Reduction of PrPC in human cerebrospinal fluid after spinal cord injury

Abstract

It has been estimated that cerebrospinal fluid (CSF) contains approximately 80 proteins that significantly increase or decrease in response to various clinical conditions. Here we have evaluated the CSF protein PrPC (cellular prion protein) for possible increases or decreases following spinal cord injury. The physiological function of PrPC is not yet completely understood; however, recent findings suggest that PrPC may have neuroprotective properties. Our results show that CSF PrPC is decreased in spinal cord injured patients 12 hours following injury and is absent at 7 days. Given that normal PrPC has been proposed to be neuroprotective we speculate that the decrease in CSF PrPC levels may influence neuronal cell survival following spinal cord injury.

Introduction

The cellular prion protein, PrP^C, is an evolutionarily conserved protein whose biological function remains elusive (reviewed in refs. 1 and 2). Following synthesis in the rough endoplasmic reticulum, PrP^C is constitutively transported to the cell surface where it remains attached to the plasma membrane by a C-terminal glycosylphosphatidyl (GPI) anchor;3 this localization facilitates its association with dynamic lipid raft membrane domains.4 PrP^C is subject to N-linked glycosylation and non-, mono- and di-glycosylated versions of PrP^C are simultaneously found in the cell.5 In addition to N-linked glycosylation sites, PrP^C contains a single intramolecular C-terminal disulfide bond. As new PrP^C is synthesized, the cell must also make triage/quality control decisions, either refolding or destroying misfolded PrP^C in the proteosome or lysosome; however, the precise mechanism(s) that control steady state PrP^C levels are not yet known. PrP^C is highly expressed thoughout the central nervous system,6 but is also found in many non-neuronal tissues such as cardiac muscle, lymphoid tissue and testes.7 PrP null mice are resistant (see below) to infectious prions8 but are otherwise healthy with only subtle phenotypic abnormalities.1

Prion disease such as variant Creutzfeld Jakob disease and Kuru are fatal neurodegenerative disorders associated with the conversion of the cellular prion protein (PrP^C) into an alternatively folded disease specific isoform (PrP^Sc).9 Currently, no effective treatment for prion disease is known. Either sporadic protein misfolding or discrete genetic mutations can result in the initial conversion of normal cellular PrP^C, a protein of predominantly α helical structure, into a protein of predominantly β-sheet structure that is insoluble in detergents and partially resistant to protease K, PrP^Sc. PrP^Sc is a self propagating infectious protein conformation that progressively converts endogenous cellular PrP^C into the abnormal PrP^Sc conformation.9 Hence, unlike other neurodegenerative disorders (e.g., Huntington’s disease, Alzheimer disease), ingested PrP^Sc is believed to spread from gut to brain so that prion diseases are transmissible within species and sometimes between species (reviewed in ref. 10). PrP^Sc replication is posttranslational, however, at present there is conflicting evidence regarding the exact cellular compartment(s) in which PrP^Sc replicates. Expression of the normal cellular form of prion, PrP^C, is essential for susceptibility to prion disease as well as replication of infectivity since PrP knockout mice are disease resistant.8 Scrapie in sheep, chronic wasting disease (CWD) in elk and mule deer, bovine spongiform encephalopathy (BSE) in cattle and variant Creutzfeldt-Jacob disease (vCJD) in humans represent prion diseases that are caused by oral exposure to transmissible spongiform encephalopathy (TSE) agents. Despite concentrated efforts prompted by the threat of potential prion diseases to human health, the normal physiological function of PrP^C has remained enigmatic. Moreover, the PrP^Sc triggered cascade of pathogenic events underlying prion-mediated neurotoxicity remains to be characterized.

Progressive misfolding of endogenous PrP^C into disease associated PrP^Sc, excessive accumulation of prion aggregates and amyloid-like plaques, disruption of normal cellular function, neuronal loss and histopathological spongiform change are the signposts of prion disease development. Molecular chaperones have been prominent among the cellular factors speculated to influence prion disease progression. From a cell biological perspective, the initial conversion of normal cellular PrP^C to the aberrant PrP^Sc conformation, the subsequent propagation of PrP^C to PrP^Sc, the formation of oligomeric species/prion fibrils and the inability to eliminate toxic PrP^Sc can be attributed to a failure of the cellular protein quality control (i.e., the cellular chaperone machinery). Accordingly, if the collective function of cellular chaperones in health individuals is sufficient to prevent accumulation of misfolded prions, it follows that when protein quality control is compromised prion disease would progress.

