Peroxide-Induced Liberation of Iron from Heme Switches Catalysis during Luminol Reaction and Causes Loss of Light and Heterodyning of Luminescence Kinetics.

The peroxidation of luminol yields bright luminescence when the reaction is catalyzed by heme proteins. However, an excess of peroxide leads to less light and altered luminescence kinetics, an effect commonly referred to as "suicide inactivation". The aim of this study is to present the molecular processes causing this effect. A comprehensive set of data reported here demonstrates that suicide inactivation is due to a peroxide-induced liberation of iron from its coordinating porphyrin. Liberated iron launches catalysis of the reaction at much lower efficiency. The light-yielding efficiencies of different organic and inorganic catalysts are precisely quantified and compared. It is shown that the catalysis by free iron involves superoxide. This is explained by the formation of a ferryl-oxo-iron complex. In this context, a complete reaction mechanism involving a modified Fenton-Haber-Weiss cycle is proposed for the first time. The switch from the highly efficient biogenically catalyzed luminescence to a less efficient inorganically catalyzed reaction is accompanied by a transition from "flash-type" to "glow-type" luminescence kinetics. Ethylenediaminetetraacetic acid-mediated chelation of iron is used to demonstrate this effect and to separate both kinetics. The explanation of kinetic heterodyning is underpinned by mathematical modeling. The results are able to explain the as yet unexplained phenomena discussed in the less recent literature and to settle disputes about them. It is concluded that peroxide concentrations exceeding the level tolerated by the catalyzing heme protein negatively impact performance and precision of luminol-based assays, where the light yield is used as a quantitative measure for analyte concentrations.

Introduction

Chemiluminescence assays are among the most sensitive quantification methods allowing the detection of diverse analytes in concentrations as low as picomolar or even attomolar. In particular, assays based on luminol, one of the most important chemiluminescent compounds,[1] are used for a broad variety of analytes (ref (2) and more refs in ref (3)), and even forensic investigators take advantage of the luminol-based detection of blood.[4] However, circumspection is necessary with assay design when the luminol reaction is employed, and the light yield (photon counts) is used to determine analyte concentrations. The systematic experimental survey reported here investigates the molecular causes of the so-called “suicide inactivation” and the changes in luminescence yield and kinetics linked with it. The focus is on the liberation of iron from its porphyrin coordination during luminol reactions catalyzed by heme-based catalysts. In a complementary study, the effects of enhancers, inhibitors, and additives able to partially repress the undesired iron liberation are analyzed and discussed in detail.[3]

The Luminol Reaction

5-Amino-2,3-dihydro-1,4-phthalazinedione also known as luminol (LH2) emits blue light when oxidized in an alkaline environment in the presence of a strong oxidizing agent such as hydrogen peroxide (H2O2). The oxidizing peroxide donates the oxygen needed for the reaction and converts luminol into aminophthalate (AP*), which is the light-emitting product.

The luminol reaction is complex because a catalyst is indispensable for detectable light emission. Biogenic porphyrins carrying a central iron (heme) typically serve as catalysts. The oxidation state switching of the central iron is between III (ferric state) and IV (ferryl state) and traverses through a three stage cycle, the so-called peroxidative cycle, when a hydrogen donor (AH) and a peroxide (H2O2) are present (details are given in the Supporting Information of Plieth 2018[3]).

Peroxidases are widely used for luminol-based applications. In this context, luminescence catalyzed by peroxidases is called enzyme-catalyzed chemiluminescence (eCL). Apart from genuine peroxidases, there are many other heme proteins which also exhibit a peroxidase activity. For the luminol reaction, free iron porphyrin, without a protein shell (i.e., hemin), is also sufficient to catalyze the luminescent reaction.[5−8]

Even dissolved free iron ions, not coordinated by a porphyrin system, and also many other transition metals are able to catalyze the luminol–H2O2-reaction and thereby produce some detectable light.[9−11] This is called inorganically catalyzed chemiluminescence (iCL) to distinguish it from organic catalysts. Both the light yield efficiency of iCL and its catalytic mechanism are different from those of the peroxidative cycle. Here, the light-yield efficiencies of diverse catalysts are calculated and compared, and a mechanism is proposed, describing how the iCL-yielding luminol reaction is catalyzed by non-porphyrin iron.

The intricate dependency of the reaction on peroxide concentration, pH, and the presence of various catalysts is investigated. It is demonstrated that the light yield from luminol reactions driven by heme compounds is severely impaired when the reducing substrate (luminol) is not properly balanced with the oxidizing substrate (H2O2). Several lines of evidence are provided that this impairment is due to liberation of catalytically active iron ions from their porphyrin coordination. This liberation is accompanied by a switch from highly efficient eCL to less efficient iCL. Thus, often and unintentionally, both iCL and eCL occur within the same reaction and reveal mixed (i.e., heterodyned) chemiluminescence kinetics. It is experimentally demonstrated by diverse means how the distinct kinetic components of eCL and iCL can be separated, analyzed, and mathematically modeled.

Results

Catalysis of Luminol eCL by Horseradish Peroxidase Is Hampered by H2O2-Induced Peroxidase Inhibition

H2O2 dilution series were prepared under different pH conditions, and luminescence was recorded several minutes for each dilution to demonstrate the peroxidase inhibition at high [H2O2]. The integrated light yield plotted against the H2O2 concentration (Figure ) shows that light yield does not strictly increase with increasing [H2O2]. There is rather a maximum around 1 mM of H2O2 above which less light is produced because of horseradish peroxidase (HRP) inactivation by excess peroxide.

