Post-translational modifications in PrP expand the conformational diversity of prions in vivo

Misfolded prion protein aggregates (PrPSc) show remarkable structural diversity and are associated with highly variable disease phenotypes. Similarly, other proteins, including amyloid-β, tau, α-synuclein, and serum amyloid A, misfold into distinct conformers linked to different clinical diseases through poorly understood mechanisms. Here we use mice expressing glycophosphatidylinositol (GPI)-anchorless prion protein, PrPC, together with hydrogen-deuterium exchange coupled with mass spectrometry (HXMS) and a battery of biochemical and biophysical tools to investigate how post-translational modifications impact the aggregated prion protein properties and disease phenotype. Four GPI-anchorless prion strains caused a nearly identical clinical and pathological disease phenotype, yet maintained their structural diversity in the anchorless state. HXMS studies revealed that GPI-anchorless PrPSc is characterized by substantially higher protection against hydrogen/deuterium exchange in the C-terminal region near the N-glycan sites, suggesting this region had become more ordered in the anchorless state. For one strain, passage of GPI-anchorless prions into wild type mice led to the emergence of a novel strain with a unique biochemical and phenotypic signature. For the new strain, histidine hydrogen-deuterium mass spectrometry revealed altered packing arrangements of β-sheets that encompass residues 139 and 186 of PrPSc. These findings show how variation in post-translational modifications may explain the emergence of new protein conformations in vivo and also provide a basis for understanding how the misfolded protein structure impacts the disease.


Results
Diverse prion strains in GPI-anchorless mice show a convergence of clinical and histological features. The incubation period, clinical disease, and brain regions targeted (lesion profile) vary widely among mice infected with different mouse-adapted prion strains. To determine how the disease phenotype changes upon adaptation to a GPI-anchorless state, we serially-passaged RML, 22 L, ME7 and mCWD prions in Tg(GPI-PrP) mice, and refer to the resulting GPI-anchorless prions as GPI -RML, GPI − 22 L, GPI − mCWD, and GPI − ME7. RML, 22 L, and ME7 are biologically cloned prion strains originating from sheep scrapie 36 , whereas mouseadapted mCWD originates from chronic wasting disease prions in mule deer 33 . The incubation period of the original strains in WT mice ranged from 141 to 550 days post-inoculation (dpi)] and was consistent within a strain, but varied significantly among the strains (Fig. 1A) (mean 254 ± 42 dpi; P < 0.0001, ANOVA). The first 1-3 passages were previously reported for GPI − RML, GPI − 22 L, and GPI − mCWD 32 . However, here we found after 4-5 serial passages of all four strains in the GPI-anchorless mice, the incubation period converged to approximately 168 ± 4 dpi, although significant differences between the incubation periods for most groups were noted ( Fig. 1B) (GPI − RML: 152 ± 5 dpi; GPI − 22 L: 176 ± 1 dpi; GPI − ME7: 182 ± 2 dpi; GPI − mCWD: 160 ± 7 dpi; for four strain comparison, p < 0.005, ANOVA).
Prion strains show a striking diversity in the PrP Sc aggregate morphology and distribution throughout the brain. We and others previously found that RML, 22 L, and ME7 infection of WT mice resulted in fine to punctate PrP Sc aggregates deposited diffusely in different brain regions (Fig. 1C), whereas mCWD formed large dense plaques in the corpus callosum, meningeal vessels, and periventricular regions. In the GPI-anchorless state, all four strains converged to similar large dense plaques within and surrounding blood vessels (Fig. 1C) and ventricles in most brain regions, and did not change over 4-5 passages. Plaques showed subtle differences in having sharp or indistinct plaque borders, depending on the strain. All four GPI-anchorless vasotropic strains showed plaques with little to no spongiform degeneration (Fig. 1D).
Although all four prion strains induced disease following similar incubation periods and showed similar angiocentric plaques in the GPI-anchorless mice, the fibril structure may differ among the strains. Luminescent conjugated polymers (LCP) are amyloid-binding molecules used to distinguish prion strains, as the emission spectra varies depending on the structure of the amyloid bound 37 . We compared the emission spectra of the LCP, polythiophene acetic acid (PTAA), bound to prions in brain sections. After three serial passages, GPI − RML, GPI − ME7, Scientific RepoRts | 7:43295 | DOI: 10.1038/srep43295 and GPI − mCWD prion plaques showed a similar green-shifted PTAA emission spectra that differed significantly from that of GPI − 22 L, suggesting that certain anchorless PrP fibrils were structurally distinct ( Fig. 2A).
