Axonal prion protein is required for peripheral myelin ... - Nature

a r t ic l e s

Axonal prion protein is required for peripheral myelin maintenance

© 2010 Nature America, Inc. All rights reserved.

Juliane Bremer1, Frank Baumann1, Cinzia Tiberi1, Carsten Wessig2, Heike Fischer1, Petra Schwarz1, Andrew D Steele3, Klaus V Toyka2, Klaus-Armin Nave4, Joachim Weis5 & Adriano Aguzzi1 The integrity of peripheral nerves relies on communication between axons and Schwann cells. The axonal signals that ensure myelin maintenance are distinct from those that direct myelination and are largely unknown. Here we show that ablation of the prion protein PrPC triggers a chronic demyelinating polyneuropathy (CDP) in four independently targeted mouse strains. Ablation of the neighboring Prnd locus, or inbreeding to four distinct mouse strains, did not modulate the CDP. CDP was triggered by depletion of PrPC specifically in neurons, but not in Schwann cells, and was suppressed by PrPC expression restricted to neurons but not to Schwann cells. CDP was prevented by PrPC variants that undergo proteolytic amino-proximal cleavage, but not by variants that are nonpermissive for cleavage, including secreted PrPC lacking its glycolipid membrane anchor. These results indicate that neuronal expression and regulated proteolysis of PrPC are essential for myelin maintenance. Prions cause transmissible neurodegenerative diseases associated with accumulation of PrPSc, a misfolded and aggregated form of the cellular prion protein PrPC. The brains of humans or animals affected with prion diseases show characteristic histopathological changes, but the molecular pathogenesis of these diseases is largely unknown. PrPC, in particular its central domain, is well conserved between species, suggesting that it has an important function1. In view of the role of PrPC in mediating the toxicity of PrPSc and of amyloid β (Aβ)2,3, deciphering the molecular basis of its function might uncover pathways that are active in neurodegeneration. The mouse Prnp gene has been ablated using many strategies, and the most prominent phenotype of PrPC-deficient mice (henceforth collectively referred to as Prnp−/− mice) is their resistance to prion infections4. Further subtle phenotypes, some of which are controversial 5, include neurophysiological abnormalities 6,7 and astrogliosis8,9. A late-onset peripheral neuropathy has been identified in PrP Cdeficient Nagasaki (PrnpNgsk/Ngsk) and Zürich-I (Prnpo/o) mice4,10. This indicates that PrPC might have a role in peripheral neuropathies—highly prevalent human diseases whose treatment is often unsatisfactory. We have therefore revisited the status of peripheral nerves in mice lacking PrPC. RESULTS Peripheral polyneuropathy in Prnp−/− mice At 60 weeks of age, all investigated PrPC-deficient mice (Prnpo/o; n = 52) showed chronic demyelinating polyneuropathy (CDP; Fig. 1). CDP was 100% penetrant and conspicuous in all investigated peripheral nerves (sciatic and trigeminal nerves, dorsal and ventral spinal

roots). Axon morphometry showed that large fibers were predominantly affected (Fig. 1c). Identical pathologies were detected in sciatic nerves of Prnpo/o mice on a B6/129Sv hybrid background (data not shown), Prnpo/o mice backcrossed to Balb/c mice for >17 generations (Fig. 1a), PrnpEdbg/Edbg mice (inbred 129/Ola; n = 6, Fig. 1a), PrnpEdbg/Edbg backcrossed to C57Bl/6 mice for > 8 generations (n = 2, data not shown), and PrnpGFP/GFP mice carrying a targeted substitution of Prnp with enhanced green fluorescent protein (EGFP) 11 (n = 3; Fig. 1a). No signs of neuropathy were detected in age- and strain-matched wild-type mice (C57Bl/6, B6/129Sv, Balb/c, 129/Ola; Fig. 1 and data not shown). Genetic reintroduction of PrPC via crosses to tga20 mice12 (n = 3), or introduction of one hemizygous Prnp allele (n = 4), fully prevented the polyneuropathy (Fig. 1a,c). Overexpression of the PrP paralog Dpl induces neurodegeneration in certain Prnp−/− strains13, but Dpl can partly compensate for loss of function in Prnp−/− mice13. To uncover any potential contribution of Dpl to the CDP, we investigated Prno/o mice lacking both Prnd (the gene for Dpl) and Prnp14. The ultrastructural features of the CDP in 60-week-old Prno/o mice were similar to those of Prnpo/o mice, whereas mice selectively lacking Prnd15 showed normal sciatic nerve morphology (Supplementary Fig. 1). In contrast to what we found for PrnpNgsk/Ngsk mice, we did not detect any Prnd mRNA in sciatic nerves of Prnpo/o mice (Supplementary Fig. 1). CD68 immunostaining highlighted macrophages ingesting myelin debris of degenerating nerve fibers (‘digestion chambers’). These were significantly more prevalent in sciatic nerves of 10-, 30- and 60-week-old Prnpo/o mice than in those of age-matched wild-type mice (Fig. 1b,d), indicating that the onset of polyneuropathy was much earlier than had previously been understood.

1Institute

of Neuropathology, University Hospital of Zürich, Zürich, Switzerland. 2Department of Neurology, University of Würzburg, Würzburg, Germany. 3Division of Biology, California Institute of Technology, Pasadena, California, USA. 4Department of Neurogenetics, Max-Planck Institute of Experimental Medicine, Göttingen, Germany. 5Institute of Neuropathology, Medical Faculty, Rheinisch-Westfälische Technische Hochschule (RWTH) University Aachen, Aachen, Germany. Correspondence should be addressed to A.A. ([email protected]). Received 9 November 2009; accepted 16 December 2009; published online 24 January 2010; doi:10.1038/nn.2483

310

VOLUME 13 | NUMBER 3 | MARCH 2010 nature NEUROSCIENCE

a r t ic l e s a Semithin 60 weeks

Figure 1 Peripheral polyneuropathy in Prnpo/o mice. (a) Toluidine blue–stained semithin cross sections of sciatic nerves of wild-type (Balb/c and 129/Ola wt), Prnpo/o (Balb/c), Prnp+/− (Balb/c), tga20 (B6/129Sv), PrnpEdbg/Edbg (129/Ola) and PrnpGFP/GFP (C57Bl/6) mice, all around 60 weeks of age. (b) CD68immunostained longitudinal sections of sciatic nerves. More digestion chambers with macrophages and myelin debris are visible in Prnpo/o (Balb/c) than in wild-type (Balb/c) nerves at 60 and 10 weeks of age, respectively. (c) Axonal density within nerves (number of axons per mm2) was quantified morphometrically and plotted against the cross-sectional areas (µm2) of axons (axonal density-size distribution). Error bars, s.e.m. (d) Quantification shows significantly more digestion chambers in sciatic nerves of 30- and 60-week-old Prnpo/o mice than in wild-type mice. All scale bars are 50 µm.

Prnpo/o

129/Ola wt

nodes of Ranvier (Fig. 2e). Macrophages occasionally invaded the space between b myelin sheaths and axons (Fig. 2f). In wild-type Remak bundles, several unmyelinated nerve fibers were ensheathed by single Schwann cells (Fig. 2g). By contrast, Prnpo/o Remak bundles often showed reduced numbers of unmyelinated fibers combined with abnormal branching of Schwann cell processes (Fig. 2h). Collagen pockets, possibly indicating axon loss, were present in Prnpo/o nerves but absent from wild-type Prnpo/o Prnp o/o Wild type Wild type 60 weeks 10 weeks nerves (data not shown). P = 0.0006 P = 0.03 Electron microscopy of 30-week-old c 104 d Wild type 70 Prnpo/o mice revealed morphological alteratga20 60 Prnp+/o tions and g-ratio deviations that were less 3 10 50 pronounced than, but qualitatively simi40 2 lar to, those of 60-week-old Prnpo/o mice 10 30 (Supplementary Fig. 2). These alterations 20 10 were absent from 10-week-old mice with the o/o 10 Prnp exception of rare digestion chambers (Fig. 1b). 0 1 Expression and phosphorylation of Akt and Age (weeks): 30 30 60 60 15 25 35 45 55 65 75 2 Axonal area (µm ) Prnp: +/+ o/o +/+ o/o Erk kinases, which are crucially involved in myelination16, were unaffected in 10- and 30-day-old Prnpo/o nerves (Supplementary Fig. 3), and there was no The polyneuropathy of Prnpo/o mice is demyelinating Sciatic nerves of 60-week-old Prnpo/o mice showed characteristic change in the expression of the peripheral myelin proteins CNPase, ultrastructural signs of demyelination. Whereas the ratio between MPZ/P0, MAG and PMP22 (Supplementary Fig. 3). myelin thickness and axonal diameter was constant in sciatic nerves We then induced myelination in cultured dissociated mouse dorsal of wild-type mice (Fig. 2a), Prnpo/o nerve fibers had thinned myelin root ganglia (DRG). Morphologically normal myelin was formed in sheaths despite normal axonal morphology (Fig. 2b). Quantification both wild-type and Prnpo/o cultures. Myelin basic protein (MBP)of the g-ratio (axonal diameter/fiber diameter) revealed a higher per- positive segments were similar in density (data not shown), but their centage of fibers with thinned myelin (g-ratio >0.81) in Prnpo/o than length was slightly reduced in Prnpo/o cultures in two independent in wild-type mice (Supplementary Fig. 2 and see below). Several experiments (Supplementary Fig. 3). Cell survival in Prnpo/o cultures, axons were covered by only a single layer of myelin, which might rep- as determined by propidium iodide incorporation, was not reduced resent incipient remyelination. ‘Onion bulbs’, resulting from repeated (data not shown). Nerve conduction velocities (NCV) were significantly reduced in cycles of demyelination and remyelination and eventual Schwann cell degeneration, encircled 2.1 ± 0.54% of fibers in 4 of 4 Prnpo/o nerves Prnpo/o mice at 11, 28 and 53 weeks of age (Fig. 3a). F-wave latencies but no fibers of wild-type nerves (Figs. 2b,c and see below; P = 0.029). (FWL) were marginally increased in 11- and 28-week-old Prnpo/o Demyelinated axons were occasionally surrounded by remyelinating mice and significantly prolonged in 53-week-old mice (Fig. 3b). There Schwann cells with prominent rough endoplasmic reticulum (Fig. 2d). was no significant reduction in compound muscle action potential Some Prnpo/o nerve fibers were covered by abnormally thick myelin (CMAP) amplitudes in Prnpo/o mice (28 weeks: Prnpo/o, 10.5 ± 2.6 mV, sheaths resulting from focal myelin folding, often associated with versus wild type, 10.8 ± 3.1 mV; 53 weeks: Prnpo/o, 10.6 ± 2.1 mV, +/o