Figure 1

Dependence of luminol eCL on hydrogen peroxide concentration and pH with HRP as the catalyst. Luminol luminescence was recorded at various peroxide concentrations and different pH values. Luminescence, integrated over the first 10 min of the reaction, is plotted against [H2O2]. Reactions were started by injecting a mix of catalyst and luminol into a H2O2 solution to give the following assay conditions: 100 mM Tris/HCl; pH as indicated in the inset; 1 μg/mL of HRP, which corresponds to a concentration of 23 nM Fe (pFe = 7.6); [LH–] = 250 μM; [H2O2] as given on the abscissa; (A) 1 mM CaCl2 was allowed because the HRP is a Ca2+-sensitive peroxidase. (B) Same experiments were performed in the absence of Ca2+ [i.e., in the presence of 1 mM ethylenediaminetetraacetic acid (EDTA)]. Data are averages of four technical replicates. Error bars represent the standard deviation (SD). The SD is below symbol size, where no error bar can be seen. These data are presented in the Supporting Information (Figure S1.1) on a log–log scale for more details.

This has been observed earlier with other peroxidase indicator substrates, and the effect is referred to as “suicide inactivation”.[12−16] It is remarkable that the suicide inactivation shifts toward higher [H2O2] (Table S1 in the Supporting Information) when the reaction is performed at higher proton concentrations (i.e., pH ≤ 9). Qualitatively, the suicide inactivation is largely independent of the presence of Ca2+ ions (Figures and S1.1). Quantitatively, however, the luminescence intensity decreases to a third when Ca2+ is sequestered by EDTA (Figure B). This is in accordance with other findings, demonstrating that plant peroxidases are calcium-dependent (e.g., refs[17−20]).

Other Iron Porphyrin Proteins Catalyzing the Luminol Reaction Are Also Inhibited by H2O2

There are many other iron-porphyrin proteins which also exhibit peroxidase activity,[21−29] and there was expectation that they are less sensitive to high [H2O2] than the HRP. Consequently, some of these proteins, namely, cytochrome c (Cyt c) and hemoglobin (Hb), were also tested for peroxide inactivation and were found to behave in a similar manner as HRP (Figure A,B). This is in line with previous studies.[22,30]

Figure 2

Dependence of luminol eCL on hydrogen peroxide concentration, pH, and the type of catalyst. Luminescence, integrated over the first 10 min of the reaction, is plotted against [H2O2] at different pH values. (A) Luminol reaction with Hb 4 μg/mL corresponding to 0.25 μM iron (pFe = 6.6) as the catalyst; (B) luminol luminescence with 10 μg/mL Cyt c corresponding to 0.81 μM Fe (pFe = 6.1); and (C) luminol luminescence with hemin (2.5 μM; pFe = 5.6) as the catalyst in the assay medium. Reactions were started by injecting a mix of catalyst and luminol into an H2O2 solution to give the following assay conditions: 100 mM Tris/HCl; pH as indicated in the insets; [LH–] = 250 μM; [H2O2] as given on the abscissa. Data are averages of four technical replicates. Error bars represent the SD. The SD is below symbol size, where no error bar can be seen. These data are presented in the Supporting Information (Figure S1.2) on a log–log scale for more details.

It is also known from previous studies[5−8] that iron porphyrin itself (hemin), without a protein shell, can produce light in the presence of luminol. Surprisingly, the catalytic activity of hemin is also inhibited by high [H2O2] (Figure C). This is the first evidence that the suicide inactivation is not the result of protein oxidation by H2O2 but rather is due to the disintegration of porphyrin and the liberation of its central iron. However, it could be argued that the inhibition of eCL at [H2O2] > 1 mM may be an intrinsic property of luminol or its oxidation products and possibly not caused by iron depletion of porphyrin. To shed light on this, experiments were performed as above but with free iron ions as catalysts (=iCL).

Catalytic Activity of Free Iron Is Enhanced by Increasing pH and [H2O2]

When iCL was screened for light productivity, it turned out that the light yield increased with higher [H2O2] and at more alkaline pH values (Figure ). There is no specific [H2O2] at which an inhibition of the catalytic activity was seen. The result is the same, no matter which inorganic iron salt was used as a source of free iron ions (Supporting Information Figure S2.1).

Figure 3

Inorganically catalyzed luminol luminescence and its dependence on the hydrogen peroxide concentration and pH with ionic iron (25 μM FeSO4; pFe = 4.6) as the catalyst. The luminescence integrated over the first 10 min of the reaction is plotted against [H2O2]. Reactions were started by injecting a mix of catalyst and luminol into a H2O2 solution to give the following assay conditions: 100 mM Tris/HCl; pH as indicated in the inset; [LH–] = 250 μM. Data are averages of four technical replicates. Error bars represent the SD. The SD is below symbol size, where no error bar can be seen. These data are presented in the Supporting Information (Figure S2.1B) on a log–log scale for more details.

Mechanism of Non-porphyrin Iron Catalysis Is Still Obscure

The main unknown here (Figure ) is the catalytic mechanism driving the iCL reaction. A peroxidase cycle as presented by Plieth (2018)[3] can be excluded because there is no porphyrin coordinating the iron. The classical Fenton–Haber–Weiss cycle (cFHW, Supporting Information Figure S2.3) involving a production of hydroxyl radicals (OH•; Supporting Information eq S2.3) could be invoked as a basis for a possible mechanism yielding luminol radicals (L•–). However, there are data and arguments against it. If the cFHW reaction was the driving mechanism, it would imply that the reaction could be inhibited by the OH•-scavengers dimethyl sulfoxide (DMSO) and mannitol.[25,31] However, data obtained in the course of this study (Supporting Information Figure S2.4) clearly indicate that both scavengers do not inhibit the luminol reaction. Thus, with the assay conditions used (pH = 9; [H2O2] > 1 mM), a cFHW-driven catalysis can be ruled out and a modified FHW mechanism is proposed instead (see the Discussion section below and Supporting Information 2.5).