Chaperones are classified into families on the basis of their molecular weight: Hsp110, Hsp70, Hsp60, Hsp40 and Hsp25. Several chaperone families have been implicated in PrP^C protein homeostasis including: Hsp110,11 Hsp60,11,12 αBcrystallin,13 Rdj2,14 Hsp40/Hsp70,15 BiP/grp94calnexin/calreticulin16 and Hsp90 (inhibitors).17 In addition, given that extracellular protein deposits are associated with many neurodegenerative diseases the notion of an extracellular counter to intracellular molecular chaperones has become a recent focus of research activity (reviewed in refs. 18–20). Although chaperones have been documented in CSF, the primary source of extracellular chaperones and the physiological significance or extracellular chaperones remain to be established. Furthermore, knockout of the transcriptional regulator of the heat shock response, HSF1 in mice result in a shorter life span than control mice after inoculation with PrP^Sc. This implicates stress induced chaperones in the survival of prion infected mice.21 Proteins like PrP^C that transit through the secretory/endocytotic pathways encounter distinct chaperones in different cellular compartments (e.g., ER resident chaperones). It follows that in a given compartment, an excess of misfolded protein or a decrease in chaperones would exacerbate protein misfolding and amyloid formation. Although chaperones are generally thought to have neuroprotective functions in maintaining protein conformation and clearing cells of misfolded proteins, it is possible that specific chaperones may override the natural energy barrier between PrP^C and PrP^Sc to participate in prion disease.

The aim of this study was to evaluate human cerebrospinal fluid (CSF) for changes in the levels of PrP^C after spinal cord injury. The CSF compartment is specialized to bathe the CNS and in addition to providing neuroprotection may be a key factor in the spread of prion disease pathogensis. Changes in the spectrum of CSF proteins reflect either changes in the biosynthesis/secretion of proteins or proteolytic degradation of proteins. Proteins may also end up in the CSF as part of cell turnover or pathological conditions (e.g., brain tumors, damage of blood-CSF barrier). Under normal conditions PrP^C is released from cells into the CSF.22–26 We evaluated PrP^C in CSF samples for two reasons: First, PrP^C has been hypothesized to be neuroprotective27–38 and it follows that changes in CSF PrP^C expression levels might influence neural cell survival. Secondly, CSF from some CJD patients is infectious39 and changes in the CSF levels of PrP^C may well alter the rate of PrP^Sc replication.

Here we provide the first evidence that PrP^C levels in the CSF are lower in all spinal cord injured patients compared to control patients and that the reduction was PrP^C specific. We also examined CSF for components of the traditional cellular quality control machinery, which may be important for the maintenance of prion conformation and neuroprotection. Hsp70, Hsc70, Hsp60 were absent in all CSF samples including neurological controls. Our results indicate that many traditional intracellular protein quality control pathways are absent in CSF following spinal cord injury and we speculate that CSF proteins like PrP^C are relatively vulnerable to misfolding resulting in the accumulation of potentially toxic species in this compartment. Surprisingly, members of the Hsp27 family, which are characterized by the presence of a homologous sequence of 80–100 residues, referred to as the α-crystallin domain, were notably increased after spinal cord injury (within 12 h of surgery) in some CSF samples. Our findings identify CSF as a unique compartment that can undergo large changes in PrP^C and Hsp27s levels.

