Research ArticlePrion Disease

Oral Treatment Targeting the Unfolded Protein Response Prevents Neurodegeneration and Clinical Disease in Prion-Infected Mice

See allHide authors and affiliations

Science Translational Medicine  09 Oct 2013:
Vol. 5, Issue 206, pp. 206ra138
DOI: 10.1126/scitranslmed.3006767


During prion disease, an increase in misfolded prion protein (PrP) generated by prion replication leads to sustained overactivation of the branch of the unfolded protein response (UPR) that controls the initiation of protein synthesis. This results in persistent repression of translation, resulting in the loss of critical proteins that leads to synaptic failure and neuronal death. We have previously reported that localized genetic manipulation of this pathway rescues shutdown of translation and prevents neurodegeneration in a mouse model of prion disease, suggesting that pharmacological inhibition of this pathway might be of therapeutic benefit. We show that oral treatment with a specific inhibitor of the kinase PERK (protein kinase RNA–like endoplasmic reticulum kinase), a key mediator of this UPR pathway, prevented UPR-mediated translational repression and abrogated development of clinical prion disease in mice, with neuroprotection observed throughout the mouse brain. This was the case for animals treated both at the preclinical stage and also later in disease when behavioral signs had emerged. Critically, the compound acts downstream and independently of the primary pathogenic process of prion replication and is effective despite continuing accumulation of misfolded PrP. These data suggest that PERK, and other members of this pathway, may be new therapeutic targets for developing drugs against prion disease or other neurodegenerative diseases where the UPR has been implicated.


Several neurodegenerative diseases including Alzheimer’s disease (AD), Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS), and prion diseases are associated with the accumulation and aggregation of misfolded disease-specific proteins in the brain. However, the mechanisms of neuronal death in these disorders, and hence the development of effective treatments, have been frustratingly elusive. We recently showed how misfolded prion protein (PrP) leads to neurodegeneration in prion disease through dysregulation of generic proteostatic mechanisms (1), a process likely to be relevant to neurodegenerative diseases more broadly.

All cells require correctly folded proteins to function efficiently. The buildup of misfolded, or unfolded, proteins constitutes a form of cellular stress detected by specific sensors in the endoplasmic reticulum (ER). Activation of these sensors gives rise to signaling cascades and the unfolded protein response (UPR), one branch of which leads to the transient shutdown of protein synthesis as a protective mechanism. This is mediated by phosphorylation (hence activation) of protein kinase RNA–like ER kinase (PERK). Phosphorylated PERK (PERK-P), in turn, phosphorylates the α subunit of eukaryotic initiation factor 2 (eIF2), which inhibits the initiation of translation. Protein synthesis is restored by activation of a feedback loop through which eIF2α-P is dephosphorylated by its specific phosphatase, GADD34 (2) (fig. S1). Increased PERK-P and eIF2α-P are found in the brains of patients with AD, PD, ALS, and prion disease (35), in mice overexpressing the human apoE4 allele associated with increased susceptibility to AD (6), and in tg4510 mice that overexpress mutant human tau protein with resultant neurodegeneration (7). However, the pathological importance of activation of this branch of the UPR in these disorders is unknown. Recently, we showed that in the brains of prion-diseased mice, the accumulation of misfolded PrP because of prion replication causes sustained overactivation of the PERK/eIF2α branch of the UPR. The resulting persistently high levels of eIF2α-P lead to neurodegeneration through sustained, uncompensated repression of protein synthesis, which is catastrophic in this context because of a critical decline in levels of key proteins, including synaptic proteins (1). Genetic manipulation (by stereotactic delivery of lentiviruses to the hippocampus) of the pathway upstream and downstream of eIF2α-P in prion-diseased mice reduced eIF2α-P and restored vital translation rates, allowing recovery of synaptic protein levels, resulting in marked localized neuroprotection and increased survival of mice with prion disease (1).

These data led us to predict that pharmacological inhibition of PERK would be neuroprotective in prion disease. We have used a recently described, orally bioavailable, potent, selective inhibitor of PERK, GSK2606414 (8), as an experimental tool for translational proof-of-principle studies in a mouse model of prion disease.


GSK2606414 penetrates the blood-brain barrier

The compound GSK2606414 (fig. S1) has an IC50 (half-maximal inhibitory concentration) of ~30 nM for inhibition of PERK-P in a cellular assay and good cellular and in vivo potency. Critically, this drug is exquisitely selective for PERK with a selectivity of >385-fold for PERK over the other eIF2α kinases and over 294 other kinases (8). We first established that the orally administered compound had good brain penetration in mice using liquid chromatography–tandem mass spectrometry (LC-MS/MS) measurements using a range of doses [as described in (8)]. Twice-daily oral dosing with 50 mg/kg gave high brain concentrations and a ratio of 0.56 for brain/plasma concentrations of the compound 14 hours after administration (n = 5); higher doses did not increase brain penetration (Table 1). We also calculated the unbound fraction of GSK2606414 in blood and brain (Fu,blood and Fu,brain) and the ratio of Fu,brain/cellular PERK IC50 as a measure of brain exposure. This confirmed that doses greater or equal to 50 mg/kg given twice daily resulted in unbound brain concentrations of the compound exceeding the cellular PERK IC50 by 43.2-fold or more. The lower dose of 10 mg/kg gave much lower brain exposure, with a Fu,brain/cellular PERK IC50 ratio of only 3.2; thus, this dose was not considered therapeutically useful (Table 1). We therefore selected treatment with 50 mg/kg of compound.