Strain-specific differences in the biochemical properties of GPI-anchorless PrP Sc . To determine whether GPI-anchorless PrP Sc differs biochemically, we compared the resistance to proteinase K (PK) cleavage, electrophoretic mobility, and conformational stability among the four anchorless prion strains. PK cleaves the relatively unstructured N-terminal part of PrP Sc , resulting in a PK-resistant PrP Sc core that varies with the PrP Sc conformation 38 . We found that the PK-resistant PrP Sc core size was similar for GPI − RML, GPI − 22 L, and GPI − mCWD prions, yet was longer for the GPI − ME7 prions (Fig. 2B). This is consistent with previous HXMS studies,  22 L, mCWD and ME7 prion strains show variable incubation periods in the WT mice, but similar incubation periods in the Tg(GPI-PrP) mice after 4-5 passages. (C) PrP Sc aggregates in the brain of prion-infected WT or Tg(GPI-PrP) mice. In WT mice (GPI + ), RML forms diffuse, patchy aggregates (arrows), 22 L forms fine aggregates on cell membranes (arrows), ME7 forms dense, punctate aggregates (arrows), and mCWD forms large, dense plaques. Corresponding anchorless prions (GPI − ) all form dense plaques that infiltrate and extend beyond a central vessel (arrows) and are present in all brain regions. Brain regions shown: RML: thalamus; 22 L: dorsal striatum; ME7: dorsal striatum; mCWD: corpus callosum; GPI-anchorless strains: cerebral cortex. (D) Prion-infected WT mice show spongiform degeneration in the brain after infection with 3 of 4 strains (arrow heads). mCWD plaques (asterisk) lead to very little spongiform change. Note the central vessel and lack of spongiform degeneration in prion-infected Tg(GPI-PrP) mice (arrow head). Scale bar = 100 μ m. N = 4-5 mice/ group. One-way ANOVA followed by Tukey's multiple comparison test for survival times, *P < 0.05; **P < 0.05; ***P < 0.001. which revealed that the PK-resistant core of PrP Sc from GPI − RML and GPI − 22 L prions starts at residue 81, whereas that of GPI-ME7 prions has an N-terminus starting at residue ~73/77 39 . In previous studies, GPI-RML PrP Sc and GPI -22 L PrP Sc were more stable in chaotropes than their respective anchored counterparts 32 . Here the conformational stability of the 3 rd passage of GPI-anchorless strains was assessed by denaturing aliquots of PrP Sc with 0 to 6 M guanidine hydrochloride (GdnHCl), digesting with PK, quantifying the PrP Sc , and calculating the [GdnHCl] 1/2 , which is the GdnHCl concentration at which half the PrP Sc remains 40 . Consistent with previous measurements 32 , we found that the absence of a GPI-anchor resulted in the formation of highly stable prions (Fig. S1A), with [GdnHCl] 1/2 values ranging from 1.87 (GPI − 22 L) to 3.51 (GPI − mCWD), and strains were significantly different from each other ( Fig. 2C and D).
The GPI-anchorless strains were also compared for their resistance to increasing concentrations of PK and, similar to the stability, GPI-anchorless prions varied from each other (Fig. 2E) and were generally more resistant to PK digestion than their GPI-anchored counterparts (Figs S1B and S2). Among the GPI-anchorless strains, GPI − mCWD and GPI − ME7 PrP Sc were more resistant to PK digestion than GPI − RML or GPI − 22 L prions (Fig. 2E). Thus, although the GPI-anchorless prions led to similar incubation periods and lesion distribution in the brain of infected mice, the differences in the LCP emission spectra, PK cleavage sites, conformational stability, and PK degradation profile collectively indicate that the GPI-anchorless fibrils remain as distinct conformational variants of PrP Sc .

C-terminus of GPI-anchorless PrP Sc is less accessible to H/D exchange following multiple
passages. For all four strains, the incubation period progressively shortened with passage in the GPI-anchorless mice, although the plaque-bound PTAA emission spectra was unchanged (Fig. S3). Previously, GPI − RML showed a gradual decrease in the conformational stability with each subsequent passage in the Tg(GPI-PrP) mice, suggesting that the prion conformation changed during three serial passages 32 . To test this possibility and gain insight into the nature of these conformational changes, we used two complementary methods: backbone amide hydrogen/deuterium exchange coupled with mass spectrometry (HXMS) and histidine hydrogen/deuterium exchange mass spectrometry (His-HXMS). These mass prions reveals that the GPI − ME7 PrP Sc has a higher molecular weight than the other three strains, indicative of a longer PK-resistant PrP Sc core fragment. (C) GPI − RML, − 22 L, − mCWD and − ME7 exposed to different concentrations of GdnHCl prior to PK digestion show significant differences in the aggregate stability among the strains. Plotted are the mean and standard error (SE) of the remaining PrP Sc as measured by ELISA from 3-4 mice per strain, each run in triplicate. (D) The [GdnHCl 1/2 ] values for each strain reveal significant differences between the strains (one-way ANOVA followed by Tukey's multiple comparison test). (E) The resistance of GPI-anchorless PrP Sc to proteinase K shows a significant differences between the strains. (two-way ANOVA followed by Bonferroni post-tests of GPI-anchorless prions revealed significant differences in GPI − RML vs GPI − mCWD at 10 μ g/ml PK**, GPI − RML vs GPI − ME7 at 10 μ g/ml PK*, and GPI − 22 L vs GPI − mCWD at 10 and 50 μ g/ml PK*). *P < 0.05; **P < 0.01; ***P < 0.001.
Scientific RepoRts | 7:43295 | DOI: 10.1038/srep43295 spectrometry based methods are one of very few tools available for structural analysis of brain-derived protein aggregates such as PrP Sc 39,41 .
In the HXMS method, one monitors the rate of H/D exchange of backbone amide hydrogens. While this exchange is very fast within the unstructured protein regions, it becomes much slower for systematically hydrogen bonded β -strands that are building blocks of amyloids 42 and constitute the proteinase-resistant core of PrP Sc 39,41,[43][44][45][46][47] . In HXMS, the exchange rates are assessed for peptic fragments that can be separated by liquid chromatography and identified by MS, providing segment-specific structural information. Figure 3A shows the extent of deuterium incorporation after 24 hours of incubation in D 2 O for PrP Sc derived from Tg(GPI-PrP) mice after the first and fourth passages of RML prions. The degree of exchange for both passages appears to be essentially identical for all peptic fragments derived from the PrP Sc region up to residue ~196. However, substantial passage-dependent differences could be detected within the C-terminal segment between residues ~197 and 223, with the fourth passage GPI − RML PrP Sc consistently showing higher protection against H/D exchange compared to the first passage prions. As discussed previously 39,41 , these C-terminal region-specific differences may reflect factors such as different proportions of residues involved in β -strands and turns between them and/or packing differences between individual β -strands.
