The Extended Cell Panel Assay Characterizes the Relationship of Prion Strains RML, 79A, and 139A and Reveals Conversion of 139A to 79A-Like Prions in Cell Culture

Abstract

Three commonly used isolates of murine prions, 79A, 139A, and RML, were derived from the so-called Chandler isolate, which was obtained by propagating prions from scrapie-infected goat brain in mice. RML is widely believed to be identical with 139A; however, using the extended cell panel assay (ECPA), we here show that 139A and RML isolates are distinct, while 79A and RML could not be distinguished. We undertook to clone 79A and 139A prions by endpoint dilution in murine neuroblastoma-derived PK1 cells. Cloned 79A prions, when returned to mouse brain, were unchanged and indistinguishable from RML by ECPA. However, 139A-derived clones, when returned to brain, yielded prions distinct from 139A and similar to 79A and RML. Thus, when 139A prions were transferred to PK1 cells, 79A/RML-like prions, either present as a minor component in the brain 139A population or generated by mutation in the cells, were selected and, after being returned to brain, were the major if not only component of the population.

INTRODUCTION

Prions, the causative agents of transmissible spongiform encephalopathies, consist mainly or entirely of aggregated isoforms of the normal host protein PrP^C, designated PrP^Sc. PrP^Sc may occur in a proteinase K (PK)-resistant form, designated rPrP^Sc or PrP^res, or in a PK-sensitive form, sPrP^Sc or PrP^sen. Interestingly, prions may present in the form of different strains, whose PrP^Sc differ in regard to conformation but not to amino acid sequence (reviewed in reference 21).

Three commonly investigated isolates of murine prions—79A, 139A, and RML—were derived from populations that originated when prions from pooled scrapie-infected sheep brains (SSBP/1) were passaged through goats (the “drowsy goat” line) into mice. Because these isolates were never directly obtained from scrapie-infected sheep, they are believed to have originated in the goats used for transmission, which suffered from unrecognized “goat scrapie” (5). One goat-to-mouse transfer experiment gave rise to 79A prions, and another gave rise to the so-called Chandler isolate, from which both 79A and 139A prions were obtained (5). The Chandler isolate, transferred to the Rocky Mountain Laboratory by W. J. Hadlow in the early 1960s, gave rise to the (uncloned) RML isolate, believed to be the same as 139A (11; R. Kimberlin, personal communication). 79A and 139A were classically distinguished by their very different incubation times in VM (Prn^b) mice, but to our knowledge, RML was not characterized by the incubation time method.

The cell panel assay (CPA) distinguishes RML, 22L, ME7, and 301C prions by their differential ability to chronically infect the four cell lines PK1, R33, CAD5, and LD9 (17). The discriminatory power of the CPA has been potentiated by the introduction of the strain- and cell line-specific inhibitors swainsonine (Swa), kifunensine (Kifu), and castanospermine (1). The resulting extended cell panel assay (ECPA) has enabled us to demonstrate that, contrary to previous belief, RML and 139A prions are distinct entities, while 79A and RML prions are not distinguishable. Moreover, we found that transferring 139A prions from mouse brain to the N2a neuroblastoma PK1 cell line and back to mouse brain reproducibly gave rise to prions distinct from 139A but indistinguishable from 79A or RML.

MATERIALS AND METHODS

Cells.

The isolation of N2a-PK1 cells (here called PK1 cells) (14), CAD5 cells (here called CAD cells) (17), LD9 cells (17), and R33_2H11 cells (18) has been described previously. All cell lines were propagated in OBGS (Opti-MEM [Invitrogen] containing 4.5% bovine growth serum [HyClone, Logan, UT], 50 units penicillin/ml, and 50 μg streptomycin/ml [Invitrogen]). Uninfected cells were maintained for nine serial passages by 1:10 splits before being replaced by freshly thawed cells.

Prion strains.

RML (RML 1856-II) was obtained from the Prion Unit, University College London; 79A, 139A, and 22L were received from the TSE Resource Centre, Compton, Newbury, United Kingdom. All prion strains were propagated in C57BL/6 mice (from Charles River Laboratories, Wilmington, MA).

Prion propagation in mice.