Table 1. GSK2606414 concentrations in the brain and plasma and ratio of unbound drug.

Concentrations of GSK2606414 in the brain and plasma in mice were measured by LC-MS/MS 14 hours after oral administration of the compound (10, 50, and 150 mg/kg) as two doses 8 hours apart in each case. The brain/plasma ratio of the drug, concentration of unbound fraction (Fu) Fu, and ratio of Fu,brain to cellular PERK inhibition IC50 were calculated. Doses of 50 mg/kg twice daily gave good brain penetration (brain/plasma ratios >0.5) and Fu,brain/IC50, PERK, cell greater than the known effective IC50 value (~30 nM) in our previous cellular assay (8). n = 5 for all doses. NQ, not quantifiable.

View this table:

GSK2606414 reverses cognitive deficits and prevents clinical disease in prion-infected mice

We tested the effects of administration of GSK2606414 to prion-infected mice. Clinical prion disease in mice has a stereotypical course and incubation period according to both prion strain and mouse strain used and is diagnosed by a specified combination of early indicator and later confirmatory signs (Table 2), the latter of which usually occur when neuronal loss has already reached advanced stages. We tested tg37 transgenic mice used in our earlier studies to enable comparison of GSK2606414 treatment with previous interventions (911), in particular with the genetic manipulation of PERK/eIF2α-P that we recently showed to be neuroprotective (1). tg37 mice overexpress PrP threefold compared to wild-type mice (12) and hence have a more rapid prion incubation period but, in all other ways, are phenotypically equivalent to wild-type mice. Infection of tg37 mice with the Rocky Mountain Laboratory (RML) prion strain typically results in diagnostic clinical signs (scrapie incubation period) within ~82 ± 5 days (about 12 weeks) (911).

Table 2. Clinical signs of prion disease are prevented in GSK2606414-treated mice.

Prion-infected mice treated with PERK inhibitor or vehicle were scored according to recognized early indicator and confirmatory signs of prion disease (“scrapie”). The presence of two early indicator signs plus one confirmatory sign or of two confirmatory signs alone was used to diagnose clinical disease. The time to confirmatory signs is the scrapie incubation time. Mice were treated with the compound (50 mg/kg twice daily). One group was treated from 7 weeks after inoculation in the preclinical phase of the disease (n = 20); a second group was treated from 9 weeks after inoculation when early behavioral deficits were already present (n = 9). All vehicle-treated animals (n = 17) had confirmatory signs of terminal prion disease by 82 ± 5 days (~12 weeks after inoculation). None of the animals treated with the compound at either early or later stages of prion infection developed diagnostic signs of scrapie in this period. N/A, not applicable.

View this table:

Mice were intracerebrally inoculated with RML prions at ~4 weeks of age, as described (12). Animals were separated into two treatment groups (Fig. 1A). The first group (n = 20) was treated from 7 weeks after inoculation, when prion infection in the brain was established and synaptic loss was beginning, but before memory and behavioral deficits occurred and before neuronal loss and clinical signs occurred (1, 9, 10). The second group of mice (n = 9) was treated from 9 weeks after inoculation, when spongiform prion pathology was established and behavioral signs, such as reduced object recognition memory and burrowing activity, were apparent in the animals (10, 13). All animals were treated by oral gavage with GSK2606414 (50 mg/kg) twice daily. Control mice in each group were infected with RML prions and treated with vehicle alone from 7 and 9 weeks after inoculation (n = 9 and 8, respectively; total n = 17). Another group of mice received compound alone without prion infection (n = 10). All animals were assessed throughout the course of treatment or vehicle administration.

Fig. 1. GSK2606414 prevents clinical disease in prion-infected mice.

(A) Time course of treatment. Prion-infected mice were treated with GSK2606414 or vehicle from 7 weeks post-inoculation (w.p.i.) (blue) during the preclinical phase or from 9 weeks after inoculation (purple) when behavioral deficits emerged. (B) Vehicle-treated animals at 9 weeks after inoculation (red bar) have impaired object recognition memory. This was normal in mice treated with GSK2606414 from 7 weeks after inoculation (blue bar). Treatment 9 weeks after inoculation did not reverse established memory deficits (purple bar) (n = 10 for each). *P = 0.01, two-tailed Student’s t test. (C) Decline in burrowing activity in prion-infected mice occurred by 9 weeks after inoculation in vehicle-treated mice (red bar), but was prevented in mice treated with GSK2606414 from 7 weeks after inoculation (blue bars), and reversed in mice treated from 9 weeks after inoculation (purple bars) (n = 12 for each). *P < 0.01, ***P < 0.0001, two-tailed Student’s t test. (D and E) Images of representative mice showing the clinical appearance of prion disease at 12 weeks after inoculation. Vehicle-treated mice showed a rigid tail, hindlimb paralysis (D, top image, black arrow), and marked hindlimb clasping (E, left image). GSK2606414-treated mice from both early and later time points were free from clinical prion disease and showed normal posture, grooming, tail position, and hindlimb mobility (D, middle and bottom images, black arrows; E, right-hand images) (see also movies S1 to S5). All data in bar charts show means ± SEM. Controls represent mice inoculated with normal brain homogenate (black bars).

