Deduction of the evaluation limit and termination timing of multi-round protein misfolding cyclic amplification from a titration curve

Authors


Masanori Morita, Benesis Corporation, C/O Department of Neurological Science, Tohoku University Graduate School of Medicine, 2-1, Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan.
Tel: +81 22 717 8147; fax: +81 22 717 8148; email: morita-m@med.tohoku.ac.jp

ABSTRACT

In this study, the efficacy of disinfectants in reducing the partially protease-resistant isoform of prion protein was evaluated by a multi-round protein misfolding cyclic amplification (PMCA) technique. Hamster brains infected with scrapie-derived strain 263K were homogenized, treated under inactivating or mock conditions, and subjected to multi-round PMCA. Four sets of serial 10-fold dilutions of mock-treated samples were analyzed. Although considerable variability was observed in the signal patterns, between the second and sixth rounds the number of the PMCA round correlated in a linear fashion with the mean dilution factor of mock-treated, infected brains, corresponding to a log reduction factor (LRF) of 3.8–7.3 log. No signals were observed in the PMCA products amplified from normal hamster brain homogenates. The mean numbers of rounds at the first appearance of the signal for 1 M and 2 M NaOH-treated samples were 4.33 and 4, respectively. Using the linear regression line as the titration curve, the LRFs of these disinfectants were found to be 6.1 and 5.8 log, respectively; these values were not significantly different. The mean number of rounds for the alkaline cleaner and sodium dodecyl sulfate were 9 and 10.33, respectively, and were outside the range of both the linear regression line and evaluation limit. The disinfectants were considered very effective because their LRFs were ≥7.3 log. These estimations were concordant with previous bioassay-based reports. Thus, the evaluation limit seems to be valuable in some applications of multi-round PMCA, such as disinfectant assessment and process validation.

List of Abbreviations: 
263K-BH

brain homogenate prepared from 263K-infected hamster brain

BH

brain homogenate

CJD

Creutzfeldt-Jakob disease

LRF

log of reduction factor

N-BH

normal hamster brain homogenate prepared for PMCA substrate

NaOH

sodium hydroxide

Neg-BH

normal brain homogenate prepared for negative control

PK

proteinase K

PMCA

protein misfolding cyclic amplification

PrP

prion protein

PrPc

cellular isoform of PrP

PrPres

partially protease resistant isoform of PrP

PrPSc

disease associated isoform of PrP

Tris

tris-hydroxymethyl-aminomethane

The pathogenesis of prion disease involves conversion of a cellular isoform of the prion protein (PrPc) into a disease-associated isoform of PrP (PrPSc) (1). In healthcare facilities, inactivation or removal of prions is a vital measure for preventing iatrogenic transmission of prion disease. Bioassay, the only method for measuring residual infectivity after inactivation, has some drawbacks. Bioassays are expensive and time-consuming. Assessment of toxic reagents requires enormous dilution, which also reduces sensitivity. Sensitivity in animal models does not always reflect sensitivity in transmission between humans. Therefore, development of sensitive in vitro methods that can serve as surrogate or supplemental measures for bioassays is necessary.

Protein misfolding cyclic amplification is an in vitro method for producing large amounts of PrPres (2). PMCA has a sensitivity >4000 times greater than that of bioassays (3), and amplified products have been shown to be infectious (4). PMCA has been used to assess the efficacy of PrPSc removal, and the reported results have been concordant with the results of bioassays to assess inactivation efficacy (5, 6). However, many bioassay-proven disinfectants have not yet been assessed by PMCA. Moreover, before this method can be applied in practice, many aspects must first be quantified; for example, the termination timing of multi-round PMCA and its quantitative capabilities.

In the present study, multi-round PMCA was applied to assess the LRFs of PrPres by four agents, namely, an alkaline cleaner (mip-PC-M; Ecolab, Tokyo, Japan); 3% SDS; 1 M NaOH, (which is recommended by the World Health Organization) (7); and 2 M NaOH. The efficacy of these agents has previously been assessed by bioassays (8–10).