Mice were anesthetized by isoflurane inhalation and inoculated in the prefrontal cortex with 20- to 30-μl samples. In some experiments, volumes or concentrations were adjusted to deliver similar amounts of infectivity (determined by standard scrapie cell assay [SSCA] on CAD cells) or of PrP^res (determined by Western blot analysis). Clinical signs of disease were cessation of nesting, ruffled coat, lateral deviation with medial pronation of hind limbs, hind limb weakness, myoclonus, urinary incontinence with lesions in the vaginal area, hunched back, decreasing activity with increasing periods of lethargy, weight loss, squinty eyes, bruxing, shivering, or trembling.

When clinical signs reached the terminal stage, the animals were euthanized by CO₂ asphyxiation followed by cervical dislocation. The brains were collected, and 10% homogenates in phosphate-buffered saline (PBS) were prepared as described previously (18).

Concentrated CM.

Prions secreted by infected cells were recovered from conditioned medium (CM). It has been previously shown that the CPA results obtained with prions from cell lysates and those obtained with prions from CM are similar (see supporting online material in reference 15). CM was cleared at 500 × g for 5 min, and prions were ultracentrifuged for 2 h onto a 10-ml sucrose cushion (20% sucrose [wt/wt] in 1× TNE buffer [25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA]), in a Ti45 rotor (Beckman Coulter) at 35,000 rpm and 4°C. The resulting pellet was suspended in OBGS to a 100× to 300× concentration of the original volume.

SSCA.

Typically, six serial 1:5 dilutions of the prion preparation (brain homogenate or concentrated conditioned medium) were added in triplicate to 96-well plates and 5,000 cells were added to each well. Triplicate wells with uninfected cells served as background control; another set of triplicates contained the highest concentration of inoculum used and 10 μg pentosan polysulfate/ml (Bene PharmaChem GmbH & Co. KG, Geretsried, Germany) to inhibit prion replication (3) and to assess the possible persistence of the inoculum. After 4 days, the cells were split 1:5 to 1:8, depending on their growth rate. After reaching confluence following the third split, 20,000 cells/well were transferred into wells of preactivated Multiscreen IP 96-well 0.45-μm filter plates (Millipore). Supernatants were drained by vacuum, the plates were dried at 50°C for at least 1 h, and the samples were subjected to the PK–enzyme-linked immunosorbent spot (ELISPOT) assay directly or after storage at 4°C.

PK-ELISPOT assay.

Samples were incubated for 90 min at 37°C with 70 μl of 1 μg proteinase K (Roche)/ml lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% sodium deoxycholate, 0.5% Triton X-100). All further steps were carried out at room temperature. The samples were washed twice with PBS and denatured with 120 μl of 3 M guanidinium thiocyanate in 10 mM Tris-HCl, pH 8.0, for 10 min. After four washes with distilled water (dH₂O), samples were incubated for 1 h with 0.5% nonfat dry milk in TBS (10 mM Tris-HCl, pH 8.0, 150 mM NaCl), followed by 1 h of incubation with 70 μl of 0.7 μg humanized anti-PrP antibody D18 (22)/ml of 1% nonfat dry milk in TBST (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% Tween 20). After four washes with TBST, 70 μl of alkaline phosphatase (AP)-conjugated anti-IgG (1:5,000; Southern Biotechnology Associates, Birmingham, AL) in 0.5% nonfat dry milk in TBST was applied for 1 h. Wells were washed four times with TBST. Then, the whole plate was immersed once in TBS and dried. Signals were visualized with the AP conjugate substrate kit (Bio-Rad), and PrP^res-positive cells (“spots”) were counted using the Bioreader 5000-Eb cytometer (BioSys).

ECPA.

Prion strains can be distinguished by their specific ability to infect a panel of different cell lines, as monitored by the SSCA (17). Prion- and cell line-specific inhibitors like swainsonine, kifunensine, or castanospermine (1) complement and extend the original panel of cell lines. In the work described here, the ECPA includes CAD cells, LD9 cells, R33_2H11 cells, and PK1 cells, as well as PK1 cells in the constant presence of swainsonine (1 μg/ml) or kifunensine (5 μg/ml). The cells' response to prions is determined by the response index (RI), the reciprocal of the dilution required to yield a certain number of positive cells (spots) per 20,000 cells. The ratio of RIs is characteristic for different prion strains (Table 1). For calculating statistics, the log values of the RIs and their ratios were used.