By 12 weeks after inoculation, all GSK2606414-treated animals in both groups (from 7 and 9 weeks after inoculation) were free of diagnostic signs of prion disease, whereas all controls were terminally sick. Thus, all vehicle-treated prion-infected animals had confirmatory clinical signs of prion disease at this time, with a mean scrapie incubation period of 82 ± 5 days (n = 17) (Table 2), consistent with RML infection in tg37 mice as previously reported (1, 911). In contrast, confirmatory clinical signs of prion disease were absent in mice treated with PERK inhibitor at both early (n = 20) and later (n = 9) stages of the disease (Table 2, Fig. 1, D and E, and movies S1 to S5). However, there were occasional nonspecific early indicator signs: 5 of 20 for the early treatment group and 5 of 9 in the later-treated group, in which there were behavioral deficits at the onset of treatment. No PERK inhibitor–treated prion-infected mice had progressed to clinical prion disease (scrapie) by the time all the vehicle-treated control animals had succumbed to their disease. Treatment with GSK2606414 from 7 weeks after inoculation prevented loss of object recognition memory in prion-infected mice, although the later treatment at 9 weeks after inoculation, when memory was already lost, did not restore object recognition memory (Fig. 1B). Further, treatment with GSK2606414 prevented the decline in burrowing behavior characteristic of early prion disease (10) in mice treated at 7 weeks after inoculation (Fig. 1C). GSK2606414 treatment reversed burrowing deficits present at the onset of drug administration in mice treated at 9 weeks after inoculation (Fig. 1C). All terminally sick controls and asymptomatic GSK2606414-treated animals were sacrificed at 12 weeks after inoculation to assess the effects of treatment on morphological and biochemical outcomes apart from a small cohort that were used for longevity analysis. All mice were analyzed after sacrifice for the effects of the PERK inhibitor on neuronal numbers and prion pathology, for the biochemical effects on the levels of PERK-P, eIF2α-P, and on other branches of the UPR pathway, and for the effects on global rates of protein synthesis in the brain, as well as on PrP levels.

However, despite the lack of confirmatory clinical signs of prion disease, we were unable to perform survival studies on the remaining cohort of mice treated with GSK2606414 because these animals had a cumulative weight loss equivalent to 20% of body mass soon after 12 weeks after inoculation (fig. S2A), which, according to UK Home Office regulations, means that they had to be culled. We also found mildly elevated blood glucose at this stage, with blood concentrations of 10 to 15 mM glucose in GSK2606414-treated animals, compared to 8 to 10 mM in vehicle-treated control animals, but well below the diabetic range in mice (>22 mM) (14) (fig. S3). Both effects are likely due to systemic effects of PERK inhibition, particularly on pancreatic function (see Discussion). Despite the weight loss, all animals were otherwise overtly well and active, with no other systemic signs and no other indication for culling. An uninfected group of older mice aged 6 months, used as a small pilot control group (n = 4) to test the effects of the compound in older animals, did not show weight loss after 7 weeks of treatment (fig. S2B).

GSK2606414 is neuroprotective in mice with prion disease

By examining brains histopathologically, we observed neuroprotection at 12 weeks after inoculation throughout the brains of PERK inhibitor–treated mice, consistent with a lack of clinical symptoms (Fig. 2, A and B, right-hand panels) (n = 5). This was in contrast to vehicle-treated mice, in which neuronal loss and spongiform degeneration were extensive (Fig. 2, A, ii and vi, and B, ii, vii, x, and xiv). In both groups of treated animals, the neuronal ribbon of hippocampal regions CA1–4 was protected and did not degenerate, as was observed in vehicle-treated animals by 12 weeks after inoculation (Fig. 2A, compare iii, vii and iv, viii with ii and vi). CA1 pyramidal neuron counts from prion-infected GSK2606414-treated mice at early and later time points were equivalent to those in non–prion-infected mouse brains, whereas in vehicle-treated animals, CA1 pyramidal neuron counts had declined to less than 30% of control values (Fig. 2C) (n = 4; P < 0.0001). The neuroprotective effects of GSK2606414 were also reflected in the degree of astrocytosis in prion-diseased brains. Astrocytes proliferate in response to diseased neurons. Both astrocyte number and activation were reduced in animals treated with GSK2606414 compared to vehicle-treated mice (Fig. 2A, ix to xii). Treatment with GSK2606414 also resulted in minimal spongiform degeneration (characteristic of prion neuropathology) throughout the mouse brain, although this was less marked in animals treated later, at 9 weeks after inoculation, presumably because a degree of spongiform pathology was already present at the time of treatment in this group (Fig. 2, A, iii, vii and iv, viii, B, right-hand panels, and D). The data reflect neuroprotective activity of the compound throughout the whole brain, as opposed to localized effects because of anatomically targeted genetic manipulations in our previous studies (1). GSK2606414 treatment had no effect on neuronal numbers and hippocampal morphology, on UPR proteins, or on burrowing behavior in uninfected control mice (fig. S4).

