Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death Rescue

Resource

Beyond DPPH: Use of Fluorescence-Enabled

Inhibited Autoxidation to Predict Oxidative Cell

Death Rescue

Graphical Abstract

Authors

Ron Shah, Luke A. Farmer,

Omkar Zilka, Antonius T.M. Van Kessel,

Derek A. Pratt

Correspondence

dpratt@uottawa.ca

In Brief

Shah et al. develop a ﬂuorescence-

enabled inhibited autoxidation (FENIX)

approach that permits accurate

quantitation of radical-trapping

antioxidant activity in phospholipid

bilayers. The methodology is far superior

to existing assays (e.g., DPPH) and

enables reliable prediction of the anti-

ferroptotic potency of redox-active

compounds. The approach is easily

amenable to high-throughput screening.

Highlights

d

FENIX enables screening or counter-screening of lipid

peroxidation inhibitors

Equally simple to the DPPH assay, but relays information on

kinetics and stoichiometry

FENIX in liposomes predicts the anti-ferroptotic potency of

antioxidants in cells

H-bonding to the phospholipid head group attenuates RTA

activity in lipid bilayers

Shah et al., 2019, Cell Chemical Biology 26, 1–14

November 21, 2019 ª 2019 Elsevier Ltd.

https://doi.org/10.1016/j.chembiol.2019.09.007

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Cell Chemical Biology

Resource

Beyond DPPH: Use of Fluorescence-Enabled Inhibited

Autoxidation to Predict Oxidative Cell Death Rescue

Ron Shah, Luke A. Farmer, Omkar Zilka, Antonius T.M. Van Kessel, and Derek A. Pratt¹^,²^,*

1

Department of Chemistry and Biomolecular Sciences, University of Ottawa, Ottawa, ON K1N 6N5, Canada

²Lead Contact

*Correspondence: dpratt@uottawa.ca

https://doi.org/10.1016/j.chembiol.2019.09.007

SUMMARY

product, may slow the onset and development of degenerative

disease. Despite considerable investigation, the use of RTAs in

‘‘Antioxidant activity’’ is an often invoked, but gener- commercial and/or industrial applications has not translated to

ally poorly characterized, molecular property. results in the clinic. Put simply, there has been overwhelming ev-

idence against antioxidant intervention as a useful strategy for

degenerative disease treatment and/or prevention in humans

Several assays are available to determine antioxidant

activity, the most popular of which is based upon the

(

Lonn, 2005; Semba et al., 2014; Sesso et al., 2012).

At the root of many clinical trials of antioxidant treatment and/or

ability of a putative antioxidant to reduce 2,2-di-

phenyl-1-picrylhydrazyl. Here, we show that the re-

sults of this assay do not correlate with the potency

of putative antioxidants as inhibitors of ferroptosis,

the oxidative cell death modality associated

with (phospho)lipid peroxidation. We subsequently

supplementation is the characterization of potential antioxidants

using assays of antioxidant activity. The most common assays

involve the decolorization of a highly absorbing oxidant by reduc-

tion with a putative antioxidant. Examples include 2,2-diphenyl-

1-picrylhydrazyl (DPPH) (Blois, 1958), the ferric reducing ability

describe our efforts to develop an approach that of plasma (Benzie and Strain, 1996), oxygen radical absorbance

0

quantiﬁes the reactivity of putative antioxidants capacity (ORAC) (Cao et al., 1993), and 2,2 -azino-

with the (phospho)lipid peroxyl radicals that propa- bis(3-ethylbenzothiazoline-6-sulfonic acid (ABTS) (Re et al.,

1

999). The limitations of these assays have been elaborated

gate (phospho)lipid peroxidation (dubbed FENIX

ﬂuorescence-enabled inhibited autoxidation]). The

upon elsewhere. Most importantly, they fail to provide information

about the reactivity of a potential inhibitor toward the kinetically

relevant oxidizing species––a propagating lipid peroxyl radical

[

results obtained with FENIX afford an excellent cor-

relation with anti-ferroptotic potency, which facili-

tates mechanistic characterization of ferroptosis

inhibitors, and reveals the importance of H-bonding

interactions between antioxidant and phospholipid

that underlie both the lackluster antioxidant activity

(i.e., kinh; Figure 1A) (Ingold, 1961; Ingold and Pratt, 2014). This

is where the divergence between industrial compound develop-

ment and potential preventive/therapeutic compounds origi-

nates. Industrial antioxidants, such as hindered phenols and aro-

matic amines, are developed and tested in a context that is highly

of phenols under physiologically relevant conditions similar to their eventual application––and where the experimental

and the emergence of arylamines as inhibitors of methods measure relevant properties, such as oxidation product

choice.

formation or viscosity changes (Gray, 1978; Naidu et al., 1984;

Shah and Pratt, 2016). This underscores the need for methods

that assay RTA activity under physiologically relevant conditions.

Of course, it must be acknowledged that modeling a lubricant or

polymer is far easier than a cell, tissue, or whole organism.

The recent characterization of ferroptosis (Dixon et al., 2012;

INTRODUCTION

Lipid peroxidation, the radical-mediated chain reaction that con-

verts lipids into lipid hydroperoxides (and a myriad of other Friedmann Angeli et al., 2014; Stockwell et al., 2017), a form of

oxidation products) has been implicated in the pathogenesis of regulated necrosis associated with the accumulation of (phos-

virtually every type of degenerative disease (Barnham et al., pho)lipid oxidation products, and the development of new cell-

2

004; Finkel and Holbrook, 2000; McIntyre and Hazen, 2010; based tools to study it, have provided an unprecedented

Di Paolo and Kim, 2011). The current molecular level under- capability to investigate the link between lipid peroxidation and

standing of lipid peroxidation (simpliﬁed mechanism, Figure 1A) disease. Ferroptosis can be induced in a variety of ways, but

(

Yin et al., 2011; Zielinski and Pratt, 2017) has been established the quintessential––and arguably most direct––method is to

largely on the basis of detailed chemical studies of hydrocarbon interfere with the cell’s ability to detoxify (phospho)lipid hydro-

autoxidation, the process primarily responsible for the oxidative peroxides. Left to accumulate, the hydroperoxides can undergo

degradation of petroleum-derived products (Denison and Harle, one-electron reduction by ferrous iron, or other one-electron re-

1

949; Frank, 1950). Therefore, it has long been thought that ductants, to produce lipid-derived alkoxyl radicals that can

radical-trapping antioxidants (RTAs) (Ingold, 1961; Ingold and initiate further lipid peroxidation chain reactions. Glutathione

Pratt, 2014), which are preservatives added to essentially peroxidase 4 (GPX4), a selenoprotein that catalyzes the gluta-

every type of hydrocarbon-based commercial and/or industrial thione-mediated reduction of hydroperoxides to alcohols, is

Cell Chemical Biology 26, 1–14, November 21, 2019 ª 2019 Elsevier Ltd.

1

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Figure 1. Lipid Peroxidation, its Inhibition by RTAs, and the Inadequacy of the DPPH Assay to Account for the Anti-ferroptotic Activity of RTAs

(

A) Simpliﬁed mechanism of lipid peroxidation (autoxidation) and its inhibition by RTAs.

B) Inhibition of GPX4 leads to the induction of ferroptosis.

C) The DPPH assay (inset) and trends in publications with the keyword ‘‘DPPH’’ in either the article title or abstract. Data obtained by searching the keyword

DPPH on Scopus. The results were exported into Excel and plotted using GraphPad Prism.

(

D) Structures of purported natural product ‘‘antioxidants’’ and some related derivatives used in this study.

E) Structures of potent ferroptosis inhibitors.

F) The titration of DPPH (100 mM) with compounds 1–12. Error bars represent the standard deviation obtained from at least three independent measurements.

G) Plot of IC50 for DPPH quenching versus EC50 for inhibition of RSL3-induced ferroptosis in Pfa1 mouse embryonic ﬁbroblasts (MEFs). Data in light blue and

2

NH) in Figure S1A (R = 0.01). Data points labeled with an asterisk did not inhibit ferroptosis up to 10 mM.

brown correspond to the diarylamines (Ar

2

Data are presented as mean ± SD. See also Figure S1.

the unique enzyme that can operate directly on phospholipid hy- 2017; Weiwer et al., 2012]) (Figure 1B), deletion of the gene en-

droperoxides––and is therefore generally considered the master coding it (Friedmann Angeli et al., 2014; Seiler et al., 2008), or

regulator of ferroptosis (Conrad and Friedmann Angeli, 2015; suppression of its expression, present the ideal basis for cell-

Friedmann Angeli et al., 2014; Seibt et al., 2019).

based assays of inhibitors of lipid peroxidation. Indeed, such

The inhibition of GPX4 (e.g., with RSL3 [Dixon et al., 2012; assays have been used to screen compound libraries to identify

Yang and Stockwell, 2008] or ML210 [Viswanathan et al., some of the most potent ferroptosis inhibitors to date (e.g.,

2

Cell Chemical Biology 26, 1–14, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

ferrostatin-1 [Dixon et al., 2012; Skouta et al., 2014] and liprox-

TheDPPHassay was carried out ina96-wellmicroplatewherein

statin-1 [Friedmann Angeli et al., 2014]) (Figure 1E). However, various concentrations of compounds were incubated with

because cellular lipid peroxidation can be modulated in other 100 mM DPPH in ethanol for 30 min before determination of the

ways (e.g., the incorporation of [non]-oxidizable polyunsatu- absorbance at 517 nm using a microplate reader (Cos et al.,

rated fatty acids in the lipid bilayer) (Doll et al., 2017; Magta- 2003). The half maximal inhibitory concentration (IC50) values

nong et al., 2019), these cell-based assays need to be were calculated by ﬁtting the percentage of the DPPH radical re-

complemented by a method that can unambiguously resolve maining after the incubation period, as is customary (Figures 1F,

intervention in the propagation of lipid peroxidation. Such a S1B, and S1C) (Blois, 1958; Chen et al., 2013; Kedare and Singh,

method would enable resolution of inhibitors that do not act 2011).ThemostpotentRTAinthisassaywasthepolyphenolicnat-

as antioxidants, providing insight on other mechanisms that ural product epigallocatechin gallate (6), with an IC50 of 2.9 mM,

may contribute to ferroptosis and/or its regulation. Building while glutathione (3) was the least potent with an IC50 of 445 mM.

upon previous work wherein we showed that a small amount These results are consistent with literature precedent, wherein it

of an autoxidizable chromophore (i.e., STY-BODIPY or PBD- is reported that the efﬁcacy for DPPH quenching increased in

BODIPY) can be used to monitor hydrocarbon autoxidations the following order: glutathione (3) > hydroquinone (10) > catechol

and their inhibition by RTAs (Haidasz et al., 2016), we now (9) > ascorbic acid (2) z epigallocatechin gallate (6) (Xie and

report a microplate-based assay that enables screening for, Schaich, 2014). Surprisingly, we found that the IC50 values for

and assessment of, RTA activity in phospholipid bilayers. We glutathione (3) and N-acetylcysteine (12) were 10-fold different,

relay its development in an effort to highlight the important as- despite the fact that both molecules have the same redox-active

pects that must be considered to accurately determine antiox- thiol moiety. It is reported that natural product resveratrol (4) and

idant efﬁcacy in a cellular context. These aspects clarify why its corresponding tert-butylated analog (8) have different RTA ac-

phenolic antioxidants are not particularly effective under phys- tivities, yet they had very similar IC50 values (Matsuura et al., 2015),

iologically relevant conditions, why aromatic amines are suggesting that the assay is not sensitive enough to discern alkyl

strongly preferred, and serve to guide future design and/or opti- substituent effects. It was also difﬁcult to observe any trends in

mization of ferroptosis inhibitors.

results with arylamines––the IC50 values were in a very narrow

range (18–35 mM) despite the fact that they are reported to have

vastly different RTA activities (Shah et al., 2017).

RESULTS

The potencies of these compounds as ferroptosis inhibitors

were then determined in mouse embryonic ﬁbroblasts (Fried-

mann Angeli et al., 2014) treated with the GPX4 inhibitor RSL3

The DPPH Assay Fails to Predict Cytoprotective Potency

of Putative Antioxidants

Among the aforementioned measures of antioxidant activity, the (Yang and Stockwell, 2008) (Figure S1D). The half maximal effec-

DPPH assay (Figure 1C) is arguably the most popular method. tive concentration (EC50) values are plotted alongside the IC50

Introduced in 1922 (Goldschmidt and Renn, 1922), DPPH was values obtained from the DPPH assay in Figure 1G, wherein no

mainly used in electron paramagnetic resonance spectroscopy correlation is evident. Most naturally occurring phenolic antioxi-

as a calibrant and standard. It was not until 1958 that it was dants (1, 4, 6, and 7) were modest inhibitors of cell death with

used to determine antioxidant activity (Blois, 1958). Ever since, EC50 > 1 mM, whereas the synthetic derivatives (8–11) were

the use of DPPH to determine antioxidant activity has seen a dra- more potent with EC50 values <450 nM. As expected from previ-

matic increase that roughly coincides with the recognition of ous work, aromatic amines, including Fer-1, Lip-1, and phenox-

‘

‘oxidative stress’’ as being relevant in human health. Despite azine (PHOXN) were the most potent compounds. On the other

the well-documented shortcomings of the DPPH assay (Li and hand, water-soluble compounds (2, 3, 5, and 12) failed to rescue

Pratt, 2015; Amorati and Valgimigli, 2015; Foti, 2015; Niki, cells from ferroptosis (up to 10 mM), presumably due to their

2

010), its popularity for the assessment of antioxidant activity inability to traverse the lipid bilayer (ClogP < –0.5). Of course,

prompted us to assess its use for the prediction of ferroptosis this is not captured in the DPPH assay. Overall, it is clear that,

inhibitor potency. In fact, others have used the results of DPPH despite its popularity, the DPPH assay is a poor predictor of fer-

assays to provide insight on the mechanism of inhibition of fer- roptosis inhibition as demonstrated by the clear lack of a corre-

2

roptosis by small molecules (Dixon et al., 2012; Doll et al., lation in Figure 1G (R = 0.01).