Table 1.

Prion characterization by the ECPA^d

Inoculuma	Log RIb						Log[RICAD/RIX]					Log[RIPK1/RIX]
	CAD	LD9	2H11	PK1	PK1 + Swa	PK1 + Kifu	LD9	2H11	PK1	PK1 + Swa	PK1 + Kifu	PK1 + Swa	PK1 + Kifu
C57/PK1[79A] clone 11	6.72	4.62	≪3	5.12	<3.36	≪3	2.1	≫3.77	1.59	>3.35	≫3.77	>1.76	≫2.12
C57/PK1[79A] clone 14	6.83	4.5	≪3	4.92	3.25	≪3	2.33	≫3.83	1.9	3.58	≫3.83	1.67	≫1.92
C57/PK1[139A] clone 15	6.87	4.55	≪3	4.8	3.13	≪3	2.32	≫3.87	2.07	3.74	≫3.87	1.67	≫1.80
C57/PK1[139A] clone 18	6.77	4.63	≪3	4.88	<3.23	≪3	2.14	≫3.77	1.88	>3.54	≫3.77	>1.66	≫1.88
C57[RML]	6.84	4.43	≪3	4.68	<3.26	≪3	2.42	≫3.84	2.16	>3.58	≫3.84	>1.41	≫1.68
C57[79A]c	6.83	4.42	≪3	4.93	3.28	≪3	2.41	≫3.83	1.9	3.55	≫3.83	1.66	≫1.93
C57[139A]	6.79	4.38	≪3	4.84	3.8	4.65	2.41	≫3.79	1.96	3	2.14	1.04	0.18
C57[22L]	6.31	5.7	5.38	5.71	5.62	5.22	0.61	0.93	0.61	0.69	1.09	0.08	0.48

^aThe SSCA was performed on CAD, LD9, R33_2H11, and PK1 cells, the latter in the absence or presence of swainsonine (1 μg/ml) or kifunensine (5 μg/ml). The proportion of PrP^res-positive cells was plotted against the logarithm of the inoculum dilutions as in Fig. 1.

^bRIs are the reciprocals of the dilutions required to yield 750 PrP^res-positive cells per 20,000 cells. The ratio of RIs on pairs of cell lines (or the log of the ratios) characterizes the cell tropism pattern of a prion strain. The table shows the averages (geometric means) of log RI values and log[RI ratios] determined in triplicate.

^cOne of the curves for C57[79A] in PK1 cells failed to reach 750 spots; calculations for this sample were therefore only approximate and the assay was not used for statistical evaluation.

^dSwa, swainsonine; Kifu, kifunensine.

Endpoint dilution cloning in cell culture.

PK1 cells were seeded at 100 cells/well in 96-well plates and inoculated with highly diluted 79A- or 139A-infected brain homogenates (final concentration, 10⁻⁹ and 5 × 10⁻¹⁰). The cells were repeatedly grown to confluence and split three times 1:3 followed by eight 1:10 splits; after a total of about 50 doublings, 20,000 cells from each well were subjected to the PK-ELISPOT assay and samples containing PrP^res-positive cells (spot numbers > [background plus 5 standard deviations {SDs}]) were scored as positive. The probability that under the conditions chosen a well was infected by more than one prion, P_{(n > 1)} = 1 − e^−m (1 + m), where P_n is the probability of finding n prions/well and m is the average number of prions/well, was 10⁻³ or less, as shown in Table 2. Five 79A or 139A prion clones were expanded; conditioned media were harvested, concentrated, and inoculated into C57BL/6 mice. The brain homogenates from terminally ill mice were characterized by ECPA.

Table 2.

Cloning of 79A and 139A prions in PK1 cells

Straina	Dilution (109)	No. of cells/well	No. of positive wells/total no. of wellsb	Fraction positive	P(0)c	md	mCe (104)	P(n> 1)f (104)
79A	1	100	11/240	0.046	0.954	0.047	4.7	11
	0.5	100	4/252	0.016	0.984	0.016	1.6	1.3
139A	1	100	11/240	0.046	0.954	0.047	4.7	11
	0.5	100	7/252	0.028	0.972	0.028	2.8	3.9
	0.5	100	3/462	0.0065	0.994	0.007	0.65	0.21

^aPK1 cells were seeded at 100 cells/well in 96-well plates and inoculated with estimated endpoint dilutions of 79A or 139A brain homogenates (10⁻⁹ and 5 × 10⁻¹⁰). The cells were propagated for about 50 doublings, and PrP^res-positive cells were determined by the PK-ELISPOT assay.