MATERIALS AND METHODS

Preparation of brain homogenates

A 10% (w/v) homogenate of hamster brain infected with hamster-adapted scrapie strain 263K (263K-BH) was prepared in PBS containing protease inhibitors cocktail (Roche Diagnostics, Mannheim, Germany) and sonicated with an analogue sonifier (S-250; Branson Ultrasonic, Danbury, CT, USA) for 1 min. A 10% (w/v) Neg-BH was prepared under the same conditions. Another 10% (w/v) N-BH, which was used for a PMCA substrate, was prepared by homogenization in PMCA buffer (PBS containing protease inhibitors cocktail, 1% TritonX-100, and 4 mM EDTA) followed by sonication for 1 min.

Prion inactivation procedures

263K-BH was treated with four disinfectants: an alkaline cleaner, SDS, and 1 and 2 M NaOH, as described below. Disinfectant or mock treatments of 263K-BH and disinfectant treatments of Neg-BH were performed in triplicate.

Treatment with alkaline cleaner

Ten μL of 10% 263K-BH was mixed with 90 μL of an alkaline cleaner containing 0.125 M NaOH (mip PC-M, Ecolab) and this mixture incubated for 30 min at 70°C. After incubation, the sample was neutralized to pH 7.5 with 9 μL 1 M HCl, and then 4.6 μL 1.2 M Tris-HCl (pH 8.0) was added. For a negative control, Neg-BH was treated in the same manner. For a mock treatment control, 10 μL 10% 263K-BH was mixed with 90 μL of an HCl pre-neutralized alkaline cleaner (pH 7.5) and incubated under the same conditions.

Sodium dodecylsulfate treatment

One % 263K-BH was incubated with 3.0% (w/v) SDS (total volume, 100 μL) for 5 min at 100°C and 4.6 μL 1.2 M Tris-HCl (pH 8.0) added. For a negative control, Neg-BH was treated in the same manner. For a mock treatment, 263K-BH was treated with distilled water instead of SDS.

Sodium hydroxide treatment

Sodium hydroxide treatment was performed under two conditions. Ten μL 10% of 263K-BH was mixed with 90 μL 1.1 or 2.2 M NaOH and incubated for 120 and 60 min, respectively, at 25°C. In both methods, samples were neutralized to pH 7.5 with 9 μL 10 M HCl, and 4.6 μL 1.2 M Tris-HCl (pH 8.0) was added. For a negative control, Neg-BH was treated in the same manner. A mock treatment was performed with distilled water instead of NaOH.

Multi-round protein misfolding cyclic amplification reaction

Disinfectant-treated 263K-BH (triplicate) and disinfectant-treated Neg-BH (triplicate) were diluted 1:100 in N-BH substrate, and serial 10-fold dilutions of mock-treated 263K-BH (in duplicate) were prepared by diluting with 10% N-BH substrate. The reaction mixture (total volume, 100 μL) was placed in a 0.1 mL thin-walled PCR tube with a screw cap (No. 72.733.200; Sarstedt, Numbrecht, Germany) and subjected to multi-round PMCA. A round of amplification consisted of 48 cycles of sonication (five pulses of 5 sec with 1 sec rest) and agitation (1 hr) at 37°C using a fully automatic cross-ultrasonic protein activating apparatus (ELESTEIN 070-GOT, Elekon Science, Chiba, Japan). After each round, the reaction products were diluted 1:10 in fresh 10% N-BH, after which the next round was started. For serially diluted mock-treated samples, multi-round PMCA was terminated when a signal was detected (with some exceptions).

Proteinase-K digestion and Western blotting

Samples were digested with PK for 30 min at 37°C; a concentration of 50 μg/mL was used for samples before they were subjected to PMCA, and 100 μg/mL for samples after PMCA. The digested samples were Western blotted with an anti-PrP monoclonal antibody (3F4) as described previously (11). Western blot signals were detected with a Versa Doc 5000 imaging device and signal intensities were quantified using Quantity One software (Bio-Rad Laboratories, Hercules, CA, USA).

