, we used these locked peptidomimetics in place of native substrate in kinetic assays to determine the rates of dephosphorylation as well as in X-ray crystallography to show the binding modes of substrates with different isomeric conformations. We also compared the crystal structures of CTD phosphatases bound to locked peptidomimetics with structures bound to native peptides, showing that the locked isosteres accurately mimic natural substrates. Complementary studies using ¹³C NMR to analyze isomeric states of prolines in CTD are also discussed briefly. Finally, how Pin1-mediated regulation of CTD phosphatase substrate availability can lead to differential transcription resulting in alternative cellular outcomes will be considered. While we will focus attention toward the effect of cis/trans-proline isomerization on CTD phosphatase activity and specificity, analogous strategies can be adapted for studying any protein whose function is regulated by the proline conformation.

2. USING cis/trans-LOCKED SERINE–PROLINE ISOSTERES TO STUDY ACTIVITY AND SPECIFICITY OF CTD PHOSPHATASES

Enzymatic reactions against protein substrates can often be biochemically characterized by producing a piece of the substrate protein sequence and using it in in vitro kinetic assays. However, native proline-containing polypeptides cannot be used to assess the effect of cis/trans-proline isomerization because a mixture of cis- and trans-isomers of prolines will coexist and autoconvert during the measurement. To prevent interconversion and obtain a homogenous population of specific isomers, we designed peptidomimetic compounds using unnatural proline homologues (Wang et al., 2003). In such peptidomimetics, the proline residues are replaced by locked isostere moieties in which the peptide bond connecting proline to the preceding residue is replaced by a carbon double bond either in the cis- or trans-configuration. Unlike peptidyl-prolyl bonds, a carbon double bond cannot be easily rotated preventing interconversion between cis- and trans-prolines. Combining the aspect of isomeric locking with efficient synthesis and isolation strategies for one or the other isomer essentially guarantees that the final product will consist of a stable and homogenous species of a specific isomer configuration. Thus, these compounds are invaluable tools for studying the effect of cis/trans-proline isomerization on protein function.

2.1. Design and Synthesis of cis/trans-Locked Serine–Proline Isosteres

When designing cis- and trans-locked proline peptidomimetics, we aim to alter the chemical nature of the peptidyl-prolyl bond to prevent free rotation but minimize structural alterations to the rest of the peptide. Therefore, we kept the primary sequence of the substrate peptides the same as the natural protein sequence while replacing the peptidyl-prolyl bond with a carbon–carbon double bond. In order to accommodate a double bond between the proline and its preceding amino acid, the prolyl nitrogen was substituted with a carbon and the carbonyl oxygen of the preceding residue was removed (Wang et al., 2003). Using the carbon–carbon double bond to lock the isostere has several advantages: (1) the geometry of a carbon–carbon double bond is similar to a peptide bond, but the p-orbitals of the pi-bond between double bonded carbons must remain parallel to each other and therefore cannot be rotated (resulting in a planar locked configuration); (2) carbon–carbon double bonds are more stable than carbon–nitrogen double bonds in aqueous solution, increasing their versatility in experimental design; and (3) carbon–carbon double bonded peptidomimetics are less likely to be degraded by proteases that target peptidyl-prolyl bonds. Loss of the carboxyl group and adoption of a planar conformation in the dipeptide isosteres are unlikely to significantly change conformation of the entire peptide. Overall, modifying peptidyl-prolyl dipeptides with carbon–carbon double bonds poses minimal structural disturbances to the peptide chains while securing proline conformation.

Locked XP (X for any amino acid) isosteres can be chemically synthesized and then incorporated during solid-phase peptide synthesis (Hart, Sabat, & Etzkorn, 1998). Locked proline isosteres are not yet commercially available, but there are relatively efficient synthesis strategies for their production. The key aspect in the chemical synthesis of these isosteres is obtaining sufficient quantities of product with the correct stereochemistry in the double-bond formation step. Two chemical synthesis strategies have been successfully applied to generate stereospecific XP-locked isosteres. The first scheme uses careful application of organocopper-mediated anti-S_N2′ reactions (Otaka et al., 2002; Sasaki et al., 2006). This strategy is particularly effective with the synthesis of the trans-isostere, in which an SP-like oxazolidinone (4a) is modified with a silylmethyl group in an organocopper-mediated anti-S_N2′ attack to predominantly form the trans-alkene precursor 5a (Fig. 3; Otaka et al., 2002). However, syntheses based on organocopper chemistry have very low yield in the production of _L-form X-cis-P isosteres which mimic naturally occurring amino acids (Otaka et al., 2002; Sasaki et al., 2006, Table 1). Instead, rearrangement chemistry can be used to position a double bond in the desired peptidyl-prolyl position to efficiently synthesize SP analogs in either cis- or trans-configuration (Wang et al., 2003). Synthesis of either isostere begins by making a dipeptide-like precursor through condensation of a Weinreb amide (2b and 1c) with cyclopentyllithium derived from 1-iodocyclopentene (3b) (Figs. 4 and 5). The key stereoselective step for the S-cis-P is a Still–Wittig rearrangement of the ether functional group and the cyclopentene to form ~53% of the cis-isomer (Fig. 4, 6b→7b-cis). Yield for the trans-isomer in this synthesis is poor (~28%), so instead an Ireland–Claisen rearrangement between an ester functional group and the double bond of a cyclopentene can be performed to acquire ~52% of the trans-isomer (Fig. 5, 4c→5c; Wang et al., 2003). Synthesis of the S-trans-P isosteres using either method takes eight steps with overall yields of about 17% using organocopper chemistry vs 13% for Ireland–Claisen rearrangement (Table 1). On the other hand, S-cis-P can be synthesized with Still–Wittig rearrangement with many fewer steps and higher overall yield (10 steps, 20% yield) than with organocopper chemistry (Table 1).