^bWells giving spot numbers of > (background plus 5 SDs) were scored as positive. The probability that a well was infected by more than one prion, P_{(n > 1)} = 1 − P₍₀₎ − P₍₁₎ = 1 − e^−m (1 + m), is derived from the Poisson distribution P_n = (e^−m · mⁿ)/n!, where m is the average number of prions/well and P_n is the probability of n prions/well.

^cP₍₀₎ = 1 − (positive wells/total wells).

^dm = ln[1/P₍₀₎] = average number of prions/well.

^em_C = m/100 = average number of prions/cell.

^fP_{(n > 1)} = 1 − e^−m (1 + m).

For comparison, uncloned PK1[139A] prions were prepared by inoculating PK1 cells seeded at either 100 cells/well or 5,000 cells/well with low dilutions of 139A brain homogenate (final concentrations, 5 × 10⁻⁵ and 4 × 10⁻⁵, respectively). After about 50 and 14 cell doublings, respectively, three distinct populations were expanded, concentrated conditioned medium was inoculated into mice, and the brains were processed as described above.

Infectivity determination by SSCA.

Infectivity determinations were performed by SSCA on CAD cells, relative to control brain homogenate, whose titer was determined by the mouse bioassay (18).

Western blot analysis.

To detect PrP^res by Western blot analysis, samples (3 mg total protein/ml in PBS, 0.5% Triton X-100) were digested with 25 μg proteinase K (Roche)/ml at 37°C for 1 h. Undigested controls were run in parallel. The digestion was stopped with 20 μl phenylmethylsulfonyl fluoride (PMSF) (100 mM)/ml, and the samples were denatured by being boiled in XT-MES sample buffer (Bio-Rad) at 100°C for 10 min. Electrophoretic separation (4 to 12% Criterion gel; Bio-Rad) and wet transfer (Bio-Rad) to polyvinylidene difluoride (PVDF) Immobilon membranes (Millipore) were performed by standard procedures. All further procedures were done at room temperature. Membranes were incubated for 1 h in blocking solution (5% nonfat dry milk in PBST [PBS, 0.1% Tween 20]) and immunostained with 0.7 μg humanized anti-PrP antibody D18 (22)/ml in 1% nonfat dry milk in PBST, followed by three washes with PBST and 1 h of incubation with horseradish peroxidase (HRP)-conjugated anti-IgG antibody (Southern Biotechnology Associates, Birmingham, AL; 1:15,000 in 0.5% nonfat dry milk in PBST). After three washes with PBST, chemiluminescence was induced by ECL-Plus (Pierce) and recorded by charge-coupled device (CCD) imaging (BioSpectrum AC Imaging System; UVP).

RESULTS

Many murine prion strains used in laboratories throughout the world were cloned by endpoint dilution in mice at the Neuropathogenesis Unit of the Institute for Animal Health (NPU). In this procedure, mice are injected with prion preparations diluted to the extent that only 1 in 10 or more mice becomes infected, and this procedure is repeated once or twice more. The strains 79A and 139A were derived from the so-called Chandler isolate, which originated from scrapie-infected goats; 79A, but not the 139A that we used, had been cloned at the NPU in Edinburgh, United Kingdom (H. Baybutt, personal communication). RML was also derived from the Chandler isolate and propagated for many decades in the Rocky Mountain Laboratory but was never cloned. There are contradictory opinions as to whether or not RML is similar or identical to 79A or 139A; several authors consider RML similar to or identical with 139A (11, 19), while Tremblay et al. (20) found different PrP^res brain deposition patterns for these two strains.

We compared RML, 139A, and 79A prions by the ECPA and found that, as described earlier (17), all three strains could chronically infect CAD, PK1, and LD9 cells but not R33_2H11 cells. However, swainsonine (1 μg/ml) and kifunensine (5 μg/ml) strongly inhibited chronic infection of PK1 cells by both RML and 79A prions, while infection by 139A was less inhibited by swainsonine and almost unaffected by kifunensine (Fig. 1; Table 1). Thus, 139A and RML are clearly different, while 79A and RML were indistinguishable by the criteria that we used.