RESULTS

Assessment of logs of reduction factor of the partially protease resistant isoform of prion protein by four disinfectants using multi-round protein misfolding cyclic amplification

We assessed residual amounts of PrPres after four inactivation procedures: alkaline cleaner treatment for 30 min at 70°C, 3% SDS treatment for 5 min at 100°C, 1 M NaOH treatment for 2 hr at 25°C, and 2 M NaOH treatment for 1 hr at 25°C. 263K-BH (1%) was treated with four inactivation procedures and neutralized. No signal was detected in undiluted, disinfectant-treated 1% 263K-BH, but strong signals were observed in 30-fold diluted, mock-treated 263K-BH (positive control) (Fig. 1a), indicating that the residual amounts of PrPres were below the Western blot detection limit. To eliminate inhibition by disinfectants carried over into the PMCA reaction mixture, the disinfectant-treated 263K-BHs or Neg-BHs were diluted 100-fold in 10% N-BHs, generating 10−4 dilutions of brain, and subjected to PMCA. To evaluate the inactivation efficacy or LRF of each disinfectant individually, PMCA of serial 10-fold dilutions of relevant mock-treated samples were performed simultaneously (dilution factors used are indicated in Figs. 1b–e).

Figure 1.

Western blot analysis of PrPres before and after PMCA. (a) Western blots of disinfectant-treated and mock-treated samples, containing 1% 263K-infected brain homogenate, after proteinase K digestion. Samples were undiluted (×1) or diluted 30-fold (×30) before Western blotting. The positive controls are mock-treated samples. (b–e) Western blots of PMCA products after proteinase K digestion at each round. Disinfectants include (b) an alkaline cleaner, (c) 3% SDS, (d) 1 M NaOH, and (e) 2 M NaOH. Arrowheads indicate the signal first detected from disinfectant-treated samples or from each dilution of mock-treated samples. The PMCA round number is indicated at the right top corner of each panel, and molecular mass of the protein bands (kDa) is indicated at the left of each panel. Dilution factors of mock-treated samples starting from the infected brain are provided at the top of each panel; control indicates negative control reactions. * indicates bands caused by insufficient proteinase K digestion. Connecting lines between panels indicate continuity of samples in multi-round PMCA.

Alkaline cleaner

We detected PrPres signals by Western blot in each round (Fig. 1b). Signals for two out of three alkaline cleaner-treated samples appeared in the ninth round (indicated by arrowheads in Fig. 1b). All signals for ≥−10 log dilutions of mock-treated samples appeared by the third round, and a signal for one of the −11 log dilutions appeared in the ninth round. Thus, we inferred that the residual amount of PrPres in the initial PMCA reaction mixture of alkaline cleaner-treated samples was ≤−11 log dilution. Because there was a −4 log dilution of 263K-infected brain in the initial reaction mixture, the LRF of PrPres by alkaline cleaner was calculated as ≥ 7 log.

Sodium dodecylsulfate

We detected PrPres signals by Western blot in each round (Fig. 1c). Two of the SDS-treated replicates generated signals in the tenth round, and the third generated signal in the eleventh round (indicated by arrowheads in Fig. 1c). All of the ≥−11 log dilutions generated signals by the tenth round. One −12 log replicate appeared in the fifth round and the other had not appeared by the eleventh round. Thus, we inferred that the residual amount of PrPres in the initial PMCA reaction mixture was ≤−12 log dilution. Because there was a −4 log dilution of 263K-infected brain in the initial reaction mixture, the LRF of PrPres by SDS was calculated as ≥8 log.

1 M or 2 M sodium hydroxide

We detected PrPres signals by Western blot in each round (Fig. 1d, e). Signals for two of the 1M NaOH-treated samples were detected in the fourth round, and the signal for the third replicate was detected in the fifth round (indicated by arrowheads in Fig. 1d). In the mock-treated samples, no signal appeared for the −8 log dilutions until the fifth round. Although signals for the −12 log dilutions appeared in the fifth round, signals for one of the duplicate −10 and −11 log dilutions appeared in the fourth round, while the others appeared in the sixth round (Fig. 1d). Thus, we estimated that the residual amount of PrPres in the initial PMCA reaction mixture of 1 M NaOH-treated samples was between −8 and −10 log dilutions. Because there was a −4 log dilution of 263K-infected brain in the initial reaction mixture, the LRF of PrPres by 1 M NaOH was calculated to be between 4 and 6 log.