Fig. 3.

Synthesis of the S-trans-P isostere via organocopper-mediated chemistry. A protected serine derivative (1a) is first reacted with cyclopentenyl-lithium (2a) to generate an SP-like precursor. Cyclization and protection of 3a produces 54% of the trans-oxazolidinone derivative (4a), which can then be treated with an silyl–organocopper complex ((i-PrO)Me₂SiCH₂Cu(CN)MgCl 2LiCl) to form 98% of the anti-S_N2′ product (5a). Conversion of the silyl group to a carboxyl group results in the final S-trans-P isostere precursor (7a), which can then have the tert-butyl blocking group removed to form serine.

Table 1.

Strategies for Synthesizing cis/trans Proline Dipeptide Isosteres

Synthesis Strategy	Product	Number of Steps	Overall Yield (%)
Organocoppera	S-trans-P	8	17
Organocopperb	A-cis-P	>20	<11
Ireland–Claisenc	S-trans-P	8	13
Still–Wittigc	S-cis-P	10	20

^aOtaka et al., 2002.

^bSasaki et al., 2006.

^cWang et al., 2003.

Fig. 4.

Synthesis of the S-cis-P isostere via Still–Wittig rearrangement. A protected serine derivative (2b) is reacted with 1-iodo-cyclopentene (3b) in the second synthesis step to generate an SP-like precursor. The Still–Wittig rearrangement of 6b results in two major isomeric products: ~28% in the trans-conformation (7b-trans) and ~53% in the cis-conformation (7b-cis) which is used in the subsequent reaction. Product 9b is the acid form of 8b which is converted to the final S-cis-P isostere (10b).

Fig. 5.

Synthesis of the S-trans-P isostere via Ireland–Claisen rearrangement. 1-Iodo-cyclopentene (3b) is used as a reactant in the first synthesis step to generate a proline-like precursor. In the third step of synthesis (3c → 4c), tert-butyldimethylsilyloxyacetyl chloride is used to generate the Ireland–Claisen ester precursor. The Ireland–Claisen rearrangement of 4c is carried out in two steps with activation of TMSCl followed by removal of the TBS group with tert-butylammonium fluoride (Bu₄NF) in order to generate ~52% of the trans-like precursor (5c). Subsequent oxidation and deblocking steps result in the final S-trans-P isostere (7c).

The synthesis strategies described in this section can be generally performed for any Xaa-Pro sequence and produce abundant amounts of cis- or trans-isomer with high purity. Although these isosteres are not yet commercially available, increased usage will hopefully encourage industrial chemical providers to generate locked dipeptide mimetic compounds for scientific use. Once the dipeptide mimetic precursors are synthesized and purified, they can be incorporated into routine solid-phase peptide synthesis as a unit as if they were a single amino acid. Phosphate groups can be added at this stage to the target serine/threonine dephosphorylation site. The polypeptides (incorporating X-E or X-C = C-P isosteres) are then purified through chromatography and used as substrates or ligands in characterizing the isomeric specificity of proteins of interest. For our studies, we used peptides consisting of RNAPII CTD sequences with phosphorylated S-cis/trans-P-locked moieties to determine the effect of SP motif conformation on CTD phosphatase function (Table 2).

Table 2.

Synthetic Peptide Sequences Used in Malachite Green Kinetics and X-Ray Crystallography

Peptide	Sequence
Native pS5	Ac-S-P-Y-S-P-T-S(PO3H2)-P-S-Y-S-NH2
trans-locked pS5	Ac-S-P-Y-S-P-T-S(PO3H2)ψ[(E)C = CH)]-P-S-Y-S-NH2
cis-locked pS5	Ac-S-P-Y-S-P-T-S(PO3H2)ψ[(Z)C = CH)]-P-S-Y-S-NH2
Native pS2	Ac-S-P-S-Y-S(PO3H2)-P-T-S-P-S-Y-S-NH2
trans-locked pS2	Ac-S-P-S-Y-S(PO3H2)ψ[(E)C = CH)]-P-T-S-P-S-Y-S-NH2
cis-locked pS2	Ac-S-P-S-Y-S(PO3H2)ψ[(Z)C = CH)]-P-T-S-P-S-Y-S-NH2

The symbol E represents trans and the symbol Z represents cis-conformations. The symbol ψ represents the psi bond of the SP-like isostere.