Prnp

tga20

Edbg/Edbg

Prnp

GFP/GFP

+

CD68 digestion chambers per visual field

2

CD68

Prnp

Number of axons per mm


Semithin 60 weeks

Balb/c wt

nature NEUROSCIENCE VOLUME 13 | NUMBER 3 | MARCH 2010

311


a r t ic l e s Figure 2 Ultrastructural alterations in Prnpo/o sciatic nerves. (a–h) Electron microscopy of sciatic nerves of 60-week-old Prnpo/o (b–f,h) and wild-type (a,g) mice (both Balb/c). Crosssections of wild-type nerves show normally myelinated nerve fibers (a) and regular unmyelinated axons in Remak bundles (R) (a,g). Cross-sections of Prnpo/o nerves show thinly myelinated axons, surrounded by onion bulb formations (arrows) (b,c), axon (A) surrounded by Schwann cell with prominent rough endoplasmic reticulum (arrow) and increased density of other organelles (d), as well as loss of unmyelinated axons in Remak bundles. Schwann cells ensheathing unmyelinated axons (A) frequently show abnormal branching of cytoplasmic processes (arrows) (h). Longitudinal sections of Prnpo/o nerves reveal macrophages (M) within the axon-glia interface (f) and focally folded myelin (white arrows) in the vicinity of a node of Ranvier (black arrow; e). The right internode shows thinner myelin than the left one, indicating de- and remyelination. Scale bars: a,b, 8 µm; c–h, 5 µm.

a

b

c

d

e versus wild type, 11.5 ± 3.7 mV, after distal stimulation). These combined electrophysiological and ultrastructural findings strongly support the conclusion that myelin sheaths, rather than axons, were primarily affected in Prnpo/o mice. The NCV and FWL abnormalities of 1-year-old Prnpo/o mice were abolished by transgenic expression of Prnp in tga20 mice12 (Supplementary Fig. 4). On accelerating rotarods, the latency to fall of 12-, 30- and 60-week-old Prnpo/o g mice (n = 8–11) was similar to that of strain-, age- and sex-matched wild-type mice (n = 8–10; Fig. 3c). In hot plate tests, 8 of 9 wild-type mice, but only 2 of 11 Prnpo/o mice (all 60 weeks old), licked their hind paws after 60 s (endpoint). The time lag to licking was significantly longer in Prnpo/o mice (Fig. 3d). Heat responses were also marginally delayed in 12- and 30-week-old Prnpo/o mice. However, these relatively subtle results were not confirmed in 60-week-old C57Bl/6 PrnpEdbg/Edbg mice, which fared similarly to C57Bl/6 wild-type mice in both hot plate and rotarod tests (data not shown). Grip strength was significantly weakened in 60-week-old Prnpo/o mice. Younger Prnpo/o mice (12 and 30 weeks old) also showed reduced grip strength (Fig. 3e). This indicates that both afferent and efferent fibers were functionally affected in Prnpo/o mice. Neuronal PrPC is required for myelin maintenance We then sought to define the cell types in which PrPC expression is required to prevent demyelination. We first investigated sciatic nerves of Prnpo/o mice expressing the tgNSE-PrP transgene17, which drives PrPC expression from the neuron-specific enolase promoter (Fig. 4). Although Prnp mRNA levels in tgNSE-PrP nerves were only 1.9 ± 0.2% of those in wild-type nerves (Fig. 4b), indicating that there was very little Prnp transcription by Schwann cells, their PrPC concentration approached that of wild-type nerves (113 ± 5.98 versus 132 ± 1.22 ng per mg total protein, respectively; Fig. 4a). We conclude that essentially 312

f

h

all PrPC in tgNSE-PrP mice was of axonal origin. Accordingly, PrPC immunoreactivity was found along axons of tgNSE-PrP nerves, whereas in wild-type mice it was also detectable in non-compact myelin (Fig. 4e). Semithin cross sections of sciatic nerves of 60-week-old tgNSE-PrP mice were unremarkable (Fig. 4f), and axon size morphometry, quantification of thinly myelinated nerve fibers (g-ratio > 0.81) and onion bulb formation confirmed complete rescue of CDP (Fig. 4d,h). Therefore, neuronal expression of PrPC suffices to prevent Prnpo/o polyneuropathy. To study the effects of myelin-restricted PrPC expression, we generated mice expressing PrPC under the control of the proteolipid protein (PLP) promoter, which is active in myelinating Schwann cells and oligo dendrocytes. Seven lines were derived. Line 159, henceforth referred to as tgPLP-PrP, had levels of PrPC similar to those of wild-type mice in sciatic nerves (126 ± 1.54 ng per mg total protein; Fig. 4a) despite having 12-fold higher Prnp mRNA levels (Fig. 4b), and was used for further analysis. Immunofluorescence confirmed myelinspecific expression of PrPC in tgPLP-PrP mice (Fig. 4e). Sciatic nerves VOLUME 13 | NUMBER 3 | MARCH 2010 nature NEUROSCIENCE

a r t ic l e s a

b

c n.s.

7 6 5 4 3 2 1 0 Age (weeks): 11 Prnp: +/+

n.s.

P = 0.006

11 o/o

28 +/+

28 o/o

53 +/+

53 o/o

11 o/o

28 +/+

28 o/o

53 +/+

53 o/o

200 150 100 50

0 Age (weeks): Prnp: Test:

12 +/+

12 o/o

12 +/+

12 o/o

30 +/+

30 o/o

30 +/+

30 o/o

60 +/+

60 o/o

60 +/+

60 o/o

Figure 3 Electrophysiology and behavior of Prnpo/o mice. 1 1 2 2 1 1 2 2 1 1 2 2 (a) NCVs were significantly reduced in Prnpo/o mice at 11 weeks (33.0 ± 3.5 m s−1 (Prnpo/o) versus 42.3 ± 4.4 m s−1 (wild type)), Hot plate test Grip strength test at 28 weeks (37.5 ± 1.5 m s−1 (Prnpo/o) versus 50.5 ± 2.8 m s−1 n.s. n.s. n.s. P = 0.002 P = 0.014 P = 0.036 70 (wild type)) and at 53 weeks (25.1 ± 0.87 m s−1 (Prnpo/o) versus 1.25 60 54.2 ± 5.6 m s−1 (wild type)). (b) FWLs were only marginally 1.00 50 increased in Prnpo/o mice at 11 weeks (5.6 ± 0.5 ms (Prnpo/o) versus 0.75 40 5.1 ± 0.4 ms (wild type)) and 28 weeks (4.2 ± 0.2 ms (Prnpo/o) versus 30 0.50 4.0 ± 0.3 ms (wild type)), but were significantly prolonged at 53 weeks 20 0.25 (6.0 ± 0.5 ms (Prnpo/o) versus 4.4 ± 0.4 ms (wild type)). (c) Prnpo/o 10 0 0 mice showed no difference in performance on the rotarod test when Age (weeks): 12 12 30 30 60 60 Age (weeks): 12 12 30 30 60 60 compared to wild-type mice at 12, 30 and 60 weeks of age. There was Prnp: +/+ o/o +/+ o/o +/+ o/o Prnp: +/+ o/o +/+ o/o +/+ o/o some training effect, as both groups showed slightly longer latencies o/o to fall in test 2, two days after test 1. (d) Hot plate test performance was significantly worse at 60 weeks in Prnp mice. (e) Prnpo/o mice showed significantly lower grip strength at 12 and 60 weeks. All mice were kept in the Balb/c background.