The FHW mechanism, however, involves superoxide (Supporting Information eqs S2.4 and S2.5). This can be demonstrated by the addition of superoxide dismutase (SOD) to the reaction. At high activities, this enzyme effectively neutralizes superoxide and reduces iCL (Figure ). No such inhibiting effect is seen with eCL. Rather, the biogenic luminescence increases at high [SOD] (Figure ). This effect is explained and discussed below (see the Discussion section).

Figure 4

Luminol luminescence in the presence of SOD with different catalysts as indicated in the inset. Luminescence data represent light yield (ILY) integrated over the first 15 min of the reaction. A decrease of light yield is seen with iCL (free iron as catalysts) but not with eCL (iron-porphyrin compounds as catalysts). The reactions were started by injecting H2O2 to a mix of catalyst and luminol to give the following assay conditions: Tris/HCl 100 mM; [H2O2] = 1.1 mM; [LH–] = 750 μM; [HRP] = 0.5 μg/mL; [Cyt c] = 10 μg/mL; [Hb] = 30 μg/mL; [hemin] = 1 μM; [FeSO4] = 200 μM; [FeEDDHA] = 100 μM; pH = 9. Data are normalized by the luminescence obtained without SOD. Data are means of n = 3 technical replicates. Error bars represent the SD. The SD is below symbol size, where no error bar can be seen.

There Are Significant Differences in Relative Light-Yield Efficiency between iCL and eCL

If the relative efficiency of light production (light-yield efficiency ELY) is to be compared for the different catalysts tested so far, then the catalyst concentration used is critical. ELY is defined here as the logarithm of the ratio of integrated light measured (ILY = integrated light yield) and the catalytic iron concentration [Fe] used (eq )ELY thus represents the amount of light each iron atom in the assay is able to produce, regardless of being porphyrin-coordinated or free (more details are given in Supporting Information 3; eqs S3.1–S3.3).

Standard conditions were chosen to measure ELY for diverse catalysts (Figure ). Therefore, dilution series of the catalysts were prepared, and the emitted light was recorded for each dilution.[28] The five dilutions giving the best linear correlation (i.e., maximum correlation coefficient r2) on a log–log plot (Figure A) were used to calculate ELY. These five ELY values were averaged and plotted as a bar diagram (Figure B).

Figure 5

Light-yield efficiencies of iCL and eCL catalysts under mild conditions. (A) Five different concentrations of catalysts were used to produce light under controlled mild assay conditions (100 mM Tris/HCl pH = 9; [H2O2] = 1.1 mM; [LH–] = 750 μM; catalysts as indicated in the inset). The reactions were started by injecting H2O2 into a mix of catalyst and luminol. The log of integrated light yield (ILY) of each reaction is plotted against the total iron concentration, given as pFe, in the assay. For FeEDDHA, Fe(NO3)3, and hemin, the total iron concentration equals the molarity of the respective substance. For proteins, the Fe molarities were calculated as follows according to the molecular weight of the protein and the number of Fe bound to each protein molecule: 1 μg/mL of Hb = 0.0625 μM of Fe; 1 μg/mL of Cyt c = 0.081 μM of Fe; and 1 μg/mL of HRP = 0.023 μM of Fe. (B) Efficiencies ELY for each reaction calculated according to eq were averaged for each catalyst (n = 5) and plotted as a bar chart. Error bars represent SDs.

Marked differences in light-yield efficiency between the different catalysts were observed. The light yield efficiency of hemin is around ELY = 12 and thus about two orders of magnitude greater than ELY obtained from free iron. This means that each iron atom can produce about 100 times more light when it is coordinated by porphyrin. When the catalyzing iron porphyrin has a protein shell as is the case with Hb and Cyt c, its ELY is another order of magnitude greater and emits 1000 times more light compared with free Fe. Finally, when the porphyrin-coordinated iron (heme) constitutes the central core of a genuine peroxidase as is the case with HRP, its efficiency ELY is another two orders of magnitudes higher (ELY = 15). Thus, iron embedded in HRP can produce 100 000 times more photons during the first minutes of the luminol reaction than free iron.

This experiment (Figure ) was repeated several times and with different pH values (Supporting Information Figure S3) as light yields were found to be pH-dependent (Figures –3). Hb, hemin, and free Fe clearly show that ELY is pH-dependent (Figure S3). The more alkaline the medium, the better is ELY. This is in accordance with what has been observed previously (Figures A,C and 3). HRP and Cyt c do not show such strict correlation of pH and ELY, which is also in line with prior observations (Figures and 2B).

Chelators Help To Distinguish between iCL and eCL

When the luminol reaction is catalyzed by free iron dissolved from an iron salt (FeSO4, FeCl2, or FeNO3), EDTA is able to abolish the reaction as soon as the total chelator concentration [EDTA] is above the total iron concentration [Fe] (Figure ; orange symbols). This is because all free iron ions are sequestered away by EDTA and are thus unable to catalyze the luminol reaction because they cannot form a ferryl-oxo-iron complex (for details see the Discussion section).

Figure 6

Luminol luminescence in the presence of a metal chelator (EDTA) with different catalysts as indicated in the inset. Normalized luminescence data represent light yield (ILY) integrated over the first 15 min of the reaction. Data are normalized by the luminescence obtained without EDTA. A decrease of light yield is seen with iCL (free iron as catalysts) but not with eCL (heme protein as the catalyst). Assay conditions: Tris/HCl 100 mM; pH = 9; no added Ca2+; [LH–] = 750 μM; [H2O2] = 1.1 mM; catalysts as indicated in the inset: [HRP] = 0.5 μg/mL; [Hb] = 20 μg/mL; [FeSO4] = 200 μM; [FeEDDHA] = 100 μM; [Cyt c] = 20 μg/mL. The reactions were started by injecting H2O2 to a mix of catalyst and luminol. Data are normalized by the luminescence obtained without EDTA. Data are means of n = 3 technical replicates. Error bars represent the SD. The SD is below symbol size, where no error bar can be seen.