2.2. Malachite Green Kinetic Analysis of CTD Phosphatase Activity Against cis/trans-Locked Isosteres

In order to study RNAPII CTD recognition, we generated two sets of peptide mimetics with either SP motif of the CTD replaced by locked, dipeptide isosteres (phosphoryl-S-(E, Z)C=CH)-P] (Table 2). These peptidomimetic compounds containing locked pSP isosteres can thus be used in place of natural substrates in kinetic or thermodynamic measurements to investigate whether CTD-binding proteins prefer cis- or trans-SP-motifs. Here, we used CTD peptidomimetics containing proline-locked isosteres to investigate proline isomeric specificity of different CTD phosphatases including: (1) Fcp1, which dephosphorylates pS₂ during general transcription (Ghosh, Shuman, & Lima, 2008; Hausmann & Shuman, 2002; Kobor et al., 1999); (2) Ssu72, the phosphatase responsible for pS₅ dephosphorylation during general transcription (Krishnamurthy et al., 2004; Xiang et al., 2010); and (3) Scp1, which inhibits transcription of neuronal genes and dephosphorylates pS₅ (Yeo et al., 2005; Yeo, Lin, Dahmus, & Gill, 2003; Zhang et al., 2006).

The specific activity of phosphatases is often measured by the malachite green phosphate detection method (MGPD) using short phosphorylated peptides as substrates. MGPD is a colorimetric assay based on a complex formation between malachite green molybdate (BIOMOL® Green, Enzo Life Sciences) and free orthophosphate, the absorbance of which can be easily detected at wavelengths of 620–650 nm. The overall procedure for MGPD is listed below:

Step 1:

Phosphate Standard curve determination (see note 1)

1. Allow an aliquot of Malachite Green Reagent to warm up to room temperature (see note 2).

2. Perform triplicate serial dilutions of a phosphate standard (see note 3) (usually Malachite Green Reagents will be sold along with a phosphate standard). For a 96-well clear microplate scale assay, perform eight serial dilutions in which the last tube should be solvent only acting as a blank.

3. Using a multichannel pipette, take 20 μL of each dilution and add to a clean, clear microplate.

4. Using a multichannel pipette, add 40 μL of room temperature Malachite Green Reagent (as an extra precaution, filtered tips may be used to prevent phosphate contamination).

5. Cap microplate or seal with tape. Incubate at room temperature for 15–20 min to allow green color development.

6. Measure absorbance at 620 nm (precipitation should not be observed as it indicates oversaturation of the solution and will yield inaccurate readings).

7. Average triplicate absorbance measurements and subtract the blank from each reading. Plot absorbance at 620 nm on the y-axis and [phosphate] (M) on the x-axis. Fit to a linear equation and evaluate goodness of fit.

Notes

1. Ideally, the standard curve will be performed using the protein buffer as solvent as this will provide a more comparable standard to the enzymatic reaction. For different buffers and/or old reagent, it is recommended to repeat the standard curve.

2. Malachite Green Reagent is light sensitive so using dark opaque tubes or wrapping tubes in foil is recommended.

3. Malachite Green Reagent should be highly sensitive to phosphate in the nanomolar range; however, carefully inspecting the linearity of standard curve is advised to ensure that the lower limit of phosphate concentration can be reasonably detected.

Step 2:

Enzyme activity assay: Phosphatase activity against peptidomimetic substrate

1. Allow an aliquot of Malachite Green Reagent to warm up to room temperature.

2. Prepare 10 × concentrated working stocks of buffer, substrate, and enzyme.

3. Perform a serial dilution of the 10 × substrate at a concentration that when diluted 10-fold results in a maximal rate of phosphate release for the target phosphatase^%. Leave the last dilution without enzyme as a control for background phosphate signal.

4. Mix buffer, water, and enzyme in a total volume of 36 μL each (excluding substrate) in an eight-strip tube (3 × reactions and 1 × control without enzyme for background phosphate from each substrate dilution).

5. Prepare a 96-well clear microplate with 40 μL of Malachite Green Reagent for quenching each reaction. Seal the plate with plate to prevent contamination.

6. Start reactions by multichannel pipetting 4 μL of 10 × substrate serial dilutions to each reaction and control strip tube. Incubate reactions at an appropriate temperature and time for the target enzyme for as long as the reactions remain in the steady state (see note). Cap strip tubes to prevent evaporation.