2

017; Gao et al., 2015; Liu et al., 2015; Magtanong et al.,

019). Our test set included ferrostatin-1 (Fer-1) and liproxsta- Initial Efforts to Develop a Predictive Assay

tin-1 (Lip-1), the two quintessential ferroptosis inhibitors, a The rate-limiting step in the propagation of the lipid peroxidation

variety of diarylamines (Figures 1E and S1A) that we recently chain reaction involves the abstraction of an H-atom from the

demonstrated to be equivalent to––or better than––Fer-1 and lipid by a peroxyl radical (Yin et al., 2011; Zielinski and Pratt,

Lip-1 (Shah et al., 2017), some representative phenolic antioxi- 2017) (Figure 1A). (It should be mentioned that some lipids,

dants including Nature’s premier RTA, a-tocopherol (Burton e.g. cholesterol, can also propagate by addition of a peroxyl

and Ingold, 1986), and the related, but much more potent tetra- radical to an unsaturation Zielinski and Pratt, 2019). An effective

hydronaphthyridinol derivative (Li et al., 2013; Zilka et al., 2017). RTA must compete with phospholipids for this peroxyl radical

We also included a variety of purported antioxidants of varying and yield an RTA-derived radical that is persistent (i.e., does

hydrophilicity/hydrophobicity, including ascorbic acid, gluta- not propagate the chain reaction) (Ingold, 1961; Ingold and Pratt,

thione, N-acetylcysteine, NADPH, a hydroquinone, a catechol, 2014). Thus, an assay that is based upon this kinetic competition

curcumin, resveratrol, and epigallocatechin gallate, the so- is expected to be better able to predict the potency of putative

called ‘‘green tea polyphenol’’ (Figure 1D).

RTAs in cells. A short time ago, we developed the basis for

Cell Chemical Biology 26, 1–14, November 21, 2019

3

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Figure 2. The Co-autoxidation of either STY-BODIPY or PBD-BODIPY and an Appropriate Substrate Enables the Rapid Determination of RTA

Kinetics and Stoichiometry under a Variety of Conditions

(A) The reaction of STY-BODIPY and PBD-BODIPY with peroxyl radicals can be used as a signal carrier in inhibited co-autoxidations to obtain inhibition rate

constants (kinh) and stoichiometries (n) of added RTAs. [BODIPY] = dye concentration at t = 0 (M); n = stoichiometry; [RTA] = antioxidant concentration (M); R

i

=

rate of initiation; kBODIPY = rate constant of propagation; tinh = inhibition time (s).

ꢀ

(

B) Representative co-autoxidations of PBD-BODIPY (10 mM) and 1-hexadecene (2.8 M) initiated by azobisisobutyronitrile (AIBN) (6 mM) in chlorobenzene at 37 C

gray) and inhibited by 2 mM of compounds 1–12.

2

C) Correlation of inhibition rate constants determined in chlorobenzene and EC50 values from RSL3-induced ferroptosis in Pfa1 MEFs (R = 0.29).

D) Representative co-autoxidations of STY-BODIPY (10 mM) embedded liposomes of egg phosphatidylcholine (1 mM) suspended in PBS (10 mM) at pH 7.4

ꢀ

initiated by MeOAMVN (0.2 mM) at 37 C (gray) and inhibited by 2 mM of compounds 1–12.

2

(

E) Correlation of inhibition rate constants determined in liposomes and EC50 values from RSL3-induced ferroptosis in Pfa1 MEFs (R = 0.42). Compounds below

the dashed line did not inhibit ferroptosis up to 10 mM.

Data show mean ± SD from three independent experiments. See also Figure S1.

such an assay. The experiment involves the competition of an initial rate of the inhibited autoxidation where this competition

RTA and a highly colored and highly oxidizable probe for the per- takes place (Equation 1 in Figure 2A, see Supplemental Informa-

oxyl radicals that propagate the autoxidation of a hydrocarbon tion for derivation of kinh).

substrate. Thus, by adding a small amount of this probe mole-

In preliminary work, we used this method to determine the

cule (i.e., STY-BODIPY or PBD-BODIPY, Figure 2A) to an autox- apparent inhibition rate constants (kinh) for Lip-1, Fer-1, and a va-

idizable substrate, the progress of the reaction and its inhibition riety of diarylamines that were found to be potent inhibitors of fer-

by RTAs can be readily monitored by conventional spectropho- roptosis (Shah et al., 2017; Zilka et al., 2017). We have since

tometry (Haidasz et al., 2016). Moreover, since the rate constant expanded the set of antioxidants investigated with this approach

for the reaction of the probe molecule and a peroxyl radical can to those with which we obtained the above data for reactions with

be independently determined, the rate constant for the reaction DPPH. Representative data from co-autoxidations of PBD-BOD-

of the RTA and a peroxyl radical (kinh) can be derived from the IPY and1-hexadecene in chlorobenzene are included in Figure 2B

4

Cell Chemical Biology 26, 1–14, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

(rate constants are shown in Figure S1E). It is important to note rescence as a function of dye concentration resulted in a response

4

that the autoxidations were initiated with azobisisobutyronitrile factor of 7.5 3 10 relative ﬂuorescence units (RFU)/mM (Fig-

i

and that the rate of initiation (R ), which is important in determining ure S2B). This enabled the determination of the concentration of

the stoichiometry of radical-trapping (or the number n of peroxyl the oxidized probe as a function of time, which is necessary to

radicals trapped per molecule of RTA, Equation 2 in Figure 2A, derive kinh from Equation 1 in Figure 2A. With a microplate-based

see Supplemental Information on derivation), was determined us- method available to follow reaction progress in liposome autoxi-

ing the inhibition time of 2,2,5,7,8-pentamethyl-6-chromanol dations, we surveyed a variety of conditions to establish a model

(

PMC), the truncated analog of vitamin E, for which n = 2 (Burton capable of predicting the potency of ferroptosis inhibitors in cells

and Ingold, 1986). The kinetic data roughly trend with the cytopro- using the apparent inhibition rate constants in liposomes. To

tective potencies of the test compounds, although the correlation expand on our preliminary work, we tested both water-soluble

is rather poor, with a signiﬁcant amount of scatter still evident in (2,2-azobis(2-amidinopropane)dihydrochloride (AAPH), 1 mM)

2

Figure 2C (R = 0.29). Therefore, although these experiments and lipid-soluble (MeOAMVN, 0.2 mM) initiators in 1 mM egg PC

report on the kinetics of reactions of antioxidants with peroxyl liposomes (Figures 3C and 3D, respectively). Initially, autoxida-

radicals they still lack predictive capacity when it comes to the tions inhibited using PMC (4 mM), PHOXN (4 mM), glutathione

potency of antioxidants as ferroptosis inhibitors. (GSH) (10 mM), ascorbic acid (10 mM), or NAC (50 mM) (Figure 3E,

In the aforementioned preliminary work, a slightly better correla- left and 3F, left) were carried out. Surprisingly, although the values

tion between kinh and cytoprotective potency was obtained when of kinh determined for the lipophilic RTAs (PMC and PHOXN) var-

0

the kinetics were determined from 2,2 -azobis(4-methoxy-2,4- ied with initiator, those of the water-soluble RTAs (GSH, NAC and

dimethylvaleronitrile) (MeOAMVN)-initiated co-autoxidations of ascorbic acid) did not (cf. Table S1). Since AAPH is water-soluble,

STY-BODIPY and the polyunsaturated lipids of unilamellar lipo- it is expected that the water-soluble RTAs inhibit STY-BODIPY

somes of egg phosphatidylcholine––conditions expected to oxidation by reacting directly with the initiator-derived radicals

more closely mimic the cellular membrane and the membranes before initiating lipid peroxidation, whereas the lipophilic RTAs

of subcellular compartments (Figure 2D). However, once again, require import of the radicals from the aqueous phase to the lipid

an expansion of the initial dataset to include all of the compounds phase to react, and therefore compete with lipids for peroxyl rad-

investigated above did not lead to a substantial improvement (Fig- icals. MeOAMVN, on the other hand, is considered to be a lipid-

2

ure 2E) (R = 0.42). Interestingly, while the data for many com- soluble initiator, so it is expected that the efﬁcacy with which

pounds seem to correlate better, most of the ‘‘outliers’’ segregate water-soluble RTAs inhibit the autoxidation would be substantially

into two distinct groups: water-soluble compounds that were worse than in the AAPH-initiated autoxidations. This was obvi-

measuredtobegoodRTAsinliposomes, butare ineffective incells ously not the case. Because small unilamellar vesicles have large

(

i.e., 2, 3, and 5), and water-soluble diarylamines that are among curvature resulting in increased dynamics of the bilayer, we

the most potent of RTAs, but are less potent in cells than expected wondered whether this effect was simply due to the facile diffusion

brown circles). These results suggest that the assay conditions of water-soluble RTAs into the liposomal membrane. However, a

(

developed to date are still not fully representative of the lipid per- similar experiment with 200 nm liposomes afforded equivalent re-

oxidation that drives ferroptotic cell death. Since we anticipated sults, wherein water-soluble RTAs were still highly potent in

that efforts to optimize assay parameters using the current proto- MeOAMVN-initiated autoxidations (Figure S2E).

col could prove prohibitively time consuming, we ﬁrst sought to These ﬁndings suggested that the water-soluble RTAs were re-

develop a high-throughput version of our co-autoxidation acting directly with MeOAMVN-derived radicals at the lipid-

approach to enable us to more rapidly survey reaction conditions. aqueous interface, such that these RTAs compete directly with

initiation of lipid peroxidation. Expanding the test set of com-

Optimization of Co-autoxidation Conditions

pounds investigated to the ‘‘full’’ set examined in Figures 1 and

Untilnow,reactionprogressinSTY-BODIPY(orPBD-BODIPY)co- 2, reinforced the trends observed with the small test set and our

autoxidations has been monitored by the decrease in absorbance preliminary efforts—particularly, that water-soluble compounds

of the chromophore at its lmax as it reacts with peroxyl radicals appear to be reacting directly with initiator-derived radicals (Fig-

2

(

Figure 3A). However, we anticipated that a ﬂuorescence-based ure 3G) (R = 0.40) (Shah et al., 2017). Indeed, when we calculated

method would be better suited for a more high-throughput micro- the kinetic chain lengths for both reaction systems—even in the

plate-based assay. The narrow Stokes shift in STY-BODIPY absence of RTA—they were found to be less than 1 (Table S2),

prompted to test several excitation wavelengths to determine suggesting that the oxidations were not radical chain reactions.

which would afford optimal increase in green ﬂuorescence. We

To ensure that the oxidations were being carried out under

settled onexcitationat 488 nmand emissionat518 nm(Figure 3B). radical chain condition (i.e., rate of oxidation >> rate of initiation),

Expectedly, the ﬂuorescence of the probe increased linearly such that they can be considered bona ﬁde autoxidations, we

over time when it was added as a signal carrier during the autox- increased the lipid concentration to 20 mM in egg PC while keep-

idation of 1 mM egg PC liposomes (extruded to 100 nm). To ing the rate of initiation the same (Figures 3E, right, and 3F, right).

enable quantitation of the amount of STY-BODIPY (0.2–1 mM) Doing so gave kinetic chain lengths of 8 and 12 for the AAPH-

that had been oxidized, we determined the plateau in ﬂuores- and MeOAMVN-initiated oxidations, respectively. Under these

cence as a function of [STY-BODIPY]. The plateau represents conditions, the water-soluble RTAs were far less effective than

the point at which the styryl side chain has undergone oxidation, under non-chain conditions, consistent with the phase-separa-

but the slower reaction of peroxyl radicals with the BODIPY core tion of the lipid peroxyl radicals from the water-soluble antioxi-

has yet to take place (at which point the ﬂuorescence decreases, dants. Encouraged by these results, we carried out inhibited

see Figure S2A) (Krumova et al., 2012). Plotting the maximal ﬂuo- autoxidations with the full test set of compounds under these

Cell Chemical Biology 26, 1–14, November 21, 2019

5

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Figure 3. RTA Kinetics Derived from STY-BODIPY/Phospholipid Co-autoxidations Initiated with Conventional Radical Initators do not Corre-

late well with Anti-ferroptotic Potency

(

A and B) Co-autoxidation of STY-BODIPY (1 mM)-embedded liposomes of egg PC (1 mM) suspended in PBS (10 mM) at pH 7.4 initiated with MeOAMVN (0.2 mM)

monitored by either absorbance (A) or ﬂuorescence (B).

C–F) AAPH (C) and MeOAMVN (D) radical initiators. Representative co-autoxidations of STY-BODIPY (1 mM)-embedded liposomes of egg PC (1 or 20 mM)

(

initiated by either (E) AAPH (1 mM, black) or (F) MeOAMVN (0.2 mM, black) and inhibited by PMC (4 mM, red), phenoxazine (4 mM, blue), ascorbic acid (10 mM,

cyan), glutathione (10 mM, green), or NAC (50 mM, magenta).