Results and Discussion

PrP^C is reduced in CSF after spinal cord injury. In order to investigate the possibility that CSF levels of PrP^C change, we evaluated PrP^C levels in the CSF of patients after spinal cord injury. CSF contains a wealth of diagnostic/prognostic information. Figure 1A shows the western blot analysis of PrP^C in human CSF from a control patient, and a patient at 12 h, 24 h, 2, 3, 4, 5 and 6 d after spinal cord injury. PrP^C is detected by Western analysis as 28 kDa, 32 kDa and a dominant 37 kDa PrP^C immunoreactive band, representing the un-, mono- and di-glycosylated PrP^C species respectively that are not completely resolved in Figure 1A. PrP^C in CSF may be from (a) either damaged cells or (b) hydrolysis of the GPI anchor. Figure 1B further shows PrP^C levels were lower in four spinal cord injured patients compared to control patients and that the 37 kDa band is the major species in the CSF samples. Samples taken within 12 h following injury already had lower values than controls (no injury) and PrP^C was often not detectable at 7 d following spinal cord injury. Reduction in CSF levels of PrP^C after spinal cord injury was confirmed with two distinct antibodies (6H4 Prionics and 11C6 SPI-BIO). Table 1 shows the clinical data from six controls and 13 spinal cord injured patients. We found that minocycline treatment did not alter reduction of PrP^C. Tetracycline has been reported to decrease prion infectivity in animal models in other studies.40 Variations among the CSF levels of PrP^C were observed between patients, but in all cases PrP^C was reduced following spinal cord injury. Taken together, these results confirm previous studies documenting PrP^C in human CSF24,41,42 and further show that PrP^C levels in CSF can be subject to variations in response to trauma. We specifically show that spinal cord injured patients have reduced levels of PrP^C in CSF.

14-3-3 is not a marker for spinal cord injury. Next we examined CSF for the presence of the cytoplasmic protein, 14-3-3 (Fig. 2A). 14-3-3 is an abundant protein of the central nervous system and the presence of 14-3-3 proteins in CSF samples is a clinical biomarker for extensive and rapid neuronal destruction (e.g., prion disease, dementia).43 This diagnostic test is currently used as the leading preliminary screen for prion diseases; however, it is used in combination with other diagnostic criteria due to the overlap between prion diseases and other conditions that elevate CSF levels of 14-3-3. None of the CSF samples evaluated, from either control or spinal cord injured patients, were found to have detectable 14-3-3 protein. Note that there are multiple 14-3-3 isoforms expressed in human brain and these experiments evaluated 14-3-3beta/zeta isoforms. Our results indicate that the presence of 14-3-3 is not a clinical biomarker for spinal cord injury. In contrast, CSF levels of IgG heavy chain (55 kDa) and light chain (26 kDa) were observed to increase following spinal cord injury. Figure 2A demonstrates that IgG light chain increases in CSF after spinal cord injury. We evaluated the antibodies present in the CSF patient samples for PrP^C immunoreactivity. CSF antibodies were not immunoreactive with PrP^C (data not shown).

A CSF expression of crystallin-domaincontaining- proteins is variable in patients following spinal cord injury. In contrast, Figure 2B demonstrates that members of the Hsp27 family were identified in some (but not all) CSF samples. αA/αB-crystallin immunoreactive proteins were found in several CSF samples. Also, Hsp27 immunoreactive bands were found in several CSF samples. Significant differences in Hsp25/27 expression in the brain and spinal cord is reported between mice and rats.44,45 In contrast to cellular prion protein, Hsp27 and αA/αB-crystalin levels in CSF did not correlate with spinal cord injury. αA/αB-crystallin has been implicated in the prevention of protein aggregation and fibril formation in a wide range of diseases. Although a direct PrP^C::αB-crystallin association has been documented,13 our results indicate that CSF levels of αA/αB-crystallin do not decrease like that observed for PrP^C following spinal cord injury and that αA/αB-crystallin levels in the CSF are variable between patients following spinal cord injury (Fig. 2B). Taken together these results indicate that following spinal cord injury CSF levels of PrP^C decrease, while 14-3-3 levels remain unchanged and IgG levels increase. In contrast to PrP^C, αA/αB-crystallin, another protein implicated in neuroprotection, does not show a consistent reduction in CSF levels following spinal cord injury.