The nature of structural differences between PrP Sc from the first and fourth passage of RML prions in mice expressing GPI-anchorless PrP was further probed using His-HXMS. This relatively novel approach monitors the rate of H/D exchange for C2 protons in histidine imidazole rings 48,49 . In contrast to backbone amide HXMS that probes structural organization at the level of the polypeptide backbone, His-HXMS reports on the microenvironment (water accessibility) of individual His side chains. The latter method proved very useful in studies of amyloids and related protein aggregates, providing information about packing arrangement and interfaces between β -sheets 39,41,50 . In particular, recently His-HXMS allowed detection of region-specific packing differences between two major strains of CJD prions 41 .
There are five His residues in the PK-resistant region of mouse PrP Sc . As shown in Figure 3B, four of them (His95, His110, His139 and His186) are characterized by a very similar environment in the first and fourth passage GPI − RML PrP Sc , as indicated by essentially identical exchange half-times. In contrast, pronounced differences are observed for His176. The latter side chain is substantially more protected from water in fourth passage  In contrast, GPI-mCWD led to small clusters of plaques in WT mice, differing markedly from the large, discrete dense plaques of original GPI-anchored mCWD, although both plaque types were enriched in the corpus callosum (arrows). (C) Lesion profile of WT mice infected with Pre and Post GPI − prion strains. For the Pre and Post GPI − RML, − 22 L and − ME7 infected WT mice, the severity of spongiosis, astrogliosis, and PrP Sc deposition were scored for nine brain regions (see Methods) and were nearly superimposable. For Post GPI -mCWD, the regions and severity of spongiosis, astrogliosis, and PrP Sc distribution were more widespread and severe as compared to Pre GPI − mCWD. Radial plots show the mean from 4-7 mice per strain. (D) Emission spectra of PTAA-labelled Pre and Post GPI − prions in the brain of WT mice (for 22 L and ME7) and in the brain of tga20 mice (for mCWD prions) measured at wavelengths from 500-700 nm. (E) Ratios at 538 nm/643 nm and at 538 nm/emission maximum are shown. Brain regions shown in (B): RML, Pre: thalamus, Post: dorsal striatum; 22 L, Pre and Post: cerebral cortex; ME7, Pre and Post: dorsal striatum; mCWD, Pre and Post: corpus callosum. N = 4-5 mice per group. Scale bar = 100 μ m. Logrank test for survival times, *P < 0.05; **P < 0.01; ***P < 0.001. GPI − RML PrP Sc compared to first passage GPI − RML PrP Sc , suggesting distinct packing arrangements and/or interfaces between β -sheets around residue 176.

Strain
Defining the strain recovered after passage through the GPI-anchorless prion state. To determine whether PrP Sc conformation of the original strain may be retained in the GPI-anchorless fibrils, we inoculated all four anchorless prions into WT mice. We compared the incubation period, aggregate morphology and lesion profile with those induced by the original prion strain. Passage of GPI − RML prions into WT mice led to an incubation period that was similar to that observed with the original RML prions, at 146 ± 3 days versus 147 ± 0 days, respectively, whereas GPI − 22 L and GPI − ME7 prions led to a modestly prolonged incubation period in WT mice. In contrast, GPI − mCWD prions in WT mice showed a dramatic 3-fold reduction in the incubation period when compared to the original strain in WT mice ( Fig. 4A and Table 1). Transmission barriers in prion disease are suggested by a progressive decrease in the incubation period upon serial passage. To determine whether the incubation period would decrease with serial passage, we performed two serial passages of the GPI-anchorless prions in tga20 mice, which overexpress WT PrP by approximately 4-6 fold 51 . GPI − RML and GPI − mCWD prions in tga20 mice showed only subtle differences in the incubation period between the first and second passages suggestive of a stable strain, whereas GPI − 22 L and GPI − ME7 prions showed a decrease in the incubation period upon second passage (Fig. S4).
Despite the slight change in incubation period for the 22 L and ME7 prions following passage through the anchorless state, the PrP Sc aggregate morphology and lesion profiles were identical to the original strains. Remarkably, the pathology in WT mice inoculated with GPI − mCWD prions had markedly changed, as plaques were widespread and appeared smaller and less distinctly defined, clearly differing from the sharply demarcated, dense congophilic plaques typical of mCWD (Fig. 4B). Similar histological findings were also observed in tga20 mice inoculated with GPI − mCWD (Fig. S5A), further supporting that a novel mCWD prion strain arose or was selected following passage through the GPI-anchorless state. The lesion profile of the mCWD strain also changed between the first and second passage, with increased PrP Sc deposition and gliosis on the second passage in nearly every brain region ( Fig. S5A and B). However, there were no changes in the electrophoretic mobility of the PrP Sc between the 1 st and 2 nd passage of GPI − mCWD (Fig. S5C).