The situation is different when iron is dissolved as FeEDDHA (brown symbols in Figure ). EDTA (pKFe = 25) is unable to sequester all the catalysts because EDDHA, which is the stronger chelator (pKFe = 35[32−34]), competes with it and is thereby able to keep a certain amount of Fe free in solution. Generally, the amount of free iron available to catalyze the iCL reaction is dependent on the assay conditions (pH, ionic strength, presence of divalent cations, and temperature) and the stability constants of the two competing chelates FeEDTA and FeEDDHA.[32−34]

Enzymatically catalyzed luminescence produced by HRP is clearly reduced by EDTA (green symbols in Figure ), although not totally quenched, because HRP is a Ca2+-dependent protein.[18,20,35] This agrees with data shown previously (Figure ). In contrast, Hb and Cyt c (red and pink symbols, respectively, in Figure ) appear less Ca2+-dependent and thus show negligible reduction of light yield by high [EDTA].

The EDTA effect on free iron (orange symbols in Figure ) was used to investigate the peroxide-induced suicide inactivation of eCL in more detail. When the suicide inactivation is recorded in the presence of EDTA (Figure ; red symbols), there is less luminescence at high [H2O2], whereas without EDTA or with extra divalent cations, there is increased luminescence. This indicates that the luminescence seen without the chelator at high [H2O2] is due to catalysis by free iron liberated from its porphyrin coordination by H2O2, as shown in Figure . In the presence of EDTA, the liberated iron is sequestered and does not yield light at high [H2O2]. Equivalent results were obtained with Cyt c and hemin (see Figure S4 in the Supporting Information).

Figure 7

Dependence of luminol eCL catalyzed by Hb on the presence of EDTA or divalent cations. At high [H2O2] (red shaded area), luminescence progressively originates from modified Fenton–Haber–Weiss cycle (mFHW) catalysis by free iron released from heme. This is inhibited in the presence of EDTA. Luminescence data represent light yield (ILY) integrated over the first 15 min of the reaction. Assay conditions: Tris/HCl 100 mM; pH = 9; [LH–] = 750 μM; [Hb] = 50 μg/mL; and extra Ca2+, Mg2+, or EDTA as indicated in the inset. The reactions were started by injecting a mix of catalyst and luminol into the buffer volume containing H2O2 so that the [H2O2] indicated by the abscissa was reached.

Kinetic Analyses Reveal iCL and eCL Running in Parallel and Provide More Details about the Peroxidase Suicide Inactivation and Iron Liberation

Luminescence kinetics produced by redox reactions are usually characterized as being of either glow-type or flash-type. A precise definition of how these two terms are used in the context here is given in the Supporting Information (SI5; Figures S5.1–S5.5; eqs S5.1–S5.5).

The kinetics of the light output are of the glow-type luminescence when the luminol reaction is catalyzed by free iron ions (Figure ). iCL reaches a certain level when the reaction is started by mixing luminol and iron ions with H2O2, and light output continues for many minutes without decay. eCL kinetics, in contrast, are mainly of the flash-type (Figure ). This implies that the luminescence catalyzed by heme-coordinated iron is transient and decays after it has reached a maximum amplitude. At [H2O2] below 0.5 mM, the transients are slow, although always with a continuous decay after a maximum. With higher [H2O2], above 0.5 mM H2O2, the transient amplitude is high and the kinetics are fast (i.e., the exponential decay has a short time constant). At [H2O2] above 10 mM, the eCL “flash” amplitudes are reduced.

Figure 8

Glow-type kinetics of luminol luminescence catalyzed by free iron. The reactions were started at t = 134 s by injecting a mixture of luminol and FeSO4 into assay buffer with H2O2. Assay conditions: 100 mM Tris/HCl pH = 9; [LH–] = 250 μM; [Fe] = 25 μM (pFe = 4.6). [H2O2] (μM) as indicated in the insets. (A) Kinetics given on a semi-log plot. (B) Data shown in A were normalized by the area under the curve and plotted on linear coordinates to ease direct comparison of the kinetics. This shows that there is no decrease in the luminescence within the recording period (i.e., glow-type luminescence).

Figure 9

Flash-type kinetics of luminol luminescence catalyzed by hemin and HRP. The reactions were started at t = 134 s by injecting a mixture of luminol and catalyst into assay buffer with H2O2. Assay conditions: [LH–] = 250 μM; pH = 9; [H2O2], as indicated in the insets. (A) eCL kinetics obtained with hemin (2.5 μM) on a semilog plot. (C) eCL kinetics obtained with HRP (1 μg/mL, which corresponds to 0.023 μM of iron or pFe = 7.6) on a semilog plot. (B,D) Data shown in A and B were normalized by the area under the curve and plotted on linear coordinates to ease direct comparison. This shows that there is a flash-type luminescence at all H2O2 concentrations tested. However, all kinetics also have a glow-type component incorporated, which becomes more pronounced at high [H2O2] and indicates the liberation of iron from heme and the transition from eCL to iCL. The inset in B gives a magnification of the ordinate.

On closer scrutiny, it becomes evident that with high [H2O2], the flash is always followed by a glow (Figure B,D), which resembles the iCL luminescence caused by free iron (Figure ). This indicates that there is a transition from flash-type to glow-type kinetics during the course of the reaction. The cause for this is highly likely to be the liberation of iron from its porphyrin coordination. Similar data obtained with Hb and Cyt c as catalysts are presented in the Supporting Information (Figure S6).