7. Quench reactions by adding 20 μL from each strip tube to 40 μL of Malachite Green Reagent in the microplate.

8. Reseal the microplate with tape to avoid contamination and incubate at room temperature for 10–20 min for green color development. During this time the P_i + MG complex will be formed.

9. Read the absorbance in the plate reader at 620 nm. Using the standard curve, convert absorbance to concentration of phosphate in order to calculate the rate of phosphate release for each phosphatase reaction.

Notes

Determine the linear range of the steady state and the concentration of substrate to result in maximal rate of dephosphorylation for the target enzyme by performing a time course with constant concentrations of substrate.

We used the highly sensitive malachite green assay in combination with proline-locked CTD peptides to determine the rates of CTD phosphatases against cis- and trans-specific substrates (Mayfield et al., 2015). We determined the specificity constants for each enzyme and substrate by plotting the substrate concentration against the initial rate and fitting to the classical steady-state Michaelis–Menten kinetic equation. Using MGPD, we found that Scp1 has an almost 10-fold preference for the pS₅-trans-P₆ substrate with a k_cat/K_m of 323 ± 22 × 10³ M⁻¹ s⁻¹ compared to the cis-proline substrate with a k_cat/K_m of only 38.0 ± 1.3 × 10³ M⁻¹ s⁻¹ (Mayfield et al., 2015). Similarly, Fcp1 can dephosphorylate pS₂-P₃ of the CTD in either conformation, but it has a twofold preference for the trans-locked peptidomimic compared to the cis-proline substrate (Mayfield et al., 2015). In contrast, Ssu72 had virtually no activity against the pS₅-trans-P₆ peptide and very low activity against the native peptide. Ssu72 did have robust activity against the cis-peptide with a k_cat/K_m of 5.24 ± 0.08 × 10³ M⁻¹ s⁻¹ (Mayfield et al., 2015), which supports the crystallographic data suggesting that Ssu72 is CTD phosphatase is specific for cis-proline in substrates. Combining the malachite green assay with our proline-locked peptides yielded accurate and quantitative results revealing the cis/trans-proline specificity of several CTD phosphatases as well as their activity toward each distinct isomeric substrate.

While the malachite green assay is highly sensitive, there are issues that must be addressed for successful application of this assay on studying phosphatase activity. Like all kinetic assays, the enzymatic conditions (e.g., enzyme concentration, temperature, etc.) need to be optimized in order to obtain the best signal/noise and sensitivity. The reaction time is a particular important factor to consider for this assay because the color development buffer tends to precipitate when it is saturated with the ortho-phosphate complex, leaving a narrow window for detection. Additionally, one must take into account that certain additives may interfere with the assay by either showing a strong absorbance at 620 nm or interfering with molybdate–phosphate complex formation. For example, commonly used detergents such as Triton X-100, SDS, Tween-20, and NP-40 can adversely affect signal readings at 620 nm (Carter & Karl, 1985). Divalent cations such as Zn²⁺ and Ca²⁺ can also affect signal by forming salts with phosphates which frequently precipitate due to their low solubility. Finally, the high sensitivity of the malachite green assay makes it prone to high noise interference from any source of free inorganic phosphate. While it is difficult to completely eliminate phosphate contamination, negative controls lacking enzyme and/or phosphorylated substrate should be used to subtract unwanted noise from activity curves.

2.3. X-Ray Crystallography of CTD Phosphatases Bound to cis/trans-Locked Peptides

After kinetically quantifying the proline conformation preference of several CTD phosphatases, we obtained the binding mode of locked peptidomimic substrates using X-ray crystallography. X-ray crystallography has unexpectedly been an invaluable tool for studying the substrate specificity of CTD phosphatases (Xiang et al., 2010; Zhang et al., 2006). It can directly visualize the bound substrate conformation and identify protein residues involved in substrate recognition. However, since X-ray crystallography captures the overall signal from many repeating units of protein molecules, signal from any minor conformations that exist in the crystal may not be great enough to be observed in the occupancy-weighted average. Structure determination of CTD phosphatases bound to locked peptidomimetics on the other hand allowed us to directly visualize binding modes of less favorable but still biologically relevant substrate proline-conformations.

Since locked peptidomimetics are structurally analogous to native peptides, the procedure for determining complex structures by X-ray crystallography is identical to structural determination of proteins in complex with natural peptide ligands. If the target protein turns over the substrate through enzymatic activity, a mutant variant of the target protein must be used in order to capture the binding of substrate in the active site without releasing the product. Protein crystallization is generally an empirical process often requiring much trial and error for each different protein. Müller’s recent review provides a good general procedure of the protein–ligand complex crystallization process if you are interested in a more detailed overview (Müller, 2017). To obtain protein–ligand complexes within target protein crystals, soaking is often a good strategy to start with. We begin by testing a range of ligand concentrations and soaking times with good quality crystals while watching for signs of crystal deterioration upon complex formation. Stably soaked crystals can then be cryocooled and data can be collected to obtain the structures of protein–ligand complex. Ligand binding may cause substantial conformational changes and disrupt crystal packing, so alternatively cocrystallization of the ligand and protein can be attempted. However, there is no guarantee that cocrystallization will work in the unliganded crystallization conditions thus requiring rescreening with ligand included. Incorporating ligand into protein crystals depends on both the protein and substrate, but obtaining complex structures with locked peptides should be no different than for native peptides due to the highly similarity in structure between peptidomimetic and natural peptides.