of tgMBP-PrP mice, which express PrPC from the MBP promoter18, contained about 5% of wild-type PrPC protein (data not shown). At 60 weeks of age, both tgPLP-PrP Prnpo/o mice (Fig. 4g) and tgMBPPrP Prnpo/o mice (data not shown) showed CDP qualitatively similar to that of Prnpo/o mice. TgPLP-PrP Prnpo/o mice had significantly more thinly myelinated fibers (g-ratio > 0.81) and onion bulbs than tgNSEPrP and wild-type mice, but less than Prnpo/o mice (Fig. 4h), indicating that the tgPLP-PrP transgene has a weak anti-CDP effect. Thirty-five-week-old Prnpo/o mice showed significantly prolonged FWL, which were completely normalized by the tgNSE-PrP transgene (Fig. 4c). Similarly, tgNSE-PrP expression normalized the electrophysiological alterations in 1-year-old Prnp−/− mice (Supplementary Fig. 4). The FWL of tgPLP-PrP mice were similar to those of Prnpo/o mice (Fig. 4c). These data support the unexpected idea that the demyelinating polyneuropathy of Prnpo/o mice is rescued by neuron-restricted expression of PrPC. We next investigated the effects of neuron-specific PrPC depletion on peripheral myelin homeostasis. Transgenic mice carrying a loxP-flanked Prnp minigene on a Prnpo/o background (line tg37), henceforth termed tgPrnploxP, were crossed to mice expressing Cre under the control of the neuron-specific neurofilament heavy chain promoter (tgNFH-Cre) (Fig. 5)7. Recombination efficiency was tested by in situ hybridization; whereas tgPrnploxP mice showed robust Prnp transcription in spinal cord and DRG neurons, the signal was completely abrogated from all spinal cord neurons and ~70% of DRG neurons of tgPrnploxP × tgNFH-Cre mice (Fig. 5i–l; Supplementary Fig. 4). Accordingly, total PrPC protein expression in sciatic nerves was reduced after neuronal depletion of Prnp (Fig. 5m). TgPrnploxP and tgPrnploxP × tgNFH-Cre sciatic nerves showed similar Prnp mRNA expression (data not shown), excluding the possibility that Cre geno toxicity had affected endogenous Prnp transcription by Schwann cells. TgPrnploxP × tgNFH-Cre mice developed a CDP similar to that of Prnpo/o mice, featuring thinly myelinated fibers and onion bulb formation (Fig. 5a,b,d,e,g,h). We removed PrPC from Schwann cells by breeding tgPrnploxP mice with the pan-Schwann cell deletor tgDhh-Cre (backcrossed to Prnpo/o). nature NEUROSCIENCE VOLUME 13 | NUMBER 3 | MARCH 2010

Strength (N)

e

Time (s)

d


Rotarod test

300 250

Time to fall (s)

NCV (m/s)

70 60 50 40 30 20 10 0 Age (weeks): 11 Prnp: +/+

P = 0.0009

F-wave latency (ms)

P = 0.0061 P = 0.0035

Depletion of PrPC from Schwann cells was confirmed in 6-weekold tgDhh-Cre × tgPrnploxP mice; whereas age-matched tgPrnploxP mice on the Prnpo/o background expressed PrPC in both Schwann cells and axons, tgDhh-Cre × tgPrnploxP mice expressed PrPC only along axons (Supplementary Fig. 4). Also, Prnp mRNA expression was significantly reduced in tgDhh-Cre × tgPrnploxP sciatic nerves (19 ± 11% of wild-type expression in tgPrnploxP and 6 ± 1% in tgDhhCre × tgPrnploxP mice; P = 0.016). Quantification of g-ratio and onion bulb formation in these mice revealed no differences when compared with tgPrnploxP mice (Fig. 5c,f–h), indicating that PrPC expressed by Schwann cells is not required for peripheral myelin maintenance. PrPC fragments and domains required for myelin maintenance The monoclonal antibodies POM1 and POM3, which recognize epitopes in the C terminus (around amino acids (aa) 140–152) and charged cluster of PrPC (aa 95–100), respectively19, were used to investigate the relative prevalence of full-length PrPC and of its fragments by immunoblotting. C1 is generated by α-cleavage at aa 110–112, whereas C2 is derived by β-cleavage in the octarepeat region or at position 96 (refs. 20, 21). In contrast to brain tissue, sciatic nerves contained more C1 than full-length PrPC (Fig. 6a,b). Neuronal depletion of Prnp in tgPrnploxP × tgNFH-Cre mice reduced the amount of C1 (Fig. 5m). The POM1 antibody, which recognizes all three forms of PrPC, revealed their presence in non-compact myelin, Schmidt-Lanterman incisures (SLIs) and paranodes and along axonal surfaces (Fig. 4e). Might a secreted, soluble PrPC variant rescue the CDP? We addressed this question in tgGPI−PrP mice (line 44) lacking the GPI anchor22.The concentration of GPI−PrP in sciatic nerves was determined by western blot (POM11) to be ~20–25% of wild-type fulllength PrPC (Fig. 6c). TgGPI−PrP mice lacked all C1 or C2 (Fig. 6d), indicating that membrane anchorage is essential for regulated PrP C proteolysis in peripheral nerves. Sciatic nerves of 68-week-old tgGPI−PrP mice (n = 3) showed CDP similar to that found in Prnpo/o mice (Supplementary Fig. 5). The flexible amino-proximal domain of PrP comprises a charge cluster (CC1), the octarepeat region and a central domain with a second 313

a r t ic l e s a

b

POM1

kDa

100

25

10

0.1

15 β-actin

PrPC Nfasc

o/o

NSEPrP

PLPPrP

n.s. P = 0.0048 P = 0.0002

4.0

+/+

o/o

Prnp

d

10

4

NSE-PrP PLP-PrP

tgNSE-PrP

PrPC Nfasc

tg PLP-PrP

C PrP Nfasc

o/o

Wild type tg NSE-PrP

103 tgPLP-PrP o/o Prnp 35

NSE-PrP PLP-PrP

+/+

65 45 55 Axonal area (µm2)

g

f

75

h

2.5

n.s. P = 0.0002

2.0 1.5

P = 0.0059

1.0 0.5 0.0

tgNSE-PrP

tg PLP-PrP

Wild type

Prnpo/o

Percentage of fibers with onion bulb formation

3 Percentage of fibers with g-ratio > 0.81


+/+

4.5

Prnp

C57BI/6

0.01 tga20

Cumulative number of 2 axons per mm

FWL (ms)

c 5.0

P < 0.0001

1

20

Prnp

e

P = 0.0007

37

2 P = 0.033 1

0

tgNSE-PrP

tgPLP-PrP

PrPC

Figure 4 Expression of by neurons is essential for myelin sheath maintenance. (a,b) Prnp mRNA and PrPC protein content of tgNSE-PrP and tgPLP-PrP sciatic nerves were investigated by western blot (a) and by real-time PCR (b; values on ordinate are normalized against wild-type mRNA). (c) Normal FWL in 35-week-old tgNSE-PrP mice (4.1 ± 0.09 ms (tgNSE-PrP) and 4.2 ± 0.2 ms (wild type)) and prolonged latencies in age-matched B6/129Sv Prnpo/o (4.5 ± 0.2 ms) and tgPLP-PrP mice (4.3 ± 0.2 ms). (d) Quantification of cumulative axonal density-size distribution indicates nearly normal distribution in tgNSE-PrP mice and reduced large-axon density in Prnpo/o and tgPLP-PrP mice. Error bars, s.e.m. (e) PrPC localization was studied by immunofluorescence. (f,g) Semithin sections of nerves from 60-week-old tgNSE-PrP and tgPLP-PrP mice stained with toluidine blue. (h) Percentage of fibers with g-ratio > 0.81 and onion bulb formation in wild-type, Prnpo/o, tgNSE-PrP and tgPLP-PrP mice. Error bars, s.d.

charge cluster (CC2) and a hydrophobic core (Figs. 6b and 7a). PrPC lacking the octarepeats (PrP∆C, line tgC4)23 fully rescued the CDP of Prnpo/o mice (Supplementary Fig. 5). In addition to full-length PrP∆C, tgC4 sciatic nerve lysates contained the two proteolytic fragments C1 and C2 (Fig. 7a,b). We used the ‘half-genomic’ pPrPHG backbone12 to generate tgPrP∆CC mice expressing a PrPC variant lacking the CC2 domain (aa 94–110). Transgenic offspring (line 46) were used for analysis. Sciatic nerves of tgPrP∆CC mice contained C1, albeit less than wildtype nerves, and were devoid of C2 (Fig. 7a,b). Upon crossing to Prnpo/o mice, tgPrP∆CC mice did not develop CDP at 60 weeks of age (Supplementary Fig. 5). In contrast, Prnpo/o mice expressing PrPC variants lacking aa 32–134 and aa 32–121 (termed tgF35 and tgE11, respectively)23 lacked C1 and had a CDP qualitatively similar to that of Prnpo/o mice (Fig. 7a,b and Supplementary Fig. 5). We prepared detergent-resistant membranes (DRMs) from sciatic nerves of wild-type, Prnpo/o and the various PrP deletion-mutant transgenic mice. The buoyancy of the PrP deletion mutants was similar to that of PrPC, indicating that they reside in similar microdomains (Fig. 7c and Supplementary Fig. 6). Finally, we generated transgenic mice expressing a PrP variant lacking the hydrophobic core (aa 111–134). Transgenic mice were termed tgPrP∆HC (line 1146). TgPrP∆HC × Prnpo/o mice showed a reduced life expectancy (survival 80 ± 3.5 days) and showed CDP (Supplementary 314

