Beta-endoproteolysis of the cellular prion protein by dipeptidyl peptidase-4 and fibroblast activation protein

Significance In the fatal brain disorders known as prion diseases, the cellular prion protein (PrPC) is converted into an abnormal structure by other abnormal prion protein molecules. A fragmentation process known as β-cleavage that splits PrPC into two parts is associated with prion diseases, but a clear description of the underlying cleavage mechanism is lacking. Here, we use cultured cells, cell-free systems, and mouse models to show that β-cleavage of PrPC can be performed by two proteins closely related to each other: dipeptidyl peptidase-4 and fibroblast activation protein. By applying inhibitors of these proteins to prion-infected cells, we also show that the β-cleavage activity of dipeptidyl peptidase-4 in particular may be important in the pathogenesis of prion diseases.


Cell lysis and collection of conditioned medium.
For experiments using 96-well plates, the culture medium was aspirated and replaced with ice-cold lysis buffer consisting of 50 mM Tris, pH 7.4, 150 mM NaCl, 1% (v/v) Nonidet P-40 substitute, 0.5% (w/v) sodium deoxycholate, 0.36% (w/v) SDS, 29 mM DTT, 1 mM EDTA, and a protease inhibitor cocktail. After 15 min incubation on ice, the plate was centrifuged at 1000 ⨯ g for 10 min at 4 °C and supernatants were transferred either to microcentrifuge tubes or to 96-well untreated polypropylene microplates if subsequent treatment with PNGase F was required (see relevant section). For experiments using 6-well plates, the cell monolayer was rinsed twice with ice-cold PBS before lysis using the aforementioned buffer lacking DTT and with the SDS concentration adjusted to 0.1%. After 15 min incubation on ice, lysates were scraped into tubes (combining duplicates) and clarified by centrifugation at 15000 ⨯ g for 10 min at 4 °C. Total protein concentrations of the supernatants were determined by bicinchoninic acid assay. For experiments requiring the collection of conditioned media, the culture media from replicate wells were aspirated, combined and concentrated 6-fold using Amicon Ultra-15 Centrifugal Filter Units.

Further information on mouse models and details of the tissues used in this study. Homozygous
Dpp4-null mice of the Dpp4 tm1Nwa line are fertile and healthy, with no major abnormalities observed (5). However, Dpp4-null mice do have increased levels of insulin secretion and improved glucose tolerance (5), findings that led to the development of DPP4 inhibitors for the treatment of type II diabetes in humans. The Fap em1Tcp/Ddr line of Fap-null mice were generated more recently using CRISPR/Cas9 gene editing (6). Again, homozygous mice of this line do not display any major abnormalities. However, in contrast to the Dpp4-null mice, glucose tolerance and insulin secretion were unaffected by Fap KO (6). Animal handling procedures and husbandry were in accordance with Canadian Council on Animal Care guidelines. Experimental procedures were approved by the Animal Care Committee at the University of Ottawa (AUPs 2909 and 2920). Dpp4 +/+ and Dpp4 -/littermates were housed in pairs in standard cages at 23 ˚C on a 12 h light and dark cycle. Upon sacrifice at 8-10 weeks of age, mice were perfused with icecold PBS (~10 mL) and tissue dissections were performed via midline incision. The Fap +/+ and Fap -/-EWAT and IWAT as well as a portion of the pancreas tissues analysed here derive from an earlier study (6) in which mice were either fed regular chow throughout or were raised on regular chow before being switched to a 45% high-fat diet at 12 weeks of age (this dietary intervention was incidental to the aims of the current study). Fap +/+ and Fap -/littermates were sacrificed at 40 weeks of age by CO2 asphyxiation (no PBS perfusion). EWAT were extracted from mice fed regular chow, whereas pancreas tissues and IWAT were from mice switched onto the high-fat diet. Additional Fap +/+ and Fap -/tissues (lung, kidney, spleen, brain and extra pancreas tissues) were derived from a separate colony of the same line; these mice were cared for, sacrificed, perfused with PBS and dissected as described above for the Dpp4 -/mice. In all cases, extracted tissues were snap-frozen on liquid nitrogen prior to storage at -80 °C. Further information on the tissues is provided in Table S1.
Tissue homogenization. Non-brain tissues were manually homogenized using a Dounce homogenizer in ice-cold lysis buffer consisting of 50 mM Tris, pH 7.4, 150 mM NaCl, 1% Nonidet P-40 substitute, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM EDTA and a protease inhibitor cocktail. Homogenates were clarified by centrifugation at 15000 ⨯ g for 10 min at 4 °C. Total protein concentrations in the supernatants were determined by bicinchoninic acid assay. Brain tissues were processed in the same manner except that PBS containing 1 mM EDTA and the protease inhibitor cocktail was used in place of lysis buffer and the homogenates were not clarified prior to use.