2

(G and H) Correlation of inhibition rate constants determined in either 1 mM (R = 0.40) (G) or 20 mM (R = 0.58) (H) liposomes and EC50 values from RSL3-induced

ferroptosis in Pfa1 MEFs. Compounds below the dashed line did not inhibit ferroptosis up to 10 mM.

Data show mean ± SD of n = 3 wells from a 96-well plate from one representative of three independent experiments. See also Figure S2 and Table S1.

conditions (Figure 3H). Surprisingly, the resultant correlation of philic RTAs may still be a problem. Indeed, we found that no

2

EC50 with kinh improved only marginally (R = 0.58) relative to consumption of STY-BODIPY took place in MeOAMVN-treated

that obtained when using data from the oxidations carried out liposomes, which were subsequently puriﬁed by size-exclusion

under non-chain conditions (Figure 3G), suggesting that inter- chromatography (SEC) (see Figure 4B), indicating that

ception of initiator-derived radicals by water-soluble or amphi- MeOAMVN is poorly incorporated into lipid bilayers and must

6

Cell Chemical Biology 26, 1–14, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Figure 4. Optimized FENIX Assay Conditions Yields Kinetic Data that Predicts the Potency of RTAs to Subvert Ferroptosis

(

A) The DTUN radical initiator.

B) Puriﬁcation of liposomes by size-exclusion chromatography following initiator additions removed MeOAMVN, but not DTUN, indicating that MeOAMVN is not

efﬁciently incorporated into the liposome.

C) Representative co-autoxidations of STY-BODIPY (1 mM)-embedded liposomes of egg PC (1 or 20 mM) initiated by DTUN (0.2 mM, black) and inhibited by

either PMC (4 mM, red), phenoxazine (4 mM, blue), ascorbic acid (10 mM, cyan), glutathione (10 mM, green), or NAC (50 mM, magenta).

(

2

(

D) Correlation of inhibition rate constants determined in liposomes and EC50 values from RSL3-induced ferroptosis in Pfa1 MEFs (R = 0.84). Compounds below

the dashed line did not inhibit ferroptosis up to 10 mM.

E) Model of liposome and associated kinetics accounting for the difference in reactivity of MeOAMVN- and DTUN-derived initiating radicals. Data show mean ±

(

SD of n = 3 wells from a 96-well plate from one representative of three independent experiments.

See also Figure S3 and Table S1.

reside primarily in the interfacial region. Coextrusion of al., 1993) before the widespread adoption of MeOAMVN (Nogu-

MeOAMVN with egg PC liposomes followed by SEC resulted in chi et al., 1998). The commercial availability of MeOAMVN, its

similar rates of STY-BODIPY consumption to liposomes to which well-characterized kinetics of radical generation in multiple sys-

MeOAMVN was added directly (without SEC), suggesting that tems, and ease of handling (it is crystalline at room temperature)

MeOAMVN can be efﬁciently incorporated in this manner. How- has led to its adoption as the initiator of choice for lipid peroxida-

ever, initiator addition after extrusion is preferable to prevent tion studies. Since hyponitrites decompose to yield alkoxyl rad-

autoxidation during liposome formation, handling, and storage icals, which are the initiating radicals when lipid hydroperoxides

(

of refrigerated stock liposome suspensions).

decompose, we reasoned that a lipophilic analog of DTBN may

be more suitable for controlled initiation of lipid peroxidation in

liposomes. Therefore, we synthesized di-tert-undecyl hyponi-

trite, hereafter DTUN (Figure 4A), by alkylation of silver hyponi-

Use of a Lipophilic Hyponitrite Initiator Completes a

Predictive Assay

Hyponitrites, such as di-tert-butyl hyponitrite (DTBN), were often trite with tert-undecyl iodide (see Supplemental Information for

used to initiate lipid autoxidation (Barclay et al., 1986; Kigoshi et details, Figure S3A). DTUN was conﬁrmed to be an efﬁcient

Cell Chemical Biology 26, 1–14, November 21, 2019

7

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

initiator (see Figures S3B–S3D); the inhibition periods of PMC-in- infra); Snelgrove et al., 2001.) Taking into account the relative

hibited autoxidations of 1-hexadecene yielded rates of radical concentration of PMC (4 mM) and polyunsaturated lipid

ꢁ

6

ꢁ1

initiation from which ek

was within experimental error of the value of 6.7 3 10

we found for DTBN. The value for MeOAMVN under the same DTUN-derived alkoxyl radicals will react with lipid 30-fold faster

d

= 8.3 3 10

s

was derived—which (150 mM; linoleate makes up ca. 15% of the lipid in egg phos-

ꢁ6

ꢁ1

s

phatidylcholine, which is 1 mM), the kinetics predict that the

ꢁ

5

ꢁ1

conditions was measured to be 2.4 3 10

ment with the literature (Figure S3E) (Noguchi et al., 1998). In

s

, in good agree- than with PMC.

ꢁ

6

ꢁ1

, expectedly slower than in solu-

liposomes, ek

d

= 1.2 3 10

s

H-Bonding of RTAs and Phospholipid Headgroups

Attenuates Reactivity in a Predictive Fashion

tion due to the increased microviscosity of the medium, and

ꢁ1

s ) (Figures

ꢁ6

similar to that measured for MeOAMVN (1.0 3 10

The signiﬁcant improvement in the cell potency versus RTA ac-

tivity correlation observed when the kinetics were determined

in liposomes instead of solution prompted us to consider the

origin of this difference in more detail. We anticipated that it

must be related to the dramatic suppression of reactivity

observed upon moving from solution to liposomes, e.g., for

S3F and S3G). Interestingly, like MeOAMVN, DTBN was not

efﬁciently incorporated into liposomes, because we found sub-

stantially less consumption of STY-BODIPY in DTBN-treated li-

posomes that were subsequently puriﬁed by SEC (see Figures

S3H–S3J). In contrast, DTUN-treated liposomes oxidized to a

similar extent before or after puriﬁcation (Figure 4B). The rate

constant for the decomposition of DTBN is too low to measure

under our reaction conditions (Figure S3L). Addition of DTBN

as an aliquot or before liposome extrusion results in immediate

precipitation due to limited solubility in buffer. Previous experi-

ments by Barclay et al., 1986 and Noguchi et al. (1998) were

carried out in the presence of multilamellar (PLPC or DLPC) lipo-

somes, where both DTBN and lipids were vortexed together in

buffer to generate liposomes. This was likely done to ensure

that the liposomes are formed around DTBN, effectively incorpo-

rating it into the lipid bilayer. However, this procedure results in

multilamellar liposomes with a large polydispersity index, making

standardization and reproduction of data very challenging.

DTUN-initiated co-autoxidations of STY-BODIPY and egg

phosphatidylcholine were carried out initially with the same

6

ꢁ1 ꢁ1

4

ꢁ1 ꢁ1

PMC, kinh = 3.8 3 10 M

s

and 2.9 3 10 M s in chloro-

benzene and egg PC liposomes, respectively. Although other

suggestions have been made to account for this observation

(

vide infra), we hypothesized that H-bonding between the RTA

and the phospholipid head group renders it unreactive to propa-

gating lipid peroxyl radicals (Figure 5A).

Ingold has clearly shown that H-bonding can slow the rate of

H-atom transfer reactions because, while engaged in an

H-bond, the H-atom is inaccessible to abstraction by radical spe-

cies (Figure 5B) (Litwinienko and Ingold, 2007; Snelgrove et al.,

2001). Thus, the kinetics of H-atom transfer in H-bonding media

follow a simple pre-dissociation model and can be expressed

as in Equation 3 in Figure 5B. This expression is often shown as

a linear free energy relationship, wherein the rate constant of the

S

‘

‘training set’’ of compounds from Figures 3E and 3F under H-atom transfer in a given solvent ðkin hÞ is related to the rate con-

0

in

both chain and non-chain conditions (Figure 4C). In contrast to stant in a non-H-bonding solvent ðk Þ by a term composed of

h

H

the results of the AAPH- and MeOAMVN-initiated autoxidations, the product of Abraham’s H-bond acidity ða Þ (Abraham et al.,

2

the water-soluble RTAs did not inhibit autoxidation to a signiﬁ-

cant extent under either set of conditions. These results are

similar to those obtained from MeOAMVN-initiated autoxida-

tions carried out under chain conditions. Autoxidations under

chain conditions, which require a large amount of lipid, are

both expensive and more challenging to carry out due to the

increased optical density of the solution, which lowers signal

quality. Since the foregoing suggested that the RTA kinetics

determined from the DTUN-initiated autoxidations are effectively

independent of chain length (lipid concentration), we assayed

the full test set of compounds in 1 mM PC liposomes. The addi-

tion of DTUN does not affect the liposomal particle size (see Fig-

ure S3K). The correlation of the kinh values determined under

these conditions and the EC50 values obtained in cells was

H

2

1

989) and basicity ðb Þ (Abraham et al., 1990) parameters of the

H-atom donor and solvent, respectively (Equation 4) (Snelgrove

et al., 2001).

The H-bond accepting ability of egg phosphatidylcholine was

19

determined by the F NMR method devised by Gurka and Taft

(

1969) (Equation 5, Figure S4A), who showed that the chemical

shift difference between 4-ﬂuorophenol and 4-ﬂuoroanisole in

a non-H-bonding solvent (e.g., CCl ) is related to the formation

4

constant of the H-bonded dimer between 4-ﬂuorophenol and

an added H-bond acceptor (Figure 5C). This can subsequently

H

be recast in terms of b₂to facilitate comparison with other

H-bond acceptors without needing to remeasure their individual

equilibrium constants. The resultant value of is quite large—for

H

2

comparison, b2 = 0.38 and 0.45 for water and an ester, respec-

excellent (Figure 4D, R = 0.84).

tively—the other possible H-bond acceptors in the liposome

H

The fact that DTUN-initiated autoxidations yield similar data

at either high or low concentrations of lipid can be rationalized

on the basis of available kinetic data (Figure 4E). Using PMC to

illustrate this point, alkoxyl radicals abstract bis-allylic H-atoms

^s^y^s^t^e^m^.^M^o^r^e^o^v^e^r^t^h^e^d^e^t^e^r^mⁱⁿ^e^d^b2 is consistent with that of

a phosphate triester (0.77), as reported by Abraham et al.

(1990). This value, when used with Equation 4 in Figure 5B, pre-

6

ꢁ1 ꢁ1

dicts a drop in kinh ^f^o^r^P^M^C^ð^a ^H2 = 0:38Þ on going from chloro-

with k = 9.0 3 10 M

stract an H-atom from PMC with k = 1.1 3 10 M

s

(Barclay et al., 1989), while they ab-

7

ꢁ1 ꢁ1

H

s

. (The benzene ðb₂= 0:09Þ to egg PC bilayers ðb₂= 0:79Þ of 161-

rate constant for reactions of PMC with alkoxyl radicals in phos- fold, in very good agreement with the experimental observation

pholipid bilayers was estimated using the rate constant ob- of a 213-fold decrease in reactivity.

tained in chlorobenzene from the following reference and the

H

To demonstrate how profoundly the H-bonding interaction be-

Ingold-Abraham equation [Equation 4], assuming the a of tween an RTA and phospholipid head group can affect RTA ki-

2

H

PMC = 0.38 and b₂of egg phosphatidylcholine of 0.79 (vide netics, we explored the medium dependence of the reactivity

8

Cell Chemical Biology 26, 1–14, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Figure 5. Hydrogen-Bonding Interactions Dramatically Suppresses the Reactivity of RTAs in Phospholipid Bilayers

(

A) Formation of H-bonded complexes between a-tocopherol and the phosphatidylcholine head group are proposed to slow its reaction with peroxyl radicals in

lipid bilayers.

B) The equilibrium between ‘‘free’’ RTA (A–H) and RTA participating in a 1:1 hydrogen bond complex, with hydrogen bond accepting solvent S. Assuming the 1:1

complex does not react with peroxyl radicals, the kinetics of radical-trapping can be described as in Equations 3 or 4 (Snelgrove et al., 2001).

C) The equilibrium between free 4-ﬂuorophenol and 4-ﬂuorophenol participating in a 1:1 hydrogen bond complex with the head group of phosphatidylcholine

(

19

4

along with representative F NMR spectra of 4-ﬂuorophenol and 4-ﬂuoroanisole (0.01 M) in CCl in the presence of none (top) or 30 mM (bottom) egg PC used to

H

2

derive b of the phospholipid. Equations 5 and 6 can be used to determine the b , where A

0

is the concentration of 4-ﬂuorophenol, B

0

is the concentration of egg

2

PC, D is the difference in chemical shift seen between 4-ﬂuorophenol and 4-ﬂuoroanisole at a concentration of HBA saturation (B

0

= 0.20 M), and d is the dif-

ference in chemical shift seen between 4-ﬂuorophenol and 4-ﬂuoroanisole at the dilute concentrations of egg PC.

1

D and E) (D) Structures of differently substituted phenols along with their calculated ( H NMR) and (E) predicted a values.

H

(

2

F) Correlation of logk^Sand b₂for compounds 13 ( ), 14 ( ), 15 ( ), 16 ( ), 17 ( ), and 18 ( ) in solvents with increasing hydrogen bond basicity.