Hsp40s, Hsc70, Hsp90 and Hsp110 are absent from CSF obtained from control and spinal cord injured patients. The relevance of molecular chaperones to diseases involving protein misfolding remains a central biological question. In addition to prion diseases a number of other protein conformation disorders are associated with amyloid deposits (e.g., Alzheimer disease, Down syndrome)20 in the absence of proteosome inhibition. The traditional view is that chaperones interact with hydrophobic residues associated with partially unfolded unstructured chain segments, to guard against various folding traps such as aggregation and other nonproductive interactions that result in disease. Current evidence suggests that different chaperones follow distinct strategies to fold nacent proteins, protect, refold mature proteins and eliminate misfolded proteins. Once the native conformation is achieved, chaperones are thought to dissociate from the protein. While typically thought to be neuroprotective, it is possible that chaperones may protect proteins that contribute to diseases processes (e.g., toxic conformations or conformations that elicit an autoimmune response). To begin to address the role of protein chaperones in maintaining function we evaluated CSF of patients after spinal cord injury for Hsp40, Rdj2, Hdj2, Hsp110, Hsp90, Hsp70, Hsc70 and Hsp60. Figure 3 shows that these traditional cellular chaperones were not found in the CSF of spinal cord injured patients. ATP levels in CSF are lower than cellular levels,20 further suggesting that the ATP dependent intracellular chaperone machinery that PrP^C encounters during its transit through the secretory/endocytotic pathway does not maintain PrP^C conformation in the CSF. Taken together, although several members of the traditional ATP-dependent chaperone machinery have been implicated in PrP^C conformational regulation,19 with the exception of the Hsp27 family (see below), they do not appear to play a role in CSF in the patients evaluated in this study.

Methods

Subject enrollment. Between 2004 and 2009 subjects presenting with spinal cord injury (SCI) with motor deficit to the Foothills Hospital spine service and presenting within 12 h of their injury were offered enrolment in a clinical trial examining minocycline and spinal cord perfusion pressure (SCPP) augmentation (ClinicalTrials.gov identifier NCT00559494). Following informed consent, all subjects received a lumbar indwelling catheter for measurement of spinal fluid pressure and for sampling of CSF. Surgical decompression and stabilization was performed within 24 h of injury. Of note, no subjects were treated with corticosteroids. All protocols were approved by the University of Calgary Conjoint Health Research Ethics Board.

Minocycline treatment. Subjects were randomized (1:1) to intravenous minocycline or placebo. The first 10 subjects were randomized to placebo (equivolume of normal saline) versus minocycline (Wyeth Pharmaceuticals) 200 mg intravenously twice daily. Subsequent subjects were randomized to placebo versus minocycline 800 mg intravenous loading dose followed by twice daily dosing tapering each dose by 100 mg to 400 mg that was then maintained for the remainder of time. Study treatment was continued for 7 days.

Perfusion pressure augmentation. An indwelling lumbar catheter inserted at the L4/5 level was transduced to measure spinal cerebrospinal fluid pressure. An indwelling arterial catheter placed in the radial artery at the wrist was similarly transduced for blood pressure. SCPP (mean arterial pressure minus cerebrospinal fluid pressure) was electronically monitored in real time. If the SCPP fell below 75 mmHg at any time during the first seven days, (except during a surgical procedure), the patient was randomly assigned to guideline directed blood pressure management (control) versus SCPP augmentation. Subjects assigned to control were administered crystalloid fluid and if necessary inotrope therapy (norepinephrine) to maintain mean arterial pressure above 65 mmHg. Subjects assigned to SCPP augmentation were administered crystalloid fluid and if necessary inotrope therapy (norepinepherine) to maintain SCPP greater than 75 mmHg. Treatment was continued until the end of day 7.

CSF collection. CSF samples (up to 10 mL each) were collected from the indwelling lumbar catheter starting at the time of enrolment before first administration of the study intervention and for the subsequent 7 d scheduled around the twice daily study drug administration (30 min infusion) as follows: ½ hour before drug infusion, ½ hour after drug administration and 6 h after drug administration essentially providing samples every 6 h. Samples were centrifuged at 2,000 rpm for 5 min to separate cellular matter and the supernatant was flash frozen and stored at −80°C in aliquots.

Neurological control samples (10 mL) were collected from six subjects by lumbar puncture under general anaesthesia prior to microdiscectomy surgery. These individuals had evidence of symptomatic degenerative disc disease refractory to non-surgical management for greater than 12 wks but no evidence of central nervous system pathology on history and physical exam.

