There’s more to enzyme antagonism than inhibition

Abstract

A native enzyme’s usual assurance in recognizing their physiological substrate(s) at the ground state and on going to the transition state can be undermined by interactions with selected small molecule antagonists, leading to the generation of abnormal products. We classify this mode of enzyme antagonism resulting in the gain-of-nonnative-function as paracatalytic induction. Enzymes bound by paracatalytic inducers exhibit new or enhanced activity toward transformations that appear aberrant or erroneous. The enzyme/ paracatalytic inducer complex may take up native substrate but then bring about a transformation that is chemically distinct from the normal reaction. Alternatively, the enzyme / paracatalytic inducer complex may exhibit abnormal ground state selectivity, preferentially interacting with and transforming a molecule outside the physiological substrate scope. Paracatalytic inducers can be cytotoxic, while in other cases they divert enzyme activity toward transformations that appear adaptive and even therapeutically useful. In this perspective, we highlight two noteworthy examples from recent literature.

1. Introduction

Small molecule discovery programs for enzyme antagonists generally seek tight binding compounds that turn enzymatic activity off. The stronger the inhibition, the more promising the hit. In this Perspective we draw attention to a lesser-known class of antagonists that function by turning enzymatic activity on. The activity that gets turned on is abnormal or nonnative. An enzyme in complex with an antagonist of this type, which we categorize as a paracatalytic inducer, has their specificity modified, thereby activating a reaction pathway separate from the genetically encoded transformation.¹

“Para” can mean alongside or accompanying, like paratope, in addition to, faulty or flawed, like parapraxis. Paracatalytic induction manifests in altered stereo, chemo or regiospecificity or by elevated enzyme affinity toward atypical substrate(s).¹ Because these scenarios represent a gain of function, experimental approaches to identify paracatalytic inducers are generally turn-on assays, detecting enhanced rather than suppressed function. It may be worth differentiating paracatalytic inducers from enzyme antagonists that act through substrate-dependent inhibition mechanisms. A recent example of the latter class of molecules is remdesivir, an antagonist of viral RNA-dependent RNA polymerase (RdRp) that terminates transcription soon after incorporation into the product strand.³ Remdesivir is not a paracatalytic inducer as it stops rather than diverts enzymatic activity. Paracatalytic inducers exhibit a kind of “turnover”, rerouting the antagonist-bound enzyme over multiple (para)catalytic cycles.

The list of paracatalytic inducers is short but growing. Historical examples include the analgesic, aspirin, a paracatalytic inducer of Cox-2³ and the antibiotics, odilorhabdins and paromomycin, enhancers of ribosome miscoding during protein synthesis,^4-5 along with molecules under pre-clinical evaluation, including ADEPs⁶ and dioctatin⁷ that hijack the degradative action of the ClpP proteases. More recently, small molecules have been designed to induce aberrant ground state specificity of phosphatases,⁸ ubiquitination and deubiquitination enzyme complexes, e.g. PROTACs⁹ and DUBTACS,¹⁰ along with RIBOTACs¹¹ that induce targeted RNA degradation.

Our interest in paracatalytic induction developed by chance through a small molecule screen for conventional inhibitors of hedgehog protein autoprocessing.¹² In native hedgehog autoprocessing, also called cholesterolysis, an enzymatic domain within hedgehog binds to a cholesterol molecule and activates the lipid to cleave an internal peptide bond, releasing a carboxy sterylated hedgehog polypeptide.¹³ While screening for inhibitors, we discovered small molecule antagonists that enhanced sterol-independent autoprocessing. This is an apparently nonnative reaction where water replaces cholesterol as the substrate nucleophile. Hydrolytic autoprocessing is a slow side reaction with hedgehog, proceeding at a rate 1/1000 of the cholesterol reaction.¹⁴ The most effective of the paracatalytic inducers accelerated hydrolytic autoprocessing by 225-fold, bringing the side reaction nearly on par with cholesterolysis. Mutagenesis and modeling suggest that the antagonists are bound by hedgehog allosterically,¹⁵ although precise structural details remain to be elucidated.