H

inh

G) Predicted and measured EC50 values from RSL3-induced ferroptosis in Pfa1 MEFs for phenols 13–18. Predicted EC50 values were obtained using the line of

best ﬁt from Figure 4C: |logEC50| = 1.529 (logkinh)–0.0509.

Data represent mean ± SD from three independent experiments. See also Figures S4 and S5.

of a group of structurally related phenols (Figure 5D) that differ cal Properties Predictor to be: 1.5 (13), 2.5 (14), 2.5 (15), 3.3 (16),

slightly in their inherent H-atom transfer reactivity, but, due to 4.0 (17), and 4.9 (18)), which ensures that the reactivity differ-

differing steric environments of the reactive phenolic H-atom, ences do not derive simply from different partitioning between

exhibit substantial differences in their H-bond acidity (a^H₂ range the lipid and aqueous phases. The results are shown in Figure 5F

from 0.59 to 0.16; Figure 5E; Figures S5A–S5I). All of the (see Figure S4B for data).

compounds are lipophilic with logP > 1.5 (the logP values for

The medium effects are dramatic—and clearly depend linearly

these compounds were estimated using the ChemDraw Chemi- on the b2 of the solvent as predicted by Ingold (see Figure S4B).

H

Cell Chemical Biology 26, 1–14, November 21, 2019

9

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Moreover, they clearly also depend on the H-bond donating abil- can be rendered more informative by determining the kinetics

H

ity of the phenol a . For example, while 4-methoxyphenol of the reaction, but this generally requires specialized equipment

H

2

H

ða = 0:59Þ and 2,6-dimethyl-4-methoxyphenol ða₂= 0:42Þ (a stopped-ﬂow spectrophotometer) (Foti et al., 2008). More-

2

differ in reactivity by a factor of ꢂ3.5 in chlorobenzene, the over, because these measurements are carried out in homoge-

difference in egg PC liposomes is nearly 30-fold. According to neous solution, and the identity of the solvent can dramatically

H

Equation 4, the slopes of the log kinh=b₂correlations should inﬂuence both the rates and the mechanism of the reaction (Lit-

H

2

H

correspond to ꢁ8.3a . Indeed, we generally ﬁnd good agree- winienko and Ingold, 2003, 2007), it is unlikely that the results will

ment between the a₂values derived from the slopes and the be predictive of what transpires in a cell or tissue.

values we determined directly by NMR spectroscopy (Figure 5E), Our recently implemented co-autoxidation approach using

demonstrating the robustness of Ingold’s relationship. STY-BODIPY as a signal carrier served as an ideal starting

Interestingly, while the kinetic data obtained for the unhin- point for the implementation of a far more predictive alternative

dered phenols 13, 14, and 15 in liposomes fall on the antioxidant activity assay to DPPH (and presumably related ex-

H

log kinh=b₂correlation, as the bulk around the phenolic O-H in- isting methods, although this was not rigorously investigated).

creases further as in 16, 17, and 18, the deviation from the cor- It should be pointed out that the assay can also be carried out

581/591

relation increases. Indeed, the most hindered phenol (18) is with the commercially available C11-BODIPY

cellular lipid

almost as reactive in the egg PC liposomes as in chloroben- peroxidation reporter (Drummen et al., 2002)—which served as

zene—a stark difference from the dramatic drop in reactivity of the inspiration for the initial development of PBD-BODIPY—

the unhindered compounds in the different media (ꢂ3 orders of and which yields essentially indistinguishable results (see Sup-

magnitude). This represents the limitation of this very useful plemental Information for examples, Figures S2C, S2D, and

linear free energy relationship: because the a^H₂ of the phenols S2F). Our efforts to generate a model system capable of predict-

were determined using an unhindered H-bond acceptor ing the potency of RTAs in cells highlights the importance of

H

(

DMSO) and the b of PLPC was determined with an unhindered selecting both an appropriate reaction environment and an

2

H-bond donor (4-ﬂuorophenol), approximation of the H-bonding appropriate radical initiator. Indeed, the results of experiments

equilibrium between a hindered H-bond donor and a hindered initiated by MeOAMVN, the conventional choice as a lipid-solu-

H-bond acceptor is likely to be overestimated. Nevertheless, ble initiator, fail to provide a good correlation with cell-based

as expected based on the data in Figure 4C, the kinetics deter- results—yielding data which overestimate the reactivity of wa-

mined in liposomes using the DTUN initiator correctly predict the ter-soluble or amphiphilic RTAs. This is a particularly signiﬁcant

potency of these compounds in cells (Figure 5G; Figure S5J).

problem at low concentrations of lipid, where there is no radical

chain reaction, such that the added RTAs simply react with initi-

ator-derived radicals (Figure 3G). When the concentration of lipid

is increased such that a radical chain reaction takes place in the

DISCUSSION

Lipid peroxidation has been implicated in disease pathogen- absence of RTA, uncertainty remains over the precise location of

esis for decades, but attempts to identify, design, or develop the somewhat amphiphilic—and not at all lipid-like—initiator.

compounds to target aberrant lipid peroxidation for therapeu- Therefore, some apparent inhibition of lipid peroxidation by wa-

tic or preventive purposes have largely failed. This has promp- ter-soluble or amphiphilic RTAs may still derive simply from

ted many to believe that lipid peroxidation is not a part of direct reactions with initiator-derived radicals. This is implied

disease etiology, but instead simply a hallmark of pathogen- from the poor correlation of water-soluble RTA activity and

esis. Alternatively, it could be argued that the antioxidant anti-ferroptotic activity (Figure 3H).

compounds that have been the subject of the great multitude

of clinical trials aimed at suppressing ‘‘radical damage,’’ such lipid concentration; it decreases with higher concentrations of

as dietary (poly)phenols, were not good choices, simply lipid where the lipid competes with it for chain-propagating per-

p

The rate of consumption of STY-BODIPY (R ) depends on the

convenient ones.

Lipids are the primary cellular target of radical damage due to as the difference in R

both their high reactivity toward H-atom abstraction and their MeOAMVN to DTUN. The ratios of R

oxyl radicals. This indirectly reports on the efﬁciency of initiation

increases upon going from AAPH to

under chain to non-chain

p

high local concentration in lipid bilayers, which facilitate propa- conditions are ꢂ2, 7, and 8 for AAPH, MeOAMVN and DTUN,

gation of the autoxidative chain reaction. Therefore, assessment respectively. The newly developed lipophilic hyponitrite initiator

of putative antioxidants should be made on the basis of their abil- DTUN addresses the shortcomings of the assay based upon

ity to compete with lipids for peroxyl radicals in lipid bilayers. Un- MeOAMVN. The lengthy hydrocarbon side chains ensure that

fortunately, this is not common. Instead, surrogate experiments radical generation occurs exclusively in the lipid, and the greater

have been developed to assess antioxidant activity, such as the reactivity of the alkoxyl radicals derived therefrom ensures they

DPPH assay (Amorati and Valgimigli, 2018; Huang et al., 2005). react with the lipids, which are in far greater abundance than

Although the DPPH radical is isoelectronic with peroxyl radicals, the RTA (Figure 4E). Therefore, the inhibition observed in the

the IC50 values obtained from this assay are clearly a poor pre- presence of added RTAs—and apparent kinh derived there-

dictor of the ability of a given compound to inhibit lipid peroxida- from—results from their reaction with lipid peroxyl radicals.

tion in cells (cf. Figure 1G). Indeed, in addition to being the wrong These advantages are clearly borne out in the excellent correla-

type of radical, the kinetics of H-atom transfer from the antioxi- tion between the logkinh values determined using these condi-

dant to the radical are not determined; the IC50 values obtained tions and the potency of the compounds in cells (Figure 4C). A

from this assay simply reﬂect the position of the H-atom transfer further advantage of the use of DTUN to initiate lipid peroxidation

equilibrium between DPPH and the test compound. The assay is its mechanistic similarity to what presumably takes place

10 Cell Chemical Biology 26, 1–14, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Figure 6. FENIX Resolves RTA and Non-RTA Inhibitors of Ferroptosis

(A) Structures of various classes of ferroptosis inhibitors and (B) their potency to inhibit RSL3-induced ferroptosis in Pfa1 MEFs and reactivity as inhibitors of lipid

peroxidation overlaid on the correlation of EC50 and kinh from Figure 4C. Compounds on the dotted line did not inhibit liposome autoxidation. Data represent

mean ± SD from three independent experiments.

during ferroptosis, wherein (phospho)lipid hydroperoxides that

At ﬁrst glance, moving from homogeneous organic solution

escape detoxiﬁcation by GPX4 can undergo Fenton-type reac- to lipid bilayers to assay RTA activity seems like an obvious

tions with low-valent iron (or other one-electron reductants). and necessary step to obtain data that would be useful in pre-

Although different in the structure of the alkyl side chains, the dicting potency in cells. Barclay ﬁrst quantitated the difference

lipid-derived alkoxyl radicals that result are expected to have in the rate constants for the reaction between RTAs and peroxyl

essentially identical reactivity to the alkoxyl radicals, which result radicals upon moving from organic solution to lipid bilayers

from the decomposition of the hyponitrite.

(Barclay et al., 1989, 1990, 1999). He, and others, generally

In addition to offering a more rigorous high-throughput alterna- attributed this observation to one of three factors: (1) the local-

tive to existing antioxidant activity assays, this approach enables ization of the RTA, particularly its partitioning between the lipid

researchers to reliably rule out RTA activity when attempting to and aqueous phase, (2) increased van der Waals interaction be-

elucidate the mode of action of compounds which protect tween aliphatic side chains on the RTA and the lipids, slowing

against ferroptosis or related forms of oxidative cell death (e.g., their encounter with propagating radicals (Niki and Noguchi,

oxytosis). A recent example that demonstrates this point relates 2004), and (3) H-bonding between the RTA and water at the

to the role of lipoxygenase-catalyzed lipid peroxidation in ferrop- lipid-aqueous interface. The importance of localization is abun-

tosis. Lipoxygenases (LOXs) had been implicated in ferroptosis dantly clear from the low reactivity of water-soluble compounds

based largely upon the observation that LOX inhibitors are able (2, 3, 5, and 12) toward lipid peroxyl radicals. The role of dy-

to subvert ferroptosis. Characterization of the RTA activity of namics is evident upon comparing the reactivity of compounds

these LOX inhibitors enabled us to conclude that LOX activity that differ only in alkyl substitution, e.g., 1 and PMC: despite

could only sensitize cells to ferroptosis, since LOX inhibitors lack- having similar reactivity in solution, PMC has a 2.6-fold higher

ing RTA activity had no effect on cell survival—even in LOX-over- reactivity in lipid bilayers than 1. Although H-bonding between

expressing cells (Shah et al., 2018). Indeed, when the kinh is the RTA and water is insufﬁcient to explain the drop in reactivity,

determined for these redox-active LOX inhibitors (NDGA, Zileu- we have now shown that H-bonding to the phospholipid head

ton, and PD146176) using the optimized assay protocol which af- group is far stronger, and accounts for the massive drop in

forded the EC50 versus logkinh correlation in Figure 4, we ﬁnd that reactivity upon moving from solution to liposomes. These re-

these data also lie on the correlation (Figure 6)—fully consistent sults highlight an element that is overlooked in the design of

with our assertion that they protect cells from ferroptosis simply RTAs and/or anti-ferroptotic agents: H-bond acidity. Further-

as RTAs (Shah et al., 2018). In addition to LOX inhibitors, we also more, it enables the prediction of the cytoprotective potencies

assayed the nitroxides lipo-TEMPO and phenoxazine-N-oxyl of putative RTAs from the results of inhibited autoxidations or

(

PHOXNO) (Griesser et al., 2018), since nitroxides have been organic substrates in homogeneous organic solvents—no cell

suggested to inhibit ferroptosis via a mechanism involving mito- culture required! H-Bond acidity underlies the lackluster activity

chondrial ROS (Krainz et al., 2016). Again, we ﬁnd these to be of phenols, the quintessential naturally derived antioxidants of

competent RTAs, with kinh values that correlate with their po- much fanfare, as RTAs in a cellular context, and may contribute

tencies in cells, implying that they act merely as RTAs to prevent to the disappointing results of phenolic antioxidants in the many

ferroptosis. In contrast, when other known ferroptosis inhibitors, clinical trials in which they have been examined as potential

such as the iron chelators desferrioxamine (DFO) and ciclopirox- therapeutic or preventive agents. Moreover, it underpins the

amine (CPX), the deuterated polyunsaturated fatty acid identiﬁcation of aromatic amines, such as Fer-1 (Dixon et al.,

d

6

-arachidonic acid and a thiazolidinedione ACSL4 inhibitor, 2012) and Lip-1 (Friedmann Angeli et al., 2014) in high-

were assayed, no RTA activity could be detected, consistent throughput screens of compound libraries for inhibitors of

ferroptosis. Despite the fact that they are far less inherently

with the fact that they act by other mechanisms.

Cell Chemical Biology 26, 1–14, November 21, 2019 11

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

reactive than many phenols (including a-tocopherol), aromatic

amines are simply far less H-bond acidic than phenols. It will

be interesting to see how compounds which are excellent

RTAs under relevant conditions (including cell models of ferrop-

tosis) fare in clinical trials.

SUPPLEMENTAL INFORMATION

Supplemental Information can be found online at https://doi.org/10.1016/j.

chembiol.2019.09.007 .