Fig. 5). In addition, terminally sick tgPrP∆HC mice developed CNS white-matter vacuolation and astrogliosis in cerebellum, brain stem and corpus callosum (Supplementary Fig. 7). Brains of tgPrP∆HC mice lacked all C-terminal fragments (data not shown). Immunological and further morphological aspects of CDP To ascertain whether the immunomodulatory functions of Prnp can cause CDP, we crossed Prnpo/o to Rag1−/− mice lacking B and T cells. At 60 weeks of age, Prnpo/o Rag1−/− mice (but not Rag1−/− Prnp+/+ mice) showed CDP similar to that of Prnpo/o mice (Fig. 7d,e). Therefore B and T cells were not required for CDP pathogenesis. Accordingly, immunohistochemical stains did not show increased CD3+ or B220+ lymphocytes in sciatic nerves of 60-week-old Prnpo/o mice (data not shown). Saltatory conduction in peripheral nerve fibers relies on the proper arrangement of the nodes of Ranvier and paranodes. By immunofluor escence, we showed that sodium channels, neurofascin, the paranodal proteins Caspr and JamC, and the extracellular protein versican were normally distributed in Prnpo/o mice (Supplementary Fig. 3). JamC staining revealed increased SLI density along internodes of 8-monthold Prnpo/o mice. In nerves from age-matched tgNSE-PrP mice, SLI density was similar to that of wild-type nerves, whereas in nerves from tgPLP-PrP × Prnpo/o mice it was similar to that of Prnpo/o nerves (Supplementary Fig. 8). VOLUME 13 | NUMBER 3 | MARCH 2010 nature NEUROSCIENCE

a r t ic l e s

c

loxP

x tg NFH-Cre

m

lo tg xP P x rnp lo tg N xP tg FHP C x rnp lo re tg D xP hh -C re

tgPrnp

0

o

0

75

0.5

np

k

0.5

P = 0.012

1.0

o/

loxP

1.0

1.5

Pr

tgPrnp

x tg Dhh-Cre

tg

45 55 65 Axonal area (µm2)

P = 0.019 P = 0.011

pe

1

1.5

Percentage of fibers with onion bulb formation

2.0

3

35

loxP

tgPrnp loxP tgPrnp loxPx tgDhh-Cre loxP x tgNFH-Cre tgPrnp

h

Wild type loxP tgPrnp tgPrnp loxP x tg Dhh-Cre tgPrnp loxP x tg NFH-Cre Prnp o/o

ild

10

tgPrnp

f

Percentage of fibers with g-ratio > 0.81

4

10 Cumulative number of axons per mm2

β-actin 37 kDa 25 kDa 20 kDa

Figure 5 Neuron-specific but not Schwann cell–specific depletion of PrPC induces polyneuropathy. Prnpo/o mice carrying a loxP-flanked Prnp transgene, termed tgPrnploxP, were crossed to tgNFH-Cre expressing Cre in neurons or to tgDhh-Cre expressing Cre in Schwann cells. (a–h) Morphological analysis of sciatic nerves at 60 weeks of age. (b) Toluidine blue-stained semithin sections showing CDP in tgPrnploxP × tgNFH-Cre mice with neuronal PrPC depletion. (e) Electron microscopy showing onion bulb formation in a tgPrnploxP × tgNFHCre mouse. (a,c,d,f) By contrast, tgPrnploxP on a Prnpo/o background and tgPrnploxP × tgDhh-Cre showed normal morphology of sciatic nerves in semithin sections (a,c) and electron microscopy (d,f). Scale bars: a, 50 µm; d–f, 2 µm. (g) Quantification of cumulative axonal density-size distribution showing reduction of large axons in tgPrnploxP × tgNFH-Cre as in Prnpo/o nerves. Error bars, s.e.m. (h) Percentage of fibers with g-ratio > 0.81 and onion bulb formation in tgPrnploxP × tgNFH-Cre mice was significantly increased compared to tgPrnploxP × tgDhh-Cre and tgPrnploxP mice. Error bars, s.d. (i–l) Expression of Prnp mRNA was analyzed by in situ hybridization in 60-week-old tgNFH-Cre and tgPrnploxP × tgNFH-Cre mice using a Prnp antisense probe. TgPrnploxP mice express Prnp in dorsal root ganglia (i) and spinal cord neurons (j). Dashed line: border between spinal white and gray matter. Scale bars: i, 200 µm; j, 500 µm. After recombination, Prnp was undetectable in ~70% of DRG (k) and in all spinal cord neurons (l). (m) Western blot of PrPC expression in tgPrnploxP mice and following conditional depletion of Prnp. Samples were treated with PNGase or left untreated; antibody: POM1.

nature NEUROSCIENCE VOLUME 13 | NUMBER 3 | MARCH 2010

lo xP

np

Pr

tg

tg

Pr

np

o/

o

DISCUSSION CDP was associated with four indeβ-actin pendently targeted Prnp knockout fl lines, including pure 129/Ola PrnpEdbg/Edbg C2 mice. Prnp-flanking genes in PrnpEdbg/Edbg C1 and PrnpGFP/GFP mice did not cause CDP, as PNGase + both strains were made in 129/Ola-derived embryonic stem cells, and 129/Ola wild-type mice did not develop any neuropathy. Breeding to various strains and genome-wide STR analyses failed to identify any role for genes outside the Prnp locus in the pathogenesis of CDP. Prnpo/o and PrnpEdbg/Edbg mice suffered from CDP despite normal expression of Dpl, indicating that Dpl upregulation did not cause the polyneuropathy. CDP was present in mice lacking both Prnp and Prnd14 but absent from mice selectively lacking Prnd15. Therefore Dpl, unlike PrPC, is not required for the maintenance of peripheral nerves. The first signs of CDP (mild electrophysiological alterations, reduced grip strength and CD68+ digestion chambers) were detected around 10 weeks of age, immediately after completion of peripheral myelination. Accordingly, ultrastructural examinations confirmed the initial formation of morphologically normal myelin in the absence of PrPC. We conclude that demyelination is early but not congenital. Phosphorylation of Akt and Erk, which is regulated during myelination16, was indistinguishable between Prnpo/o and wild-type myelinating nerves. Assuming that any differences have not been obfuscated by neuronal Akt and/or Erk, these results indicate that PrPC, even if it modulates Akt, Fyn, cAMP and

pe ty

Genomic analyses We analyzed DNA short tandem repeats (STRs) in 37 mice for the presence of mouse strain–specific polymorphisms that could contribute to the observed phenotypes. Twenty-two STRs covered chromosome 2 (on which Prnp resides) and 184 STRs covered the rest of the genome. We were unable to identify any genetic association other than Prnp that would explain the CDP phenotype. The genetic markers not flanking the Prnp locus were similar in mice with and without neuropathy. This was true for mice on the mixed background B6/129Sv (tgNSE-PrP, Prnpo/o, tgPLP-PrP, tgDelC4, tgPrP∆CC, tgF35) and for mice on FVB (tgPrnploxP, tgPrnploxP × tgNHF-Cre, tgPrnploxP × tgDhh-Cre; Supplementary Fig. 9). When comparing wild-type and Prnpo/o mice on the Balb/c background, STR polymorphisms delineated a genomic region of ~14 cM segregating with the Prnpo allele and CDP. As this region was preserved in Prnpo/o mice whose neuropathy was suppressed by an unlinked Prnp transgene (for example, tgNSE-PrP), it is unlikely that it determined the CDP phenotype (Supplementary Figs. 10–12).