Fig 1.

Prion characterization by the ECPA. CAD cells (red), PK1 cells (blue), LD9 cells (purple), R33_2H11 cells (green), and PK1 cells in the presence of swainsonine (dashed light blue) or kifunensine (dashed gray) were exposed to serially diluted prion samples as indicated. The proportion of PrP^res-positive cells was plotted against the logarithm of the inoculum dilutions. The cells' response to a prion sample is defined by the response index (RI), the reciprocal of the dilution required to yield an arbitrary number, here 750, of infected cells per 20,000 cells. The RI ratios or the logarithm of the RI ratios on different cells defines the cell tropism pattern of the sample. (A) C57[RML] and C57[79A] prions show similar cell tropism patterns; in particular, both strains are unable to infect R33_2H11 cells (R33_2H11 incompetent) and are strongly inhibited by swainsonine and kifunensine. C57[139A] prions are also R33_2H11 incompetent, but they are less inhibited by swainsonine and completely unaffected by kifunensine. For comparison, C57[22L] prions can infect R33_2H11 cells (R33_2H11 competent); they are unaffected by swainsonine or kifunensine. 79A and 139A prions that were cloned and propagated in cell culture and then returned to C57BL/6 mice are Swa sensitive and kifunensine sensitive; the cell tropism pattern is indistinguishable from that of C57[RML] and C57[79A] but very different from that of C57[139A] prions. Log RIs and log[RI ratios] of the complete ECPA are shown in Table 1. (B) The log[RI ratios] ± SDs of PK1 and PK1 plus swainsonine (light blue) and of PK1 and PK1 plus kifunensine (gray) are shown in a bar chart. *, one of the curves representing C57[79A] in PK1 cells failed to reach 750 spots for technical reasons; calculations for this sample were therefore only approximate and the assay was not used for statistical evaluation.

We next undertook to clone 79A and 139A by endpoint dilution in cultured cells rather than in mice. PK1 cells were distributed into 96-well plates at 100 cells/well, infected with 79A or 139A brain homogenate at a high dilution (10⁻⁹ or 5 × 10⁻¹⁰), and propagated for about 50 doublings, and the individual wells were assayed by PK-ELISPOT assay. Table 2 shows that the average number of prions per cell in the various experiments ranged from 0.65 × 10⁻⁴ to 4.7 × 10⁻⁴ and that the probability that a well was infected by more than one prion was between 10⁻³ and 10⁻⁴ for 79A and between 10⁻³ and 10⁻⁵ for 139A.

C57BL/6 mice inoculated intracerebrally (i.c.) with concentrated conditioned medium of 79A- or 139A-infected PK1 clones (designated PK1[79A] and PK1[139A], respectively) (Table 3) succumbed to scrapie disease with the symptoms usually seen in animals infected with RML, 79A, or 139A (see Materials and Methods). The ECPA patterns of brain homogenates from mice inoculated with original brain-derived C57[79A] and from mice inoculated with PK1[79A]-derived prion populations were indistinguishable (Fig. 1 and Tables 1 and 4). Unexpectedly, however, the brains infected with conditioned medium of PK1-passaged 139A (C57/PK1[139A] prions) presented an ECPA pattern that was distinctly different from that of the original brain-derived 139A (C57[139A]) prions: their propagation on PK1 cells was strongly inhibited by kifunensine, while that of the original C57[139A] prions was virtually unaffected (Fig. 1; Table 1). Moreover, C57/PK1[139A] prions showed enhanced Swa sensitivity compared to that of C57[139A] prions (Fig. 1 and 2; Tables 1, 4, and 5). Thus, the 139A prions that had been cloned in PK1 cells and subsequently propagated in brain differed from the original 139A prions and resembled brain-derived RML or 79A prions. To ascertain whether the clones that we picked might have been aberrant, we infected PK1 cells with high multiplicities of C57[139A] prions, expanded three distinct cell populations, and inoculated four mice each with conditioned media of the resulting prion populations (Table 3). As in the first experiment, the C57/PK1[139A] prions resembled RML or 79A prions and not 139A prions (Fig. 2B and Table 5).