The statement that the glow component in the kinetics of eCL originates from liberated iron can easily be verified when the kinetics are compared with kinetics recorded in the presence of a chelator (Figure ). Without EDTA, eCL always occurs along with biphasic flash-glow kinetics (Supporting Information eq S5.4; Figure S5.5) when [H2O2] is high. However, in the presence of EDTA, only monophasic flash-type kinetics can be seen, ending in darkness. Divalent cations seem to assist the liberation of iron from its porphyrin coordination (Figure blue and green curves). Similar data obtained with Cyt c and hemin as catalysts are presented in the Supporting Information (Figure S6) and confirm that EDTA degenerates biphasic to monophasic kinetics.

Figure 10

Kinetics of luminol eCL with Hb as the catalyst in the presence of EDTA or divalent cations. (A) At [H2O2] = 100 mM, free iron is liberated from heme. The liberated iron catalyzes the chemiluminescence reaction via the mFHW mechanism and produces glow-type kinetics. In the presence of EDTA (red curve), the glow-type component in the kinetics is abolished and only the flash component remains. (B) Data from A presented on a log–log plot. Assay conditions: Tris/HCl 100 mM; pH = 9; 750 μM luminol; [Hb] = 50 μg/mL; and extra Ca2+, Mg2+, or EDTA, as indicated in the insets. The reactions were started by injection of H2O2 to a final concentration of [H2O2] = 100 mM. Kinetics are from the data set already presented in Figure . Data are normalized by the peak value.

It has been shown previously[3] that luminol at high concentrations prevents the suicide inactivation to some extent and exerts a protective effect on the heme group. This is also reflected in the kinetics (Supporting Information Figure S8) and provides another indication that H2O2 induces liberation of iron and an eCL-to-iCL transition. The lower the luminol concentration in the assay, the more pronounced are luminescence kinetics, revealing catalysis by free iron when [H2O2] is high (Supporting Information Figure S8D).

Soret Band Monitoring Verifies the Peroxide-Induced Destruction of Porphyrin

Heme compounds typically absorb light in the range between 390 and 450 nm. Each compound has a characteristic so-called Soret peak in this range.[36−38] The Soret peak can be used to monitor the intactness of the respective porphyrin group. Here (Figure A), it is demonstrated that the Soret peak of Hb disappears with increasing concentrations of peroxide, indicating the destruction of the iron-coordinating system.

Figure 11

Absorbance of Hb depending on [H2O2]. (A) Absorbance was scanned around the typical Soret band of Hb (i.e., λ = 412 nm) in the presence of H2O2 at concentrations (μM) as indicated by the inset. (B) A (412 nm ± 10) plotted against the respective [H2O2]. The EC50 was obtained by four-parametric logistic curve fitting. Data in B represent averages of n = 5; error bars indicate the SD. Assay conditions: 100 mM Tris/HCl pH = 9 at 28 °C; [Hb] = 1.25 mg/mL; [H2O2] in (μM) as indicated by the abscissa; 1 h preincubation at 28 °C before spectral recording.

The Soret-peak absorbance (λabs = 412 nm for Hb) plotted against [H2O2] (Figure B) indicates an EC50 in the range of 1 mM < [H2O2] < 10 mM. Similar data were obtained with Cyt c and hemin (Supporting Information Figures S9.1 and S9.2). The EC50 values of the Soret-peak reduction (Figure B, Supporting Information Figures S9.1B and S9.2B) roughly coincide with the suicide inactivation reaction (Supporting Information Figures S1.1 and S1.2B; Table S1) observed before. This coincidence is another indication that the suicide inactivation of luminescence is caused by disintegration of porphyrin and the liberation of iron from it.

A H2O2-induced liberation of the whole intact iron porphyrin complex from the heme proteins is unlikely because in this case, the absorbance would not vanish with increasing [H2O2] but rather display a Soret band broadening to become the spectrum seen with hemin (Figure S9.2).

Summary of Results

The data presented here highlight the intertwined processes that occur during the luminol reaction: the reduction of eCL (i.e., the so-called suicide inactivation; Figures and 2, and Supporting Information Figures S1.1 and S1.2) observed with high peroxide concentrations ([H2O2] > 1 mM) is due to H2O2-induced disintegration of the heme catalyst, as demonstrated by Soret band extinction (Figure ; Supporting Information Figures S9.1 and S9.2) and the concomitant liberation of free iron. Free iron, when liberated from heme, launches iCL of glow-type kinetics and with low light-yield efficiency (Figure ; Supporting Information Figure S3). Thus, suicide inactivation leads to a transition from eCL to iCL (Figures –10, and Supporting Information Figures S6 and S7) during the running reaction and to a heterodyning of flash-type and glow-type luminescence kinetics (Figure ; Supporting Information Figure S6). The glow component in these mixed kinetics can be either reduced by sufficiently high luminol concentrations (Supporting Information Figure S8) or completely abolished by EDTA-mediated iron chelation (Figure ; Supporting Information Figure S7).

