1 Introduction

Protein misfolding cyclic amplification (PMCA) was developed with the objective of detecting very small amounts of protease-resistant prion protein (PrPSc), a surrogate marker of transmissible neurodegenerative diseases such as Creutzfeldt Jakob Disease (CJD) and bovine spongiform encephalopathy. The process of PMCA involves seeding natively-folded PrPC (from whole brain, cell culture, or recombinant expression) with a minute quantity of infectious PrPSc, followed by repeated cycles of incubation and sonication [1]. A templating interaction of PrPC with PrPSc causes seeded protein misfolding and elongation of the PrPres polymer. The 20 kHz sonication step in PMCA was introduced to fragment the growing polymer, exposing more PrPSc to the substrate and producing an exponential increase in prion replication [2]. PMCA is therefore viewed as a means to accelerate the conversion process that occurs in vivo, enabling the presence of PrPSc in the original titre to be inferred upon immunodetection of the amplified PrPres. Contamination of otherwise non-infectious samples, or the de novo (spontaneous) generation of infectious PrP from non-infectious substrate, can produce false positives when using PMCA to screen for prion disease. De novo prion formation [3,4,5,6], as opposed to propagation/amplification of an existing prion [7, 8], requires RNA and lipid as essential cofactors, and has not been observed in vitro in the absence of sonication. The molecular mechanism underlying this phenomenon remains unknown, but we have proposed that it involves a confluence of structural (membrane-mimetic) and chemical (free radical) conditions present during the sonication step of PMCA [9]. We previously demonstrated that these radicals and reactive oxygen species (ROS) created during 20 kHz sonication lead to formation of PrP- and RNA-centred radicals, RNA oxidation, protease-resistant covalent PrP-RNA adducts [9]. In this study, we sought to quantify oxidative damage to these and other components of the PMCA substrate (including Tris, Triton X-100) by measuring their conversion to aldehydes by sonolytically-produced hydroxyl radicals. The results demonstrate that even bystander molecules undergo radical damage during the PMCA method, which can lead to a number of cross-linking interactions that may underpin the stochastic creation of synthetic PrP-containing molecules that possess infectivity to wild type mice (de novo prions). Notably, the formation of aldehydes offers a means to calibrate the power delivered to substrates by individual sonicators, which can improve the correlation of results between laboratories and may also correlate the frequency of de novo events with the delivered sonicator power and oxidative damage.

2 Materials and Methods

2.1 Reagents

Tris-buffered saline (TBS; 10 mM Tris, 150 mM NaCl, pH 7.5) was obtained from G-BioSciences. Bis–Tris (bis(2-hydroxyethyl)amino-tris(hydroxymethyl)methane), acetylacetone and 2-Oleoyl-1-palmitoyl-sn-glycero-3-phospho-rac-(1-glycerol) sodium salt (POPG), ammonium acetate and acetic acid were obtained from Sigma-Aldrich. Peroxide-free 10% w/v aqueous Triton X-100 (octylphenol ethylene oxide condensate) solution was obtained from Amresco. 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) was obtained from Alexis Biochemicals and used without further purification. Bacterially-expressed α-folded recombinant murine PrP23–231 was obtained from Prionatis. Mouse total RNA (whole brain) was purchased from Agilent Technologies. Nuclease-free water was from Qiagen. Thin-walled, RNAase-free 0.2 mL PCR tubes were obtained from Scientific Specialties Inc. Deionised water had a resistivity of 18 MΩ cm (Milli-Q®, Millipore).

2.2 Sample Preparation

Samples contained either buffer alone, Triton X-100 in buffer, complete PMCA substrate preparation used for creating de novo prions from recombinant PrP. The PMCA substrate preparation followed the original protocol used for de novo prion experiments [3, 4, 9] except for the use of total RNA from whole mouse brain in place of total RNA from mouse liver, and the preparation of SUVs by extrusion through a 50 nm polycarbonate membrane (Avanti Polar Lipids); final reagent concentrations: [PrP23–231] = 0.2 μM, [RNA] = 0.03 mg/mL, [POPG] = 6 μM, [Triton X-100] = 4.3 mM (0.28% w/v), [Tris HCl] = 10 mM, [NaCl] = 150 mM, pH 7.5. All samples had 100 μL volume and were contained in sterile PCR tubes.