PNGase F treatment.
Cell lysates or tissue homogenates were deglycosylated using PNGase F when PrP C was the target for immunoblotting or capillary western assays. All reagents used were from a New England Biolabs kit. Samples in tubes were incubated at 98 °C for 10 min in the presence of Glycoprotein Denaturing Buffer. The denatured samples were deglycosylated by 1 h incubation at 37 °C with 5-15 units/µL PNGase F (depending on total protein concentration) in the presence of GlycoBuffer 2 and 1.25% (v/v) Nonidet P-40. For cell lysates in 96-well plates, the 98 °C incubation step in the presence of Glycoprotein Denaturing Buffer was omitted.
Edman sequencing. Samples were denatured and reduced, separated by SDS-PAGE and transferred to membranes as previously described (1), except for the following changes: i) Immobilon-PSQ polyvinylidene fluoride membranes were used to enhance retention of low-MW fragments; and ii) a splitbuffer transfer system consisting of 60 mM Tris, 40 mM N-cyclohexyl-3-aminopropanesulfonic acid, pH 10, and either 15% (v/v) ethanol (anode) or 0.1% SDS (cathode) was used to minimize glycine contamination. After transfer, membranes were rinsed once with H2O, incubated for 1 min in Coomassie stain (0.3% [w/v] Coomassie Brilliant Blue R-250, 45% [v/v] methanol, and 10% [v/v] glacial acetic acid), and were destained by several 5 min washes in 50% methanol. Membranes were air-dried and the bands of interest excised using a razor blade. Edman sequencing was performed by staff at The Protein Facility of the Iowa State University Office of Biotechnology. Briefly, each membrane fragment was loaded into a Shimadzu PPSQ-53A instrument and subjected to 10 cycles of Edman degradation. A Wakopak Wakosil PTH-GR column was used with a flow rate of 300 µL/min and a gradient elution approach. The major labelled amino acid in each cycle was determined by manual inspection of the absorbance spectrum (269 nm) by staff at the facility.    charts showing that levels of C2 relative to total PrP in RK13 cells were lower when MoDPP4 or MoFAP were transiently co-expressed with G/S-switch PrP compared with WT MoPrP (one-sample t-tests; n = 3 independent experiments; **, p < 0.01), and that this effect was not due to differences in MoDPP4 or MoFAP expression levels. Data are shown as means ± S.D. Different contrast settings were applied to each part of (C, F), as indicated by the dividing lines. WT MoPrP lanes in (C) display signals from the same capillaries as the equivalent lanes of Fig. 2A. The Sha31 anti-PrP antibody was used for obtaining the panel (C) image.      blotting (B, D) using the Sha31 anti-PrP antibody. Conventional western blots were used for kidney and brain tissues because the somewhat lower sensitivity of the capillary western system prevented accurate quantification of PrP C signals in general (kidney) or C2 signals specifically (brain). Data in the charts are shown as means ± S.D. No significant effect of Dpp4 genotype on PrP C fragmentation was detected in any tissue, as determined by unpaired two-sample t-tests (n = 5-10; p > 0.05). Vertical dividing lines indicate that samples were analysed on different capillary western plates or western blots. Panel (C) shows data from the same spleen homogenates analysed either in two batches ("split") or in a single assay ("combined"). No total protein image could be obtained for the "combined" analysis of spleen homogenates due to the limited number of capillaries that can be analysed simultaneously . For panel (D), a pool of all the samples was included in quadruplicate on each blot to control for inter-blot variability; C2 signals relative to total PrP are expressed as percentages of the mean value obtained from the sample pool. Asterisks indicate samples that were not included in quantification due to air bubbles overlapping with bands.  blotting (B, C) using the Sha31 anti-PrP antibody. The quantification data displayed in Fig. 5 derive from these images. Vertical dividing lines indicate that samples were analysed on different capillary western plates or western blots. Conventional western blots were used for kidney and brain tissues because the somewhat lower sensitivity of the capillary western system prevented accurate quantification of PrP C signals in general (kidney) or C2 signals specifically (brain). In panel (C), the C2 bands were quantified from the longer exposure and the FL and C1 bands from the shorter exposure. No total protein image could be obtained for EWAT (E) due to a recurring problem with electrophoretic separation specific to these samples. (H) Chart re-capitulating the C2 data from Figs. 5D, F and G, but with the sex of each sample marked (blue = WT, male; orange = WT, female; green = KO, male; gold = KO, female). EWAT data are not shown here since all EWAT samples were male. No sex-specific differences could be detected. (I) Confirmatory western blot of EWAT homogenates (Sha31 antibody; +PNGase F) and chart showing a significant Fap KO-induced reduction in C2 levels similar to that observed by capillary westerns. Data are shown as means ± S.D. (Mann-Whitney U test; n ≥ 6; **, p < 0.01). (J) Capillary western data showing FAP expression levels in various WT and Fap-null mouse tissues. It was not possible for all tissues of each genotype to be from the same individual mouse, but two sets of tissues analysed in independent experiments (separated by the dividing line) gave similar results.   confirming that DPP4 is detected in the cultures and that its expression level may be increased by linagliptin treatment (n = 3 technical replicates). (C) Representative confocal images and chart showing co-localization of DPP4 and PrP C immunostaining in (uninfected) primary cerebellar glial cultures. Data are shown as means ± S.D, with glial fibrillary acidic protein (GFAP) used as a control; co-localization of DPP4 with the astrocyte marker GFAP would not be expected because GFAP is expressed in the cytoplasm (n = 16 images from a single experiment). Scale bar = 20 µM.   1 Open Biosystems reagents are now sold by Horizon Discovery but this vector was purchased before that change occurred. 2 Used for the initial compound screen as well as being the source of E64d, pepstatin A, marimastat and PD 151746 protease inhibitors used in later experiments.