Discussion

Genuine Peroxidases and “Pseudo-Peroxidases”

The most prominent catalysts of the luminol reaction are heme peroxidases, which evolved in nature to reduce peroxides with the help of reducing organic substrates. Other heme compounds such as Hbs, cytochromes, and even hemin also have peroxidase activities (Figures and 2). They have been referred to as “pseudo-peroxidases” to distinguish them from genuine peroxidases.[39] Their pH dependencies can be used to display this distinction experimentally. The HRP-driven luminol reaction exhibited a maximum light yield at pH ≈ 8 (Figure ). However, with other heme compounds, more light is produced at pH > 9 and much less when pH < 9 (Figure ). This can be explained by the chemistry of the luminol reaction (details in ref (3)), which prefers high pH with an apparent pK at 8.2.[40] The maximum activity of HRP at pH ≈ 8 detected here (Figure ) indicates that a genuine peroxidase has a protein shell which shields the central catalytic heme core from the conditions of the bulk buffer and provides it with the specific local environment necessary for its specific enzymatic activity. Such a local environment is needed by the HRP for the oxidation of a broad spectrum of substrates. Pseudo-peroxidases like Hb and Cyt c evolved in nature to perform different tasks. Consequently, they have no protein shell to shield the heme for peroxidation of diverse substrates, and their pseudo-peroxidase activities (Figure A,B) reflect a pH and peroxide dependency similar to that seen with hemin (Figure C). In terms of peroxidase activity, it can be concluded that pseudo-peroxidases fulfill no more than just an “eggcup function” to hold the catalytic heme in place and have neither optimized nor specific peroxidase function as is the case of genuine peroxidases like HRP.

Another property of HRP that makes it distinct from pseudo-peroxidases is the calcium dependency of its catalytic efficiency (compare Figure A,B), which is not observed with Cyt c and Hb (Figures and 7; Supporting Information Figure S4).

As a control, non-heme iron proteins like ferritin and ferredoxin were tested for their ability to catalyze the luminol reaction (Supporting Information Figure S2.2). In comparison with the light production catalyzed by free iron ions (Figure ; Supporting Information Figure S2.1), these proteins have no activity beyond that of the iron ions they harbor.

Peroxidase Suicide Inactivation

Usually, the reaction rate of enzyme-catalyzed reactions saturates with increasing substrate concentrations, according to the Michaelis–Menten–Henri model.[41] This is different with heme peroxidase-catalyzed reactions. With increasing [H2O2], the reaction rate reaches a maximum and is inactivated by higher [H2O2] (Figures and 2). This suicide inactivation is typical for heme peroxidases and has already been investigated photometrically with various indicator substrates.[12,14−16,42] Diverse inactivation mechanisms have been proposed.[43] The most likely one is liberation of iron, as substantiated by spectroscopic data.[12,13,44] This view is supported by the luminometric approach reported here (Figures –3, and Supporting Information Figures S1.1, S1.3, and S2.1) and validated by experiments which involve the chelator EDTA (Figures , 7, and 10, and Supporting Information Figures S4 and S7) and Soret peak monitoring (Figure ; and Supporting Information Figures S9.1 and S9.2). These data also confirm the peroxide-induced iron release from Hb demonstrated earlier.[45]

Hemin is able to catalyze the eCL reaction (Figure C) and exhibits a similar inhibition pattern at high [H2O2], like the (pseudo-)peroxidases (Figures and 2A,B). The light maximum obtained at around 1 mM H2O2 (Figures , 2, and 7; and Supporting Information Figures S1.1, S1.2, and S4) is hence a feature of heme and less a feature of the protein to which it is bound.

Apart from the peroxidative cycle (ref (3) suppl.), HRP and possibly other heme-based catalysts may also operate an oxygenation cycle with compound III as an intermediate.[16,46] This has been proposed for plant peroxidases by Kimura et al. (2014),[46] particularly for the presence of free ferrous iron ions and dissolved oxygen, at high [H2O2] and under alkaline conditions. For the suicide inactivation as shown here (Figures , 2, and 7; and Supporting Information Figures S1.1, S1.2, and S4), the latter two conditions are given and free iron is produced during suicide inactivation, as demonstrated. Hence, an involvement of compound III is likely.

When HRP, immobilized on a collagen fibril membrane, was removed during the luminol reaction, no cessation of the luminescence was seen. Instead, a continuing glow-type luminescence was observed.[47] This finding can now be easily explained by the suicide inactivation experiments reported here: iron was liberated from membrane-bound HRP by H2O2 and remained in the reaction as a catalyst after the membrane together with the attached HRP apoprotein was removed.

Light-Yield Efficiency and Kinetics Allow a Differentiation between eCL and iCL

The data presented here allow a clear differentiation between eCL and iCL. Maximum light yields of iCL and eCL can only be obtained under specific conditions (Figures , 2, and 3; and Supporting Information Figures S1.1, S1.2, and S2.1) and the reactions exhibit different kinetics (Figures and 9; and Supporting Information Figure S6). iCL requires harsh conditions ([H2O2] > 10 mM and pH > 10; Figure ) and is of low efficiency under mild conditions (Figure ; Supporting Information Figure S3). iCL kinetics are of the glow type (Figure ) because a reaction steady state is reached before one of the substrates (luminol or H2O2) is consumed.

eCL, in contrast, produces a maximum quantum yield under mild conditions ([H2O2] ≤ 1 mM and pH ≤ 9). eCL kinetics are mainly of the flash type (Figure ; and Supporting Information Figure S6) because the heme-catalyzed reaction is of high efficiency (Figure ; and Supporting Information Figure S3). Flash-type kinetics always suggest that at least one essential reaction component is consumed before a reaction equilibrium is reached. In the case of the luminol reaction, this is either H2O2 or luminol. However, with eCL, the catalyst can also become inactivated when [H2O2] is too high. Taken together, three different effects can cause the decay of light emission, resulting in flash-type kinetics of eCL.

The comparison of the eCL catalysts’ efficiencies (Figure ; and Supporting Information Figure S3) reveals that hemin is by an order of magnitude less efficient than the heme proteins Hb and Cyt c. This difference is largely independent of the pH (Supporting Information Figure S3). A possible explanation of this effect may be the dimerization of free hemin which occurs in aqueous solutions.[48] Such dimerization can alter the local electronic structure at the Fe center and hence inhibit its catalytic efficiency.[49] This, in consequence, causes less light when luminol is the substrate of the catalyzed reaction.