2.3 Sonication

Sample tubes were sealed around the lid using Parafilm (Bemis Company, USA) and placed in a plastic microcentrifuge tube adapter (part # 444) positioned above the microplate horn accessory (part # 431MPH) of a model S-4000 sonicator (Misonix, USA) (Fig. 1a). Four replicate sample tubes were positioned in the centre of the adapter tray with one empty sample position separating each tube (Fig. 1a). The horn reservoir was filled with 200 mL distilled water and the PCR tubes positioned ca. 1 mm above the surface of the horn (Fig. 1b). Sonication powers in the range 0–300 W were applied by adjusting the amplitude setting on the Misonix display to achieve the approximate power desired. To mimic the PMCA procedure, 48 cycles of 20 kHz sonication were applied for 30 s every 30 min (Fig. 1c), corresponding to one round of PMCA. The average power delivered over the entire 48 cycles was determined by dividing the total energy delivered (read from the Misonix display) by the sonication time (48 × 30 s = 1440 s). After the completion of the cyclic sonication, sample tubes were transferred to − 80 °C storage until further analysis.

Fig. 1
figure 1

a Experimental setup of the ultrasonic processor (Misonix, S-4000) used for protein misfolding cyclic amplification (PMCA), with polypropylene PCR tubes positioned in the centre of the Perspex tube holder (Misonix, #444). b Tubes containing 100 μL substrate were positioned approximately 1 mm above the surface of a microplate horn (Misonix, #431MPX, 15 cm diameter) filled with 200 mL deionised water, and c 48 cycles of 20 kHz sonication (30 s) and incubation (29.5 min) were applied (equivalent to one round of PMCA). d Hantzsch reaction between an aldehyde (RCHO), a β-diketone (R1CHOCH2CHOR2), an amine (R3NH2), producing a coloured (Amax = 412 nm). To measure aldehydes produced during PMCA, acetylacetone (R1=R2=CH3) and ammonium acetate (R3=H) were used as the Hantzsch reagent

Full size image

2.4 Aldehyde Measurement

Two independent methods were used. The first employed a proprietary fluorescent formaldehyde detection kit (Invitrogen, EIACH2O) and the second used a colorimetric/fluorometric assay based on the established Hantzsch reaction (Fig. 1d) [10,11,12].

Using the proprietary formaldehyde detection kit, standard samples (0–200 μM) were freshly prepared in deionised water using the supplied 2 mM formaldehyde standard. Fifty microliters of each sample (in quadruplicate) or standard (in duplicate) was combined with 25 μL proprietary detection reagent in the wells of a black, half-volume, 96-well plate. The plate was incubated at 37 °C for 30 min, then fluorescence of each well was measured using a microplate reader (FLUOstar Optima; BMG Labtech) fitted with 450 nm excitation and 510 nm emission filters. Formaldehyde concentration of each sample was determined from the average fluorescence of the four replicate wells, with reference to a calibration curve generated from the average fluorescence of formaldehyde standards measured in duplicate wells. Each assay was repeated three times (n = 3).

For detection of aldehydes using the Hantzsch assay, the Hantzsch reagent was prepared by combining 1.54 g ammonium acetate, 28.6 μL glacial acetic acid and 20.5 μL acetylacetone with deionised water to a final volume of 10 mL (final concentrations of 2 M ammonium acetate, 0.05 M acetic acid, 0.02 M acetylacetone). Ten microliters from each of the four replicate samples were combined in a 1.5 mL microcentrifuge tube, then 20 μL Hantzsch reagent was added. Forty microliters of formaldehyde standards (prepared using the same 2 mM standard supplied with the fluorescence detection kit) were also combined with 20 μL Hantzsch reagent. The tubes were incubated in 37 °C water bath for 40 min, then the absorbance at 412 nm was read using a SPECTROstar Nano UV–Vis spectrometer (BMG LABETCH) with a 50 μL disposable cuvette (UVette, Eppendorf). The aldehyde concentration in each sample was determined using a calibration curve generated from the formaldehyde standards. Interference from alkylamines in the sample (buffer molecules, protein, and nucleic acid) is low because of their relatively slow rate of reaction compared with excess ammonium ions [10].