Table 3.

Inoculation of C57BL/6 mice with brain-derived or cell-propagated 79A and 139A prions

Inoculum	Preparation	Doseb (10−2 IU or PrPres units)	DPId (avg ± SD)	No. of mice affected/no. inoculated
Expt 1a
C57[79A]	10−3 BH	37	160 ± 2	4/4
C57[139A]	10−2 BH	432	156 ± 2	4/4
PK1[79A] clone 15	100× CM	3.3	167 ± 3	4/4
PK1[79A] clone 14	100× CM	104	154 ± 12	4/4
PK1[79A] clone 13	100× CM	123	156 ± 14	4/4
PK1[79A] clone 12	100× CM	16	156 ± 2	4/4
PK1[79A] clone 11	100× CM	78	156 ± 3	4/4
PK1[139A] clone 12	100× CM	113	156 ± 3	4/4
PK1[139A] clone 13	100× CM	254	153 ± 0	4/4
PK1[139A] clone 15	100× CM	303	153 ± 8	4/4
PK1[139A] clone 18	100× CM	113	149 ± 7	4/4
PK1[139A] clone 17	100× CM	106	161 ± 4	4/4
Expt 2c
C57[139A]BH1	6.9 · 10−4 BH	100	167 ± 20	4/4
C57[139A]BH2	8.2 · 10−4 BH	99	173 ± 11	4/4
PK1[139A]2F3	50× CM	57	157 ± 8	4/4
PK1[139A]6F11	50× CM	197	148 ± 11	4/4
PK1[139A]1D6	50× CM	42	156 ± 1	4/4
PK1[139A]wp/5000	50× CM	67	157 ± 14	4/4
PK1[139A]wp/strong	50× CM	128	161 ± 7	4/4
PK1[139A]wp/weak	50× CM	4	158 ± 9	4/4
PK1-1 (uninfected cells)	50× CM		>436	0/4
PK1-2 (uninfected cells)	50× CM		>436	0/4

^aC57BL/6 mice were inoculated intracerebrally with either 20 μl homogenate of brains (BH) infected with “authentic” brain-derived 79A or 139A (C57[79A] or C57[139A]) or with 20 μl of 100-fold-concentrated conditioned medium (CM) of PK1 cells infected with endpoint-diluted 79A or 139A prions (PK1[79A]_cloneX and PK1[139A]_cloneY), respectively.

^bFor experiment 1, infectivity was determined by the SSCA on CAD cells. Doses are shown as infectivity units (IU). For experiment 2, doses are shown as arbitrary PrP^res units, determined by Western blot assay.

^cC57BL/6 mice were inoculated with 139A brain homogenates (C57[139A]_BH1 and C57[139A]_BH2) or with 50-fold-concentrated conditioned medium (CM) from different lines of 139A-infected PK1 cells: PK1[139A]_2F3, PK1[139A]_6F11, and PK1[139A]_1D6 were infected with endpoint-diluted 139A and represent 139A clones while PK1[139A]_wp/5000, PK1[139A]_wp/strong, and PK1[139A]_wp/weak were infected at high multiplicities and represent 139A populations (wp, whole population); CMs from uninfected PK1 cells were injected as negative controls.

^dDPI, days after inoculation, when terminal stage was reached.

Table 4.

139A prions cloned in PK1 cells and passaged through brain differ from the original 139A prions but are indistinguishable from brain-derived 79A or 79A prions cloned in cell culture and then passaged through brain

Inoculuma	Log RIPK1		Log RIPK1+Swa		Log[RIPK1/RIPK1+Swa]	SD
	Avg	SDb	Avg	SD
C57[79A]	5.91	0.11	3.73	0.16	2.18	0.19
C57[139A]	5.41	0.30	4.46	0.26	0.95	0.40
C57/PK1[79A] clone 15	6.00	0.03	3.81	0.13	2.18	0.13
C57/PK1[79A] clone 14	6.03	0.10	3.68	0.35	2.35	0.36
C57/PK1[79A] clone 13	5.68	0.07	3.63	0.12	2.05	0.14
C57/PK1[79A] clone 12	5.92	0.03	3.66	0.08	2.25	0.09
C57/PK1[79A] clone 11	6.05	0.18	3.95	0.18	2.10	0.25
C57/PK1[139A] clone 12	5.82	0.19	3.73	0.13	2.09	0.23
C57/PK1[139A] clone 13	6.12	0.17	3.93	0.15	2.19	0.23
C57/PK1[139A] clone 15	5.83	0.08	3.78	0.12	2.05	0.15
C57/PK1[139A] clone 18	5.75	0.14	3.25	0.17	2.51	0.22
C57/PK1[139A] clone 17c	5.61	0.17	3.07	0.03	2.55	0.17