P x rnp lo tg N xP FH -C re

PNGase –

l

ild

j

W


x tg NFH-Cre

e

g

i

loxP

W

d

tgPrnp

ty

b

np

loxP

Pr

tgPrnp

a

315

–

sn –

b

POM1 brain + – + fl

POM11 POM3

kDa PrPc full length 37 C2 fragment 25 C1 fragment

C2

aa 23

20 C1 15

c

POM1

CC2 HC H1

OR

CC1

H2 H3

GPl

PNGase –

SS SS

51

SS

90 112 134

kDa 37

PNGase +

178 213 231

Pr

d

25 β-actin

Pr –/– n Pr p +/+ n tg p +/– G P Pr l –P n r Pr p –/– P n Pr p +/+ n tg p +/– G Pl – P Pr np rP Pr –/– np Pr +/+ n tg p +/– G Pl – Pr P

PNGase

POM3 brain + – +

POM11

sn

np

a

Pr n Pr p –/– n Pr p +/– n Pr p 25% n Pr p 12. 5 tg np 6. % G 13% P – Pr l P r n Pr p –/– P np + Pr /– n Pr p 25% np Pr 12. 5 n tg p 6.1 % G 3% Pl – Pr P

a r t ic l e s

kDa

tg E11 (∆32–121) tg F35 (∆32–134) tg PrP∆HC (∆111–134)

+ –

No No

Normal Normal

SS

–

–

Yes

~30

SS

–

–

Yes

~14

–

–

Yes

~11

SS

51

90 112 134

tg E11

tg C4

tg PrP∆CC

Prnp +/+

–/–

tg F35

kDa 37 25 20

tgC4

SS

+ +

PNGase +

Prnp +/+

SS

PNGase –

POM1

tg C4 (∆32–93)

Peripheral Life expectancy C1 C2 neuropathy (weeks) + + No Normal

Prnp

GPl

SS

tg PrP∆CC (∆94–110)

aa 23

H2 H3

tg F35

CC2 HC H1

tg E11

OR

tgC4

CC1

POM1

Wild type

tg PrP∆CC

b

+/+

a

Prnp –/–

adaxonal myelin into the axon-glia interspace. Alternatively, PrPC might have an additional, cell-autonomous role in Schwann cells which could be related to the localization of SLIs. Tissue-specific expression patterns of transgenes do not always accurately reflect those of the parental genes. We therefore investigated whether PrPC expression in the above mice was as intended. As peripheral nerves consist mostly of Schwann cell bodies and neuronal processes, their extracts contain predominantly Schwann cell mRNA, whereas neuronal mRNA is localized mainly in DRGs and spinal cord neurons. Quantitative RT-PCR analyses revealed abundant Prnp transcripts in tgPLP-PrP nerves, but little in tgNSE-PrP sciatic nerves. By contrast, PrP protein levels in tgPLP-PrP and tgNSE-PrP nerves were similar to those in wild-type nerves. Hence, transgenic expression in tgNSE-PrP nerves is largely restricted to the neurons, and suppression of polyneuropathy is not due to illegitimate PrPC expression in Schwann cells. Furthermore, in situ hybridization confirmed that the loxP-flanked Prnp transgene was excised in most tgPrnploxP × tgNFHCre neurons, whereas the similar levels of Prnp mRNA in tgPrnploxP × tgNFH-Cre and tgPrnploxP in sciatic nerves excludes the possibility of ectopic recombination events. Prnp

Erk1/2 (refs. 24–28), exerts its effects on myelin through different pathways. The clinical manifestations of the polyneuropathy were limited to reduced grip strength and nociception. These phenotypes are consistent with incomplete demyelination and might explain some of the reported behavioral abnormalities of Prnpo/o mice29,30. As PrPC-deficient sciatic nerves showed classical ultrastructural and electrophysiological features of demyelinating neuropathies, we were surprised to discover that its prevention required expression of PrPC by neurons but not by Schwann cells. This unexpected result was confirmed in five genetic paradigms: tissue-restricted Prnp expression in neurons or Schwann cells, and tissue-specific ablation of Prnp by neuron- or Schwann-cell-restricted Cre transgenes. Therefore, peripheral myelin requires neuronal PrPC in trans, indicating that PrPC is involved in directional communication from axons to Schwann cells (Supplementary Fig. 13). Although the tgPLP-PrP transgene partially rescued CDP, the lack of pathology after Schwann cell–specific depletion argues against an essential role for PrPC in Schwann cells. The partial rescue might be explained by the excessive myelin-specific expression of PrPC in tgPLPPrP mice. Some of this surplus PrPC could be transferred to neurons by ‘GPI painting’31, or myelinotrophic fragments could be released from

15 β-actin

178 213 231

d

c tg F35

tg E11

tg PrP∆CC

tgC4

+/+

Prnp

Prnp

–/–

Figure 7 Role of N-terminal domains and lymphocytes in the pathogenesis of Prnpo/o polyneuropathy. (a) Transgenic constructs, presence of PrP C-terminal fragments in the transgenic mice, presence of peripheral neuropathy, and kDa life expectancy. (b) Mice expressing deletion mutants of PrP were analyzed –/– +/+ Rag1 Prnp for PrP expression and processing in sciatic nerves. Western blots before and 37 e after PNGase treatment of lysates are shown. Asterisk marks full-length PrP; 25 hash marks C2 fragment; circle marks C1 fragment. A western blot with better 20 separation of full-length and C2 fragment for tgC4 mice is shown (all lanes are from the same blot, but intervening lanes were deleted for clarity). (c) DRMs were 15 prepared from sciatic nerves of transgenic and wild-type mice and subjected to –/– o/o Flotillin Rag1 Prnp step density gradient centrifugation. Fractions containing DRM were analyzed by western blot for presence of PrP C and flotillin as a control. Western blots of the fractions from this experiment are shown in Supplementary Figure 6. (d,e) Toluidine blue-stained semithin cross sections of sciatic nerves of 60-week-old Rag1−/− × Prnp+/+ mice showed normal morphology of the sciatic nerve (d) and peripheral neuropathy in Rag1−/− × Prnpo/o mice (e). POM1


POM1

PNGase – PNGase + PNGase + Figure 6 PrPC expression and proteolytic processing in sciatic nerves of wild-type and tgGPI−PrP mice. 37 Western blot analysis comparing PrPC protein expression in the sciatic nerve with that in the brain of wild25 type mice, using two different monoclonal antibodies, POM1 and POM3, for detection. (a) After PNGase 20 treatment of protein lysates, two bands were recognized by POM3 antibody, whereas POM1 detected three 15 bands. (b) This band pattern is explained by the localization of antibody epitopes and by the existence of β-actin three PrP isoforms, each containing an intact C terminus: full-length PrP, C2 fragment and C1 fragment. (c) TgGPI−PrP mice were analyzed for PrP expression with POM11 antibody, before and after PNGase treatment in comparison to serially diluted Prnp+/− sciatic nerve lysates. (d) For analysis of PrP processing, we used antibody POM1 which detects all holo-PrP C and all C-terminal fragments. POM1 detected GPI−PrP to a lesser extent than POM11. However, no formation of C-terminal fragments was observed, even after very long exposures (right).

316



a r t ic l e s Long-term axonal integrity depends on glial products including CNPase, myelin peroxisomes, myelin-associated glycoprotein and ciliary neurotrophic factor32. In addition, the anterograde myelin sheath degradation that occurs after axonal cuts indicates that myelin health depends on axonal signals whose molecular nature is largely unknown. The evidence reported here identifies PrP C as a crucial neuronal mediator of peripheral myelin maintenance. Whereas neuronal NRG1 type III regulates Schwann cell development and myelination through ErbB receptors on myelinating cells33, PrPC seems to fulfill a different role: rather than disrupting primary myelination, its absence impairs long-term myelin maintenance. Therefore, these two proteins are unlikely to use identical transduction pathways. However, PrPC modulates the activity of BACE34, which cleaves NRG1 type III and regulates myelination35,36, raising the question of whether PrPC influences myelin maintenance through BACE and NRG1 type III. We failed to detect any difference in NRG1 expression between Prnpo/o nerves and wild-type nerves (Supplementary Fig. 14) or in PrPC catabolism between BACE1−/− nerves and wildtype nerves (data not shown). PrPC might interact with myelin components directly or through other axonal proteins (Supplementary Fig. 13). Some of the reported PrPC interacting proteins have roles in myelin homeostasis37, and represent possible candidates for mediation of its myelinotrophic effects. Which structural motifs of PrPC are involved in myelin maintenance? The octapeptide repeat region was not required for myelin maintenance, whereas mice expressing PrP lacking the central domain (aa 94–134) developed CDP38. The hydrophobic core, but not the charge cluster (CC2), of this central PrPC domain was essential for peripheral myelin maintenance. PrPC undergoes regulated proteolysis in late secretory compartments20,39–41. We observed an association between the presence of CDP and lack of the C1 fragment in sciatic nerves. All transgenic mice showing CDP (tgE11, tgF35, tgGPI−PrP, tgPrP∆HC) lacked C1, whereas all PrP mutants in which the CDP was rescued (tgC4 and tgPrP∆CC) produced abundant C1. Cleavage of PrPC appears therefore to be linked to its myelinotrophic function. This conjecture might also explain the requirement for membrane anchorage of PrPC uncovered in tgGPI−PrP mice, as anchorless PrPC did not undergo regulated proteolysis. Analogously to NRG1, cleavage of axonal PrPC might expose bioactive fragments, which in turn might transmit signals to Schwann cells (Supplementary Fig. 13). Alternatively, the cleavage might suppress a PrPC-mediated signal, whose abrogation could result in uncontrolled signaling and toxicity. Theoretically, axonal PrPC might have indirect effects on myelin. However, this would require lack of PrPC to damage axons to the extent that it affects myelin maintenance, while being morphologically and electrophysiologically undetectable, and this is unlikely. Prion diseases mainly affect the CNS, but we did not detect myelin degeneration in optic nerves, corpus callosum or spinal cords of 60-week-old Prnpo/o mice (Supplementary Fig. 15). Nevertheless, subliminal myelin pathologies might extend to central myelin in Prnp−/− mice8, and transgenic mice expressing toxic PrPC mutants show both peripheral and central myelinopathy17,38. PrPC deficiency was reported to affect synaptic function6,7. However, the amplitudes of foot muscle CMAPs following distal stimulation were not significantly altered in 53-week-old Prnpo/o mice, arguing against an important presynaptic defect in neuromuscular synaptic transmission. It has been suggested that PrPC has various roles in immunity42, and lymphocytes are important in mouse models of hereditary demyelinating neuropathies. As the CDP in our mutant mice was not modulated by removal of Rag1, lymphocytes are not involved in its pathogenesis. nature NEUROSCIENCE VOLUME 13 | NUMBER 3 | MARCH 2010