ACKNOWLEDGMENTS

SIGNIFICANCE

This work was supported by grants from the Natural Sciences and Engineering

Research Council of Canada (NSERC) and the Canada Foundation for Innova-

tion. R.S. and A.T.M.V.K. acknowledge NSERC for support via post-graduate

and undergraduate scholarship programs, respectively.

Quantiﬁcation of ‘‘antioxidant activity’’ is thought by many

to be a routine experiment. However, investigators gener-

ally employ surrogate ‘‘antioxidant assays’’ which do not

provide the kinetic information necessary to provide truly

meaningful insight on the reactivity of putative antioxi-

dants. Despite being the most popular of these assays by

a considerable margin, the DPPH assay is not at all predic-

tive of the potency of ferroptosis inhibitors – and should not

be used to inform on radical-trapping antioxidant activity in

cellular contexts. Instead, we show that the newly-devel-

oped ﬂuorescence-enabled inhibited autoxidation (FENIX)

assay is entirely predictive of cell potency. As such, results

obtained from this assay can be used to provide insight on

the mechanisms of ferroptosis modulators and which

properties are important in the design of novel inhibitors

of lipid peroxidation and/or ferroptosis. We expect this

approach to be of very broad interest since the inhibition

of lipid peroxidation is not only key to the inhibition of

oxidative cell death, but also in various other pathological

contexts. One of the key insights derived from the valida-

tion of FENIX is that phenols are, in general, relatively

poor inhibitors of (phospho)lipid peroxidation, and that

H-bonding between the phenol and the phospholipid

headgroup is primarily responsible. This phenomenon

may underlie the failure of the hundreds of clinical trials

carried out with phenolic antioxidants.

AUTHOR CONTRIBUTIONS

D.A.P. designed and directed the research, and R.S., L.F., O.Z. performed the

experimental work and contributed to project design. A.T.M.V.K. contributed

preliminary experimental results. R.S., L.A.F., and D.A.P. wrote the

manuscript.

DECLARATION OF INTERESTS

There are no competing interests.

Received: July 30, 2019

Revised: August 21, 2019

Accepted: September 9, 2019

Published: September 26, 2019

REFERENCES

Abraham, M.H., Grellier, P.L., Prior, D.V., Duce, P.P., Morris, J.J., and Taylor,

P.J. (1989). Hydrogen bonding. Part 7. A scale of solute hydrogen-bond acidity

based on log K values for complexation in tetrachloromethane. J. Chem. Soc.

Perkin Trans. 2 , 699–711.

Abraham, M.H., Grellier, P.L., Prior, D.V., Morris, J.J., and Taylor, P.J. (1990).

Hydrogen bonding. Part 10. A scale of solute hydrogen-bond basicity using log

K values for complexation in tetrachloromethane. J. Chem. Soc. Perkin Trans.

2

, 521.

Amorati, R., and Valgimigli, L. (2015). Advantages and limitations of common

testing methods for antioxidants. Free Radic. Res. 49 , 633–649.

STAR+METHODS

Amorati, R., and Valgimigli, L. (2018). Methods to measure the antioxidant

activity of phytochemicals and plant extracts. J. Agric. Food Chem. 66 ,

3324–3329.

Detailed methods are provided in the online version of this paper

and include the following:

Barclay, L.R.C., Baskin, K.A., Locke, S.J., and Vinqvist, M.R. (1989). Absolute

rate constants for lipid peroxidation and inhibition in model biomembranes.

Can. J. Chem. 67 , 1366–1369.

d KEY RESOURCES TABLE

d LEAD CONTACT AND MATERIALS AVAILABILITY

d EXPERIMENTAL MODEL AND SUBJECT DETAILS

d METHOD DETAILS

Barclay, L.R.C., Baskin, K.A., Dakin, K.A., Locke, S.J., and Vinqvist, M.R.

(1990). The antioxidant activities of phenolic antioxidants in free radical perox-

idation of phospholipid membranes. Can. J. Chem. 68 , 2258–2269.

B DPPH Assay

Barclay, L.R.C.C., Edwards, C.E., and Vinqvist, M.R. (1999). Media effects on

antioxidant activities of phenols and catechols. J. Am. Chem. Soc. 99 ,

B Inhibited Autoxidation of 1-Hexadecene

B Inhibited Autoxidation of Egg-PC Liposomes on UV-Vis

B Inhibited Autoxidation of Egg-PC Liposomes on Plate

Reader

6

226–6231.

Barclay, L.R.C., Kong, D., and VanKessel, J. (1986). Autoxidation of phospha-

tidylcholines initiated with dicumylhyponitrite in solution and in bilayers. Can.

J. Chem. 64 , 2103–2108.

B Method to Calculate n and kinh

B Determination of a^H₂ by H NMR

B Inhibition of Ferroptosis Induced by Gpx4 Inhibition

with (1S,3R)-RSL3

1

Barnham, K.J., Masters, C.L., and Bush, A.I. (2004). Neurodegenerative dis-

eases and oxidative stress. Nat. Rev. Drug Discov. 3 , 205–214.

Benzie, I.F.F., and Strain, J.J. (1996). The ferric reducing ability of plasma

(

FRAP) as a measure of ‘‘antioxidant power’’: the FRAP assay. Anal.

B Determination of Kinetic Chain Length

Biochem. 239 , 70–76.

H

^B^D^e^t^e^r^mⁱⁿ^a^tⁱ^oⁿ^o^f^a2 by Kinetic Solvent Effect

Blois, M.S. (1958). Antioxidant determinations by the use of a stable free

H

19

^B^D^e^t^e^r^mⁱⁿ^a^tⁱ^oⁿ^o^f^b2 for PLPC by F NMR

B Synthesis and Characterization

radical. Nature 181 , 1199–1200.

Breslow, R., Groves, K., and Mayer, M.U. (2002). Antihydrophobic Cosolvent

Effects for Alkylation Reactions in Water Solution, Particularly Oxygen versus

Carbon Alkylations of Phenoxide Ions. J. Am. Chem. Soc. 124 , 3622–3635.

d QUANTIFICATION AND STATISTICAL ANALYSIS

d DATA CODE AND AVAILABILITY

12

Cell Chemical Biology 26, 1–14, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Burton, G.W., and Ingold, K.U. (1986). Vitamin-E––application of the principles

of physical organic-chemistry to the exploration of its structure and function.

Acc. Chem. Res. 19 , 194–201.

Huang, D., Boxin, O.U., and Prior, R.L. (2005). The chemistry behind antioxi-

dant capacity assays. J. Agric. Food Chem. 53 , 1841–1856.

Ingold, K.U. (1961). Inhibition of the autoxidation of organic substances in the

Cao, G., Alessio, H.M., and Cutler, R.G. (1993). Oxygen-radical absorbance

liquid phase. Chem. Rev. 61 , 563–589.

capacity assay for antioxidants. Free Radic. Biol. Med. 14 , 303–311.

Ingold, K.U., and Pratt, D.A. (2014). Advances in radical-trapping antioxidant

chemistry in the 21st century: a kinetics and mechanisms perspective.

Chem. Rev. 114 , 9022–9046.

Chen, Z., Bertin, R., and Froldi, G. (2013). EC 50 estimation of antioxidant activ-

ity in DPPH* assay using several statistical programs. Food Chem. 138 ,

4

14–420.

Conrad, M., and Friedmann Angeli, J.P. (2015). Glutathione peroxidase 4

Gpx4) and ferroptosis: what’s so special about it? Mol. Cell. Oncol. 2 , e995047.

Kedare, S.B., and Singh, R.P. (2011). Genesis and development of DPPH

method of antioxidant assay. J. Food Sci. Technol. 48 , 412–422.

(

Kigoshi, M., Sato, K., and Niki, E. (1993). Oxidation of lipids induced by diocta-

decyl hyponitrite and di-t-butyl hyponitrite in organic solution and in aqueous

dispersions. Effects of reaction medium and size of radicals on efﬁciency of

chain initiation. Bull. Chem. Soc. Jpn. 66 , 2954–2959.

Cos, P., Hermans, N., Calomme, M., Maes, L., De Bruyne, T., Pieters, L.,

Vlietinck, A.J., and Vanden Berghe, D. (2003). Comparative study of eight

well-known polyphenolic antioxidants. J. Pharm. Pharmacol. 55 , 1291–1297.

Denison, G.H., and Harle, O.L. (1949). Oxidation of lubricating oils at high tem-

Krainz, T., Gaschler, M.M., Lim, C., Sacher, J.R., Stockwell, B.R., and Wipf, P.

(2016). A mitochondrial-targeted nitroxide is a potent inhibitor of ferroptosis.

ACS Cent. Sci. 2 , 653–659.

peratures. Ind. Eng. Chem. 41 , 934–937.

Di Paolo, G., and Kim, T.-W. (2011). Linking lipids to Alzheimer’s disease:

cholesterol and beyond. Nat. Rev. Neurosci. 12 , 284–296.

Krumova, K., Friedland, S., and Cosa, G. (2012). How lipid unsaturation, per-

oxyl radical partitioning, and chromanol lipophilic tail affect the antioxidant ac-

tivity of a -tocopherol: direct visualization via high-throughput ﬂuorescence

studies conducted with ﬂuorogenic a -tocopherol analogues. J. Am. Chem.

Soc. 134 , 10102–10113.

Dixon, S.J., Lemberg, K.M., Lamprecht, M.R., Skouta, R., Zaitsev, E.M.,

Gleason, C.E., Patel, D.N., Bauer, A.J., Cantley, A.M., Yang, W.S., et al.

(2012). Ferroptosis: an iron-dependent form of nonapoptotic cell death. Cell

149 , 1060–1072.

Li, B., and Pratt, D.A. (2015). Methods for determining the efﬁcacy of radical-

Doll, S., Proneth, B., Tyurina, Y.Y., Panzilius, E., Kobayashi, S., Ingold, I.,

Irmler, M., Beckers, J., Aichler, M., Walch, A., et al. (2017). ACSL4 dictates fer-

roptosis sensitivity by shaping cellular lipid composition. Nat. Chem. Biol.

trapping antioxidants. Free Radic. Biol. Med. 82 , 187–202.

Li, B., Harjani, J.R., Cormier, N.S., Madarati, H., Atkinson, J., Cosa, G., and

Pratt, D.A. (2013). Besting vitamin E: sidechain substitution is key to the reac-

tivity of naphthyridinol antioxidants in lipid bilayers. J. Am. Chem. Soc. 135 ,

1

3 , 91–98.

Drummen, G.P.C., Van Liebergen, L.C.M., Op den Kamp, J.A.F., and Post, J.A.

2002). C11-BODIPY581/591, an oxidation-sensitive ﬂuorescent lipid peroxi-

1

394–1405.

(

dation probe: (micro)spectroscopic characterization and validation of method-

Litwinienko, G., and Ingold, K.U. (2003). Abnormal solvent effects on hydrogen

atom abstractions. 1. The reactions of phenols with 2,2-diphenyl-1-picrylhy-

d

ology. Free Radic. Biol. Med. 33 , 473–490.

drazyl (DPPH ) in alcohols. J. Org. Chem. 68 , 3433–3438.

Dryzhakov, M., Hellal, M., Wolf, E., Falk, F.C., and Moran, J. (2015). Nitro-as-

sisted Brønsted acid catalysis: application to a challenging catalytic azidation.

J. Am. Chem. Soc. 137 , 9555–9558.

Litwinienko, G., and Ingold, K.U. (2007). Solvent effects on the rates and mech-

anisms of reaction of phenols with free radicals. Acc. Chem. Res. 40 , 222–230.

Finkel, T., and Holbrook, N.J. (2000). Oxidants, oxidative stress and the

Liu, Y., Wang, W., Li, Y., Xiao, Y., Cheng, J., and Jia, J. (2015). The 5-lipoxyge-

nase inhibitor zileuton confers neuroprotection against glutamate oxidative

damage by inhibiting ferroptosis. Biol. Pharm. Bull. 38 , 1234–1239.

biology of ageing. Nature 408 , 239–247.

d

Foti, M.C. (2015). Use and abuse of the DPPH radical. J. Agric. Food Chem.

6

3 , 8765–8776.

Lonn, E. (2005). Effects of long-term vitamin E supplementation on cardiovas-

cular events and cancer: a randomized controlled trial. J. Am. Med. Assoc.

293 , 1338–1347.

Foti, M.C., Daquino, C., Mackie, I.D., DiLabio, G.A., and Ingold, K.U. (2008).

Reaction of phenols with the 2,2-diphenyl-1-picrylhydrazyl radical. Kinetics

and DFT calculations applied to determine ArO-H bond dissociation enthalpies

and reaction mechanism. J. Org. Chem. 73 , 9270–9282.

Magtanong, L., Ko, P.J., To, M., Cao, J.Y., Forcina, G.C., Tarangelo, A., Ward,

C.C., Cho, K., Patti, G.J., Nomura, D.K., et al. (2019). Exogenous monounsat-

urated fatty acids promote a ferroptosis-resistant cell state. Cell Chem. Biol.

Frank, C.E. (1950). Hydrocarbon autoxidation. Chem. Rev. 46 , 155–169.

2

6 , 420–432.e9.

Friedmann Angeli, J.P., Schneider, M., Proneth, B., Tyurina, Y.Y., Tyurin, V.A.,

Hammond, V.J., Herbach, N., Aichler, M., Walch, A., Eggenhofer, E., et al.