^a79A and 139A prions were cloned by endpoint dilution in PK1 cells. Five clones of each were expanded; concentrated conditioned medium was prepared from each and inoculated into C57BL/6 mice. Brains were harvested from terminally ill mice, and the prions were characterized by triplicate SSCA on PK1 cells in the absence or presence of 1 μg swainsonine (Swa)/ml.

^bRIs are the reciprocals of the dilutions required to yield 750 PrP^res-positive cells per 20,000 cells. Average log RI is the average (geometric mean) of the log RI values; SD is the standard deviation. Inhibition is expressed as the logarithm of the ratio of average RIs, in the absence and presence of swainsonine, in the column “Log[RI_PK1/RI_PK1+swa]” ± SD. One-way analysis of variance (ANOVA) with Tukey's multiple comparison test of the log[RI ratios] ± SDs showed that C57[139A] prions that had been passaged in PK1 cells and subsequently in C57BL/6 mice were significantly (***, P < 0.001) more susceptible to inhibition by swainsonine than were the original C57[139A] prions and thus resembled C57[79A] and C57/PK1[79A] in that way.

^cOne of the three curves representing C57/PK1[139A] clone 17 in PK1 plus swainsonine did not reach 750 spots. Calculations for this sample were therefore only approximate and were not statistically evaluated.

Fig 2.

“Authentic” brain-derived 139A prions differ from 139A prions propagated in PK1 cells and returned to brain. (A) The averages (geometric means) ± SDs of the log[RI ratios] of PK1 and PK1 plus Swa of three C57[RML] and four C57[79A], C57[139A], and C57/PK1[139A] brain homogenates were used to analyze and compare the different prion samples by t test evaluation: the inhibitory effects of swainsonine on prion replication in PK1 cells were the same for C57[79A], C57[RML], and C57/PK1[139A] prions but were significantly different from that on C57[139A] prions. (B) The log[RI_PK1/RI_PK1+Swa] values for cloned C57/PK1[139A]_clone and uncloned C57/PK1[139A]_wp were indistinguishable but differed significantly from that of “authentic” C57[139A]. See Table 5 for details. ns, not significant.

Table 5.

Comparison of “authentic” RML, 79A, and 139A and 139A propagated in PK1 cells and returned to brain

Inoculum	Sample	Avg log RIPK1b	Avg log RIPK1+Swa	Log[RIPK1/RIPK1+Swa]	Avg log[RIPK1/RIPK1+Swa]	SD
Expt 1a
C57[RML]	a	5.71	4.03	1.68	1.73	0.21
	b	4.96	3.41	1.55
	c	5.32	3.36	1.96
C57[79A]	a	5.13	3.40	1.73	1.65	0.14
	b	5.01	3.46	1.55
	c	5.70	4.17	1.52
	d	6.07	4.26	1.81
C57[139A]	a	5.17	4.07	1.10	0.83	0.25
	b	5.11	4.33	0.77
	c	4.83	3.90	0.92
	d	5.58	5.06	0.51
C57/PK1[139A]	a	4.90	3.15	1.75	1.76	0.12
	b	5.28	3.50	1.78
	c	5.44	3.53	1.91
	d	5.12	3.51	1.61
Expt 2c
C57[139A]	a	5.79	4.75	1.05	1.06	0.15
	b	6.28	5.24	1.05
	c	6.48	5.25	1.22
	d	5.97	5.15	0.83
	e	6.33	5.18	1.15
C57/PK1[139A]clone	a	5.91	4.56	1.35	1.51	0.11
	b	5.86	4.41	1.45
	c	6.43	4.89	1.53
	d	6.35	4.74	1.61
	e	6.53	4.91	1.62
C57/PK1[139A]wp	a	6.21	4.73	1.48	1.53	0.05
	b	6.30	4.73	1.57
	c	6.55	5.00	1.55

^aBrain-derived RML (C57[RML]), 79A (C57[79A]), and 139A (C57[139A]) prions were compared to prions from brains infected with 139A prions that had been propagated in PK1 cells (C57/PK1[139A]) by SSCA in triplicate on PK1 cells with or without 1 μg swainsonine (Swa)/ml.