The combined results of restricting expression of PrPC to neurons and of selectively depleting PrPC from neurons indicate that the expression of PrPC by neurons is essential for the long-term integrity of peripheral myelin sheaths. Not only was the trophic function of PrPC exerted in trans, but it also correlated with the proteolytic processing of PrPC in diverse transgenic mouse models. These findings identify PrPC as a critical messenger of transcellular axomyelinic communication and indicate that regulated proteolysis of axonal PrPC might expose domains that interact with Schwann cell receptors. Clarifying the molecular basis of these phenomena might lead to a better understanding of peripheral neuro pathies—particularly those of late onset—and might help to uncover new therapeutic targets for these common, debilitating disorders. Methods Methods and any associated references are available in the online version of the paper at http://www.nature.com/natureneuroscience/. Note: Supplementary information is available on the Nature Neuroscience website. Acknowledgments We thank M. Delic, R. Moos, D. Goriounov, C. Tostado, H. Mader, K. Nairz, M. Bieri, N. Wey and D. Meijer for methodological advice and technical help. G. Mallucci provided tgNFH-Cre and tgPrnploxP mice, D. Meijer provided tgDhh-Cre mice, J.C. Manson provided PrnpEdbg/Edbg mice, J. Collinge provided Prnpo/o FVB mice, S. Lindquist and W. Jackson provided PrnpGFP/GFP mice and P. Saftig and A. Rittger provided nerves from BACE1−/− mice. We thank W.B. Macklin for the PLP plasmid; T. Rülicke for pronuclear injections and H. Welzl, I. Drescher and S. Wirth for help with behavioral tests. J.A. Girault and M.T. Dours-Zimmermann donated anti-paranodine/ Caspr and anti-versican antibodies, respectively. We thank B. Seifert for statistical consulting. A.A. received an ERC Advanced Investigator Grant and grants from the European Union (PRIORITY and LUPAS), the Novartis Foundation and the Swiss National Foundation. J.B. received a Career Development award from the University of Zürich. C.W. and K.V.T. were supported by the Departmental Research Fund. AUTHOR CONTRIBUTIONS J.B. and A.A. designed the study and wrote the manuscript. J.B., F.B., C.T., C.W., H.F., P.S., A.D.S., K.V.T. and J.W. did the experiments. J.B., F.B., C.T., C.W., H.F., A.D.S., K.V.T., K.-A.N., J.W. and A.A. analyzed the data. COMPETING INTERESTS STATEMENT The authors declare no competing financial interests. Published online at http://www.nature.com/natureneuroscience/. Reprints and permissions information is available online at http://www.nature.com/ reprintsandpermissions/. 1. Aguzzi, A., Baumann, F. & Bremer, J. The prion’s elusive reason for being. Annu. Rev. Neurosci. 31, 439–477 (2008). 2. Brandner, S. et al. Normal host prion protein necessary for scrapie-induced neurotoxicity. Nature 379, 339–343 (1996). 3. Laurén, J., Gimbel, D.A., Nygaard, H.B., Gilbert, J.W. & Strittmatter, S.M. Cellular prion protein mediates impairment of synaptic plasticity by amyloid-beta oligomers. Nature 457, 1128–1132 (2009). 4. Büeler, H. et al. Normal development and behavior of mice lacking the neuronal cell-surface PrP protein. Nature 356, 577–582 (1992). 5. Lledo, P.M., Tremblay, P., Dearmond, S.J., Prusiner, S.B. & Nicoll, R.A. Mice deficient for prion protein exhibit normal neuronal excitability and synaptic transmission in the hippocampus. Proc. Natl. Acad. Sci. USA 93, 2403–2407 (1996). 6. Collinge, J. et al. Prion protein is necessary for normal synaptic function. Nature 370, 295–297 (1994). 7. Mallucci, G.R. et al. Postnatal knockout of prion protein alters hippocampal CA1 properties, but does not result in neurodegeneration. EMBO J. 21, 202–210 (2002). 8. Nazor, K.E., Seward, T. & Telling, G.C. Motor behavioral and neuropathological deficits in mice deficient for normal prion protein expression. Biochim. Biophys. Acta 1772, 645–653 (2007). 9. Steele, A.D., Lindquist, S. & Aguzzi, A. The prion protein knockout mouse: a phenotype under challenge. Prion 1, 83–93 (2007). 10. Nishida, N. et al. A mouse prion protein transgene rescues mice deficient for the prion protein gene from Purkinje cell degeneration and demyelination. Lab. Invest. 79, 689–697 (1999). 11. Heikenwalder, M. et al. Lymphotoxin-dependent prion replication in inflammatory stromal cells of granulomas. Immunity 29, 998–1008 (2008). 12. Fischer, M. et al. Prion protein (PrP) with amino-proximal deletions restoring susceptibility of PrP knockout mice to scrapie. EMBO J. 15, 1255–1264 (1996).

317


a r t ic l e s 13. Moore, R.C. et al. Ataxia in prion protein (PrP)-deficient mice is associated with upregulation of the novel PrP-like protein doppel. J. Mol. Biol. 292, 797–817 (1999). 14. Genoud, N. et al. Disruption of Doppel prevents neurodegeneration in mice with extensive Prnp deletions. Proc. Natl. Acad. Sci. USA 101, 4198–4203 (2004). 15. Behrens, A. et al. Absence of the prion protein homologue Doppel causes male sterility. EMBO J. 21, 3652–3658 (2002). 16. Ogata, T. et al. Opposing extracellular signal-regulated kinase and Akt pathways control Schwann cell myelination. J. Neurosci. 24, 6724–6732 (2004). 17. Radovanovic, I. et al. Truncated prion protein and Doppel are myelinotoxic in the absence of oligodendrocytic PrPC. J. Neurosci. 25, 4879–4888 (2005). 18. Prinz, M. et al. Intrinsic resistance of oligodendrocytes to prion infection. J. Neurosci. 24, 5974–5981 (2004). 19. Polymenidou, M. et al. The POM monoclonals: a comprehensive set of antibodies to non-overlapping prion protein epitopes. PLoS One 3, e3872 (2008). 20. Watt, N.T. & Hooper, N.M. Reactive oxygen species (ROS)-mediated beta-cleavage of the prion protein in the mechanism of the cellular response to oxidative stress. Biochem. Soc. Trans. 33, 1123–1125 (2005). 21. Mangé, A. et al. Alpha- and beta- cleavages of the amino-terminus of the cellular prion protein. Biol. Cell 96, 125–132 (2004). 22. Chesebro, B. et al. Anchorless prion protein results in infectious amyloid disease without clinical scrapie. Science 308, 1435–1439 (2005). 23. Shmerling, D. et al. Expression of amino-terminally truncated PrP in the mouse leading to ataxia and specific cerebellar lesions. Cell 93, 203–214 (1998). 24. Mouillet-Richard, S. et al. Signal transduction through prion protein. Science 289, 1925–1928 (2000). 25. Schneider, B. et al. NADPH oxidase and extracellular regulated kinases 1/2 are targets of prion protein signaling in neuronal and nonneuronal cells. Proc. Natl. Acad. Sci. USA 100, 13326–13331 (2003). 26. Toni, M. et al. Cellular prion protein and caveolin-1 interaction in a neuronal cell line precedes fyn/erk 1/2 signal transduction. J. Biomed. Biotechnol. 2006, 69469 (2006). 27. Chen, S., Mange, A., Dong, L., Lehmann, S. & Schachner, M. Prion protein as trans-interacting partner for neurons is involved in neurite outgrowth and neuronal survival. Mol. Cell. Neurosci. 22, 227–233 (2003).