Proposal: a Modified FHW-Cycle (mFHW) Drives iCL

Although the efficiency of free iron catalysis (iCL) is several orders of magnitude lower at pH = 9, when compared with porphyrin-coordinated iron (Figure ), the luminol reaction can be used to assay trace amounts of iron or other transition metals.[10,50−52] However, the precise mechanism of the iron-catalyzed luminol–H2O2 reaction has not yet been elucidated. An attempt to fill this gap is summarized in Figure . Its deduction, all relevant reaction equations, and references are provided in the Supporting Information (SI 2.3; eqs S2.1–S2.5; SI2.5; eqs S2.6–S2.8).

Figure 12

Possible mechanisms explaining how the iCL luminol reaction is catalyzed by free iron through an mFHW. (A) Free iron in aqueous solution is able to catalyze the luminol reaction. At high pH, an oxo-ferryl iron complex (FeIV) is formed in the presence of H2O2. FeIV has the oxidation power to abstract electrons from luminol (LH–). The resulting ferric iron (FeIII) is recycled, either by H2O2 producing superoxide (O2•–) or by superoxide producing molecular oxygen (Haber–Weiss path in pink). (B) Alternatively, superoxide can possibly react with the luminol radical (L•–) (yellow path), directly leading to the light donor (AP*) without an azaquinone (AQ) intermediate.

The two mechanisms (Figure A,B) just differ in the final steps of the light-yielding reaction, and they may run alternatively or even in parallel. Luminol radicals can exchange electrons with each other (Figure A) to produce AQ, which reacts with peroxide anion to AP* to yield light, and they can react with superoxide (Figure B) to directly form AP* without the loop way via AQ.

Role of Superoxide in iCL

The proposed mechanism depicted in Figure B, however, favors the interaction of the luminol radical (L•−) with superoxides (O2•–).[40,53−56] The short-lived endoperoxide (details in ref (3)) is thus necessary to form the excited AP* directly, with no AQ intermediate (Figure B; refs (51) and (57)) according to eq

Scavenging of superoxide has a negative effect on iCL (Figure ), suggesting that superoxide is indeed involved in iCL. There are three possible ways to explain this inhibiting effect of SOD:

SOD abrogates the second route of ferric iron (FeIII) reduction (Haber–Weiss path in pink; Figure ), thus limiting the pool of ferrous iron (FeII) and in consequence, lowering the amount of active ferryl oxo-iron (FeIV).

When there is no second route of FeIII → FeII reduction (pink in Figure ), FeIII accumulates and can precipitate as insoluble FeIII hydroxide, a process which is promoted by the alkaline and aerobic molecular environment.[58,59] Thus, the reaction suffers from catalyst depletion when SOD is present. Such depletion is reduced in the presence of EDDHA (brown symbols in Figure ) as this iron chelator partially prevents FeIII hydroxide precipitation.

With no superoxide, luminescence can only be produced by L•– via the AQ intermediate (Figure A). The alternative reaction (yellow in Figure B), that is, the interaction of superoxide with L•− which shortcuts the loop via AQ, is blocked by SOD.

Role of Superoxide in eCL

The eCL reactions catalyzed by HRP, Cyt c, and hemin show the opposite effect when SOD is involved. The light yield increases by up to 40% in the presence of SOD (Figure ). This is probably an effect caused by dissolved molecular oxygen (O2) in the assay, which is present at a concentration of about 245 μM under equilibrium conditions, at 28 °C and at normal atmospheric pressure (http://water.usgs.gov/software/DOTABLES/) and thus at a concentration in the same order of [H2O2] and [LH–]. Peroxidases are also able to drive an oxygenation cycle,[46,60] as already mentioned above. Thereby, superoxide (O2•–) is produced at the price of a reduced peroxidase cycle turnover and reduced L•– formation. In the presence of SOD, however, the O2•– produced is converted into H2O2 which promotes the peroxidase cycle and thus increases L•– production and generates more light.

Kinetic Analyses of Luminol Luminescence Reveal Heterodyning of iCL and eCL

The definition and distinction between glow-type and flash-type luminescence and their kinetic modeling by mixed exponentials is explained in the Supporting Information (SI 5; Figures S5.1–S5.5; eqs S5.2–S5.5). The kinetics of glow-type luminescence can be modeled with a single exponential (Supporting Information eq S5.2) leveling off when t → ∞. The glow kinetics produced with free iron ions (Figure ) are well-represented by this single-exponential model (Supporting Information Figure S5.2).

Usually, pure flash-type kinetics can be described by a single exponential decay (Supporting Information eq S5.3; Figure S5.3). Here, however, the flash-type kinetics are biphasic (Figure ; and Supporting Information Figure S6) and consequently require at least two exponentials (Supporting Information eq S5.5; Figure S5.5). The flash kinetics of hemin, HRP, and Hb are to some extent reflected by such a dual-exponential function (Supporting Information Figure S5.5). Heme-catalyzed luminescence ceases within the first minute of the reaction with a time constant of τ ≈ 30 s (Figure ; and Supporting Information Figure S6). This is followed by a glow-type luminescence which may last many minutes. This strongly suggests that iCL and eCL occur in parallel. eCL proceeds with flash kinetics, and during this process, heme is destroyed by H2O2, liberating iron ions and so launching iCL with glow kinetics. This view is substantiated by the results depicted in Figure and in Supporting Information Figure S7, where reactions performed at high [H2O2] and in the presence or absence of a strong iron chelator (EDTA; pKFe = 25) are compared. In the presence of EDTA, a second exponential in the kinetics representing the glow is missing (Figure ; and Supporting Information Figure S7); consequently, there is less luminescence at high [H2O2] (Figure ; and Supporting Information Figure S4).