The PTAA emission spectra for the WT mice infected with GPI − 22 L or GPI − ME7 prions were nearly indistinguishable from that of the original 22 L and ME7 strains ( Fig. 4D and E), while no PTAA labelling of the GPI − RML in WT mice was detected, as previously observed for RML 32 . In contrast, the GPI − mCWD in tga20 mice showed a significantly red-shifted emission spectra of PTAA bound to PrP Sc as compared to that of the original mCWD strain in tga20 mice ( Fig. 4D and E). Consistent with the stability and glycoprofile results for three strains, the resistance to PK degradation also reverted back to the low PK resistance observed for the original WT strains, RML, 22 L, and ME7 (Pre GPI − versus Post GPI − , at 10 μ g/ml PK, P < 0.001; at 50 μ g/ml PK, P < 0.05; Two-way ANOVA followed by Bonferroni post-tests) ( Fig. 5D and S2B). The mCWD strain became more resistant to PK degradation after passage through the anchorless state, again suggesting that the mCWD prions had either formed a new structure, or that an mCWD substrain was selected by GPI-anchorless PrP C (Pre GPI − versus Post GPI − mCWD at 10, 50, and 100 μ g/ml PK: P < 0.001; at 250 μ g/ml PK, P < 0.01; Two-way ANOVA followed by Bonferroni post-tests) (Fig. 5D). Finally, we measured the ratio of soluble to insoluble PrP Sc for each strain before and after passage through the GPI-anchorless state. Although there were no differences in the ratio of soluble to insoluble PrP Sc for RML, 22 L, or ME7 strains, mCWD prions became significantly more soluble following passage through the GPI-anchorless state ( Fig. 5E and F).
His-HXMS analysis of GPI-anchored prion strains. The biochemical assays suggest that the PrP Sc aggregate structure of at least some prion strains could be altered by pre-passage through mice expressing GPI-anchorless PrP. To further explore this issue, we performed His-HXMS measurements on PrP Sc from the original (GPI-anchored) prions and the same strains after passage through the GPI-anchorless state. These experiments revealed that, for most strains studied, this pre-passage has a significant effect on the environment of His side chains in PrP Sc aggregates (Fig. 6), indicating differences in PrP Sc structure at the level of packing arrangements and interfaces between β -sheets (see discussion of the His-HXMS method above). These differences are particularly pronounced around His139 for mCWD prions and, to a somewhat lesser degree, ME7. Structural differences are also detectable by His-HXMS between pre-and post-GPI − RML PrP Sc , even though they appear to be more modest and localized to a more C-terminal part of the protein (i.e., around His176 and His186). In contrast, no such differences could be detected for PrP Sc associated with 22 L prions.

Discussion
Protein aggregates, including those composed of tau, amyloid-β , and α -synuclein can form conformationally distinct structures [11][12][13][14][15] , yet the factors contributing to this conformational diversity are poorly understood. The prion protein also misfolds into a wide range of conformations, often with multiple forms co-occurring within an individual [6][7][8][9][10] . Here we show that four prion strains serially passaged in GPI-anchorless mice led to fibrillar, angiocentric prions associated with a nearly identical clinical and pathological disease phenotype. However, these four fibrillar strains differed biochemically, suggesting that the secondary and tertiary structure varied within the fibrils, in agreement with IR studies 30 .
On first passage into the GPI-anchorless mice, all four strains switched to a new pathological phenotype, with prions arranged in large angiocentric plaques throughout the brain, as previously reported 28,32 . Anchorless plaques are composed of fibrils 29 , suggesting that the GPI-anchor obstructs fibril assembly in WT mice. Each subsequent prion passage in the GPI-anchorless mice led to a decrease in the incubation period and conformational stability of PrP Sc , although the plaques appeared unchanged histologically and by PTAA emission spectral analysis. HXMS experiments revealed a higher degree of protection against H/D exchange in the C-terminal region of PrP Sc near the N-glycan sites (N180, N196) after multiple passages, suggesting that the C-terminal region had become more ordered with passage. Furthermore, differences were observed in the packing of the β -sheets around His176 (i.e., in the vicinity of one of the glycosylation sites). Thus, the poorly glycosylated state of the GPI-anchorless prions may enable a more ordered packing arrangement in the distal C-terminus.
To determine whether the PrP Sc structure of the original strain was maintained within the fibril, GPI-anchorless prions were passaged back to WT mice. Two strains, RML and 22 L, showed little or no change following passage through an anchorless state. The incubation period, pathology, biochemical profile and certain biophysical characteristics of PrP Sc such as conformational stability were nearly identical to those in the corresponding original strains and were similar to an earlier passage of GPI-RML into WT mice 32 , even though the His-HXMS data suggest modest packing alteration for RML PrP Sc after passage through the anchorless state. These observations, together with the finding that global secondary structures of anchored and anchorless RML and 22 L prions are similar 30 , suggest that these anchorless prions maintain structural features of the original RML and 22 L strains, even though they assemble into fibrils. The morphological differences between the GPI-anchored and anchorless PrP Sc (subfibrils versus fibrils) are likely related to distinct packing arrangements of β -sheets.
The post GPI − ME7 prion strain subtly differed from the original strain, but these differences were less dramatic compared to those observed for mCWD. The incubation period had become prolonged and the PrP Sc stability was lower than for the original ME7, suggesting that the structure had changed. This structural alteration of PrP Sc was further confirmed by His-HXMS data that indicate distinct packing arrangements, especially around His139. In this context, it should be noted that an elegant report by Baron and colleagues using infrared (IR) spectroscopy on highly purified anchored and anchorless ME7 PrP Sc noted subtle differences in the secondary structure of ME7 following conversion into GPI-anchorless ME7, indicating that the anchorless state may alter the structure of certain prion strains 30 . The structure of GPI-anchored ME7 prions emerging upon passage of GPI-anchorless ME7 in WT mice was not examined by IR spectroscopy.
Mahal and colleagues also performed two serial passages of RML and ME7 in GPI-anchorless mice and found that passage of anchorless RML into WT mice led to a novel strain, based on Cell Panel Assay (CPA) studies of cell tropism 31 . In the present study, differences in the aggregate morphology or lesion profile in mice infected with GPI -RML were not detected, however there were subtle differences in the His-HXMS of the aggregates that may translate into the differences reported with the CPA. Both studies were in agreement in that the incubation period, western blotting, and conformational stability assay were unchanged in anchorless RML-infected WT mice. The CPA comparison of the original and Post-GPI passaged ME7 strain showed no differences in the strain, whereas we report subtle differences in the biological readout (incubation period) and in the conformational stability of the prions, as well as differences in the His-HXMS of pre-and post-GPI passaged ME7 around His139.