318

28. Santuccione, A., Sytnyk, V., Leshchyns′ka, I. & Schachner, M. Prion protein recruits its neuronal receptor NCAM to lipid rafts to activate p59fyn and to enhance neurite outgrowth. J. Cell Biol. 169, 341–354 (2005). 29. Meotti, F.C. et al. Involvement of cellular prion protein in the nociceptive response in mice. Brain Res. 1151, 84–90 (2007). 30. Nico, P.B. et al. Altered behavioral response to acute stress in mice lacking cellular prion protein. Behav. Brain Res. 162, 173–181 (2005). 31. Liu, T. et al. Intercellular transfer of the cellular prion protein. J. Biol. Chem. 277, 47671–47678 (2002). 32. Nave, K.A. & Trapp, B.D. Axon-glial signaling and the glial support of axon function. Annu. Rev. Neurosci. 31, 535–561 (2008). 33. Britsch, S. The neuregulin-I/ErbB signaling system in development and disease. Adv. Anat. Embryol. Cell Biol. 190, 1–65 (2007). 34. Parkin, E.T. et al. Cellular prion protein regulates beta-secretase cleavage of the Alzheimer′s amyloid precursor protein. Proc. Natl. Acad. Sci. USA 104, 11062–11067 (2007). 35. Hu, X. et al. Bace1 modulates myelination in the central and peripheral nervous system. Nat. Neurosci. 9, 1520–1525 (2006). 36. Willem, M. et al. Control of peripheral nerve myelination by the beta-secretase BACE1. Science 314, 664–666 (2006). 37. Rutishauser, D. et al. The comprehensive native interactome of a fully functional tagged prion protein. PLoS One 4, e4446 (2009). 38. Baumann, F. et al. Lethal recessive myelin toxicity of prion protein lacking its central domain. EMBO J. 26, 538–547 (2007). 39. McMahon, H.E. et al. Cleavage of the amino terminus of the prion protein by reactive oxygen species. J. Biol. Chem. 276, 2286–2291 (2001). 40. Sunyach, C., Cisse, M.A., da Costa, C.A., Vincent, B. & Checler, F. The C-terminal products of cellular prion protein processing, C1 and C2, exert distinct influence on p53-dependent staurosporine-induced caspase-3 activation. J. Biol. Chem. 282, 1956–1963 (2007). 41. Walmsley, A.R., Watt, N.T., Taylor, D.R., Perera, W.S. & Hooper, N.M. Alpha-cleavage of the prion protein occurs in a late compartment of the secretory pathway and is independent of lipid rafts. Mol. Cell. Neurosci. 40, 242–248 (2009). 42. Isaacs, J.D., Jackson, G.S. & Altmann, D.M. The role of the cellular prion protein in the immune system. Clin. Exp. Immunol. 146, 1–8 (2006).



ONLINE METHODS Mice. We housed mice and performed animal experiments in accordance with the Swiss Animal Protection Law and in compliance with the regulations of the veterinary office (canton Zurich). We included the following mice in our study: Prnpo/o (ref. 4) on a mixed C57Bl/6;129Sv background or backcrossed to Balb/c >17 generations; PrnpEdbg/Edbg (ref. 43) and controls (both 129/Ola), or PrnpEdbg/Edbg backcrossed >8 generations to C57Bl/6; tgGPI−PrP (line 44)22; tgNFH-Cre; tgPrnploxP (ref . 7); and PrnpGFP/GFP (ref. 11). We kept tga2012, tgNSE-PrP, tgMBPPrP17, tgC4, tgE11, tgF35 (ref. 23), Prno/o (ref. 14) and Prnd−/− (ref. 15) on a C57Bl/6; 129Sv background. Except for tgGPI−PrP, tgE11 and tgF35, we maintained all mice homozygous for the respective transgenes. We crossed Rag1−/− (ref. 44) with Prnpo/o to obtain Rag1−/− Prnpo/o. We bred tgDhh-Cre45 to FVB Prnpo/o and subsequently to tgPrnploxP mice. We either kept tgPrnploxP homozygous or crossed them to Prnpo/o (FVB). To generate tgPLP-PrP, we amplified the Prnp open reading frame by PCR using the following primers: forward (fw): AscI-PrP 5′-GGG GGC GCG CCA ATT TAG GAG AGC CAA GCA GA-3′; reverse (rev): PmeI-PrP: 5′-CAG GTT TAA ACC ACG AGA ATG CGA AGG AAC A-3′. We then cloned it with AscI and PmeI into a PLP promoter cassette. Linearized transgenic constructs (ApaI/SacII) for generation of tgPrP∆HC and tgPrP∆CC were based on pPrPHG12. We subcloned a PmeI/NheI fragment of pPrPHG into pMECA. The resulting plasmid was used as a template to create Prnp∆HC and Prnp∆CC. We used the following primer sets: for Prnp∆HC, the sets RAMP fw: 5′-CTA TCA GTC ATC ATG GCG AAC C-3′ and RAMP111 rev: 5′-CCA AAA TGG ATC ATG GGC CTC ACA TGC TTG AGG TTG GTT T-3′, and 5′-RAMP111–134 fw: 5′-AAA CCA ACC TCA AGC ATG TGA GGC CCA TGA TCC ATT TTG G-3′ and RAMP rev: 5′-CAT CAT CTT CAC ATC GGT CTC G-3′; and for Prnp∆CC, the sets RAMP fw (see above) and RAMP111–93 rev: 5′-GCT GCC GCA GCC CCT GCC ACA CCC CCT CCT TGG CCC CAT C-3′, and RAMP93–111 fw: 5′-GAT GGG GCC AAG GAG GGG GTG TGG CAG GGG CTG CGG CAG C-3′ and RAMP rev (see above). We mixed the resulting PCR product pairs, either for Prnp∆HC or Prnp∆CC. We added the flanking primers RAMP fw and RAMP rev, amplified the fusion products, and cloned them first into a pMECA pPrPHG subcloning vector (XmaI/BstEII) and then into a pPrPHG plasmid (PmeI/NheI). Linearized transgenic constructs were injected into fertilized Prnp+/+ or Prnp+/− oocytes (C57Bl/6;DBA/2), bred to Prnpo/o (B6/129Sv) and maintained as heterozygous lines. We identified transgene-positive mice by PCR using the following primers: tgPLP-PrP (fw: 5′-TCA TTT TTA AGA ATG GGA CAG CTG G-3′; rev: 5′-TTT GCT GGG CTT GTT CCA CT-3′); tgPrP∆CC and tgPrP∆HC (fw5′- CAA CCG AGC TGA AGC ATT CTG CCT-3′; rev: 5′-CCT GGG ACT CCT TCT GGT ACC GGG TGA CGC-3′). Genome-wide STR analysis. We analyzed 206 distinct STRs using fluorescently labeled primers (Supplementary Table 1). We purified genomic DNA from tail biopsies, amplified by PCR using PCR colorless Taq mastermix (Promega), and fluorescently labeled primer mix (FAM, VIC, NED; Applied Biosystems). Diluted adducts were added to Hi-Di Formamide (Applied Biosystems) containing GeneScan 600LIZ size standard (Applied Biosystems). Following denaturation, we subjected the samples to a 16-capillary sequencer 3130xl (Applied Biosystems). We performed analysis, allele-calling, binning and calibration of mouse strains manually and in combination with in-house–developed software (Applied Biosystems). Ultrastructural investigations. We anesthetized and transcardially perfused mice with PBS followed by 3.9% glutaraldehyde in 0.1 M sodium phosphate buffer, pH 7.4. We embedded the tissues in Epon using standard procedures and stained semithin sections (500 nm) with toluidine blue. We mounted ultrathin sections on copper grids coated with Formvar membrane and contrasted with uranyl acetate and lead citrate. We examined the specimens using a Philips CM12 electron microscope (FEI) operating at 80 kV. We took pictures with a Gatan Bioscan 1K × 1K digital camera (Gatan GmbH). For morphometric analysis of axon sizedensity distribution, we photographed semithin sections of at least three mice per genotype and analyzed the images using a semi-automatized software developed in house. We plotted axonal density cumulatively against axonal size. We assessed and quantified g-ratio and onion bulbs on electron micrographs using Analysis software (Olympus) and homebred scripts. Immunohistochemistry. We fixed sciatic nerves in 4% formalin and embedded them in paraffin, or snap froze nerves in OCT medium. We incubated