Fig. 2. GSK2606414 prevents spongiosis, gliosis, and neurodegeneration in prion-infected mice.

(A). Representative images of hematoxylin and eosin–stained hippocampal sections from uninfected control mice (i and v), prion-infected animals treated with vehicle (ii and vi), and prion-infected animals treated with GSK2606414 from 7 weeks post-inoculation (w.p.i.) (iii and vii) or from 9 weeks after inoculation (iv and viii). Vehicle-treated animals showed marked neuronal loss in the CA1–4 region of the hippocampus and extensive spongiosis (ii and vi, black arrowheads and arrows). GSK2606414 treatment (vii and viii) prevented neuronal loss and reduced spongiosis (the typical vacuoles of prion disease indicated by arrows in vi). Glial fibrillary acidic protein (GFAP) immunostaining for astrocyte activation showed that this was greatly reduced in treated mice, confirming the neuroprotective effects of the compound (ix and x). (B) Neuroprotection after treatment with GSK2606414 was observed in other mouse brain regions. Vehicle-treated mice showed spongiform changes in cortex, thalamus, brainstem, and cerebellum (ii, vi, x, and xiv). These changes were absent in mice treated with GSK2606414 from 7 weeks after inoculation (iii, vii, xi, and xv) and were reduced in mice treated with the drug from 9 weeks after inoculation (iv, viii, xii, and xvi) (n = 3 to 5 for each condition). (C) Number of CA1 pyramidal neurons remaining after GSK2606414 treatment. Extensive neuronal loss occurred in vehicle-treated prion-infected mice but not drug-treated animals. Neurons were counted in five sections for each of four mice in each group. ***P < 0.0001, two-tailed Student’s t test. (D) Assessment of spongiosis in sections of mouse brain: −, absent; +, mild; ++, moderate; +++, severe. Scoring was done in brain sections from five mice (8 to 10 sections per mouse). Scale bars, 200 μm except for (A), panels v to x, where scale bars are 50 μm.

GSK2606414 inhibits PERK phosphorylation in brains of prion-infected mice

We found that PERK-P (Fig. 3A) and eIF2α-P (Fig. 3B) levels were reduced in both groups of prion-infected animals receiving the compound (early and late treatment) compared to those receiving vehicle alone, consistent with inhibition of PERK phosphorylation by GSK2606414 (n = 4). As a result, global translation rates were restored in mice treated with the compound compared to vehicle-treated mice (Fig. 3C) in which translation was repressed because of UPR activation resulting from prion infection, as we have previously observed (1). Thus, inhibition of PERK phosphorylation by GSK2606414 prevented UPR-mediated translational inhibition in prion-diseased mice.

Fig. 3. GSK2606414 inhibits PERK phosphorylation and blocks eIF2α-P–mediated decline in protein synthesis in prion-infected mice.

(A and B) Treatment of prion-infected mice with GSK2606414 from 7 and 9 weeks post-inoculation (w.p.i.) resulted in reduced PERK-P (A) and eIF2α-P (B) at 12 weeks after inoculation, consistent with inhibition of PERK. In vehicle-treated mice at the same time point, PERK-P and eIF2α-P levels did not decrease. Representative immunoblots of hippocampal lysates and bar charts quantitating relative levels of proteins (in three independent samples) are shown. (C) Protein synthesis rates in hippocampal slices, determined by [35S]methionine incorporation into protein, showed ~60% reduction in prion-infected vehicle-treated mice. In contrast, GSK2606414 treatment resulted in maintenance of normal global translation rates (n = 3 for each group). (D) GSK2606414 treatment reduced ATF4 and CHOP proteins in prion-infected mice, but vehicle did not, consistent with inhibition of PERK by the compound. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; NBH, normal brain homogenate. (E) Schematic of the three branches of the UPR showing PERK-P/eIF2α-P, ATF6, and IRE1-XBP1 branches and downstream effector pathways. (F) Representative Western blot showing that administration of GSK2606414 does not alter cleavage of full-length ATF6 (flATF6) to its nuclear fragment (nATF6), which occurs equally in all samples except human embryonic kidney (HEK) 293 negative controls. (G) Reverse transcription–polymerase chain reaction (PCR) of XBP1 transcripts. There was no splicing of XBP1 in prion-infected mice with or without GSK2606414 treatment (only full-length uXBP1 transcripts were detected). Control lanes contain RNA from N2A cells with (+) or without (−) tunicamycin (Tm) treatment and show spliced XBP1 (sXBP1) transcripts with tunicamycin treatment. All data in bar charts show means ± SEM. *P < 0.05, two-tailed Student’s t test.