Here we highlight recent reports on two other enzymes classes where novel small molecule antagonists have been identified that exhibit hallmarks of paracatalytic induction.¹ The first involves the DNA repair enzyme, OGG-1, where an antagonist induces new repair chemistry.¹⁶ The second example introduces bitopic kinase antagonists that redirect phosphorylation toward nonnative or “neo” substrates.^17,18

2. Antagonists of ogg-1 induces extra dna repair chemistry

Eukaryotic 8-oxoguanine DNA glycosylase (OGG-1) is responsible for clearing mutagenic 8-oxoguanine bases from nuclear and mitochondrial genomes.^19-21 In the absence of OGG-1, the oxidatively damaged purine accumulates and DNA polymerases may recognize it as thymine, leading to T-A transversions. OGG-1 has been subjected to small molecule screens in search of inhibitors as well as activators. Inhibitors of OGG-1, which would suppress DNA repair, hold clinical interest as sensitizers of radiation and chemotherapy for cancer.^22-25 Activators of OGG-1, by contrast, are pursued as treatments to counter oxidative DNA damage from hyper-inflammatory episodes and exposure to chemical oxidizers.^16,26

In recent work, Michel et al. reported the surprising mechanistic basis for one OGG-1 activator, C/TH10785. The molecule was identified in an earlier HTS campaign by Bohr and colleagues.²⁶ The Bohr screen selected small molecules that enhanced OGG-1 turnover in a gain-of-function biochemical assay. Hits were assessed subsequently in cell-based protection assays following paraquat exposure. Michel et al. resynthesized and retested all five biologically active OGG-1 activators.¹⁶ The new study found C/TH10785 (Fig. 1A, hashed box) as the most promising, first by thermal shift analysis and then by a panel of DNA repair assays. Analytical experiments and structural studies uncovered the mechanism of OGG-1 activation. When bound by C/TH10785, the enzyme does indeed exhibit accelerated DNA repair but through a surprising means, involving the incorporation of an elimination step into an already complex reaction sequence.²⁷

Fig. 1.

Enzyme antagonists that reshape specificity. A. Native DNA repair by OGG-1 (glycosylase, β lyase) and small-molecule induced non-native DNA repair activity of OGG-1, which includes glycosylase, β lyase and new δ lyase steps. Bonds broken in the native reaction pathway, green; Additional bonds broken in paracatalytic pathway, red. Hashed box: OGG-1 activator, C/TH10785. Four products formed by OGG-1 when bound by C/TH10785 are shown in the lower box. B. Molecular structure of bifunctional PHICS for inducing AMPK catalyzed neophosphorylation of the transcription factor, BRD4. C. PHICS mechanism of action. Ternary complex formed from AMPK, BRD4 and PHICS, induces neophosphorylation at 5 sites on BRD4.

OGG-1 in the native DNA repair reaction combines nucleophilic catalysis and general acid/base catalysis to bring about no less than five bond making/breaking events per turnover, releasing three products: two DNA strands and the 8-oxoguanine base. The enzyme’s glycosylase activity removes 8-oxoguaine; this is followed by a slower β lyase activity, cutting the DNA backbone at the 3′ side of the lesion. The substrate and end products from the glycosylase/β lyase reactions are shown in Fig. 1 A.

OGG-1’s mechanism features product assisted catalysis where the cleaved 8-oxoguanine base is enlisted by OGG-1 as a co-factor to enhance the ensuing β lyase step.²⁷ The activator C/TH10785 elaborates on product assisted catalysis. The authors propose that a nitrogen atom of the modified purine functions as a dual-purpose general acid/base catalyst. First C/TH10785 accelerates the β lyase activity of OGG-1, analogous to the behavior of 8-oxoguanine. In an ensuing step, the bound antagonist catalyzes a challenging δ elimination. The δ lyase activity, which appears unprecedented in previous studies on OGG-1, severs the DNA backbone at the 5′ side of the lesion. Antagonist-bound OGG-1 thereby breaks one substrate molecule into four products: 2 DNA strands, the 8-oxoguanine base and an unsaturated aldehyde (Fig. 1A, lower box). Along with the biochemical and structural analyses, elegant fluorescence microscopy experiments were carried out that support the altered mechanism and enhanced DNA repair by OGG-1 when C/TH10785 was added to cultured mammalian cells. This study by the Helleday group illustrates the versatility of small molecule enzyme antagonism while also showing that enzymatic side reactivity can be beneficial.²⁶

3. Antagonists to override kinase substrate specificity

Kinases are evergreen targets for small molecule discovery, an enterprise largely focused on blocking oncogenic phosphorylation.²⁸ Choudhary and colleagues have introduced novel kinase antagonists called PHICS, phosphorylation inducing chimeric small molecules that function by enhancing non-native activity.¹⁸ PHICS are reminiscent of molecular glues and PROTACs, mentioned earlier.²⁹ This new class of bifunctional molecule combines an allosteric kinase activator with a binding element specific for a user-defined, substrate protein (Fig. 1B). The two moieties are joined by a tether of variable length and polarity. AMP activated protein kinase (AMPK) and protein kinase C (PKC) were targeted by PHICS; and two proteins, BRD4 and BTK, were evaluated as potential PHICS-induced neosubstrates.