2.5 Statistical Analyses

GraphPad Prism version 8.4.2 for Windows (GraphPad Software, San Diego, California USA, www.graphpad.com) was used for all statistical comparisons and segmental regression analysis. Residual analysis of some samples failed tests of normality required for parametric analysis of variance (ANOVA). Therefore, nonparametric two-way ANOVA was performed using an aligned rank transform followed by a full-factorial ANOVA of the transformed data, as described by Wobbrock et al. [13]. In contrast to parametric methods, the aligned rank transform does not permit cross-factor pairwise multiple comparisons; therefore, post hoc comparisons were made only within a single factor (no interaction). Values quoted in the text correspond to mean ± SEM unless stated otherwise.

3 Results

3.1 PMCA Produces Formaldehyde from Tris Buffer

Hydroxyl radicals have been reported to degrade a range of buffer ions containing hydroxymethyl or hydroxyethyl groups into formaldehyde [14]. We have previously demonstrated that hydroxyl and hydrogen radicals are produced during 20 kHz sonication [9]. Therefore, we began by applying 24 h of cyclic sonication (30 s) and incubation (29.5 min) to Tris-buffered saline (equivalent to one round of PMCA). As anticipated, formaldehyde was detected at the end of the round (Fig. 2a). Although the concentration of 33 ± 1 μM formaldehyde is low compared with the total Tris concentration (10 mM), it is 2–3 orders of magnitude higher than the concentration of PrP used for PMCA assays [1, 3].

Fig. 2
figure 2

Formaldehyde measured after 24 h incubation at 37 °C with (+) or without (−) cyclic sonication (30 s at 300 W every 30 min) of a Tris-buffered saline (TBS) and b bis–Tris-buffered saline (BTBS) (n = 3, mean ± SD). Using two-way non-parametric ANOVA, the main effects of cyclic sonication (F1,8 = 26.31, p < 0.0009) and buffer type (F1,8 = 26.31, p < 0.0009) are equally significant, and their interaction is also significant (F1,8 = 26.05, p < 0.0009)

Full size image

The mechanism of formaldehyde formation from Tris remains unclear, however the primary amine group has been identified as a site of hydroxyl radical attack. To assess the role of the amine, TBS was substituted with bis–Tris buffered saline (BTBS; 10 mM bis–Tris, 150 mM NaCl, pH 7.5). Bis–Tris was previously reported to produce negligible formaldehyde in the presence of a Fenton system [14]; however, in the present case, sonication of bis–Tris generated 20 ± 1 μM formaldehyde (Fig. 2b), suggesting that the tertiary amine of bis–Tris is merely a less facile site of hydroxyl radical attack or that different free radical pathways are possible during 20 kHz sonication. Interestingly, the addition of the radical-trapping agent DMPO (100 mM) to TBS resulted in a higher yield of formaldehyde than sonication of TBS alone (Figure S1). This suggests that either a minor fraction of DMPO is itself degraded to a (form)aldehyde product or that DMPO intercepts a fraction of the hydroxyl radicals that would otherwise oxidise the TBS-derived formaldehyde to formic acid [15]. We do not concern ourselves further with such matters here, instead focussing on the empirical observation that the buffer used to create synthetic de novo prions from recombinant PrP [3, 4] is not an inert bystander during cyclic 20 kHz sonication.