With a double-mixing stopped flow system, Saha et al. (2003) showed that flash kinetics occurred when an iron porphyrin catalyst was mixed with luminol prior to the addition of H2O2, whereas glow kinetics were obtained when H2O2 and the catalyst were mixed before the reaction was started with luminol.[61] A probable explanation for this result is that the latter course of action allowed iron to be liberated from the catalyst to produce a glow, whereas the former injection sequence started the luminescent reaction with intact peroxidase before iron liberation. Similar experiments were performed in the course of this study with a plate reader-based assay. The Supporting Information (Figure S10) verifies the findings of Saha et al. (2003)[61] and underpins the conclusions about iron liberation drawn above.

Destruction of Heme by H2O2 as Monitored by Soret Band Absorbance

The destruction of the iron porphyrin complex by high [H2O2] mentioned above is directly monitored here by the Soret peaks absorbance. The Soret peaks of all investigated heme compounds disappear with effective peroxide concentrations (EC50) in the range between 1 and 10 mM, giving 50% absorbance reduction (Figure and Supporting Information Figures S9.1 and S9.2). This effect is in agreement with the range of the IC50 values (Table S1) found for the suicide inactivation effect of luminescence (Figures and 2, and Supporting Information Figures S1.1 and S1.2) and provides further evidence that heme-based catalysis of the luminol reaction is transformed into a free iron-based catalysis when [H2O2] > 1 mM.

Conclusions

Two different catalytic mechanisms can occur simultaneously during a luminol reaction set up with a heme compound as the catalyst. The first is catalysis with high light-yield efficiency and flash-type kinetics based on the heme compound and a peroxidative cycle (eCL). The other is catalysis based on a ferryl-oxo-iron complex with low efficiency and glow-type kinetics (iCL). When [H2O2] is above [luminol] at pH = 9, iron is liberated from its porphyrin coordination during the reaction. This may occur unintentionally in quantitative luminol-based assays and thus impair their reproducibility and precision. Thus, the design of quantitative assays based on the luminol reaction requires careful considerations of disturbing side effects and has to comply with a delicate balance between luminol and H2O2 in order to avoid iron liberation from heme and to obtain a long-lasting luminescence suitable for quantitative intensity recording.

Impurities in the matrix of the samples to be assayed may also influence the reliability of luminol-based assays. In particular, the presence of reducing agents has a pronounced effect on light yield and kinetics, as has been demonstrated previously.[3]

Experimental Section

Chemicals

The following chemicals were used: calcium chloride (CaCl2·6H2O), Roth #T886; Cu/Zn-SOD from bovine liver, Sigma #S8409; Cyt c from equine heart, Sigma #C2506 and BioChemica/Fluka #30400; DMSO, Roth # 4720; EDTA, Aldrich #E2,628-2; FeEDDHA 138, Duchefa #F0527; ferritin, Sigma #F4503; ferrous sulfate (FeIISO4·7H2O), Roth #P015 and Merck #3965; ferredoxin from Chlamydomonas reinhardtii; ferric nitrate (FeIIINO3·9H2O), Fluka #44949; heminchloride, Roth#7629; Hb from bovine blood, Sigma #H2500; hydrochloric acid (HCl), Roth #4625; hydrogen peroxide (H2O2), Roth #8070 and Merck # 1.08597; luminol, Roth # 4203; mannitol, Roth #4175; peroxidase from horseradish (HRP), Sigma #P6140; potassium hydroxide (KOH), Roth #5658; and Tris ultrapure, ICN Biomedicals #77861.

Stock Solutions and Buffers

10× Tris-Stock: Tris/HCl 1 M, adjusted at 28 ± 2 °C to desired pH; Tris working buffer: Tris/HCl 100 mM diluted 1:10 from Tris-Stock; KOH 5 M stock in H2O; luminol stock 2 M in 5 M KOH; TriLu buffer: 1 mM luminol in Tris working buffer diluted from luminol stock; and starter solution for luminol reaction: H2O2 in Tris working buffer.

Specific Instrumentation and Materials

A plate reader Infinite M200 PRO with an injection unit (Tecan, Crailsheim, Germany) was used for optical readout, and 96-well microtiter plate (MTP) white flat bottom (Greiner BioOne #655075) and 96-well MTP transparent flat bottom (Sarstedt #82.1581) were used for luminescence and absorbance recording, respectively.

Methods

Luminescence and absorbance recording was performed at 28 °C. Prefilled MTPs were incubated for 1 h at 28 °C with sporadic shaking prior to recording. Luminol reactions were started by dispensing the complementary assay compounds to prefilled wells.

To study luminol-derived light yield and its dependence on hydrogen peroxide concentration and pH, four H2O2-dilution series (1:2) were established on an MTP. Dilution series were set up on two rows (150 μL in each well). After incubation (1 h at 28 °C), reactions were started by dispensing 50 μL of the catalyst dissolved in TriLu buffer.

To investigate catalytic efficiencies, white MTPs were prefilled with dilution series of the respective catalyst in TriLu buffer (150 μL in each well) and reactions were started by injecting 50 μL of H2O2 starter solution.

For absorbance measurements (Soret peak monitoring), transparent MTPs were prefilled with H2O2 dilution series as described above and the respective heme compound was added to the concentrations, as indicated in the figure insets. MTPs were incubated for 1 h at 28 °C with sporadic shaking prior to absorbance recording.

Detailed assay conditions for each experiment are given in the figure legends.