PHICS were assessed initially by AlphaScreen and by co-immunoprecipitation to quantify kinase/substrate co-localization potential. Concentration response curves indicated formation of the intended ternary complex with a maximum at ~1 μM PHICS. In those experiments, the authors also explored the influence of tether length in the PHICS, which proved to be an important variable for empirical optimization. Next and most importantly, PHICS were demonstrated to scaffold kinase/substrate interactions that supported phosphorylation. Induced phosphorylation was established for the nonnative substrate, BRD4, by Western blot and by mass spectrometry. In the presence of PHICS-1, hyper-phosphorylation of BRD4 by AMPK was observed, with at least 5 different sites of modification, all residing in predicted unstructured segments of BRD4 (Fig. 1C). Cellular potency was also evaluated with a third generation PHICS, PHICS-3, and the induction of phosphorylation was established for AMPK toward the nonnative substrate, BTK, in cultured mammalian cells.

The kinases studied here were excellent test subjects for PHICS. The AMPK and PKC enzymes appear intrinsically tolerant toward substrate variation. AMPK for example phosphorylates more than 100 different cellular proteins.³⁰ Additionally, AMPK and PKC harbor allosteric pockets and well characterized allosteric activators exist for each enzyme, including high resolution structures of kinase/activator complexes.

Those advantages notwithstanding, pitfalls can still be envisioned. Sequestration of kinase and substrate in separate organelles hampered PHICS-1 and −2 effectiveness when added to cells. A second limitation of heterobifunctional molecules like PHICS derives from the “hook effect”, so called because of the distinctive ascending and descending limbs in concentration–response plots. The ascending limb (increasing activity) with PHICS was observed at concentrations ≤1 μM and represents the intended behavior: PHICS co-localizing kinase and substrate in a productive ternary complex; the descending limb of the plot, which became apparent with PHICS ≥ 1 μM, reflects the formation of non-productive binary complexes: PHICS-kinase and PHICS-substrate. The ternary-to-binary switch is interesting and can be modelled mathematically.³¹ The biological consequences of activating AMPK or PKC by PHICS without colocalizing to the intended substrate could pose dosing challenges, although a recent follow-up paper from the group using the next generation of PHICS seems to allay this concern.¹⁷

PHICS offer an innovative means to manipulate kinase signaling pathways and may lead to novel modes of therapy.¹⁷ The probes bind wild-type enzyme and neo-substrate together in a manner that overrides native specificity. Just as in the OGG-1 study, PHICS reprogramming was translatable from biochemical assays into the cell. Unlike the antagonist of OGG-1, PHICS were prepared by rational design, grounded in the “proximity effect” of physical organic chemistry.^29,32 It seems likely that similar approaches could be applied to alter the specificity enzymes that catalyze related group transfer reactions, not only other kinases but methyl transferases, acetyl transferases etc., and extend further the example set forth with PROTACs.

4. Summary: Antagonism by paracatalytic induction, bind-to-change

The small molecules highlighted in this Perspective antagonize the physiological function of wild-type enzymes by inducing atypical transformations. Effects may be mediated by binding at the enzyme active site, as in C/TH10785, or through allosteric interactions, as in PHICS. The antagonists allow their target enzyme to take up substrate and catalyze reactions albeit with altered specificity. This is a key distinction with conventional activators that speed up the native reaction³³ and with conventional antagonists where all enzyme activity is knocked out. We have suggested the term paracatalytic induction for antagonism that results in the gain of nonnative activity, and have discussed two general classes of paracatalytic inducer: (i) a molecule that promotes transition state ambiguity, whereby the stereo, regio or chemo selectivity of the enzyme is altered to enable nonnative bond making or breaking, e.g. C/TH10785; (ii) a molecule that promotes substrate ambiguity, where the enzyme’s ground state selectivity is subverted to enable the transformation of nonnative substrates, e.g. PHICS.¹ Changing an enzyme’s chemical mechanism or an enzyme’s substrate selectivity are tasks often left to protein engineering and computational design (for recent examples,^34,35). The paracatalytic induction discussed here were brought about by small molecules and affected unmutated enzymes. These are noteworthy studies and invite exploration by enzymologists and by antagonist designers: if it can be done by mutation, it can be done by a small molecule.

Abbreviations:

ClpP

casein-like protease proteolytic core

ADEP

acyl-depsipeptide

PROTAC

proteolysis-targeting chimera

DUBTAC

deubiquitinase-targeting chimeras

RIBOTAC

ribonuclease-targeting chimeras

OGG-1

8-oxoguanine glycosylase

C/TH10785

a small molecule activator of OGG-1

PHICS

phosphorylation inducing chimeric small molecules

AMPK

AMP activated protein kianse

PKC

protein kinase C

BRD4

bromodomain-containing protein 4

BTK

Bruton’s tyrosine kinase