3.2 Formaldehyde Production from Tris Buffer Correlates with Ultrasonic Power

We next assessed the formaldehyde production as a function of sonication power. Non-negligible formaldehyde levels were measured only when the average sonicator power exceeded a value of 160 ± 10 W/cm2, above which the formaldehyde concentration increased monotonically (Fig. 3). This value is ascribed to the threshold for achieving transient cavitation, which is responsible for sonochemical effects [16]. Although many factors affect the threshold power (temperature, heat capacity of dissolved gases, solution viscosity, surface tension, vapour pressure, the geometry and thickness of sample tubes, and their positioning relative to the transducer), the measured transition point is comparable with powers typically required to achieve cavitation at 20 kHz [17], and with the minimum average power necessary for efficient amplification of PrPSc using the same ultrasonicator [18]. Comparison of segmental regression analyses yielded no difference between the Hantzsch assay (Fig. 3a) and the proprietary fluorometric assay (Fig. 3b). The relatively large variation in the measured formaldehyde concentration at 200 W suggests that operation close to cavitation threshold produces an outcome that is highly sensitive to the precise sample positioning. Figure S2 shows the same data together with the 95% confidence intervals, from which it is clear that PMCA experiments operating near the cavitation threshold can provide a major source of inter-assay variation.

Fig. 3
figure 3

Formaldehyde measured after 24 h of cyclic 20 kHz sonication (30 s at 250 W) and incubation (29.5 min) of TBS at 37 °C using different ultrasonic powers (n = 3, mean ± SD). Results from a the Hantzsch assay are compared with those from b a proprietary formaldehyde detection kit. The solid and dashed lines show the segmental linear regression and the associated 95% confidence intervals; the cavitation threshold is (a) 160 ± 9 W and (b) 161 ± 9 W, above which formaldehyde is produced at (a) 0.22 ± 0.02 μM/W and (b) 0.23 ± 0.02 μM/W. Goodness of fit parameter: (a) R2 = 0.927. (b) R2 = 0.933

Full size image

3.3 Aldehyde Production from Substrate Molecules During PMCA

The Hantzsch reaction proceeds for all aldehydes (Fig. 1d), including those formed during the oxidative damage to substrate molecules used for de novo creation of synthetic prions (PrP, nucleic acid, detergent, lipid). Triton X-100 is a non-ionic surfactant that is incorporated into PMCA protocols to promote solubility of misfolded PrP [3, 19]. With the possible exception of buffer molecules, it is usually present at the highest concentration of any cofactor. It can also contain up to 2% impurities from carbonyl compounds such as carboxylic acids, ketones or aldehydes. Therefore, we determined the aldehyde content of our source of Triton X-100 at the concentration used for creating synthetic de novo prions (0.28% w/v [3, 4]) in TBS. Untreated Triton X-100 was found to contain a relatively modest concentration (6.7 ± 0.2 μM) of aldehydes (Fig. 4). Surfactant molecules are known to accumulate radially at the gas–liquid interface of cavitation bubbles [20], exposing Triton X-100 molecules to the highest concentration of free radicals. Consistent with a high degree of oxidative degradation, 185 ± 16 μM aldehydes were measured after 24 h cyclic 20 kHz sonication (30 s at 250 W) and incubation (29.5 min) of 0.28% w/v Triton X-100 (Fig. 4). The untreated PMCA substrate (PrP, RNA, POPG, Triton X-100 in TBS) contained 16 ± 12 μM aldehydes (Fig. 4), which increased to 380 ± 63 μM after 24 h cyclic sonication. This value is approximately twice that measured for Triton X-100 alone, suggesting that half of the aldehyde content arises from oxidative damage to PrP, RNA, and POPG.

Fig. 4
figure 4

Comparison of aldehyde concentration measured after 24 h incubation at 37 °C with (+) or without (−) cyclic sonication (30 s at 300 W every 30 min) of a TBS, b 0.28% w/v Triton X-100 in TBS, and c PMCA substrate (0.2 μM PrP23–231, 0.03 mg/mL total RNA, 6 μM POPG, 0.28% w/v Triton X-100, TBS pH 7.5). Total aldehydes were measured as equivalents of formaldehyde using the Hantzsch reaction (n = 3, mean ± SD). Using two-way nonparametric ANOVA, the main effects of cyclic sonication (F1,12 = 38.15, p < 0.0001) and sample composition (F2,12 = 51.16, p < 0.0001) are significant, and their interaction is also significant (F2,12 = 54.50, p < 0.0001). Within-effects analysis of sample composition (averaged over the levels of sonication): TBS vs Triton X-100, p = 0.0008; TBS vs PMCA substrate, p < 0.0001; Triton X-100 vs PMCA substrate, p = 0.0008 (Tukey multiple comparison tests)