Sample analysis. Prior to all sample preparation procedures, all samples were blinded. Forty microliters of each CSF sample was fractionated by SDS-PAGE and transferred from polyacrylamide gels to nitrocellulose (0.45 µm) in 20 mM Tris, 150 mM glycine and 12% methanol. Membranes were blocked with 4% milk solution (prepared in PBS with 0.1% Tween 20) and incubated with primary antibody for 2 h at room temperature or overnight at 4°C. The membranes were washed in blocking solution and incubated with horseradish peroxidase-coupled secondary antibody. The signal was developed using West Pico Pierce reagent (Pierce Biotechnology, Inc.,) and exposed to Kodak film.

Cytosolic and membrane fractions of rat brains were prepared as previously described49 and PC12 cells were prepared as previously described.50

Antibodies. Anti-PrP^C monoclonal were from Prionics (6H4) or SPI-BIO (11C6) as indicated. Anti-Hsp70 mouse monoclonal (SPA 810), anti-Hsp110 rabbit polyclonal (SPA 1101), anti- Hsp40 rabbit polyclonal (SPA 400), anti-Hsp90 rat monoclonal (SPA 835), anti-Hsp25 rabbit polyclonal (SPA 801), anti-Hsp27 mouse monoclonal (SPA 800), anti-Hsp20 rabbit polyclonal (SPA 803), anti-14-3-3 mouse monoclonal (KAM CC012E) and αA/αB-crystallin were from Assay Designs. Anti-Hsp70/Hsc70 mouse monoclonal (H5147) and Hsp60 (H4149) was from Sigma.

Summary

A main observation of this study is that based on CSF analysis of PrP^C, spinal cord injured patients could clearly be distinguished from neurological controls that show no evidence of central nervous system pathology. While suggested roles for PrP^C include cell signaling, differentiation and adhesion, an increasing number of studies have implicated PrP^C in neuroprotection against oxidative stress, ischemia and toxic protein aggregation.27–36,38,46 Given the proposed neuroprotective role for PrP^C, low levels of PrP^C in CSF following spinal cord injury may well create a window in which neurons are susceptible to damage. CSF reduction in PrP^C following spinal cord injury is specific (Figs. 1 and 2). For example, expression of proteins with αA/αB crystallin domains were observed to increase, decrease or remain unchanged following spinal cord injury depending on the patient. What regulates PrP^C levels in CSF? The primary determinants of the level of PrP^C in the CSF are not yet established but almost certainly involve a balance between folding, trafficking, release, retrotranslocation and degradation. Future investigations will establish if CSF levels of PrP^C are reduced in neurological situations other than spinal cord injury. In agreement with our study, Meyne et al.47 recently showed that total prion levels are reduced in the CSF in patients with a number of neurological diseases. Cellular levels of PrP^C are subject to rapid change, for example PrP^C is reported to be upregulated after focal cerebral ischemia34 and disparities between the PrP mRNA and cellular PrP^C protein expression have been documented.48 Understanding the precise mechanisms that underlie abrupt changes in the levels of PrP^C in CSF, such as the large reduction observed after spinal cord injury, should offer significant insights into the physiological role of prions.

The CSF compartment is specialized to bathe the CNS and thereby has the capacity to deliver both neuroprotective proteins (e.g., PrP^C) and toxic proteins that transmit diseases (e.g., PrP^Sc). CSF from some CJD patients is infectious,39 indicating the presence of the toxic prion template in human CSF. However, infectivity of human CSF from CJD patients is not 100%. It is tempting to speculate that changes in the levels of PrP^C in CSF may support this range of sensitivity to PrP^Sc infection. While high levels of PrP^C in CSF may be neuroprotective as outlined above, it would also support the accumulation of the pathogenic isoform. This raises the possibility that reducing PrP^C in CSF would reduce sensitivity to PrP^Sc infection. Furthermore, Castagna et al.24 have previously documented extensive PrP^C fragmentation using two-dimensional electrophoresis of human CSF. Their data indicate that measurements of the full length proteins (e.g., Fig. 1) underestimates total PrP^C in CSF. These PrP^C fragments differ in glycosylation and N terminal truncation and are proposed to be generated during the extracellular/intracellular cycling of PrP^C a process also implicated in the neuronal spread of toxic prions. It is possible that extensive PrP^C fragmentation would influence PrP^Sc’s infectivity. Thus, in the case of prion disease, changes in CSF levels of PrP^C may well influence disease progression following PrP^Sc exposure, however there is currently no direct support for this notion. Future studies will have to explore whether high CSF levels of PrP^C do in fact effect prion disease progression.