One replicate each of the 2 M NaOH-treated samples generated signals in the third, fourth, and fifth rounds (indicated by arrowheads in Fig. 1e). One each of the −8 or −9 log-diluted mock-treated samples generated signals in the second and third rounds. All of the −10 to −12 dilutions generated signals between the third and fifth rounds. One −13 log-diluted replicate generated a signal in the fifth round (Fig. 1e). Thus, we estimated that the residual amount of PrPres in the initial PMCA reaction mixture of 2 M NaOH-treated samples was between −8 and −13 log dilutions. Because there was a −4 log dilution of 263K-infected brain in the initial reaction mixture, the LRF of PrPres by 2 M NaOH was calculated to be between 4 and 9 log. We performed all experiments in duplicate and generated the same results on both occasions.

Titration curve of protein misfolding cyclic amplification

The rough estimates described above are due to considerable variability in the rounds at which signals for the diluted mock-treated samples appeared. Signals for samples of higher dilutions did not always appear later than did those for lower dilutions (Fig. 1b–e). We concluded that accurate assessment of the efficacy of disinfectants would require a greater number of replicate samples for each serial 10-fold dilution.

We then combined the data from the four mock treatments, which we had performed under different conditions. We first compared residual amounts of PrPres between mock-treated 263K-BH before PMCA to study the direct effect of mock treatments (Fig. 2a). Signal intensities of PrPres in samples treated with distilled water for 5 min at 100°C (mock treatment for SDS) were about one-fifth of those in untreated samples (controls). Signal intensities for the remaining three mock treatments were comparable to those in the controls. We then compared amounts of amplified PrPres between mock-treated 263K-BH after the first round of PMCA to determine whether the slight difference in PrPres in the starting material affected the PMCA product yield (Fig. 2b). Signal intensities in the mock-treated alkaline cleaner and SDS experiments were comparable to those in the controls, while the signals for 1 M and 2 M NaOH were about half those in the controls. Thus, the slightly different amounts of PrPres in the starting materials did not affect the amplified product yield after a single round of PMCA. We then determined whether slight differences in amounts of PrPres in the starting materials affected signal appearances during repeated PMCA rounds; rounds at which signals for 10-fold serial dilutions of four mock treatments first appeared (indicated by arrowheads in Fig. 1b–e) are shown in Fig. 3a. If the SDS mock treatment affected PMCA performance, signal appearances for SDS mock-treated samples should have been delayed. However, no remarkable delay was observed, indicating that slight differences in the amount of PrPres after four mock treatments did not affect signal appearances in multi-round PMCA. Thus, we concluded that combined statistical analysis of the four sets of mock treatment data was reasonable.

Figure 2.

The effect of mock treatments on PrPres amplification by PMCA. (a) Western blot analysis of residual PrPres in samples after no treatment or four different mock treatments. Physical conditions of mock treatments are indicated above the bars. The Y axis indicates relative signal intensities (percentage) versus the mean signal intensities of the controls. (b) Western blot analysis of amplified PrPres after the first round of PMCA of 100-fold diluted samples shown in (a). Physical conditions of mock treatments are indicated above the bars. The Y axis indicates relative signal intensities (percentage) versus the mean signal intensities of the controls.

Figure 3.

Determination of the PMCA titration curve. (a) The rounds at which signals were first detected for each dilution factor of four mock treatments (indicated with arrowheads in Figs. 1b–e) are depicted. Open circles, alkaline cleaner; filled circles, 3% SDS; open triangles, 1 M NaOH; and filled triangles, 2 M NaOH. For example, in the second round of mock treatment for alkaline cleaner, one signal was detected for the 10−6, 10−7, and 10−9 log dilutions (Fig. 1b). Thus, in the row corresponding to “PMCA round” 2, one open circle is depicted on each line corresponding to “log 10 of dilution factors” of −6, −7, or −9. The mean value and SD of dilution factors of each round were calculated. *, number of PrPres positive samples per total number of samples tested; , mean and SD not calculated because no signals appeared in the first and eighth rounds; , SD not calculated because the number of signals was <3 in the seventh, ninth, tenth, and eleventh rounds. (b) Filled circles and error bars indicate the mean value and SD of the dilution factors, respectively. Error bars are not shown after the seventh round because the number of signals detected was <3. Solid line, linear regression line; dashed line, fitting curve; vertical lines, means of the rounds at which signals were first detected.