Matsuura, B.S., Keylor, M.H., Li, B., Lin, Y., Allison, S., Pratt, D.A., and

Stephenson, C.R.J. (2015). A scalable biomimetic synthesis of resveratrol di-

mers and systematic evaluation of their antioxidant activities. Angew. Chem.

Int. Ed. 54 , 3754–3757.

(

2014). Inactivation of the ferroptosis regulator Gpx4 triggers acute renal failure

in mice. Nat. Cell Biol. 16 , 1180–1191.

Gao, M., Monian, P., Quadri, N., Ramasamy, R., and Jiang, X. (2015).

Glutaminolysis and transferrin regulate ferroptosis. Mol. Cell 59 , 298–308.

McIntyre, T.M., and Hazen, S.L. (2010). Lipid oxidation and cardiovascular dis-

ease: introduction to a review series. Circ. Res. 107 , 1167–1169.

Goldschmidt, S., and Renn, K. (1922). Zweiwertiger Stickstoff: U¨ ber das a ,

a -diphenyl-b -trinitrophenyl-hydrazyl. (IV. Mitteilung u € ber Amin-Oxydation).

Berichte Der Dtsch. Chem. Gesellschaft A B Ser. 55 , 628–643.

Naidu, S.K., Klaus, E.E., and Duda, J.L. (1984). Evaluation of liquid phase

oxidation products of ester and mineral oil lubricants. Ind. Eng. Chem. Prod.

Res. Dev. 23 , 613–619.

Gray, J.I. (1978). Measurement of lipid oxidation: a review. J. Am. Oil Chem.

Soc. 55 , 539–546.

Niki, E. (2010). Assessment of antioxidant capacity in vitro and in vivo. Free

Radic. Biol. Med. 49 , 503–515.

Griesser, M., Shah, R., Van Kessel, A.T., Zilka, O., Haidasz, E.A., and Pratt,

D.A. (2018). The catalytic reaction of nitroxides with peroxyl radicals and its

relevance to their cytoprotective properties. J. Am. Chem. Soc. 140 ,

Niki, E., and Noguchi, N. (2004). Dynamics of antioxidant action of vitamin E.

Acc. Chem. Res. 37 , 45–51.

3

798–3808.

Noguchi, N., Yamashita, H., Gotoh, N., Yamamoto, Y., Numano, R., and Niki,

E. (1998). 2,2’-Azobis (4-methoxy-2,4-dimethylvaleronitrile), a new lipid-solu-

ble azo initiator: application to oxidations of lipids and low-density lipoprotein

in solution and in aqueous dispersions. Free Radic. Biol. Med. 24 , 259–268.

Gurka, D., and Taft, R.W. (1969). Studies of hydrogen-bonded complex forma-

tion with p-ﬂuorophenol. IV. The ﬂuorine nuclear magnetic resonance method.

J. Am. Chem. Soc. 91 , 4794–4801.

Haidasz, E.A., Van Kessel, A.T.M., and Pratt, D.A. (2016). A continuous visible

light spectrophotometric approach to accurately determine the reactivity of

radical-trapping antioxidants. J. Org. Chem. 81 , 737–744.

Re, R., Pellegrini, N., Proteggente, A., Pannala, A., Yang, M., and Rice-Evans,

C. (1999). Antioxidant activity applying an improved ABTS radical cation decol-

orization assay. Free Radic. Biol. Med. 26 , 1231–1237.

Cell Chemical Biology 26, 1–14, November 21, 2019 13

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Ramirez, F., Chen, E.H., and Dershowitz, S. (1959). Reaction of Trialkyl

Phosphites with p-Benzoquinone and with Other Symmetrically Substituted

p-Quinones. A New Synthesis of Hydroquinone Monoalkyl Ethers. J. Am.

Chem. Soc. 81 , 4338–4342.

Stockwell, B.R., Friedmann Angeli, J.P., Bayir, H., Bush, A.I., Conrad, M.,

Dixon, S.J., Fulda, S., Gasc o´ n, S., Hatzios, S.K., Kagan, V.E., et al. (2017).

Ferroptosis: a regulated cell death nexus linking metabolism, redox biology,

and disease. Cell 171 , 273–285.

Seibt, T.M., Proneth, B., and Conrad, M. (2019). Role of GPX4 in ferroptosis

Viswanathan, V.S., Ryan, M.J., Dhruv, H.D., Gill, S., Eichhoff, O.M., Seashore-

Ludlow, B., Kaffenberger, S.D., Eaton, J.K., Shimada, K., Aguirre, A.J., et al.

(2017). Dependency of a therapy-resistant state of cancer cells on a lipid

peroxidase pathway. Nature 547 , 453–457.

and its pharmacological implication. Free Radic. Biol. Med. 133 , 144–152.

Seiler, A., Schneider, M., F o¨ rster, H., Roth, S., Wirth, E.K., Culmsee, C.,

Plesnila, N., Kremmer, E., R a˚ dmark, O., Wurst, W., et al. (2008). Glutathione

peroxidase 4 senses and translates oxidative stress into 12/15-lipoxygenase

dependent- and AIF-mediated cell death. Cell Metab. 8 , 237–248.

Weiwer, M., Bittker, J.A., Lewis, T.A., Shimada, K., Yang, W.S., MacPherson,

L., Dandapani, S., Palmer, M., Stockwell, B.R., Schreiber, S.L., et al. (2012).

Development of small-molecule probes that selectively kill cells induced to ex-

press mutant RAS. Bioorg. Med. Chem. Lett. 22 , 1822–1826.

Semba, R.D., Ferrucci, L., Bartali, B., Urp ı´ -Sarda, M., Zamora-Ros, R., Sun, K.,

Cherubini, A., Bandinelli, S., and Andres-Lacueva, C. (2014). Resveratrol levels

and all-cause mortality in older community-dwelling adults. JAMA Intern. Med.

Xie, J., and Schaich, K.M. (2014). Re-evaluation of the 2,2-diphenyl-1-picrylhy-

drazyl free radical (DPPH) assay for antioxidant activity. J. Agric. Food Chem.

1

74 , 1077–1084.

Sesso, H.D., Christen, W.G., Bubes, V., Smith, J.P., MacFadyen, J., Schvartz,

M., Manson, J.A.E., Glynn, R.J., Buring, J.E., and Gaziano, J.M. (2012).

Multivitamins in the prevention of cardiovascular disease in men: the physi-

cians’ health study II randomized controlled trial. J. Am. Med. Assoc. 308 ,

6

2 , 4251–4260.

Yang, W.S., and Stockwell, B.R. (2008). Synthetic lethal screening identiﬁes

compounds activating iron-dependent, nonapoptotic cell death in onco-

genic-RAS-harboring cancer cells. Chem. Biol. 15 , 234–245.

1751–1760.

Shah, R., and Pratt, D.A. (2016). Determination of key hydrocarbon autoxida-

Yin, H., Xu, L., and Porter, N.A. (2011). Free radical lipid peroxidation: mecha-

tion products by ﬂuorescence. J. Org. Chem. 81 , 6649–6656.

nisms and analysis. Chem. Rev. 111 , 5944–5972.

Shah, R., Margison, K., and Pratt, D.A. (2017). The potency of diarylamine

radical-trapping antioxidants as inhibitors of ferroptosis underscores the role

of autoxidation in the mechanism of cell death. ACS Chem. Biol. 12 ,

Zhao, W., Huang, L., Guan, Y., and Wulff, W.D. (2014). Three Component

Asymmetric Catalytic Ugi Reaction

Substrate Mediated Catalyst Assembly. Angew. Chem. Intl. Ed. 53 ,

436–3441.

– Concinnity from Diversity via

2538–2545.

3

Shah, R., Shchepinov, M.S., and Pratt, D.A. (2018). Resolving the role of lipox-

Zilka, O., Shah, R., Li, B., Friedmann Angeli, J.P., Griesser, M., Conrad, M., and

Pratt, D.A. (2017). On the mechanism of cytoprotection by ferrostatin-1 and li-

proxstatin-1 and the role of lipid peroxidation in ferroptotic cell death. ACS

Cent. Sci. 3 , 232–243.

ygenases in the initiation and execution of ferroptosis. ACS Cent. Sci. 4 ,

387–396.

Skouta, R., Dixon, S.J., Wang, J., Dunn, D.E., Orman, M., Shimada, K.,

Rosenberg, P.A., Lo, D.C., Weinberg, J.M., Linkermann, A., et al. (2014).

Ferrostatins inhibit oxidative lipid damage and cell death in diverse disease

models. J. Am. Chem. Soc. 136 , 4551–4556.

Zielinski, Z.A.M., and Pratt, D.A. (2017). Lipid peroxidation: kinetics, mecha-

nisms, and products. J. Org. Chem. 82 , 2817–2825.

Snelgrove, D.W., Lusztyk, J., Banks, J.T., Mulder, P., and Ingold, K.U. (2001).

Kinetic solvent effects on hydrogen-atom abstractions: reliable, quantitative

predictions via a single empirical equation. J. Am. Chem. Soc. 123 , 469–477.

Zielinski, Z.A.M., and Pratt, D.A. (2019). H-atom abstraction vs. addition: ac-

counting for the diverse product distribution in the autoxidation of cholesterol

& its esters. J. Am. Chem. Soc. 141 , 3037–3051.

14 Cell Chemical Biology 26, 1–14, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE

SOURCE

IDENTIFIER

Chemicals, Peptides, and Recombinant Proteins

DPPH (2,2,-diphernyl-1-picrylhydrazine)

a-tocopherol

Sigma-Aldrich

D9132; CAS: 1898-66-4

258024; CAS: 10191-41-0

430676; CAS: 950-99-2

A5960; CAS: 50-81-7

PHR1359; CAS: 70-18-8

R5010; CAS: 501-36-0

10128023001; CAS: 606-68-8 (anhydrous)

70935; CAS: 989-51-5

C1386; CAS: 458-37-7

N/A

Sigma-Aldrich

PMC (2,2,5,7,8-pentamethyl-6-chromanol)

Ascorbic acid

Sigma-Aldrich

Glutathione (reduced form)

Resveratrol

Sigma-Aldrich

NADH disodium salt (reduced form)

EGCG

Sigma-Aldrich

Cayman Chemical

Curcumin

Sigma-Aldrich

Di-tert-butylresveratrol

Reference (Matsuura et al., 2015)

Sigma-Aldrich

3,5-Di-tert-butylcatechol

D45800; CAS: 1020-31-1

112976; CAS: 88-58-4

N/A

2,5-Di-tert-butylhydroquinone

Sigma-Aldrich

C15-THN

Reference (Li et al., 2013)

Sigma-Aldrich

NAC (N-acetyl-L-cysteine)

Liproxstatin-1

A7250; CAS: 616-91-1

Lip-1

Reference (Friedmann Angeli et al., 2014)

Reference (Skouta et al., 2014)

Reference (Haidasz et al., 2016)

Sigma-Aldrich

Ferrostatin-1

Fer-1

STY-BODIPY

PBD-BODIPY

AAPH (2,2’-azobis(2-amidinopropane)

dihydrochloride)

440914; CAS: 2997-92-4

MeOAMVN (2,2’-azobis(4-methoxy-2,4-

dimethylvaleronitrile)

Wako Pure Chemical Corporation

CAS: 15545-97-8

DTUN

This paper

N/A

4

-hydroxyanisole

Sigma-Aldrich

M18655; CAS: 150-76-5

N/A

2,5-Dimethyl-4-methoxyphenol

Reference (Breslow et al., 2002; Ramirez et al.,

1

959; Zhao et al., 2014)

Reference (Breslow et al., 2002; Ramirez et al.,

959; Zhao et al., 2014)

2,6-Dimethyl-4-methoxyphenol

N/A

1

2

,6-Diethyl-4-methoxyphenol

This paper

N/A

,6-Diisopropyl-4-methoxyphenol

,6-Di-tert-butyl-4-methoxyphenol

This paper

N/A

Sigma-Aldrich

251062; CAS: 489-01-0

d

6

-arachidonic acid

Gift from Mikhail S. Shchepinov (Retrotope, Inc.)

Cayman Chemical

N/A

NDGA

70300; CAS: 500-38-9

Zileuton

Cayman Chemical

10006976; CAS: 111406-87-2

PD146176

Cayman Chemical

10010518; CAS: 4079-26-9

CPX (ciclopirox olamine)

DFO (deferoxamine mesylate salt)

Pioglitazone

Sigma-Aldrich

C2162700; CAS: 41621-49-2

Sigma-Aldrich

D9533; CAS: 138-14-7

Cayman Chemical

71745; CAS: 111025-46-8

PHOXN (Phenoxazine)

lipo-TEMPO

Sigma-Aldrich

263893; CAS: 135-67-1

Reference (Griesser et al., 2018)

Reference (Zilka et al., 2017)

Reference (Griesser et al., 2018)

Reference (Shah et al., 2017)

N/A

RSL3

DHQN

PHOXNO

Arylamines (S1-S12)

(1S,3R)-RSL3

Reference (Friedmann Angeli et al., 2014; Skouta

et al., 2014)

(Continued on next page)

Cell Chemical Biology 26, 1–14.e1–e7, November 21, 2019 e1

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Continued

REAGENT or RESOURCE

Experimental Models: Cell lines

Pfa1 cells

SOURCE

IDENTIFIER

N/A

Reference (Friedmann Angeli et al., 2014; Seiler

et al., 2008)

Software and Algorithms

ChemDraw Ultra, Version 14.0

Perkin Elmer

http://www.perkinelmer.com/category/

chemdraw

Prism, Version 7.0

BioTek Gen5.1

GraphPad Software

BioTek Synergy H1 Software

https://www.graphpad.com/scientiﬁc-

software/prism/

https://www.biotek.com/products/

software-robotics-software/gen5-

microplate-reader-and-imager-software/

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for resources and reagents should be directed to and will be fulﬁlled by the Lead Contact Derek A.