Full size image

4 Discussion

Ultrasonication exposes PrP and cofactor molecules to primary hydroxyl and hydrogen radicals, together with a range of secondary radicals and oxidants, at or near the surface of cavitation bubbles. The hydroxyl radical is a highly reactive oxidant that can undergo addition reactions (hydroxylation) with C=C and C=N bonds (such as those found in purines of nucleic acid) and abstract H atoms from C–H and N–H groups, yielding carbon- and nitrogen-centred radicals [21]. The complexity of the ensuing radical cascades increases in proportion to the number of reactants and their molecular weight, making it impossible to fully understand the advanced oxidation processes that occur [22,23,24]. However, it is clear that PMCA exposes PrP to more than the fragmenting and templating steps that are believed to underpin the process of PrPres.

Here, we focus on aldehydes produced by cyclic sonication, which can be measured using the Hantzsch reaction. Hydroxyl radicals produce formaldehyde from many Good’s [14] buffers, including Tris and HEPES, both of which are used as PMCA conversion buffers [3, 19]. In the present case, approximately 5% of the total aldehydes produced during PMCA can be ascribed to formaldehyde derived from TBS alone (Fig. 4a), with formaldehyde concentration monotonically increasing as a function of sonicator power above the cavitation threshold (Fig. 3). Approximately half of the total aldehyde can be attributed to oxidation of Triton X-100 (Fig. 4b); hydroxyl radical addition to the hydrophobic octylphenol head group of Triton X-100 and abstraction of hydrogen atoms from the hydrophobic poly(ethylene oxide) chains have been reported [25], although precise oxidation products were not identified. The remaining proportion of aldehydes produced during PMCA results from oxidation of protein, nucleic acid, and lipid (Fig. 4c). Oxidation of protein side chains leads to formation of carbonyls (aldehydes and ketones) [26, 27]. Hydroxyl radicals can cleave the glycosyl bond of nucleic acid, creating an apurinic/apyrimidinic site whose resonance structures involve an open-chain aldehyde [28]. Similar to Triton X-100 [20], POPG will accumulate radially at the gas–liquid interface of cavitation bubbles, exposing the fatty acid chains to hydroxyl radicals [9] that can readily react with the double bond of the oleoyl chain to yield an aldehyde [29].

In some PMCA protocols, a higher concentration of 1% w/v Triton X-100 is used [19]. In such cases, aldehyde production from the detergent could exceed that produced from all other PMCA substrate molecules. Ethylenediaminetetraacetic acid (EDTA) is included in some protocols for creating synthetic de novo prions [5]. Here, hydroxyl radicals attack can EDTA, yielding nitrogen- and carbon-centred EDTA radicals that lead to formation of a number of products including formaldehyde and the aldehyde [(2-Oxoethyl)imino]diacetic acid [30].

Both aldehydes and ketones can undergo reversible nucleophilic addition reactions with amines (e.g. lysine ε-amino group, endo- and exocyclic amino groups of nucleobases [31,32,33]) to form imines [34] that can react with a nucleophile from a second molecule to yield a reversible covalent cross-link [35,36,37,38]. The nucleophilic addition reactions underpinning imine formation are favoured at a mildly acidic pH [34]. Oxidation of dissolved N2 in aerated solutions forms nitric oxide and nitrogen dioxide radicals that react with hydroxyl radicals to produce nitrous (HNO2) and nitric (HNO3) acid, respectively [39, 40]. Lower pH can also be promoted by the formic acid produced from the oxidation of formaldehyde by hydroxyl radicals [15]. Because these reactions take place near the gas–liquid interface, where the transient hydroxyl radicals are most concentrated, this can reduce the local pH to a value that promotes imine formation and cross-linking even if the bulk pH remains relatively stable. The low pH at the interface may also promote PrP misfolding [41, 42] and prion formation [43].