To study the correlation between dilution factor and round number, the mean and standard deviation of dilution factors for each round were calculated (the bottom line of Fig. 3a) and plotted (Fig. 3b). Between the second and sixth rounds, PMCA round strongly and linearly correlated with the mean of dilution factors (r2= 0.924; Fig. 3b, solid line). The linear regression line extends from the −7.8 to −11.3 log dilutions. We did not observe a linear correlation after the seventh round (Fig. 3b, dashed line). Round number means for alkaline cleaner-, SDS-, 1 M NaOH-, and 2 M NaOH-treated samples were 9, 10.33, 4.33, and 4, respectively (shown as vertical lines in Fig. 3b). As the latter two are included within the linear regression range (Fig. 3b), amounts of residual PrPres were estimated to be 10.1 and 9.8 log dilutions for 1 M and 2 M NaOH, respectively, using the linear regression line as a titration curve. Because there was a −4 log dilution of 263K-infected brain in the initial reaction mixture, the LRFs of PrPres by 1 M and 2 M NaOH were calculated to be 6.1 and 5.8 log, respectively. These results do not appear to differ significantly, particularly in light of the large variances. As the former two are outside the range of the linear regression line, we judged that residual PrPres in samples treated with alkaline cleaner or SDS are <11.3 log dilutions, which is the edge of the titration curve (Fig. 3a, bottom line). The LRFs of PrPres by these disinfectants were >7.3 log.

DISCUSSION

We used multi-round PMCA to estimate minute amounts of PrPres in 263K-infected hamster brain treated with four disinfectants. As expected, multi-round PMCA showed higher sensitivity, we could therefore perform it more quickly than a standard bioassay. The method was sensitive enough to detect signals from −13 log-diluted samples. This sensitivity is comparable to that reported in the literature (3). As the exact nature of the seed has not yet been elucidated, the relationship between seed and infectivity, or even that between seed and PrPres, is not fully understood. Therefore, we estimated the amounts of PMCA seed in samples treated with disinfectants, and calculated the LRF as the difference between the estimated amount of PMCA seed and the total PrPres input in the initial PMCA reaction mixture.

The LRF of alkaline cleaner assessed by PMCA was ≥7.3 log (Fig. 3b). Baier et al. observed no infectivity in 263K-infected hamster brain homogenate treated with alkaline cleaner (12). Thus, the deduced LRFs by PMCA is consistent with assessment by bioassays. The LRF of SDS assessed by PMCA was ≥7.3 log (Fig. 3b). Tateishi et al. observed no infectivity in mouse-adapted CJD strain treated with SDS (10). As the prion strains used in PMCA and bioassay were different, direct comparison is difficult, however the tendency of the effect of SDS treatment on either strain is similar. The LRF of 1 M NaOH was 6.1 log in this study (Fig. 3b). The infectivity in 263K-infected hamster brain homogenate treated with 1 M NaOH was <103.8 (calculated to be >5.5 LRF) (15). The inactivation efficacy of 1 M NaOH on CJD Fukuoka-1 strain as determined by mouse bioassay has been reported to be 4–6 (9, 13). In one report, the authors did not observe infectivity in 263K-infected hamster brain homogenate treated with 1 M NaOH (14), but the sensitivity of the bioassay was reduced in that study because the samples were diluted to reduce NaOH toxicity. Thus, we consider that our estimation is concordant with bioassay-determined LRFs. The LRF of 2 M NaOH in this study was 5.8 log (Fig. 3b). The inactivation efficacy of 2 M NaOH assessed by bioassay is somewhat enigmatic. Mice inoculated with CJD Fukuoka-1 strain treated with 2 M NaOH developed disease after a longer incubation time than did mice inoculated with samples treated with 1 M NaOH (9). Residual infectivity in 263K-infected hamster brain homogenate treated with 2 M NaOH for 120 min was 104.2 (calculated to be 5.1 LRF), but in samples treated with 1 M NaOH for 60 min, the residual infectivity was ≤103.8 (15). Nevertheless, the LRF for 2 M NaOH estimated by PMCA appears consistent with the LRF assessed by bioassay. Collectively, the PrPres LRFs for at least three procedures tested in this study were concordant with the LRFs of infectivity. These results suggest that PMCA can be used as a surrogate for, or supplement to, bioassay (5).