For soaking experiments with Ssu72, we began by crystallizing inactive variants of Drosophila Ssu72 (C13D/D144N)–symplekin complexed to substrate peptides. Symplekin is a scaffold protein bound to Ssu72 during transcription, which has been shown to stabilize Ssu72 (Luo et al., 2013; Xiang et al., 2010). We have shown that the association between Ssu72 and symplekin binding does not compromise the enzymatic activity but significantly improves the diffraction limits of crystals (Luo et al., 2013). Preformed crystals of Ssu72–symplekin were transferred to crystal trays containing a serial dilution of locked peptidomimetic compounds (highest concentration at 2 mM) in mother liquor (Mayfield et al., 2015). During soaking, Ssu72–symplekin crystals appeared to be stable and showed no sign of visible deterioration. The crystals were harvested and cryoprotected after incubation at room temperature overnight. In cases where protein crystals lose diffraction after compound soaking, several strategies should be attempted to obtain optimal diffraction of the complex crystals: (1) reducing the ligand concentration; (2) reducing soak time (even soaking less than a minute may successfully incorporate the ligand); and (3) including additives that may help improve the stability of a protein crystal upon ligand complexation (Hassell et al., 2007). If the data resolution is good but the ligand density is weak, increasing the ligand concentration and soaking times may improve complex formation. Obtaining optimal protein–ligand complex structures via crystal soaking requires a delicate balance between ligand binding and crystal maintenance.

Guided by these strategies, we obtained the complex structure of Ssu72–symplekin bound to the cis-proline isostere compound at high resolution (2.95 Å; Fig. 6A) but not of the trans-proline peptidomimic (Mayfield et al., 2015). Comparison of the structures of Ssu72–symplekin bound to the pS₅-cis-P₆ isostere or native pS₅P₆ revealed near identical binding modes (Fig. 6B and C), demonstrating that the cis-locked isostere is a great mimic for cis-proline. Careful inspection of the substrate-bound structure of Ssu72 provides a plausible reason for its specificity toward cis-proline. Effective substrate recognition by Ssu72 requires phosphoserine occupancy at its phosphate-binding site as well as the flanking proline residue in an adjacent hydrophobic pocket (Fig. 6D). P₆ of the CTD fits tightly in the hydrophobic pocket consisting of residues L₄₅, P₄₆, L₈₂, and M₈₅ of Ssu72, but only the cis-isomer seems to be compatible with the cavity shape. When we tried to model a trans-proline into the substrate, unfavorable steric clashes occurred with active site residues L₈₂ and M₈₅. Overall, the structure of Ssu72–symplekin bound to the cis-proline peptidomimic replicates the native substrate-binding mode very well, solidifying the hypothesis that Ssu72 is a cis-proline-dependent phosphatase.

Fig. 6.

Crystal structures of Ssu72–symplekin bound to pS₅-cis-P₆ peptidomimetic or native pS₅ peptide. (A) Overall structure of Ssu72 (light blue) complexed to symplekin (purple) and pS₅-cis-P₆ peptidomimic (sand; PDB: 4YGX). (B) Structure of cis-locked peptidomimic bound at the Ssu72 active site. The Van der Waals exterior of Ssu72 is represented as a transparent gray surface. The 2Fo-Fc map representing electron density is contoured to 1σ and illustrated as a blue mesh around the CTD peptidomimic. (C) Overlay of Ssu72 bound to pS₅-cis-P₆ peptidomimic and Ssu72 bound to native pS₅ peptide (green and gray, respectively; PDB: 3O2Q). (D) Hydrophobic pocket of Ssu72 consisting of residues L₄₅, P₄₆, L₈₂, and M₈₅ (orange) that contribute specificity for cis-P₆ substrates.