We tested other downstream effects of inhibition of PERK. Normally, phosphorylation of eIF2α by PERK results not only in inhibition of global translation (Fig. 3C) but also in the preferential translation of specific mRNAs, including ATF4, with subsequent up-regulation of specific proteins, including ATF4 and C/EBP homologous protein (CHOP) (Fig. 3E). We found that increased ATF4 and CHOP induced by prion-mediated UPR activation were reduced by GSK2606414, and that ATF4 and CHOP levels were reduced on treatment with the compound (Fig. 3D), consistent with inhibition of PERK activity. We then examined the other two branches of the UPR, ATF6 activation, and IRE1-mediated XBP1 splicing (Fig. 3E), activation of which has been reported in brains of patients and in mice with prion disease (1518). Also, XBP1 splicing has been reported to be protective against PrPC aggregation in vitro (19), although XBP1 deficiency does not alter the course of prion infection in mice clinically or pathologically (20). We tested the effect of GSK2606414 on these branches of the UPR to confirm its specificity for the PERK pathway. Administration of GSK2606414 did not inhibit ATF6 cleavage (Fig. 3F) or affect splicing of XBP1 (Fig. 3G), consistent with its described specificity for PERK (Fig. 3, A to D) (8).

GSK2606414 treatment restores synthesis of vital synaptic proteins

GSK2606414 restored global protein synthesis rates because of PERK inhibition in prion-infected animals, in contrast to vehicle-treated mice in which translation was repressed (Fig. 3C). As a result, vital pre- and postsynaptic protein levels were maintained in mice treated with the compound at both early and late time points, whereas synaptic protein levels fell markedly because of UPR overactivation in prion-infected mice treated with vehicle (Fig. 4, A and B, red bars, and fig. S5), as we have previously observed (1). GSK2606414 treatment did not affect the levels of total PrP and protease-resistant PrPSc, which were equivalent in all prion-infected groups, both vehicle and early or late GSK2606414-treated prion-infected animals (Fig. 4, C and D). This is as expected because PERK activation/inhibition occurs downstream of prion replication (fig. S1) and because PrP mRNA is among those mRNAs that escape eIF2α-P translational repression due to the structure of its 5′ untranslated region (1, 21, 22). It is well established that subclinical infection (prion replication without neurotoxicity) can exist with very high levels of PrPSc but without clinical signs (including behavioral changes) or neuronal loss (9, 2325). The continued accumulation of PrPSc in the presence of the PERK inhibitor confirms this phenomenon: that neuroprotection and lack of clinical signs can coexist with ongoing prion replication when the toxic effector pathway downstream of prion replication is blocked.

Fig. 4. GSK2606414 inhibition of translational shutdown restores synthesis of key synaptic proteins.

(A and B) GSK2606414 treatment at both stages of disease resulted in near-normal levels of (A) the presynaptic protein SNAP25 and (B) the postsynaptic protein PSD95 compared to reduced levels of these proteins in vehicle-treated animals. Bar graphs depicting relative levels are shown (n = 3 for each treatment). Representative Western blots are shown in fig. S5. (C and D) Total PrP (C) and PrPSc (D) levels detected after proteinase K (PK) digestion at 12 weeks post-inoculation (w.p.i.) were equivalent in prion-infected mice treated with vehicle and GSK2606414 (both early and late time points) because PERK inhibition is downstream of prion replication (see schematic, fig. S1). Representative immunoblot and bar charts quantitating relative levels of protein in three independent samples are shown. All analyses were carried out on mice at 12 weeks after inoculation. All data in bar charts show means ± SEM. *P < 0.05, two-tailed Student’s t test. n.s., not significant. All immunoblots were performed on hippocampal lysates. Control samples are from mice inoculated with normal brain homogenate.


We have found that inhibition of PERK using the compound GSK2606414 prevents activation of the UPR branch that mediates prion neurotoxicity. GSK2606414 prevented PERK-P/eIF2α-P–mediated translational failure that leads to neurodegeneration and clinical signs in prion-diseased mice. The compound was highly effective in mice, with excellent blood-brain barrier penetration and effective concentrations of unbound drug in the brain after oral dosing (Table 1). The findings support the concept that PERK inhibition is a target for drug discovery in the treatment of prion disorders, downstream and independently of the central pathogenic process of prion replication.

The work builds on our previous findings, whereby we showed how the accumulation of misfolded PrP in prion disease activates this branch of the UPR in a sustained, dysregulated way, leading to failure of critical protein synthesis and, hence, neurodegeneration. Focal genetic manipulation of the pathway upstream of eIF2α (by lentivirally mediated knockdown of PrP in the mouse hippocampus, abrogating the rise in PrP) prevented activation of PERK and eIF2α phosphorylation (see fig. S1). Manipulating the pathway downstream of eIF2α-P by overexpressing its specific phosphatase GADD34 increased eIF2α-P deactivation. Both approaches resulted in focal prevention of prion-induced UPR-mediated translational failure and rescue from neurodegeneration, and increased survival. Indeed, we also showed that the opposite strategy of blocking the dephosphorylation of eIF2α-P using the drug salubrinal (26) exacerbated neuronal death and accelerated time to death (1).