Pratt (dpratt@uottawa.ca ).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Pfa1 cells were generated as described in (Seiler et al., 2008) and shared by the laboratory of Dr. Marcus Conrad. Pfa1 cells were

ꢀ

cultured at 37 C in a 5% CO

2

atmosphere unless otherwise indicated. Pfa1 cells were cultured in DMEM with 10% FBS, 1%

1

003 non-essential amino acid solution, 1 mM sodium pyruvate and 1% penicillin-streptomycin. Cells were passaged by dissoci-

ation with 0.05% trypsin and 0.2% EDTA every two days and reseeded at a dilution of 1:20. All cell experiments were carried out

as a minimum of three biological replicates.

METHOD DETAILS

DPPH Assay

Test compounds (or vehicle) at various concentrations were dissolved in ethanol and added to clear bottom 96-well plates. To this

was added a solution of DPPH dissolved in ethanol to achieve a ﬁnal concentration of 100 mM. Samples were placed in the BioTek

Synergy H1 plate reader and mixed vigorously for 60 s. The samples were incubated at room temperature for 30 min and the absor-

bance at 517 nm was measured (Cos et al., 2003). Samples were subtracted for background (ethanol only) and normalized to starting

DPPH absorbance. The IC50 values were obtained by ﬁtting the results to a dose-response curve in GraphPad Prism. The results

represent the mean ± S.E. of three independent experiments.

Inhibited Autoxidation of 1-Hexadecene

A 3.5-mL quartz cuvette was charged with PhCl (0.44 mL), 1-hexadecene (2.00 mL). The cuvette was preheated to 37 C in a ther-

ꢀ

mostatted sample holder of a UV-vis spectrophotometer and allowed to equilibrate for approximately 15 min. To the cuvette was

added PBD- BODIPY (12.5 mL of a 2.00 mM stock solution in 1,2,4-trichlorobenzene) and AIBN (50 mL of a 300 mM stock solution

in chlorobenzene). The solution was thoroughly mixed prior to monitoring the uninhibited co-autoxidation via the disappearance

of the PBD-BODIPY probe at 588 nm for 10 min to ensure the reaction was proceeding at a constant rate. Finally, the antioxidant

under investigation was added (10.0 mL of a 1.0 mM solution in chlorobenzene), the solution was mixed thoroughly, and the absor-

bance readings were resumed. The resulting Abs vs time data were processed as previously reported (Haidasz et al., 2016). The rate

-9

-1

-1 -1

i

of initiation (R = 1.3 3 10 M s ) and second order rate constant for propagation for the dye (kPBD-BODIPY = 3792 M s ) necessary to

compute stoichiometric data (n) and inhibition rate constants (kinh) were determined using PMC as a standard, which has an estab-

lished stoichiometry of 2.

Inhibited Autoxidation of Egg-PC Liposomes on UV-Vis

To a cuvette of 2.34 mL of 10 mM PBS at pH 7.4 was added liposomes (125 mL of 20 mM stock in PBS at pH 7.4) and the solution was

ꢀ

equilibrated for 5 minutes at 37 C (liposome formation from egg PC was described elsewhere).(Li et al., 2013) The cuvette was

blanked and 10 mL of 2 mM STY-BODIPY in DMSO was added followed by 10 mL of 0.05 M MeOAMVN in acetonitrile and the solution

was thoroughly mixed. After 5 minutes, an aliquot of RTA stock solution in DMSO was added and the loss of absorbance at 565 nm

followed. The inhibition rate constant (kinh) and stoichiometry (n) was determined for each experiment according to Equations S2 and

S1, respectively. The resulting Abs vs time data were processed as previously reported (Haidasz et al., 2016). The rate of initiation

-

9

-1

-1 -1

(

R

i

= 2.5 3 10 M s ) and second order rate constant for propagation for the dye (kSTY-BODIPY = 894 M s ) necessary to compute

e2 Cell Chemical Biology 26, 1–14.e1–e7, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

stoichiometric data (n) and inhibition rate constants (kinh) were determined using PMC as a standard, which has an established stoi-

chiometry of 2. Kinetic data are given as averages of three independent measurements.

Inhibited Autoxidation of Egg-PC Liposomes on Plate Reader

To a black 96-well polypropylene plate (Nunc), was added a solution containing liposomes (1 mM for non-chain; 20 mM for chain

conditions) and either one of STY-BODIPY (1 mM) or C11-BODIPY (1 mM) in PBS at pH 7.4 to a ﬁnal volume of 295 mL (It is suggested

to make a larger volume in falcon tube, pre-mix and aliquot 295 mL to each well). For liposome generation from eggPC please see: (Li

et al., 2013). This was followed by the addition of inhibitors (2 mL aliquots) at desired concentrations to the appropriate wells. The plate

ꢀ

was incubated for 10 minutes at 37 C in the BioTek Synergy H1 plate reader followed by a vigorous mixing protocol for 5 minutes. The

plate was ejected from the plate reader and the autoxidation was initiated by the addition of a 3 mL aliquot of either AAPH (1 mM in

PBS at pH 7.4), MeOAMVN (0.2 mM in MeCN) or DTUN (0.2 mM in EtOH), followed by another mixing protocol for 5 minutes. The plate

ꢀ

was incubated for an additional 10 minutes to equilibrate at 37 C before data was acquired by excitation of the probes at 488 nm and

emission was measured at 518 nm (gain: 70). The data was transformed by diving the raw RFU values by the response factor of

4

.49x10 RFU/mM or 1.93x10 RFU/mM for STY-BODIPY and C11-BODIPY, respectively. The data shown is an average of three in-

4

7

dependent experiments. Settings such as read intervals and gain must be optimized for each instrument. Prior to the ﬁrst set of

experiments, the response factor must be measured for chain and non-chain conditions by carrying out a set of uninhibited autox-

idations with varying concentrations of the probe (0.2, 0.4, 0.6, 0.8, 1 mM); see Figures S2A–S2D. The response factor varies depend-

ing on the instrument and the settings being utilized. Once determined, the same response factor can then be used to transform the

data and does not need to be measured again. The inhibition rate constant (kinh) and stoichiometry (n) can be determined for each

experiment according to Equations S2 and S1, respectively (see example of a calculation below). The second order rate constant for

-1

-1 -1

propagation for the dye is kSTY-BODIPY = 894 M

s

for non-chain and kSTY-BODIPY = 1703 M

s for chain; kC11-BODIPY = 2000 M s

for non-chain conditions. The rate of initiation (R

i

) necessary to compute stoichiometric data (n) and inhibition rate constants (kinh

)

were determined using PMC as a standard, which has an established stoichiometry of 2.

Method to Calculate n and kinh

Variables

ꢁ1

s

k

STYꢁBODIPY = 894 M

(non-chain)

STYꢁBODIPY] = dye concentration at t = 0 (M)

n = stoichiometry

= calculated as shown

RTA] = antioxidant concentration (M)

inh = rate of inhibition, (M/s)

inh = inhibition time (s)

RTA] = antioxidant concentration (M)

[

R

i

[

R

t

[

0

1

B

RFU

C

A

1

) Covert all RFU into [ox-STY-BODIPY] ½ox ꢁ STY ꢁ BODIPYꢃ =

@

₀10RFU

1

7

:53

M

ꢀ

ꢁ

½

RTAꢃ$n

2

) Calculate the rate of initiation Ri =

by standardizing using PMC

tinh

ꢁ

½

4310 ⁶Mꢃ$2:0

ꢁ

10 ꢁ1

s

[

PMC] = 4mM, n = 2, tinh = time of inhibition = 10885 s rRi =

= 7:40 x 10

10885

Cell Chemical Biology 26, 1–14.e1–e7, November 21, 2019 e3

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

3

) Calculate stoichiometry using Equation S1

ꢀ

ꢁ

t

inh$R

i

n =

(Equation S1)

½

RTAꢃ

ꢁ

10

1

1234,7:40310

for PHOXZ, tinh = 11234 s r n =

= 2:1

4310 ⁶Mꢃ

ꢁ

½

4

) Determine Rinh from the rate of change of the inhibited period

-12

for PHOXZ Rinh = 1.11x10 M/s

5

) Calculate kinh using rearranged Equation S2

ꢀ

ꢁ

k

STYꢁBODIPY $½STY ꢁ BODIPYꢃ$R

n$½RTAꢃ$Rinh

i

kinh = ꢁ

(Equation S2)

ꢁ

ꢁ1

ꢁ6

ꢁ10 ꢁ1

8

94M ¹

s

$½1310 Mꢃ$7:40310

s

4

ꢁ1 ꢁ1

s

For PHOXZ rkinh = ꢁ

= 7:1x10 M

ꢁ6

ꢁ12

ꢁ1

2

:1$½4310 Mꢃ$1:11310 Ms

Determination of a^H₂ by H NMR

1

H

Functional groups which engage signiﬁcantly as hydrogen bond donors (having measurable a₂values) demonstrate substantially

lower-ﬁeld chemical shifts in a strongly hydrogen bonding solvent (DMSO-d ) compared to a weakly hydrogen bonding solvent

CDCl or Benzene-d and CDCl was described by Abraham et al.

). The correlation between a^H₂ values and the Dd of DMSO-d

to afford the empirical relationship Equation S3. In a previously published paper we utilized benzene-d instead of CDCl (due to

6

(

3

6

3

6

3

an unusual incompatibility between CDCl

rendering Equation S4.

3

and several of our phenoxazines/phenothiazines) which afforded a similar correlation

H

a₂= 0:133DdDMSOꢁCDCl3 + 0:0066

(Equation S3)

(Equation S4)

H

a₂= 0:134DdDMSOꢁBen ꢁ 0:1087

Inhibition of Ferroptosis Induced by Gpx4 Inhibition with (1S,3R)-RSL3

Pfa1 cells (3,000 in 100 mL) were seeded in 96-well plates and cultured overnight. The next day the media was removed, the cells were

washed twice with PBS and the cells were suspended in new media for 30 minutes before addition of RTAs and LOX inhibitors. Fer-

roptosis was induced using (1S, 3R)-RSL3 (10 mM) 30 minutes after incubation of compounds. Cell viability was assessed 6 hours

later using the AquaBluer assay (MultiTarget Pharmaceuticals, LLC) according to the manufacturer’s instructions. Cell viability

was calculated by normalizing the data to untreated controls. Experiments are carried out with six-technical replicates (n = 6 wells

of a 96-well plate) and performed independently with a minimum of three biological replicates.

Determination of Kinetic Chain Length

R

p

k

p

½substrateꢃ

½Dyeꢃ

Kinetic Chain Length =

3

R

i

k

p

ꢁ

11

5

:1 3 10

M

ꢂ

ꢃ

ꢁ

1

ꢁ1

ꢁ4

1:5 3 10 M

6

2 M

s

:8 3 10

Non ꢁ chain conditions ðAAPHÞ =

3

ꢂ

ꢃ

ꢁ

10

ꢁ1 ꢁ1

ꢁ6

5

894 M

s

1:0 3 10

M

s

ꢁ

11

3

:3 3 10

M

ꢂ

ꢃ

ꢁ

1

ꢁ1

ꢁ3

6

2 M

s

3:0 3 10

ꢂ

M

s

:6 3 10

Chain conditions ðAAPHÞ =

3

ꢃ

ꢁ

10

ꢁ1 ꢁ1

ꢁ6

4

1703 M

s

1:0 3 10

M

s

e4 Cell Chemical Biology 26, 1–14.e1–e7, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

Determination of a^H₂ by Kinetic Solvent Effect

Utilizing the Ingold-Abraham relationship (Equation S5) in the form of Equation S6, a^H₂ is determined as the intercept of Equation S7

0

inh

S

inh

H

2

H

2

logk = logk + 8:3a b

(Equation S5)

PhCl

H

2

HðPhClÞ

S

H

HðSÞ

logk_i_n_h+ 8:3a b

= logk + 8:3a b

inh 2

(Equation S6)

2

PhCl

S

inh

^l^o^g^kinh

logk

HðSÞ

H

2

ꢄ

_H_ð_P_h_C_l_Þꢅ =

ꢄ

_H_ð_P_h_C_l_Þꢅ + a

(Equation S7)

HðSÞ

8

:3 b

ꢁ b₂

8:3 b₂ꢁ b₂

2

H

19

Determination of b₂for PLPC by F NMR

The H-bond accepting solute (PLPC) was diluted in CCl

solution. To these were added 4-ﬂuorophenol and 4-ﬂuoroanisole for a ﬁnal concentration of 0.01M of each. The solutions were trans-

4

to render ﬁve dilute solutions of differing concentrations and one 0.20M

19

ferred into NMR tubes containing seal capillaries of benzene-d6 with a,a,a-triﬂuorotoluene, after which their F spectra were re-

ꢀ

corded with a Bruker Avance II 300 equipped with a thermostat (sample kept at 37 C). The 1:1 H-bond formation constant (K

f

) of

4

-ﬂuorophenol and the H-bond acceptor(s) can be determined with by Equation S8 as described by Taft and co-workers:

ꢀ

ꢁ

d

A

0

D

K

f

= ꢆ

ꢀ ꢁ ꢇꢈ

ꢀ ꢁ

ꢉ

(Equation S8)

h

i

d

A

0

1 ꢁ

B

0

ꢁ

A

0

D

0 0

Where A is the concentration of 4-ﬂuorophenol, B is the concentration of the H-bond acceptor (PLPC), D is the difference in chem-

ical shift between 4-ﬂuorophenol and 4-ﬂuoroanisole sufﬁcient to saturate H-bonding with 4-ﬂuorophenol (maximum d, [PLPC] =

0

.20M), and d is the observed difference in chemical shift between 4-ﬂuorophenol and 4-ﬂuoroanisole at the corresponding B

0

H

and A

0

. The b₂can be calculated from the measured K

f

using Equation S9.