Post-GPI-mCWD prions led to the emergence of a new strain in WT mice by all measurements. GPI − mCWD-infected WT mice showed marked alterations in the biochemical and biophysical properties of PrP Sc and profound changes in the incubation period, clinical disease, and pathology. Certain aspects of the plaque morphology and brain deposition pattern observed for the new mCWD strain were reminiscent of the original strain (for example, plaque deposition in the corpus callosum), suggesting that some structural features of the original mCWD PrP Sc might have been retained. Biochemical studies indicated that the Post GPI − mCWD PrP Sc was more soluble and PK resistant, and less stable in chaotropes, and had an altered glycosylation profile and PTAA emission spectra. Additionally, the microenvironments (water accessibility) of histidine side chains at position 139 and, to a lesser degree 186, were markedly different in the original and post GPI − mCWD PrP Sc (Fig. 6), indicating altered packing arrangements of β -sheets in the vicinity of residues 139 and 186. Thus, PrP Sc generated by the GPI − mCWD prions in WT mice assembled into a structure distinct from the original mCWD, higher PK resistance than-mCWD. Two-way ANOVA followed by Bonferroni post-tests of Pre and Post GPI − prions revealed significant differences in RML at 10 and 50 μ g/ml PK, and mCWD at 10, 50, and 100 μ g/ml and at 250 μ g/ml PK. (E) Western blots show the solubility of Pre and Post GPI − RML, − 22 L, − ME7 and − mCWD strains. S: supernatant and P: pellet. (F) Quantification of the pellet fraction for all strains (3 animals per strain). Two-way ANOVA followed by Bonferroni post-tests of Pre and Post GPI − prions revealed significant differences in the PrP Sc solubility of Pre-and Post GPI-mCWD prions. *P < 0.05; **P < 0.01; ***P < 0.001.
Scientific RepoRts | 7:43295 | DOI: 10.1038/srep43295 indicating that the variation in PTMs either generated or selected (from substrains) a new, rapidly replicating strain.
GPI-anchorless mice express poorly glycosylated PrP. Altering PrP glycans modifies the prion conformation in knock-in mice expressing zero, one, or two glycans 25 , yet interestingly, none of the mice expressing specific glycans formed strictly angiocentric plaques as observed in the GPI-anchorless mice, suggesting this vascular tropism is due to the lack of a membrane anchor. Considering that approximately 10-15% of PrP C is released from the cell membrane as GPI-anchorless PrP 21 , and prions may be switching between the anchored and anchorless state in vivo, these results suggest that GPI-anchorless PrP C molecules within the brain can expand the conformational spectrum of prions and serve as a source for new prion strains within an individual and within a population. Low levels of glycans on PrP may also play a role in the generation of new strains 27 , yet how conversion of poorly glycosylated PrP modifies the PrP Sc structure is not clear.
Certain anchored and anchorless prions (RML and 22 L) may maintain the same global secondary structure within non-fibrillar aggregates or fibrils, however the disease that ensues profoundly differs by nearly every measure, including clinical signs, organ tropism, neuroinvasive ability, plaque distribution, and biochemical properties of PrP Sc ( Table 2). As compared to the anchored prions, anchorless RML and 22 L prions (i) accumulate in many additional tissues, including kidneys and adipose tissue 28,32,52,53 , (ii) form extensive angiocentric plaques distributed throughout the brain 28 , (iii) cause a longer clinical disease phase 29 , and (iv) no longer readily spread into the CNS from extraneural sites 32,54 . Anchorless prions also accumulated to high titers in the blood and caused an amyloid-induced cardiomyopathy as well as cerebral amyloid angiopathy in the Tg(GPI-PrP) mice 52 . These findings suggest that the prion disease phenotype may be largely determined by the PrP Sc oligomerization state (fibrillar lar or nonfibrillar morphology) that, in turn, is likely dependent on packing arrangements between β -sheets.
Post-translational modifications commonly occur in protein biosynthesis and include phosphorylation, lipidation, glycosylation, amidation of the C-terminus, acetylation, and disulfide bond formation. Deamidation of asparagine and glutamine side chains is a common, nonenzymatic, spontaneous change that can lead to major changes in protein stability and alterations in fibril structure 23 . For prions, the role of N-linked glycans in defining strain characteristics for some strains has been demonstrated 25,27,55 . GPI anchors are present on approximately 250 proteins and have also been extensively shown to impact protein structure. For example, removing the GPI − anchor from diverse proteins by phosphatidylinositol-specific phospholipase C (PIPLC) significantly alters their enzymatic activity [56][57][58][59] . By identifying the effect of the PTMs on the disease outcome for four prion strains, we have revealed PTMs as a source of novel conformations of ordered protein aggregates in vivo. This finding has implications for other PTMs that may also have a similar effect on the conformation of a protein aggregate in vivo. In a broader context, one could argue that the post-translationally modified state of the protein may contribute to the remarkable diversity of protein aggregate structures derived from the same primary amino acid sequence in patients with various neurodegenerative diseases.