Those findings led us to predict that pharmacological targeting of this pathway would be neuroprotective throughout the brain, and our results using GSK2606414 confirm this. Prion diseases (and indeed the later stages of AD, various tauopathies, and PD dementia) show widespread brain pathology. Thus, global neuroprotection is desirable. The pharmacological inhibition of PERK throughout the brain demonstrated here is an advantage over focal modulation of the pathway. This approach was effective at the symptomatic stage of disease as well as the preclinical stage. Thus, mice treated at 9 weeks after inoculation that had early prion behavioral changes and early spongiform pathology showed recovery from behavioral deficits, no progression in spongiform degeneration, and protection from neuronal loss (Fig. 2). The evidence of efficacy of this strategy, however, was the absence of clinical prion disease in all treated mice at a time when vehicle-treated controls were all terminally ill with the disease (Table 2, Fig. 1, and movies S1 to S5).

However, despite neuroprotection in the brain and absence of clinical prion disease, GSK2606414-treated mice suffered some weight loss and mild hyperglycemia, likely due to systemic effects of the compound. PERK is ubiquitously expressed, and homozygous PERK−/− knockout mice show early postnatal lethality, with hyperglycemia arising from inadequate insulin levels because of pancreatic islet cell death, and exocrine pancreatic insufficiency (27). However, hemizygous PERK+/− mice have a much milder phenotype with impaired glucose tolerance, but weight and longevity are normal (27). The effects on glucose metabolism observed here due to pharmacological inhibition of PERK by GSK2606414 (blood concentrations of 10 to 15 mM glucose compared to 8 to 10 mM in controls; fig. S3) are relatively mild, similar to those in PERK+/− mice, and do not approach murine diabetic concentrations of >22 mM glucose (14). However, because of the associated weight loss, we were unable to perform survival studies due to UK Home Office regulations on animal welfare regarding body mass. The animals were not obviously thin and were otherwise well, and older mice did not show any weight loss, possibly due also to higher body mass at initiation of treatment. Given the increase in survival seen even with focal inhibition of UPR-mediated translational shutdown (1), it would be predicted that, in the absence of toxic side effects, generalized pharmacological PERK inhibition in the brain would have a beneficial effect on longevity in prion-infected mice. The degree to which any side effects are potentially treatable is also relevant in the clinical context, however, where management of insulin deficiency and glucose intolerance is routine, especially in elderly patients. Indeed, any disadvantage from these effects would need to be weighed against the potential benefit of neuroprotection, particularly in diseases such as sporadic prion disease, which is rapidly fatal in humans.

Although the immediate translational value of GSK2606414 may be limited by its systemic effects, the data provide critical proof of principle that this approach could work. Further, the concept that manipulation of a generic cellular pathway involved in protein homeostasis is neuroprotective has broad relevance (28). It has important translational implications for prion disease and other neurodegener„„„„ative disorders associated with protein misfolding, where UPR activation and raised eIF2α-P levels are found, for example, in the brains of patients with AD and PD (3, 4) and ALS (5), and in a mouse model of tauopathy (7). The key feature is prevention of a fatal reduction in protein synthesis downstream of the process of accumulation of disease-specific misfolded proteins such as PrP, amyloid β, tau, α-synuclein, or others. Drug discovery programs targeting the PERK/eIF2α branch of the UPR, and related pathways, such as increasing chaperone expression and reducing misfolded protein stress in neurons, may provide new treatments for a variety of neurodegenerative diseases.

Our study has focused on mice with rapidly evolving prion neurodegeneration. Before translation of this strategy into human patients, further development of this approach is essential, particularly because this would involve treatment for years or even decades in many cases. Drugs acting predominantly in neurons and devoid of systemic side effects are needed, and fine-tuning of both the inhibition of translational repression and its timing for maximal therapeutic benefit should be explored.


Study design

The aim of the study was to test if pharmacological inhibition of PERK is neuroprotective in prion disease. The rationale is based on our previous findings that prion replication overactivates the PERK/eIF2α branch of the UPR. The resultant sustained translational repression mediates prion neurotoxicity by causing critical decline in levels of key proteins, leading to neuronal death (1). Previously, we showed that genetic manipulation of the pathway to reduce levels of eIF2α-P restored translation rates and was neuroprotective in prion disease (1). We therefore tested a recently available pharmacological inhibitor of PERK, GSK2606414, in prion-infected mice using the same readouts as in our original study. We tested the effects of treatment on clinical, behavioral, and neuropathological readouts of prion disease, and we assessed the biochemical effects of GSK2606414 on inhibition of PERK/eIF2α, and other branches of the UPR, and its effects on global protein synthesis and levels of synaptic proteins.

All animal experiments were designed with a commitment to refinement, reduction, and replacement, minimizing the numbers of mice and suffering via emphasis on humane end points, while using biostatistical advice for optimization of mouse numbers [as used in our previously published peer-reviewed work using prion infection model (1, 912)]. Thus, for statistical validity, we used n ≥ 10 of group-housed mice for behavioral tests and n = 3 to 5 for biochemical analysis (with three replicates). For histology, previous experience has shown that group sizes of three to six animals are sufficient to detect even subtle differences (1, 912). Mice were treated in cohort sizes large enough to allow sufficient numbers for sampling at regular intervals, as specified in Materials and Methods and in the text.