ðlogK + 1:1Þ

:636

H

f

b₂

=

(Equation S9)

4

Synthesis and Characterization

Synthesis of DTUN

2

-Methyldecan-2-ol. 1-Bromooctane (19.3 g, 100 mmol, 1 eq) was added dropwise to magnesium turnings (3.04 g, 125 mmol, 1.25

eq, activated dry with iodine) suspended in anhydrous THF (100 mL) under N atmosphere at a rate that maintained a gentle reﬂux.

Upon completion of the reaction, the solution was transferred via cannula to a dry RBF under N , cooled thoroughly in an ice bath, and

2

acetone (7.0 mL, 95 mmol, 0.95 eq, dried overnight on 3A sieves) was added dropwise over about 30 min. The mixture was allowed to

warm to room temperature over 3 h, then recooled on ice, quenched with 6 mL acetic acid and concentrated in vacuo. The residue

was vigorously stirred with 100 mL 3:1 chloroform:isopropanol and 100 mL water for 10 min before extracting the aqueous layer 2x

5

0 mL 3:1 chloroform:isopropanol. The combined organics were washed 3x 100mL 10% aqueous sodium chloride, 1x 100 mL brine,

dried over Na SO , ﬁltered and concentrated to dryness. The residual oil was distilled with a short path fractionating distillation appa-

4

2

ratus collecting the middle fraction of distillate to afford 2-methyldecan-2-ol (13.95 g, 81%) as a colourless oil. Spectra were in agree-

ment with literature values.(Dryzhakov et al., 2015).

2

-Iodo-2-methyldecane. An oven dried 25 mL round-bottom ﬂask and stir bar were cooled under N

completely in foil, charged with concentrated HI (6.7 mL, ca. 2 eq) and then cooled in an ice bath. Lithium iodide (3.36g, 25 mmol,

eq) was added in portions (caution: exotherm/fumes with powdered LiI; beads are preferred). A solution of 2-methyldecan-2-ol

4.31 g, 25 mmol, 1 eq.) in methylene chloride (5 mL) was added slowly by syringe with vigorous stirring under nitrogen atmosphere.

2

atmosphere, wrapped

1

(

The ice bath was then removed and the biphasic mixture stirred for about 10 minutes, at which point the alcohol was completely

1

consumed by H NMR. The mixture was carefully transferred to separatory funnel and the bottom aqueous layer cautiously extracted

twice with methylene chloride (10 mL). The combined organic layers were washed once with 1:1 brine:water containing drops of so-

dium bisulﬁte solution until the mixture was mostly colourless. The organic extracts were fully protected from light while drying over

2 4

Na SO , ﬁltering and concentrating on a rotary evaporator followed by hivac (ﬂask backﬁlled with nitrogen) to yield the title compound

1

as a pale oil (6.63 g, 94%). The product was pure by H NMR without distillation and was stored at -20 C in a foil-wrapped vial, ideally

ꢀ

1

13

) d 1.92 (m, 6H), 1.65-1.22 (m, 14H), 0.88 (t, J = 6.9 Hz, 3H). C NMR (75 MHz,

backﬁlled with nitrogen. H NMR (300 MHz, CDCl

CDCl

55.1817.

3

+

3

) d 53.27, 50.72, 38.20, 32.02, 29.65, 29.57, 29.43, 28.63, 22.81, 14.27. HRMS: m/z Calc: C11

H23 (ESI: M-I) 155.1800 Found:

1

Cell Chemical Biology 26, 1–14.e1–e7, November 21, 2019 e5

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

2

Di-tert-undecyl Hyponitrite (DTUN). An oven dried 25 mL round-bottom ﬂask and stir bar were cooled under N atmosphere and

charged with 2-iodo-2-methyldecane (1.44g, 6 mmol) followed by degassed anhydrous methylene chloride (10 mL). The ﬂask

was completely wrapped in aluminum foil to exclude light and then cooled thoroughly in an ice bath. Freshly dried silver hyponitrite

(

1.24 g, 4.5 mmol) was added in three portions over 5 min with vigorous stirring. The ice bath was then removed and NMR analysis

indicated consumption of the tertiary iodide usually within 10-20 min. The reaction mixture was then re-cooled in an ice bath and

diluted with cold pentane (10 mL) before ﬁltering over celite. The ﬁltrate was concentrated on a rotary evaporator without heating

2

followed by hivac for about 5 min while on ice. The residual oil was crystallized overnight from methanol (ca. 20 mL, N purged)

on dry ice. The majority of supernatant was then removed by syringe, discarded, and the solid washed with three portions of cold

methanol in the same manner. The solid was allowed to thaw into a small volume of methanol (ca. 5 mL) until a solution formed.

The product was promptly isolated thereafter while being kept cold on ice at all times. The solution was extracted with three portions

2 4

of cold pentane (5 mL), the combined pentane extracts washed once with 1:1 methanol:water, dried over Na SO , ﬁltered, concen-

trated on a rotary evaporator without heating followed by high vacuum to yield DTUN (270 mg, 24%) as a colourless wax. The product

when free from residual solvent is solid on ice and rapidly melts when allowed to warm to ambient temperature. The solid was ali-

ꢀ

quoted by Pasteur pipette into tared vials and the vials were stored at -78 C. Stocks up to 0.2 M were prepared in ethanol or chlo-

1

robenzene and held on ice or dry ice just prior to use. H NMR (600 MHz, CDCl

3

) d 1.65 (m, 2H), 1.34 (s, 6H), 1.34-1.24 (m, 12H), 0.88 (t,

1

3

J = 6.6 Hz, 3H). C NMR (151 MHz, CDCl ) d 83.42, 40.63, 32.03, 30.16, 29.69, 29.42, 25.94, 24.02, 22.82, 14.27. FTIR (ATR) n: 988,

ꢁ

1

+

1

147, 1368, 1384, 1466, 2854, 2923 cm . HRMS: m/z Calc: C11

Synthesis of Phenols

,6-Diethylhydroquinone. A 100mL double neck round bottom ﬂask was charged with glacial acetic acid (35mL), concentrated sul-

furic acid (0.4 mL) and 2,6-diethylphenol (3.89g; 25.9mmol). With mixing, 30% H (13mL) was add by dropping funnel. While moni-

23 2 2

H23 (EI: M-C11H N O ) 155.1794 Found: 155.1787.

2

O

2

toring the temperature of the solution with an alcohol thermometer (afﬁxed through side neck) the mixture was gradually heated to

ꢀ

5

0 C (heat gun) after which the reaction was sufﬁciently exothermic to maintain its temperature. The reaction was kept between 55

ꢀ

and 65 C for ꢂ30 minutes after which the heat of the reaction subsided until cooling to room temperature. The reaction was poured

into water and extracted with ethyl acetate three times. The combined organic extracts were rinsed three times with 1M NaOH so-

lution, and subsequently dried with MgSO

40mL) and transferred to a 250mL round bottom ﬂask charged with methanol (25mL), water (90mL), and sodium dithionite (13g). The

ﬂask was sealed, evacuated, and backﬁlled with N and left to stir overnight. The ether was then separated, and the aqueous phases

was extracted with ether. The combined organic solutions were subsequently dried with MgSO , ﬁltered, and concentrated under

reduced pressure. The product was isolated by column chromatography (30% EtOAc in hexanes) and recrystallized out of hexanes

4

, ﬁltered, and concentrated under reduced pressure. The residue was dissolved into ether

(

2

4

ꢀ

1

affording 2,6-diethylhydroquinone as white needles (1.00g, 23.2%); mp 90-92 C; H NMR (400 MHz; DMSO-d

6

): d 8.49 (s, 1H), 7.32

13

(

s, 1H), 6.32 (s, 2H), 2.48 (q, J = 7.5 Hz, 4H), 1.08 (t, J = 7.5 Hz, 6H); H NMR (101 MHZ; DMSO-d ): d 150.2, 144.2, 131.93, 112.7, 23.0,

6

1

4.38; HRMS (EI, magnetic sector): calcd for C10

14 2

H O 166.0993, found 166.1007.

2,6-Diethyl-4-methoxyphenol (16). Methanol (15mL) and 2,6-diethylhydroquionone (0.95g; 5.7mmol) were placed in a 100mL round

bottom ﬂask equipped with a stir bar and reﬂux condenser. Carefully, 4.5mL of concentrated sulfuric acid was added and the reaction

mixture was heated to reﬂux for ꢂ4 hours. Once cooled the, the reaction mixture was poured into water followed by extraction with

4

ether (three times). The combined organic extracts were dried with MgSO , ﬁltered, and concentrated under reduced pressure. The

product was isolated by column chromatography (10% EtOAc in hexanes). Two recrystallizations out of hexanes (minimum of hex-

anes at room temperature, followed by chilling on dry ice resulted in rapid crystallization) afforded the 2,6-diethyl-4-methoxyphenol

ꢀ

as white needles (0.39g, 38%); mp 34-36 C; 1H NMR (400 MHz; DMSO-d6): d 7.58 (s, 1H), 6.49 (s, 2H), 3.65 (s, 3H), 2.54 (q, J = 7.5 Hz,

4

H), 1.10 (t, J = 7.5 Hz, 6H). 13C NMR (101 MHz; DMSO-d6): d 152.5, 145.6, 132.0, 111.6, 55.1, 23.2, 14.4; HRMS (EI, magnetic

sector): calcd for C11H16O2 180.1150, found 180.1140.

,6-Diisopropylhydroquinone. The same general procedure for the preparation of 2,6-diethylhydroquinone was used in the prep-

2

aration of 2,6-diisopropylhydroquinone from 2,6-diisopropylphenol (6.04g; 33.9mmol). The product was isolated by column chroma-

tography (30% EtOAc in hexanes) and recrystallized out of hexanes affording 2,6-diisopropylhydroquinone as white leaﬂets (1.34g,

ꢀ

2

0.4%); mp 94-96 C; 1H NMR (400 MHz; DMSO-d6): d 8.49 (s, 1H), 7.26 (s, 1H), 6.37 (s, 2H), 3.23 (sept., J = 6.9 Hz, 2H), 1.10 (d, J =

.9 Hz, 12H); 13C NMR (101 MHz; DMSO-d6): d 150.7, 142.75, 136.9, 109.4, 26.2, 23.0; HRMS (EI, magnetic sector): calcd for

6

C12H18O2 194.1306, found 194.1297.

2

,6-Diisopropyl-4-methoxyphenol (17). Methanol (20mL) and 2,6-diisopropylhydroquionone (1.25g; 6.4mmol) were placed in a

00mL round bottom ﬂask equipped with a stir bar and reﬂux condenser. Carefully, 6mL of concentrated sulfuric acid was added

1

and the reaction mixture was heated to reﬂux for ꢂ4 hours. Once cooled the, the reaction mixture was poured into water followed

by extraction with ether (three times). The combined organic extracts were dried with MgSO , ﬁltered, and concentrated under

reduced pressure. The product was isolated by column chromatography (10% EtOAc in hexanes). After drying under vacuum, the

4

ꢀ

residue was chilled (-20 C) affording 2,6-diisopropyl-4-methoxyphenol as a beige crystalline solid (1.06g, 79.5%); mp 48-50 C;

H NMR (400 MHz; DMSO-d6): d 7.53 (s, 1H), 6.51 (s, 2H), 3.66 (s, 3H), 3.28 (sept, J = 6.9 Hz, 2H), 1.13 (d, J = 6.9 Hz, 12H); 13C

NMR (101 MHz; DMSO-d6): d 153.0, 144.2, 136.9, 108.3, 55.0, 26.4, 22.9. HRMS (EI, magnetic sector): calcd for C13H20O2

08.1463, found 208.1487.

1

2

e6 Cell Chemical Biology 26, 1–14.e1–e7, November 21, 2019

Please cite this article in press as: Shah et al., Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death

Rescue, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.09.007

QUANTIFICATION AND STATISTICAL ANALYSIS

Unless otherwise indicated all experiments were carried out in three independent replicates and the mean ± SD are denoted in the

appropriate ﬁgures and their corresponding captions.

DATA CODE AND AVAILABILITY

The datasets supporting the current study have not been deposited in a public repository because of the large size and the diversity of

the results, but are freely available from the lead author on request.

Cell Chemical Biology 26, 1–14.e1–e7, November 21, 2019 e7

Article Doi

Beyond DPPH: Use of Fluorescence-Enabled Inhibited Autoxidation to Predict Oxidative Cell Death Rescue

DOI: 10.1016/j.chembiol.2019.09.007

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.chembiol.2019.09.007

Full text of DOI:10.1016/j.chembiol.2019.09.007

Products guided by the article

R&D Labs maybe for 1988-10-9

Relevant to this article

Hot Product