Unlike Ssu72, our kinetic characterization Scp1 indicated that while it heavily favors the trans-form of CTD substrates it can still efficiently dephosphorylate either isomer (Mayfield et al., 2015). With significantly lower activity against cis-proline substrates that occur much less frequently in nature, it is not surprising that the structure of Scp1 in complex with native CTD peptides only shows the trans-proline configuration (Zhang et al., 2006). To compare the binding modes of Scp1 with CTD substrates of different isomeric configurations, we obtained the complex structures of Scp1 bound to cis- or trans-locked isostere peptidomimetic compounds (Mayfield et al., 2015). We were able to obtain diffracting crystals of Scp1 D96N (inactive variant) after soaking with 1 mM of either peptidomimic for 2 h at room temperature (crystals diffract to 2.20 Å for the complex with cis- and 2.36 Å for the complex with trans-locked compounds). The structure of Scp1 in complex with trans-locked peptidomimetic compounds echoed the conformation of Scp1 bound to the natural CTD peptide, confirming again that locked isosteres accurately mimic native peptidyl-prolyl conformations (Fig. 7A and B). The comparison between the structures containing the two locked isosteres shows that there are conserved molecular interactions that are essential for Scp1 to recognize its substrates (Fig. 7C and D). One of the major sites of substrate recognition for Scp1 is its hydrophobic pocket, which appears to tightly interact with residues T₄ and P₆ of both CTD phosphomimetics (Fig. 7E). This conserved interaction demonstrates how Scp1 can still efficiently dephosphorylate substrates with P₆ in either isomeric conformation. However, unlike in Ssu72 where P₆ strongly contributes to substrate recognition by tightly binding in a deep pocket, Pro6 likely contributes little to Scp1 substrate recognition as it is situated at the rim of the active site with very few discernible interactions. In particular, locked cis-P₆ is much more solvent exposed than locked trans-Pro6 (Fig. 7D), which likely explains why Scp1 prefers trans-over cis-proline substrates. The availability of peptidomimetics with cis-locked isosteres allows us to directly visualize the binding mode of a less favored substrate conformation in our target protein that X-ray crystallography would normally not be able to detect.

Fig. 7.

Crystal structures of Scp1 bound to cis/trans-locked pS₅ CTD peptidomimetics or native pS₅ peptide. (A) Structure of Scp1 D96N (cerulean) bound to pS₅-trans-P₆ peptidomimic (dark green, PDB: 4YGY). (B) Overlay of Scp1 bound to pS₅-trans-P₆ peptidomimic and Scp1 bound to native pS₅ peptide (yellow and gray, respectively; PDB: 2GHT). (C) Structure of Scp1 D96N (violet) bound to pS₅-cis-P₆ peptidomimic (sand, PDB: 4YH1). The Van der Waals exterior of Scp1 is represented as a transparent gray surface. The 2Fo-Fc maps representing electron density are contoured to 1σ and illustrated as blue meshes around the CTD peptidomimetics. (D) Overlay of the Scp1 structures bound to either trans- or cis-locked peptidomimetics. (E) Scp1-binding pocket interactions stabilizing complexation formation with peptidomimic substrates. Residues F₁₀₆, V₁₁₈, I₁₂₀, and V₁₂₇ form a hydrophobic pocket around CTD-P₃. Y₁₅₈ interacts with the side chain of CTD-T₄ while R₁₇₈ can form hydrogen bounds with the backbone carboxyl groups of CTD-S₂/T₄.

3. DETERMINATION OF PHOSPHATASE PROLINE CONFORMATION SPECIFICITY USING ¹³C NMR

A powerful method to detect subtle structural changes in biological molecules is protein NMR, which is embraced by scientists who try to study changes in protein conformations due to its ability to accommodate and even quantify those structural dynamics. However, an obstacle to observing the proline isomeric configuration in a biological sample is the lack of an amide proton on these residues, which often impairs necessary chemical shift assignments near prolines through proton-detected NMR. In any case, the absence of an amide proton prevents the direct detection of signal from the proline peptide bond for the purpose of assigning the isomerization state, although readout of the proline Cβ and Cγ chemical shifts via the adjacent residue is possible if the amide chemical shifts of that residue have been assigned. Using these methods, structural information can also be derived on the bound state of ligand prolines, as was reported for Nrd1 binding of cis-P₆ in native pS₅ CTD peptides (Kubicek et al., 2012). Recently, advances in ¹³C NMR methodology and instrumentation have provided an opportunity to gain straightforward information on the backbone conformation of proteins, especially in the study of intrinsically disorder proteins (IDPs; Brutscher et al., 2015). One key aspect of ¹³C NMR is that it is not limited by pH as in proton NMR. Further limiting the application of traditional proton NMR in studying IDPs, the lack of unique chemical environments and rapid conformational averaging cause the amide proton chemical shift dispersion for these systems to collapse to a roughly 1 ppm window, thus foiling higher resolutions typically achievable with modern spectrometers. Simultaneously, the mostly exposed backbone protons of these proteins often rapidly exchange with solvent which causes loss in signal. ¹³C NMR improves resonance dispersion of IDPs and provides structural information of prolines which lack amide protons (Gibbs & Showalter, 2015), thus leading to many novel discoveries concerning IDPs (Gibbs et al., 2017; Lawrence & Showalter, 2012). ¹³C NMR can also be used in combination with the other forms of NMR to obtain more comprehensive and clear assignment of resonances of a protein–ligand complex. A particularly applicable review covers strategies for overcoming the technical challenges of employing ¹³C NMR for studying IDPs (Bastidas, Gibbs, Sahu, & Showalter, 2015).