In addition to providing evidence for how new strains may arise in vivo, the use of Tg(GPI-PrP) mice has proven useful in establishing the relationship between the PTMs and the assembly pathway of prion protein aggregates. Four prion strains cause a nearly identical disease in GPI − anchorless mice, yet retain structural variations within the fibrils and features of the original strain. Simply removing the GPI − anchor and most glycans from PrP C enabled fibril assembly to occur, and passage into WT mice led to apparent recovery of the original non-fibrillar prions for the RML and 22 L strains. Finally, the fibrillar GPI-anchorless RML and 22 L prions caused similar diseases in the Tg(GPI-PrP) mice, yet differed profoundly from the subfibrillar RML and 22 L prions in WT mice. This finding suggests that the PrP Sc oligomerization state (fibrillar versus non-fibrillar morphology) may be one of the important determinants governing the disease incubation period, organ tropism, and aggregate spread to the brain. Further studies are needed to determine the relationship between the oligomerization state and specific structural features such as the tertiary structure of individual monomers, their packing arrangement, and the interfaces between β -sheets. to Prnp null mice (C57BL/6 background) 28 . All prion strains have been maintained in C57BL/6 mice, with the exception of mCWD, which was maintained in tga20 mice 33 . mCWD was propagated in tga20 mice, and the 5 th passage was used. Homogenates from individual animals were used for the inoculation. WT (C57BL/6), tga20, or Tg(GPI − PrP) mice (groups of n = 4-6 mice) were anesthetized with ketamine and xylazine and intracerebrally inoculated into the left parietal cortex with 30 μ l of prion-infected brain homogenate prepared from terminally ill mice. Mice were monitored three times weekly, and prion disease was diagnosed according to clinical criteria including ataxia, kyphosis, stiff tail, hind leg clasp, and hind leg paresis. The incubation period was calculated from the day of inoculation to the day of terminal clinical disease. Mice were euthanized at the onset of terminal disease. The brain was halved, and one hemi-brain was formalin-fixed, then immersed in 96-98% formic acid for 1 hour, washed in water, and post-fixed in formalin for 2-4 days. Brains were then cut into 2 mm transverse sections and paraffin-embedded for histological analysis. The remaining hemi-brain was cut and a 2-3 mm transverse section at the level of the hippocampus/thalamus was embedded in OCT and immediately frozen on dry ice. The remaining brain sections were frozen for biochemical analyses. Mice were maintained under specific pathogen-free conditions. Histopathology and immunohistochemical stains. Four μ m sections of brain were cut onto positively charged silanized glass slides and stained with hematoxylin and eosin, or immunostained using antibodies for PrP (SAF84) or GFAP for astrocytes. For PrP staining, sections were deparaffinized and incubated for 5 min in 96% formic acid, then washed in water for 5 min, treated with 5 μ g/ml of proteinase-K for 7 min, and washed in water for 5 min. Sections were then placed in citrate buffer (pH 6) and heated in a pressure cooker for 20 min, cooled for 5 min, and washed in distilled water. Sections were blocked and incubated with anti-PrP SAF-84 (SPI bio; 1:400) for 45 min followed by anti-mouse biotin (Jackson Immunolabs; 1:250) for 30 min, followed by streptavidin-HRP (Jackson Immunolabs; 1:2000) for 30 min. Slides were then incubated with DAB reagent for 7 min and an enhancer for 2 min (Invitrogen). Sections were counterstained with hematoxylin. GFAP immunohistochemistry for astrocytes (1:500; DAKO) was similarly performed, however with antigen retrieval by PK-digestion (20 μ g/ml for 10 min at room temperature).
Western blotting and sodium phosphotungstic acid precipitation. Brain tissue was homogenized in PBS using a Beadbeater ™ tissue homogenizer. Homogenates in a Tris-based lysis buffer (10 mM Tris-HCl, 150 mM NaCl, 10 mM EDTA, 0.5% NP40, 0.5% DOC; pH 7.4) were digested with 50 μ g/ml proteinase K at 37 °C for 30 min and the reaction stopped by boiling samples for 5 min in LDS loading buffer (Invitrogen). Samples were electrophoresed in 10% Bis-Tris gel (Invitrogen) and transferred to a nitrocellulose membrane by wet blotting. Membranes were incubated with monoclonal antibody POM19 (discontinuous epitope at C-terminal domain,  amino acids 201-225 61 , a kind gift from Dr. Adriano Aguzzi) followed by incubation with an HRP-conjugated anti-mouse IgG secondary antibody. The blots were developed using a chemiluminescent substrate (ECL detection kit, ThermoScientific) and visualized on a Fuji LAS 4000 imager. Quantification of PrP Sc glycoforms was performed using Multigauge V3 software (Fujifilm). PrP Sc was concentrated from tga20 mouse brain samples by performing sodium phosphotungstic acid (NaPTA) precipitation prior to western-blotting 62 . Briefly, 100 μ l aliquots of 10% brain homogenate in an equal volume of 4% sarkosyl in PBS were digested with an endonuclease [BenzonaseR (Sigma)] followed by treatment with 20 μ g/ml proteinase K at 37 °C for 30 min. After addition of NaPTA, MgCl 2 , and protease inhibitors (Complete-TM, Roche), extracts were incubated at 37 °C for 30 min, and centrifuged at 18,000 g for 30 min at 37 °C. Pellets were resuspended in 0.1% sarkosyl prior to electrophoresis and blotting.
Conformation stability assay. Prion strain stability in GdnHCl was performed as previously described 40 .