^bThe RI was determined at 500 PrP^res-positive cells/20,000 cells and expressed as log RI. Avg log RI, averages (geometric means) of the triplicate log RI values. The inhibitory effect of swainsonine on the prion replication in PK1 cells is expressed as log[RI_PK1/RI_PK1+Swa] or averaged (geometric mean) from the four brains as “Avg log[RI_PK1/RI_PK1+Swa] ± SD.” The values are the same for C57[RML], C57[79A], and C57/PK1[139A] prions but significantly lower for C57[139A] prions (Fig. 2A).

^c“Authentic” brain-derived C57[139A] prions were compared to C57/PK1[139A] prions, either cloned in PK1 cells by endpoint dilution and returned to brain or propagated as uncloned whole population (wp) in PK1 cells and then returned to brain. Calculations were performed as described above. The inhibitory effects of swainsonine on the prion replication in PK1 cells are the same for cloned and uncloned C57/PK1[139A] prions, but they are significantly different from those for C57[139A] prions (Fig. 2B).

DISCUSSION

The ECPA clearly shows that RML and 139A strains are distinct entities but does not distinguish between RML and 79A, which may be the same or not. In view of our finding that 139A prions switch to a strain indistinguishable from 79A or RML when passaged through PK1 cells and returned to mouse brain, it seems probable that the Chandler sample transferred to the Rocky Mountain Laboratory—whatever it originally contained—ended up consisting of 79A prions, either because these were selected from a mixture of 79A and 139A or because 139A converted to 79A in the course of multiple vertical transfers. Conversion of 139A to 79A has been mentioned, but the details are not accessible to us (6), as quoted in references 12 and 8; Dickinson and Outram considered “139A” a mixture of strains (7). Although 139A has been cloned by endpoint dilution in mice (13), the sample that we obtained was uncloned.

It was recognized early on that prion strains can adapt to new hosts as a consequence of mutation and selection (2, 12, 13), mutation at that time being considered a change in the nucleic acid sequence of a virus-like entity, rather than a conformational change of PrP^Sc, as it is now (4, 21). More recently, it was shown that prions can undergo adaptation even when transferred from one cell type to another within the same species (15, 18, 21). Thus, when 22L prions were transferred from brain to PK1 cells, they gradually changed their properties, losing their ability to infect R33 cells and becoming sensitive to inhibition by swainsonine; these changes were reversed when the prions were returned to brain (15, 16, 18). We believe that prion populations are quasispecies (9, 10), consisting of a multitude of variants, the ones fittest for a particular environment predominating (4, 15, 21), and we have shown that variants arise in cloned prion populations in the course of propagation (15, 16, 18), explaining how quasispecies arise.

Why would 139A propagate stably in brain but, after being passaged through PK1 cells and returned to brain, propagate stably as 79A/RML-like prions? We consider two possibilities. (i) The 139A population contains low levels of 79A/RML prions, which, upon transfer to PK1 cells, are selected to the exclusion of 139A prions. (ii) The 139A population propagates stably in brain because it neither contains 79A/RML contaminants nor gives rise to them by mutation; however, while mutation from 139A to 79A/RML-like prions in brain may be precluded by an insurmountable free energy barrier, this can be overcome in PK1 cells, where 79A/RML-like prions can arise and be selected. In either case, PK1 cells would provide a potent selection of 79A/RML-like prions over 139A prions.

Conversion of 79A to 139A prions in mouse brain has not been described in the literature and may not occur, either because the free energy barrier is too high or because any 139A arising is outcompeted by 79A.

As mentioned above, 22L prions adapt to cells but, when returned to brain, give rise to prions indistinguishable from the original 22L (15, 18). An important conclusion from the work reported here is that transfer of prions from brain to a cell line and back can also result in a permanent strain change, so that the attractive strategy of endpoint cloning of prion strains in cells is not a viable option.