For the investigation of RNAPII CTD, the application of ¹³C NMR provides a very useful complement to the peptidomimic approaches described in Sections 2.1–2.3. In each heptad of the CTD, there are generally two critical SP motifs where reversible phosphorylation occurs during transcription, but the resonances associated with S₂P₃ and S₅P₆ are very difficult to distinguish using traditional protein NMR methodologies. By including ¹³C NMR in the study of the CTD, information on the overall conformation as well as the relative population of cis- and trans- conformations for these prolines can be determined (Gibbs et al., 2017). The yeast and human CTD sequences are highly repetitive making it difficult to sequence-specifically assign residues, even when using ¹³C NMR, but the Drosophila CTD has more divergent heptad sequences making it an ideal subject for NMR (Fig. 8). Intriguingly, developmentally complex organisms including all mammals and Drosophila have evolved multiple CTD heptads in which the S₇ position has been substituted with an asparagine residue. In the context of the Drosophila CTD sequence, pS₅P₆N₇ heptads were found to possess significantly increased cis-P₆ populations in solution using ¹³C NMR (Gibbs et al., 2017). Formation of an additional intramolecular hydrogen bond between N₇ and T₄ in solution likely helps stabilize the cis-conformation (Fig. 9), which should make these heptads favorable substrates for Ssu72. Indeed, kinetic data using pS₅ CTD peptides with N at the seventh position showed that they are better substrates for Ssu72 dephosphorylation compared to consensus CTD peptides (Gibbs et al., 2017). Collectively, peptidomimic chemistry allows for unambiguous demonstration that Ssu72 requires cis-proline in its substrates, while ¹³C NMR clearly demonstrates that certain conserved sequence variations of the CTD heptad dramatically enhance the equilibrium population of pS₅-cis-P₆ available in solution. Altogether, the advances in ¹³C direct-detection NMR and its application on CTD structure and function support the role of this technique as an invaluable tool for studying IDPs.

Fig. 8.

Comparison of complete RNAPII CTD sequences from Homo sapiens, Drosophila melanogaster, and Saccharomyces cerevisiae. Each heptad within the CTDs has been numbered and aligned to highlight residue variations from the consensus Y₁S₂P₃T₄S₅P₆S₇ sequence. Residue variations were colored as follows: orange for aromatic residues (F, Y, and W), green for hydrophobic residues (e.g., A and V), light blue for polar uncharged residues (e.g., S and T), purple for basic residues (H, K, and R), and red for acidic residues (D and E).

Fig. 9.

Structural comparison of Ssu72 (light green) bound to native CTD peptide and model of S7N peptide. (A) The cis-conformation of P₆ in the native peptide (white) is stabilized by three intramolecular hydrogen bonds between T₄, S₇, and Y₁’ (apostrophe denotes residue from the next heptad; PDB: 3O2Q). (B) Modeling N in place of S₇ in the native peptide (yellow) forms an additional hydrogen bond between the N₇ side chain and T₄ carboxyl group that may stabilize the cis-conformation of P₆. The Van der Waals exterior of Ssu72 is represented as a transparent gray surface.

4. SELECTIVE REGULATION OF PHOSPHATASE ACTIVITY BY PIN1 PROLYL ISOMERIZATION

The critical question regarding protein specificity for proline isomeric states of its substrate is how this special property can rewire down-stream signaling pathways. Differences in proline isomeric specificity of regulatory proteins can provide unique opportunities for differential regulation of signaling pathways. For proteins that can easily accommodate either proline conformation, the availability of cis- or trans-proline substrate should not significantly affect the whole pathway. On the other hand, if the protein only recognizes one isomeric proline substrate, it is likely that equilibrium interconversion between cis- and trans-proline is too slow to replenish the consumed isomeric form of substrate thus becoming the rate-limiting step of the biological pathway. This effect should be more prominent in proteins specific for cis-proline, since this conformation is the minority species in nature thus making it easier to be depleted. Under such a scenario, mechanisms that accelerate reequilibration of the cis/trans isomer ratio will increase the amount of apparent enzymatic activity. Pin1 in eukaryotes (or Ess1 in S. cerevisiae) catalyzes the conversion between cis- and trans-proline which can help more quickly equilibrate the population (Hanes, 2014; Lu et al., 2007). One example of how Pin1 can affect enzymatic function is when coincubated with Ssu72 there is a three-to fourfold increase in pS₅ CTD dephosphorylation compared to control reactions when the phosphate turnover was measured using the malachite green assay (Mayfield et al., 2015; Xiang et al., 2010). This increase in apparent Ssu72 activity is dependent on accelerated CTD proline isomerization since coincubation with an activity-deficient variant of Pin1 does not produce the same effect (Werner-Allen et al., 2011). In contrast, the activity of Scp1 and Fcp1 is not affected by Pin1 in vitro (Mayfield et al., 2015), consistent with their general activity against either cis- or trans-SP motifs.