Briefly, lysed brain homogenates were denatured in GdnHCl ranging from 0-6 M for 1 hr. Samples were then diluted with a Tris-based lysis buffer (10 mM Tris-HCl, 150 mM NaCl, 10 mM EDTA, 4% sarkosyl; pH 7.5) to 0.15 M GdnHCl and digested with PK at a ratio of 1:500 (1 μ g PK: 500 μ g total protein) for 1 hr at 37 °C. Digestion was stopped with 2 mM phenylmethylsulfonyl fluoride (PMSF) and Complete TM protease inhibitor (Roche) followed by centrifugation at 18,000 g for 1 hr. Pellets were washed in 0.1 M NaHCO 3 (pH 9.8) and centrifuged at 18,000 g for 20 min. Pellets were then denatured in 6 M guanidine isothiocyanate (GdnSCN), diluted with 0.1 M NaHCO 3 , and coated passively onto an ELISA plate. PrP was detected with biotinylated-POM1 antibody (epitope in the globular domain, amino acids 121-231 61 ), a streptavidin HRP-conjugated secondary antibody, and a chemiluminescent substrate. Each strain (n = 3-5 mice per strain) was analyzed in at least 3 separate experiments. Statistical analysis was performed using a Student's t test.
Proteinase K (PK) resistance. Brain homogenate was diluted in Tris-based lysis buffer (50 mM TrisHCl, 137 mM NaCl, and 2% sarkosyl; pH 8.0), mixed for 15 min at 25 °C at 1000 rpm, and divided into 8 tubes. Tris lysis buffer containing 2% sarkosyl was added to each tube and samples were digested with increasing concentrations of PK (10 to 3,000 μ g/ml). Samples were then incubated for 2 hr at 37 °C at 1000 rpm and the PK digestion stopped by boiling for 5 min in LDS loading buffer (Invitrogen) prior to western blot and development using POM19. PrP signals from each PK digested sample was captured and quantified using a Fujifilm LAS-4000 imager and Multigage software. Each strain was analyzed in at least 3 separate experiments using 3-4 mice per strain. PK resistance was calculated by measuring the signal relative to the undigested sample.
PrP Sc solubility assay. Brain homogenates were solubilized in 10% sarcosyl in PBS and digested with 50 μ g/mL of proteinase K (final concentration) at 37 °C for 30 min. Protease inhibitors were added (Complete TM ™ ), and samples were layered over 15% Optiprep ™ and centrifuged at 18,000 g for 30 min at 4 °C. Supernatants were removed and pellets were resuspended in PBS in a volume equivalent to the supernatant. Supernatant and pellet fractions were immunoblotted using anti-PrP antibody POM19. PrP signals were captured and quantified using the Fuji LAS 4000 imager and Multigauge V3.0 software. Brain samples from 3 mice were measured per strain.
PTAA staining and analysis of frozen tissue sections. Frozen sections from mouse brain were dried for 1 hour and fixed in 100% and 70% ethanol for 10 min each. After washing with deionized water, sections were equilibrated in 100 mM sodium carbonate at pH 10.2 for 30 min. The PTAA was diluted in the sodium carbonate buffer (1 μ g: 100 μ l buffer) and added to the sections. The sections were incubated with PTAA for 30 min at room temperature and washed with sodium carbonate buffer. The emission spectra of PTAA bound to PrP aggregates was collected using an inverted LSM 780 confocal microscope (Carl Zeiss, Oberkochen, Germany) with excitation wavelength at 488 nm and the spectra were collected from 8 individual spots within 3-5 plaques from a minimum of two different cases of each prion-infected brain.
Purification of PrP Sc for structural studies by mass spectrometry. Samples were purified from brain homogenates as previously described 63 . Briefly, 10% brain homogenate was mixed with 5% sarcosyl (final) in TEN(D) buffer (50 mM Tris-HCl, 5 mM EDTA, 665 mM NaCl, 10% sarkosyl, 0.2 mM dithiothreitol, Complete-TM protease inhibitors (PI) (Roche), pH 8.0), incubated on ice for 1 hour, and then centrifuged at 18,000 g for 30 min at 4 °C. All but 100 μ l of supernatant was removed, and the pellet was resuspended in 100 μ l of residual supernatant and diluted to 1 ml with 10% sarkosyl TEN(D). Each supernatant and pellet was incubated for 30 min on ice and then centrifuged at 22,000 g for 30 min at 4 °C. Supernatants were recovered while pellets were held on ice. Supernatants were added separately into ultracentrifuge tubes with 10% sarcosyl TEN(D) buffer containing PI and centrifuged at 150,000 g for 2.5 hours at 4 °C. Supernatants were discarded while pellets were rinsed with 100 μ l of 10% NaCl in TEND buffer with 1% sulfobetaine (SB 3-14) and PI and then combined and centrifuged at 225,000 g for 2 hours at 20 °C. The supernatant was discarded and pellet was washed and then resuspended in ice cold TMS buffer containing PI (10 mM Tris-HCl, 5 mM MgCl2, 100 mM NaCl, pH 7.0). Samples were incubated on ice overnight at 4 °C. Samples were then incubated with 25 units/ml endonuclease (benzonase, Sigma-Aldrich) and 50 mM MgCl 2 for 30 min at 37 °C at 1000 rpm followed by a digestion with 10 μ g/ml PK for 1 hr at 37 °C at 1000 rpm. PK digestion was stopped by incubating samples on ice with 2 mM phenylmethylsulfonyl fluoride (PMSF) for 15 min. Samples were incubated with 20 mM EDTA for 15 min at 37 °C at 1000 rpm. An equal volume of 20% NaCl was added to all tubes followed by an equal volume of 2% SB 3-14 buffer. For the sucrose gradient, a layer of 0.5 M sucrose, 100 mM NaCl, 10 mM Tris, and 0.5% SB 3-14, pH 7.4 was added to ultracentrifuge tubes. Samples were then carefully transferred and the tubes topped with TMS buffer. Samples were centrifuged at 200,000 g for 2 hours at 20 °C. The pellet was rinsed with 0.5% SB 3-14 in PBS. Pellets were resuspended in 50 μ l of 0.5% SB 3-14 in PBS and stored at − 80 °C.