Reactive oxygen and nitrogen species also lead to many other protein-nucleobase cross-links [44]. Peroxynitrite is formed by the reaction of nitric oxide (vide supra) with superoxide (derived from secondary reactions of hydroxyl radicals) that can diffuse away from cavitation bubbles into the bulk solution [40] and cause protein nitrosylation and RNA damage [45]. Formic acid produced by oxidation of formaldehyde [15] can also covalently modify protein side chains, particularly by O-formylation of serine [46].

4.1 Relevance to Creation of De Novo Synthetic Prions and Prion Diagnostics

Because the above phenomena are triggered by primary hydroxyl radicals, all are physiologically relevant. Although some of the formaldehyde generated during PMCA is derived from non-physiological molecules (e.g. Tris), the effects of this formaldehyde on other substrate molecules may be relevant to the spontaneous creation of de novo prions from non-infectious recPrP substrate [3]. Formaldehyde results from many physiological processes [47,48,49] and an imbalance of endogenous formaldehyde metabolism has been proposed as a trigger for age-related cognitive decline because of its ability to cause β-amyloid aggregation and tau hyperphosphorylation in vivo [50].

Because aldehydes are formed in response to free radicals at the surface of cavitation bubbles, the Hantzsch assay offers a means to correlate these sonochemical phenomena with de novo prion generation. Quaking-induced conversion (QuIC) methods, involving extended shaking of the substrate, have been used to induce the misfolding of recPrP into thioflavin-T-reactive amyloid in the presence of infectious seeds [51], but QuIC products do not possess infectivity [52]. By operating in a power range below the threshold for aldehyde production (i.e. below the cavitation threshold), mechanical fragmentation of the PrP aggregates could be achieved without the unique physicochemical phenomena that accompany cavitation. If de novo prion generation is eliminated under such conditions but does not prevent the amplification of pre-existing PrPSc seeds, then ‘false positives’ could be avoided when using PMCA as a diagnostic tool; moreover, this would provide support for the hypothesis that de novo generation of synthetic prions relies on free radical-mediated phenomena at the surface of membrane-mimetic cavitation bubbles [9].

4.2 Comparison with Other Methods for Measuring Ultrasonic Power

A calorimetric method can be used as a measure of acoustic power [53]. However, this requires the local temperature change of the sample solution to be measured, which is impractical for PMCA, where the sample tubes usually contain very small volumes, many such tubes are sonicated in parallel, and ultrasound energy is transferred indirectly via a water bath. Another surrogate method of measuring ultrasonic power relies on the reaction of sonochemically liberated H2O2 with KI, which is oxidised to I3− and spectrophotometrically detected at 353 nm [53,54,55]. The KI can be added before or after sonication [55], although its inclusion during sonication is an undesirable modification of the usual PMCA substrate preparation. Addition of KI after sonication will measure any H2O2 remaining at the end of the PMCA round, but this may not correlate with the cumulative free radical damage to substrate molecules.

The calorimetric and KI oxidation methods have been used to map the ultrasonic power at specific positions across the base of a sonicating water bath [56], and this pre-calibration was used to infer the energy delivered to recPrP samples that were subsequently positioned at those same locations. These methods were shown to be excellent measures of sonochemical efficiency, and could be used to correlate ultrasonic power with formation of specific types of PrP amyloid [56]; however, the pre-calibration provides only an indirect prediction of the energy absorbed by an individual PMCA sample, which varies stochastically. In comparison, the present method permits a direct measurement of the relative power delivered to an individual sample by quantifying the sonochemical production of aldehydes.

5 Conclusions

The ultrasonic field experienced by an individual sample during PMCA can vary significantly between and within instruments. The total aldehyde content, measured using the Hantzsch assay with either colorimetric [10] or fluorometric [11, 12] readout, can be used as a surrogate marker of ultrasonic power delivered to an individual sample. This offers a simple way to assess the importance of free radical phenomena in the creation of de novo prions, and to better correlate results obtained by different diagnostic laboratories.