By analyzing the signal patterns of mock-treated samples, we found a linear correlation between PMCA rounds and infected brain dilution factors. The linear regression line can therefore serve as a titration curve. When the mean round number of disinfectant-treated samples was inside the range of the titration curve, we could deduce the amount of PMCA seed from the curve; when it was outside the range, the estimated amount remained constant, namely ≥7.3. We thus regard the edge of the titration curve as the evaluation limit. This is similar to the concept of assay system detection limits. Thus, there are two limitations of PMCA: the detection limit and the evaluation limit, which in this study were −13 log and −11.3 log dilution, respectively. Once the estimated amount of PMCA seed has become constant, we believe that continuing multi-round PMCA is fruitless. In other words, we recommend terminating multi-round PMCA after the round corresponding to the evaluation limit has been passed. The titration curve can be used in two ways: to deduce the amount of PMCA seed and to indicate the timing of termination.

The quantitative capability of the titration curve is somewhat compromised by the variability in signal appearances (Fig. 3a and b). We speculate that the reasons for this variability are as follows: (i) the efficiency of PMCA amplifications is intrinsically variable, probably due to variability in breakage of the PMCA seed by sonication, which is difficult to control (16); and (ii) dilutions prepared from aggregated material have a dilution error or lack of uniformity in PrPres particle size and number. This lack of uniformity causes gain or loss of PMCA seed in highly diluted samples, which in turn biases the presence or absence of signal. To establish a quantitative PMCA analysis, further efforts are required to reduce variability. During the preparation of this manuscript, Chen et al. reported a linear correlation between PrPres amounts and PMCA round (17). They concluded that partial purification of 263K-infected brain homogenate is important to reduce variability in PMCA.

Another objection to the use of PMCA for assessing PrPres is contamination or de novo generation of PrPres Because PMCA is very sensitive, a trace amount of contamination yields false positive signals. We took great care to avoid contamination. Because no signal appeared in the negative controls, which do not contain 263K-BH (Fig. 1b–e), we believe that no contamination or de novo generation occurred in our experiments. A third objection to PMCA is the use of disinfectant. In a bioassay, animals cannot be inoculated with undiluted toxic materials. Similarly, materials that inhibit PMCA performance must be diluted, which reduces sensitivity. Hence, we diluted the SDS-treated samples.

Before PMCA can be used for assessing disinfectant efficacy, criteria by which efficacy is defined must be established. The presence or absence of signal seems to be an inadequate criterion, because signals eventually appeared in samples treated with all four disinfectants, disinfectants that have been proved by bioassays to possess different inactivation efficacies. The noteworthy difference between very effective disinfectants (alkaline cleaner and SDS) and moderately effective disinfectants (1 M and 2 M NaOH) was the mean round numbers at which signal first appeared. The mean round numbers of the former were outside the round corresponding to the evaluation limit, while those of the latter were within the evaluation limit (Fig. 3b). The evaluation limit may be a valid means of categorizing disinfectants.

In this study, we evaluated residual amounts of PrPres after treatment with bioassay-proven disinfectants and calculated the LRF of each disinfectant using multi-round PMCA. The evaluation limit, or the range within which PMCA rounds correlate with mean dilution factors of infected brain, seems to be valuable in practical applications of multi-round PMCA, such as assessment of disinfectants and process validation.

ACKNOWLEDGMENTS

This study was supported by the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (T.K.); a grant from the Ministry of Health, Labor and Welfare (T.K.); a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology (T.K.); and the Benesis Corporation and Mitsubishi Tanabe Pharmaceutical company (A.T, M.K., and M.M). We thank Prof. K. Doh-ura for kindly providing the 263K strain. M.K. and M.M. are affiliated with the Benesis Corporation, which manufactures human plasma fractionation products and where the alkaline cleaner tested in this study is used to clean pipelines. The fact that this alkaline cleaner is also used by our company had no influence whatsoever over our experimental data.

Ancillary