Backbone amide hydrogen/deuterium exchange mass spectrometry experiments (HXMS). To initiate deuterium labeling, 10 μ l aliquots of purified prions (~1-2 μ g) were collected by centrifugation (21,000 g, 5 min) and suspended in 100 μ l of 10 mM phosphate buffer (pH 7.3) in D 2 O. After incubation at 37 °C for 24 hours, samples were collected by centrifugation and dissociated into monomers by adding 20 μ l of an ice cold exchange quench solution (100 mM phosphate, pH 2.5) containing 7 M GdnHCl and 0.1 M Tris (2-carboxyethyl) phosphine hydrochloride. After 5 minutes incubation on ice, the solution was diluted 10 times with ice cold 0.05% trifluoracetic acid and digested for 5 min using 100 μ l of agarose-immobilized pepsin slurry (Thermo Scientific, Waltham, MA) as described previously 41 . The peptic fragments were collected in a C18 trap column, washed to remove salts, and separated on an UPLC BEH-C18 column (Waters, USA). To minimize back-exchange, separation was performed using a "rapid" gradient of 2-45% acetonitrile with a total elution time of 13 min, and both the trap and the analytical column were placed in a cooled chamber (~2 °C) integrated with a LEAP TriValve system (LEAP Technologies, USA). Separated peptides were analyzed by LC-MS/MS in ESI mode using an LTQ-Orbitrap XL mass spectrometer (ThermoElectron, San Jose, CA) directly coupled to the UPLC system. The mass spectrometer (externally calibrated using a Pierce ESI positive ion calibration solution) was operated in a data-dependent MS to MS/MS switching mode with the six most intense ions in each full MS scan subjected to MS/MS for further fragmentation. Full scan experiments were acquired in the m/z 300-1800 range at a resolution 60,000 (FWHM at m/z 400) and the subsequent MS/MS analysis was performed at 15,000 resolution. The following instrument parameters were used: capillary temperature, 250 °C; sheath gas, 10; auxiliary gas, 2; sweep gas, 1; spray voltage, 49 kV; tube lens, 110 V. The total scan cycle frequency was about 1 sec. The precursor ion isolation width was set at m/z ± 8.0, allowing transmission of the M and M+ 2 isotopic ions of the peptide for CID. The threshold intensity for the MS/MS trigger was set at 2,000 and fragmentation was carried out in the CID mode with a normalized collision energy (NCE) of 35. All data were collected in the profile mode. Chromeleon software and Xcalibur software (version 2.1, Thermo Scientific, San Jose, CA) were used for instrument control, data acquisition, and data processing. Peptide masses were calculated from the centroid of the isotopic envelope using MagTran software, and the extent of deuterium incorporation in each peptic fragment was determined from mass spectra (with a correction for back-exchange) as described previously 39 . Peptide mapping. Before H/D exchange experiments, pepsin digestion fragments were identified by a standard procedure involving separation on a C18 column coupled to a LTQ Orbitrap XL mass spectrometer and sequencing by tandem MS/MS using MassMatrix search engine 64,65 (MassMatrix Xtreme 3.0.9.7 Alpha with 4 M spectra limit; http://magneto.case.edu/mm-cgi/home.py) against the database containing the sequence of mouse prion protein. The separation was performed using a "slow" gradient of 2-45% acetonitrile with a total elution time of 23 min.The search was performed using the "NO enzyme" search parameter. Mass tolerance of ± 15 ppm and ± 0.8 Daltons was used for parent and monoisotopic fragment ions, respectively. The resulting DAT files generated by MassMatrix were used as input files for peptide identification, with a constraint that only MassMatrix ion scores greater than 20 were considered. Peptide identification was further confirmed by manual inspection. Total of 99 peptides were identified (see Table S1). The peptic digest was also analyzed using FindPept 66 on the ExPASy 67 Proteomics server (Swiss Institute of Bioinformatics), allowing identification of longer N-terminal peptides corresponding to residues 81-132, 85-132 and 89-132 (Table S1). It should be noted that only a fraction of peptic fragments identified by peptide mapping could be separated and analyzed with good signal-to-noise ratio under the conditions of rapid UPLC gradient required for HXMS expeiments to minimize back exchange.
Histidine hydrogen/deuterium exchange (His-HXMS) experiments. For these measurements, samples of purified PrP Sc from brain (~1-2 μ g) were suspended in D 2 O buffer (10 mM sodium phosphate, 10 μ M EDTA, 50 μ M Pefabloc, 1 ug/ml Aprotinin, pH 9.0). After incubation for 5 days at 37 °C, samples were collected by centrifugation and deglycosylated with PNGase. To obtain fragments containing single His residues, samples were then dissociated and digested with immobilized pepsin as described above for HXMS experiments, followed by digestion with silica-immobilized trypsin (2 μl; Princeton Separations, Inc., Adelphia, NJ). Finally, the peptic fragments were separated on an an UPLC BEH-C18 column (Waters, USA) and analyzed by mass spectrometry as described above for HXMS experiments. The following single-His fragments were used for His-HXMS analysis: His 95, GQGGGTHNQWNKPSKPK; His 110, HVAGAAAAGAVVGGLGG; His 139, MSRPMIHFGND; His 176, YSNQNNFVHD or YSNQNNFVHDCVN; His 186, QHTVTTTTK or ITIKQHTVT. The half-life (t 1/2 , days) of His exchange reaction was calculated as described previously 48,49 .