Figures and Tables

Figure 1 PrP^C levels in human CS F are attenuated following spinal cord injury. (A) Western analysis of PrP^C in human CS F from control (lane1) and from spinal cord injured patient #19 (remaining lanes) at 12 h, 24 h, day 2, day 3, day 4, day 5 and day 6 following spinal cord injury. 40 µl of the indicated CS F samples were separated by SDS PAGE, the Western blot was probed with anti-PrP^C 6H4 (Prionics) antibody and developed with SuperSignal® West Dura (Pierce). (B) Western analysis of representative human CS F from control, 12 h within spinal cord injury and 7 d after spinal cord injury (40 µl). Four patients are shown. Blots were probed with anti-PrP^C 6H4 (Prionics) antibody or anti-PrP^C 11C6 antibody (SPI-BIO).

Figure 2 14-3-3 and IgG light chain levels in CS F after spinal cord injury. (A) Western analysis of 14-3-3 and IgG light chain after spinal cord injury. Three patients are shown. 14-3-3 is absent in all CS F samples. IgG light chains are present in varying amounts in all samples. 20 µg of rat brain supernatant and pellet fractions and 20 µg of PC12 cell homogenate (middle) are shown in the right lanes. (B) Western analysis of CS F (40 µl) after spinal cord injury for αA/αB crystallin and Hsp27.

Figure 3 Molecular chaperones were not detected in CSF of control or spinal cord injured patients. Western analysis of Hsp40, Rdj-2, Hdj-2, Hsp110, Hsp90, Hsc70, Hsp70 and Hsp60 in CS F samples (40 µl) from control or spinal cord injured patients as indicated. 20 µg of rat brain supernatant and pellet fractions are shown in the right lanes.

Table 1 Clinical characteristics of the patients included in the study

Patient	PrP^C Level	αA/αB-crystallin	Hsp27	Patient age	Patient sex	Motor recovery after 6 mo	Treatment
C1				29	M
C2				82	F
C3				46	M
C4		−		39	F
C7		−	+	69	F
C8		−	+	46	M
SCI 05-12 h	↓	+	+	54	M	27	Minocycline
SCI 05-day 7	↓	−	−
SCI 11–12 h	↓	−	+	33	F		Placebo
SCI 11-day 7	↓	+	−
SCI 13-12 h	↓	−	+	18	F	Total	Minocycline
SCI 13-day 7	↓	−	−
SCI 14-12 h	↓	ND	ND	21	M		Placebo
SCI 14-day 7	↓	ND	ND
SCI 15-12 h	↓	−	+	25	M	10	Minocycline
SCI 15-day 7	↓	−	−
SCI 16-12 h	↓	+	+	20	M	Deceased at 3 weeks	Minocycline
SCI 16-day 7	↓	−	−
SCI 18-12 h	↓	−	ND	31	M	0	Minocycline
SCI 18-day 7	↓	+	ND
SCI 19-12 h	↓	−	+	47	M	21	Placebo
SCI 19-day 7	↓	−	−
SCI 20-12 h	↓	−	−	39	F	13	Placebo
SCI 20-day 7	↓	+	+
SCI 22-12 h	↓	−	−	22	M	0	Placebo
SCI 22-day 7	↓	−	−
SCI 25-12 h	↓	−	+	43	M	47	Minocycline
SCI 25-day 7	↓	−	−
SCI 28-12 h	↓	+	+	16	F	0	Placebo
SCI 28-day 7	↓	−	−
SCI 31-12 h	↓	−	+	48	M	4	Placebo
SCI 31-day 7	↓	+	−

Symbols are: (↓) decreased compared to control. (+) present, (−) absent, (ND) not determined, (M) male, (F) female, Motor recovery scores: high for good recovery and low for poor recovery.

Acknowledgements

We thank Tatania Noskova and Claudia Silva for technical contributions. This work was funded by the Alberta Prion Research Institute. J.E.A.B. is an Alberta Heritage Research Foundation for Medical Research Senior Scholar.