Prion infection of mice

All animal work conformed to UK regulations and institutional guidelines, and was performed under Home Office guidelines. tg37 (12) mice were inoculated with 1% brain homogenate of Chandler/RML (Rocky Mountain Laboratories) prions or with normal brain homogenate, as described (1).

GSK2606414 treatment

Mice were orally gavaged twice daily with GSK2606414 at 50 mg/kg suspended in vehicle (0.5% hydroxypropylmethyl cellulose + 0.1% Tween-80 in H2O at pH 4.0) or with vehicle alone (n = 17) from 7 weeks (n = 20) or 9 weeks (n = 9) after inoculation.

Detection of GSK2606414 by LC-MS/MS

Blood and brain tissue were collected 14 hours after dosing from mice treated with two doses of 10, 50, or 150 mg/kg or vehicle. Blood plasma (0.025 to 0.95 ml, exact volume measured) was diluted with water to 0.1 ml and extracted with 0.4 ml of isopropanol. After vortex mixing (10 min) and centrifugation (10,000g, 10 min), the supernatant was dried with vacuum centrifugation and reconstituted in 50 μl of methanol/isopropanol (3:1). Brain tissue (one complete half, about 0.2 g) was homogenized in 0.8 ml of isopropanol and further processed exactly as the plasma samples. GSK2606414 quantitative analysis (using external standards) was by LC-MS/MS using a 4000 QTRAP (Applied Biosystems) equipped with a turbo ion source and LC series 10 AD VP (Shimadzu). The mobile phase was a water/acetonitrile gradient modified with 0.1% formic acid using a Phenomenex Gemini column (100 × 3 mm, 3-μm particle size), which was maintained at 40°C. LC-MS/MS multiple reaction monitoring used a precursor ion of mass/charge ratio (m/z) 452 and a product ion of m/z 265 in positive electrospray ionization mode (ES+). Data analysis was performed with Analyst 1.4.1 in the quantitative mode.

Western blotting

Immunoblots of synaptic proteins, UPR proteins, and PrP were performed as described (1). Briefly, samples were homogenized using a hypotonic homogenization buffer with the addition of protease inhibitors and PhosSTOP (Roche) followed by incubation with 2× lysis buffer [100 mM tris (pH 8.0), 300 mM NaCl, 4 mM EDTA, 2 mM MgCl2, 200 mM NaF, 20% glycerol, 2% Triton X-100, 2% sodium deoxycholate, 0.2% SDS, 0.25 M sucrose]. Samples were then separated by SDS–polyacrylamide gel electrophoresis and transferred onto nitrocellulose or polyvinylidene difluoride membranes by wet blotting. Specific pre- and postsynaptic markers SNAP25 (synaptosome-associated protein 25; Abcam) at 1:10,000 dilution, VAMP2 (synaptobrevin 2; Synaptic Systems) at 1:1000 dilution, PSD95 (postsynaptic density protein 95; Millipore) at 1:1000 dilution, and NR1 (N-methyl-D-aspartate receptor 1; Sigma) at 1:1000 dilution were visualized with 1:5000 dilution of horseradish peroxidase (HRP) secondary antibodies goat anti-rabbit (1:5000) (Dako) or mouse anti-guinea pig (Dako). Antibodies for ATF6 (Abcam), total and phosphorylated forms of PERK (Cell Signaling), eIF2α (Cell Signaling), ATF4 (Abcam), and CHOP (Gentex) were used at a 1:1000 dilution, with HRP secondary antibodies goat anti-rabbit (1:5000) (Dako) and donkey anti-rabbit (1:5000) (Promega), respectively. PrPC protein was detected with the 8H4 antibody at 1:1000 dilution (Abcam). To detect PrPSc, homogenized samples were digested with proteinase K (50 μg/ml) at 37°C for 1 hour before electrophoresis (9). Membranes were then probed with 1:10,000 ICSM-35 (D-GEN) and goat anti-mouse at 1:10,000 (Dako). Secondary antibodies were detected with enhanced chemiluminescence system (ECL; Amersham). Sample loading was confirmed by detecting GAPDH (1:5000; Santa Cruz Biotechnology) or β-tubulin (1:5000; Millipore). Quantitative analyses were performed with ImageJ.

[35S]Methionine incorporation into protein for assessment of protein synthesis rates

Global translation levels were detected with [35S]methionine incorporation into protein, in acute hippocampal slices, as described (1). Hippocampal slices were prepared with a tissue chopper (McIlwain) and dissected in an oxygenated cold (2° to 5°C) sucrose artificial cerebrospinal fluid (ACSF) containing 26 mM NaHCO3, 2.5 mM KCl, 4 mM MgCl2, 0.1 mM CaCl2, and 250 mM sucrose. Slices were allowed to recover in normal ACSF buffer while being oxygenated at 37°C for 1 hour in 95% O2/5% CO2, and then incubated with 5.7 mBq of [35S]methionine label for 1 hour. Samples were washed and homogenized in 1× passive lysis buffer (Promega), and proteins were precipitated with 25% trichloroacetic acid (TCA) (Sigma). TCA lysates were then placed on Whatman filters, washed with 70% industrial methylated spirits and acetone, and then placed into scintillation cocktail buffer. Incorporation of radiolabel was measured by scintillation counting (WinSpectral, Wallac Inc.).