To show whether the specific enhancement of Ssu72 activity by Pin1 is physiologically relevant, we tested if the phosphatase activity of Ssu72 is linked to Pin1 activity in cells (Mayfield et al., 2015). We performed lipofectamine (Thermo Fisher Scientific) transfections of HEK293T cells with a pLKO.1 plasmid containing shPin1 along with packaging and envelope plasmids to produce viral particles containing the shRNA coding region. HeLa cells were infected with either shPin1 or control viral particles using 4 μg/mL of hexadimethrine bromide and selected for stable cell lines using 2 μg/mL of puromycin over 48 h. These cells could then be probed with phospho-specific CTD antibodies (Eick & Geyer, 2013) to examine whether Pin1 function is related to CTD phosphorylation levels.

Knockdown of Pin1 expression in HeLa cells resulted in the accumulation of pS₅, likely due to the reduction in apparent Ssu72 activity. In contrast, the levels of pS₂ were not affected by the changes of Pin1 protein level, which was expected since the pS₂ phosphatase Fcp1 does not depend on the presence of cis-proline substrate for physiological function (Mayfield et al., 2015). These results strongly suggest that Pin1 specifically regulates the removal of pS₅ of the CTD through controlling substrate availability for Ssu72. It has been established that the phosphatase activity of Ssu72 is necessary for accurate termination of transcription (Dichtl et al., 2002; Ganem et al., 2003; Steinmetz & Brow, 2003). This corresponds with mutations in Ess1, the yeast homolog of Pin1, resulting in transcriptional read-through that was functionally linked to Ssu72 activity (Krishnamurthy et al., 2004). Our studies on the cis-proline specificity of Ssu72 provide a molecular explanation for the transcriptional defects in these various in vivo systems (Fig. 10).

Fig. 10.

Model of Pin1 effect on Ssu72-dependent dephosphorylation of RNAPII CTD. (1) Rate of isomerization of phosphorylated SP motifs with or without Pin1. The thin arrows without Pin1 represent the slow rate of normal proline isomerization, while the bold arrows represent the accelerated rate of isomerization with Pin1. (2A) In cells with low Pin1 activity, Ssu72 activity will become limited as it consumes its preferred pS₅-cis-P₆ substrate and the relatively slow rate of isomerization limits further dephosphorylation at this site. (2B) Therefore, disruption in Pin1 and subsequently Ssu72 causes transcriptionally active RNAPII to remain hyperphosphorylated and read through termination sites. (3A) With normally functioning Pin1 catalyzing the cis/trans isomerization, Ssu72 rate will no longer be limited by a lack of pS₅-cis-P₆ substrate as trans-substrate can be converted to the cis-conformation in the appropriate time frame. (3B) Proper functioning of Pin1 and Ssu72 results in accurate transcriptional termination and thus better coordination of transcriptional cycling.

5. SUMMARY AND CONCLUSION

In this chapter, we have described in detail a method using conformation-locked proline peptidomimetics to determine the activity and specificity of proteins for cis- and trans-proline substrates. Using CTD phosphatases as an example, we showed how such compounds can be used to determine without ambiguity the substrate proline conformation preference of a protein. We accomplished this using a highly sensitive colorimetric assay to quantitatively assess enzymatic activity against either proline isomer as well as captured complex structures with their preferred isomeric peptidomimetics using X-ray crystallography. These structures showed that such proline isostere compounds are very good mimics of natural prolines; and in one case, we were able to directly observe the recognition mode of a substrate with a less favored peptidyl-proline bond conformation. This would not have been possible using X-ray crystallography analysis with natural substrates which typically obviates alternative binding modes. We have also briefly discussed alternative tools such as ¹³C NMR that can provide new information on the dynamics of proline isomerization on enzymatic function. In terms of the biological relevance of proline conformation specificity, proteins with different isomeric preferences can be differentially regulated and achieve alternative cellular outcomes. Regulating the function of proline conformation sensitive proteins with prolyl isomerases like Pin1 thus appears to be critical for normal cellular function.

ABBREVIATIONS

IDP

intrinsically disordered protein

MGPD

malachite green phosphate detection method

NMR

nuclear magnetic resonance

pS₅

phosphorylated serine 5

Pin1

peptidyl-prolyl isomerase cis–trans isomerase NIMA interacting 1

PPIase

peptidyl-prolyl cis/trans isomerase

PTM

posttranslational modification

RNAPII CTD

RNA polymerase II C-terminal domain

SP motif

serine–proline motif