doi:10.1038/nn.2483

longitudinal paraffin sections and frozen sections with the following antibodies: anti-CD68 (Serotec MCA 1957), anti-CD3 (clone SP7, NeoMarkers) or antiB220/CD45R (Pharmingen). Secondary antibodies: goat anti-rat IgG (Caltag lab R40000) followed by donkey anti-goat, conjugated with alkaline phosphatase (Jackson Labs 705-055-147) or followed by donkey anti-goat conjugated with peroxidase (Jackson Labs 705-035-147) or goat anti-rabbit conjugated with peroxidase (Jackson Labs 111-035-144). We mounted slides in DAKO aqueous mounting medium and analyzed them on an Axiophot microscope (Zeiss), equipped with a JVC digital camera (KY-F70; 3CCD). Electrophysiological investigations. We investigated motor nerve conduction in 12-, 28-, 35- or 53-week-old anesthetized mice as described46. Upon supramaximal stimulation of the tibial nerve at the ankle (‘distal’) and stimulation of the sciatic nerve at the sciatic notch (‘proximal’), we recorded compound muscle action potentials (CMAP) with needle electrodes in the foot muscles. We calculated nerve conduction velocities in m s−1 from distal and proximal latencies. We recorded F-wave latencies and took the shortest latencies upon repeated stimulation at the ankle. Western blots and ELISA. We homogenized sciatic nerves in 1% Triton X 100, 137 mM NaCl, 2 mM EDTA, 20 mM Tris HCl, pH 8, using a Polytron PT 3100 (Kinematica). We determined the protein concentration using the BCA protein assay (Pierce). We boiled proteins in LDS (Invitrogen) containing β-mercaptoethanol, treated some samples with PNGase F (NEB), and separated them on NuPAGE Novex Bis-Tris Gel (Invitrogen). We blotted the gels onto nitrocellulose membranes (Schleicher & Schuell) using XCell II Blot Module (Invitrogen) and blocked membranes with TBST containing Top-Block (Sigma), decorated with monoclonal antibody POM1 or POM319, antibodies against CNPase (Abcam), MPZ/P0, MAG (Zymed), PMP22 (Abcam), NRG-1 (C20 and H210, Santa Cruz), Erk, p-Erk, Akt, p-Akt (Cell Signaling Technology) or flotillin (BD Bioscience) followed by incubation with the secondary antimouse IgG1 (Zymed) or anti-rabbit IgG (Calbiochem). We visualized images using SuperSignal West Pico Chemiluminescent Substrate System (Pierce) and Amersham Hyperfilm ECL films (GE Healthcare). In addition, we determined the PrPC concentration by sandwich ELISA with biotinylated monoclonal antibodies as described19 (POM1 for capturing, biotinylated POM2 for detection). Extraction of detergent-resistant membranes and step density gradient centrifugation. We performed Triton X-100 extraction as described38 with 100 µg of total protein for 1 h. The total volume of the gradient was 2.2 ml; we centrifuged the gradients for 2 h, collected fractions of 200 µl and analyzed them by western blot. RNA isolation and quantitative PCR. We homogenized sciatic nerves in Trizol (Invitrogen) by using a Polytron PT 3100 (Kinematica). We extracted RNA and purified it on RNeasy columns (Qiagen). We synthesized cDNA with QuantiTect Reverse Transcription kit (Qiagen) and analyzed by real-time PCR using QuantiFast SYBR Green PCR kit (Qiagen) and 7900HT (Fast Real-Time PCR systems; Applied Biosystems). The following primers were used: Prnp fw: 5′-TGGCTACATGCTGGGGAGC-3′, Prnp rev: TTCTCCCGTCGTAATAGGC; Prnd fw: 5′-CTA CGC GGC TAA CTA TTG-3′, Prnd rev: 5′-CGC CGG TTG GTC CAC-3′; GAPDH fw: 5′-CCA CCC CAG CAA GGA GAC T-3′; GAPDH rev: 5′-GAA ATT GTG AGG GAG ATG CT-3′. Immunofluorescence. Freshly cut frozen sections (10 µm, air-dried) and dissociated mouse DRG cultures on cover slips were fixed and permeabilized using ice-cold ethanol or methanol (−20 °C). We blocked in 5% BSA and 0.3% Triton X-100 (Sigma) in PBS, 3% BSA and 3% donkey serum (Jackson ImmunoResearch Laboratories), 3% BSA and MOM Blocking reagent (Vector lab), or 2% FCS. We diluted antibodies in blocking buffer or in antibody diluent (Ventana). Antibodies against the following antigens were used: sodium channel (Sigma), PrPC (POM1)19, neurofascin (Abcam), paranodine/Caspr, versican (V1), JamC (Santa Cruz), neurofilament NF-H (Calbiochem) and MBP (Serotec). We labeled POM1 with Alexa Fluor dye 488 (Invitrogen) as described for Cy5 labeling of POM211, and used the following secondary antibodies: biotinylated anti-mouse IgG (Vector laboratories)/streptavidin Alexa 488 (Invitrogen), anti-rabbit Alexa 594 and anti-rat Alexa 488 (Invitrogen). We mounted slides in DAKO fluorescence

nature NEUROSCIENCE


mounting medium and acquired images on an Olympus BX61TRF fluorescent microscope equipped with an F-View camera. We generated merged images of the different color channels in Photoshop (Adobe) and applied modifications in contrast and brightness always to the entire image and equally to all images. The JamC-positive area was quantified with Analysis software (Olympus). Each data point represents the mean value of one mouse, which is derived from three independent visual fields per mouse. At least n = 3 mice were analyzed for each condition. In situ hybridization. We derived sense and antisense probes for Prnp from a pBluescript KS plasmid (Stratagene) containing a 290-bp Asp718/BstEII fragment of mouse Prnp. We labeled the probes with digoxygenin (DIG) using the DIG-labeling kit (Roche). We fixed freshly cut frozen sections (10 µm, air-dried) in 4% PFA in PBS. Following treatment with 0.1M HCl and acetylation, we prehybridized sections for 2 h in hybridization buffer without probe (50% formamide, 5× SSC, 5× Denhardt’s solution, 250 µg ml−1 E. coli t-RNA (Roche)). For hybridization, we added 50 ng × 50 µl−1 DIG-labeled RNA sense or antisense probe to this buffer. After denaturation of the probes, we incubated sections 30 min at 85 °C, and overnight at 58 °C. We washed slides in pre-warmed SSC of different concentrations and incubated sections with alkaline phosphatase–conjugated anti-DIG antibody (Roche) in 100 mM Tris-HCl, pH 7.5; 150 mM NaCl containing 1% blocking reagent (Roche). For detection, we used 100 mM Tris-HCl, pH 9.5; 150 mM NaCl; 50 mM MgCl2 containing 1 mM levamisole; nitroblue tetrazolium chloride (NBT); and 5-bromo-4-chloro-3-indolyl phosphate, toluidine salt (BCIP) (Sigma). We mounted slides in DAKO aqueous mounting medium and analyzed them by microscopy as described above. In vitro myelination. Cultures and analyses of dissociated mouse DRG cultures were performed as described47. Behavioral tests. For the hot-plate test, we placed mice on a hot plate (constant temperature: 52 ± 0.5 °C) and recorded latencies to lick hindpaws or to jump. If no responses occurred, we removed the mice after 60 s. We tested each animal only once; mice were not habituated to the apparatus before testing. For the grip-strength test, we picked up the mice by the tail and lowered them toward a horizontal wooden bar (diameter 3 mm). After the mice grasped the bar, we gently pulled them back until they let go. The bar was attached to a spring balance

nature NEUROSCIENCE

that allowed us to determine the force at which the mice released their grip. We tested grip strength five times in a row with 2-min intervals between trials. For the accelerating rotarod test, we placed the mice on a rotating rod with a speed slowly accelerating from 4 to 40 r.p.m. (Ugo Basile, model 47600). Maximum speed was reached after 245 s. We recorded the latencies at which the animal fell off the rod or removed animals that did not fall off after 300 s. The median latency to fall (untrained mice) was tested five times in a row with an interval of 30 min between trials. Forty-eight hours later, we tested the same mice again five times, and recorded the median latency to fall (trained mice). Statistical analysis. For comparison of normally distributed data, including density of digestion chambers, and electrophysiological measurements of Prnpo/o compared to wild-type mice, we performed unpaired, two-tailed Student’s t-tests after equality of variances was tested by the F-test. In the case of unequal variances (density of digestion chambers in 30-week-old mice), we used unpaired t-test with Welch’s correction. For comparison of JamC-positive areas, we performed a one-way ANOVA followed by Bonferroni’s post-test for multiple comparisons. For statistics on behavioral test results, and on STR analysis, we performed MannWhitney tests. For statistical analysis of fibers with high g-ratio, we performed square root transformation, followed by two-tailed Student’s t-test. For statistics on onion bulb formation, we used two-tailed Fisher’s exact test. P-values were as indicated in the figures. We used SPSS (SPSS Inc.) and Prism software (GraphPad Software) for statistical tests. Numbers following the ± sign represent standard deviation (s.d.) unless otherwise indicated.

43. Manson, J.C. et al. 129/Ola mice carrying a null mutation in PrP that abolishes mRNA production are developmentally normal. Mol. Neurobiol. 8, 121–127 (1994). 44. Mombaerts, P. et al. RAG-1-deficient mice have no mature B and T lymphocytes. Cell 68, 869–877 (1992). 45. Lindeboom, F. et al. A tissue-specific knockout reveals that Gata1 is not essential for Sertoli cell function in the mouse. Nucleic Acids Res. 31, 5405–5412 (2003). 46. Zielasek, J., Martini, R. & Toyka, K.V. Functional abnormalities in P0-deficient mice resemble human hereditary neuropathies linked to P0 gene mutations. Muscle Nerve 19, 946–952 (1996). 47. Bermingham, J.R. Jr. et al. The claw paw mutation reveals a role for Lgi4 in peripheral nerve development. Nat. Neurosci. 9, 76–84 (2006).

doi:10.1038/nn.2483