XBP1 splicing assay

Total RNA was extracted from hippocampi with the mirVana RNA/miRNA isolation kit (Ambion Inc.). RNA samples were reverse-transcribed with ImProm-II Reverse Transcriptase (Promega) by priming with oligo(dT). XBP1 mRNA was amplified with primers flanking the 26–base pair intron (5′-GGAGTGGAGTAAGGCTGGTG and 5′-CCAGAATGCCCAAAAGGATA) with Phusion High-Fidelity Taq Polymerase (New England Biolabs). PCR products were resolved on 3% agarose gels. Mouse neuroblastoma cells (N2A) were treated with tunicamycin (5 μg/ml) for 8 hours and used as a positive control for XBP1 splicing (29).


Paraffin-embedded brains were sectioned at 4 μm and stained with NeuN antibody (1:200; Millipore) for neuronal counts. CA1 pyramidal neuron counts were determined with three serial sections from three separate mice. Astrocytosis was detected with anti-GFAP polyclonal antibody (1:500; Dako). Nonspecific binding was blocked before primary antibodies with Histostain-Plus Bulk kit (Invitrogen). A biotinylated secondary antibody (Invitrogen) was used, and stain was visualized with diaminobenzidine reagent. All images were taken with AxioVision 4.8 software (Zeiss) and counted with Volocity imaging system (1).

Novel object recognition memory test

Novel object recognition memory test was performed as described (10). Briefly, mice were tested in a black cylindrical arena (69-cm diameter) mounted with a 100 light-emitting diode cluster infrared light source and a high-resolution day/night video camera (Sony). Mice were acclimatized to arena 5 days before testing. During the learning phase, two identical objects were placed 15 cm from the sides of the arena. Each mouse was placed in the arena for two blocks of 10 min for exploration of the objects with an intertrial interval of 10 min. Two hours later, one of the objects was exchanged for a new one, and the mouse was replaced in the arena for 5 min (test phase). The amount of time spent exploring all objects was tracked and measured for each animal with EthoVision software (Tracksys Ltd.) All objects and the arena were cleansed thoroughly between trials to ensure the absence of olfactory cues.


Briefly, mice were placed in a large cage with a perspex tube full of food pellets, as described (10). The natural tendency of rodents is to displace (burrow) the food pellets. The percentage of burrowing activity is calculated from the difference in the weight of pellets in the tube before and after 2 hours.

Statistical analysis

All analyses were performed with hippocampi from three to five mice in triplicate for biochemical analyses unless otherwise stated. Behavioral analyses were performed on 10 or more mice, as stated. Student’s t tests were applied to all data sets (normal distribution confirmed) with two tails (two samples; unequal variance). Statistical tests were performed with Prism v5. All data in bar charts show means ± SEM.


Fig. S1. Schematic representation of UPR signaling leading to translational repression in prion disease and point of action of GSK2606414.

Fig. S2. Body weights of prion-infected mice treated with GSK2606414 and vehicle.

Fig. S3. Blood glucose levels in prion-infected mice treated with GSK2606414 and vehicle.

Fig. S4. Biochemical, morphological, and behavioral characterization of uninfected mice treated with GSK2606414.

Fig. S5. Additional synaptic protein levels in prion-infected mice treated with GSK2606414 and vehicle.

Movies S1 to S5. GSK2606414 prevents clinical signs of prion disease in mice.


  1. Acknowledgments: We thank L. Onions, A. Smart, K. White, M. G. Martin, J. Edwards, and Division of Biomedical Services staff for technical assistance. We thank GSK Oncology for supply of the compound. Funding: This work was funded by the Medical Research Council, UK. Author contributions: J.A.M. performed oral gavage and clinical assessment of mice, and histological and biochemical analyses, aided by M.H. (all procedures). C.M. performed behavioral testing. H.R. helped with biochemical analyses and N.V. performed prion inoculations. J.M.A. provided compound and expertise on its use. C.A.O., D.A.B., and P.M.F. did pharmacokinetic/dynamic analyses. A.E.W. contributed to planning of experiments. G.R.M. conceived and directed the project. J.A.M. and G.R.M. wrote the paper. All authors contributed to discussion, analysis of data, and the final draft of the paper. Competing interests: J.M.A. is an employee of GlaxoSmithKline with equity holdings and stock options. GlaxoSmithKline holds patents for PERK inhibitors including GSK2606414: U.S. Pat. Appl. Publ. (2012), US 20120077828 A1 PCT Int. Appl. (2011), WO 2011119663 A1. The other authors declare no competing interests. Data and materials availability: Data and materials will be available upon request to G.R.M. (
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article