PCSK9 reduces the protein levels of the LDL receptor in mouse brain during development and after ischemic stroke.

Proprotein convertase subtilisin/kexin type 9 (PCSK9) plays a major role in cholesterol homeostasis through enhanced degradation of the LDL receptor (LDLR) in liver. As novel inhibitors/silencers of PCSK9 are now being tested in clinical trials to treat hypercholesterolemia, it is crucial to define the physiological consequences of the lack of PCSK9 in various organs. LDLR regulation by PCSK9 has not been extensively described during mouse brain development and injury. Herein, we show that PCSK9 and LDLR are co-expressed in mouse brain during development and at adulthood. Although the protein levels of LDLR and apolipoprotein E (apoE) in the adult brain of Pcsk9(-/-) mice are similar to those of wild-type (WT) mice, LDLR levels increased and were accompanied by a reduction of apoE levels during development. This suggests that the upregulation of LDLR protein levels in Pcsk9(-/-) mice enhances apoE degradation. Upon ischemic stroke, PCSK9 was expressed in the dentate gyrus between 24 h and 72 h following brain reperfusion. Although mouse behavior and lesion volume were similar, LDLR protein levels dropped ∼2-fold less in the Pcsk9(-/-)-lesioned hippocampus, without affecting apoE levels and neurogenesis. Thus, PCSK9 downregulates LDLR levels during brain development and following transient ischemic stroke in adult mice.


Mouse ischemic surgery
Transient middle cerebral artery occlusion (tMCAO) was performed on two-to three-month-old WT and Pcsk9 Ϫ / Ϫ mice (weight 20-25 g). Under temporary 2% isofl urane anesthesia, unilateral transient focal cerebral ischemia was induced by tMCAO for 1 h, followed by different reperfusion periods (6,24, 72 h and 1 week) ( 19,20 ). Body temperature was maintained at 37°C using a heating pad and an infrared heating lamp. Under an operating microscope, a 12 mm-long 6-0 silicon-coated monofi lament suture (Doccol Corporation) was inserted through the left proximal external carotid artery into the internal carotid artery, and then into the circle of Willis, thus occluding the middle cerebral artery. After 1 h of tMCAO, the fi lament was withdrawn, blood fl ow restored to normal, and wounds sutured. The same procedure was performed on sham-operated mice, where the monofi lament was not inserted. All animals were allowed ad libitum access to water and food before and after surgery. They were fasted for 3 h before euthanization to standardize LDLR levels in all mice. For in situ hybridization, brains were removed after 6, 24, 72 h and 1 week of brain reperfusion. To quantify neurogenesis in the dentate gyrus, bromodeoxyuridine (50 mg/kg) (BrdU; Sigma-Aldrich) was injected intraperitoneally 24 h postsurgery, twice a day (with an 8 h interval) for two days. To quantify the lesion size and BrdU-positive cells, mouse brains were removed after 72 h reperfusion, perfused with 50 ml of PBS, frozen in isopentane at Ϫ 30°C, and cut into coronal cryosections (17 m thick).

Neurological scores
An expanded six-point scale was used as described previously ( 21 ). Behavioral assessments were made at 1, 24, 48, and 72 h after reperfusion. Neurological defi cits were scored as follows: 0 = normal; 1 = mild turning behavior with or without inconsistent curling when picked up by the tail, <50% attempts to curl to the contralateral side; 2 = mild consistent curling, >50% attempts to curl to the contralateral side; 3 = strong and immediate consistent curling, mouse holds a curled position for more than 1-2 s, with its nose almost reaching tail; 4 = severe curling progressing into barreling, loss of walking or righting refl ex; 5 = comatose or moribund state. At least 8 mice per group were evaluated for each experiment, and scores were averaged for statistical analysis.

Nissl staining
Brain infarct was assessed by Nissl staining. The relative size of the infarct was measured using image software from the Scion Corporation (Frederick, MD), calculated in arbitrary units (pixels), and expressed as a percentage of the contralateral nonlesioned area (100%) for each section. expressing enhanced green fl uorescent protein (EGFP) under the control of the Pcsk9 promoter revealed the presence of EGFP in nerve fi bers within the olfactory bulb, which is innervated by the RE-OP (National Institutes of Health GENSAT Project) ( 14 ).
Although the role of PCSK9 in the brain during mouse development ( 1 ) has not been extensively investigated, its overexpression in primary neuronal cultures obtained at E12.5 has been shown to enhance the recruitment of undifferentiated progenitor cells into the neuronal lineage ( 1 ). Contrary to PCSK9 knockdown in zebrafi sh, which results in early death and an extensive disorganization of the central nervous system (CNS) ( 15 ), PCSK9 knockout mice are viable ( 5,6 ). Furthermore, we did not observe any gross alterations in adult Pcsk9 Ϫ / Ϫ mice in the cerebellum, hippocampus, or cortex ( 16 ). In pathological situations, such as induction of neural apoptosis by serum withdrawal, PCSK9 is upregulated ( 17 ), and overexpression of PCSK9 in cultured cerebellar granular neurons induces cell death ( 18 ), suggesting that PCSK9 may be involved in neural apoptotic processes.
In the present study, we show that PCSK9 and LDLR mRNAs are co-expressed in the same cell layer within the telencephalon at E12.5, the cerebellum at P7, and at adulthood in the RE-OP. As in liver, PCSK9 also enhances LDLR protein degradation during brain development. In contrast, at adulthood within the RE-OP and olfactory bulb, LDLR protein levels are not affected by PCSK9. To investigate the role of PCSK9 following brain injury, we induced a transient ischemic stroke in adult mice ( 19,20 ) and analyzed the expression of PCSK9 in the dentate gyrus from 6 h to 1 week following injury. The data showed that the upregulated PCSK9 reduced LDLR protein levels in the lesioned dentate gyrus without signifi cantly affecting de novo neurogenesis. We also showed that protein levels of apoE were decreased in Pcsk9 Ϫ / Ϫ mice during development but not at adulthood or following transient ischemic stroke.

Animals
Wild-type (WT) C57BL/6J mice, Ldlr Ϫ / Ϫ C57BL/6J mice (#002207) and apoE Ϫ / Ϫ C57BL/6J mice (#002052) were obtained from The Jackson Laboratory and bred in house. Pcsk9 Ϫ / Ϫ and transgenic mice overexpressing V5-tagged PCSK9 in the liver were described previously ( 6 ) and were backcrossed for 10 generations to the C57BL/6J genetic background. The mice were housed in the Clinical Research Institute of Montreal (IRCM) animal facility on a 12 h light/dark cycle. All mouse experimentations were approved by the IRCM bioethics committee for animal care.

Tissue collection
E12.5, P7, and adult (three-month-old) mice were euthanized with 2% isofl urane. For Western blot analyses, mouse brains at E12.5, cerebella at P7, RE-OP, and adult olfactory bulbs were dissected and frozen in isopentane at Ϫ 30°C. For Nissl staining and LDLR immunofl uorescence, E12.5 embryonic, P7, and adult brains were frozen at Ϫ 30°C in isopentane and cut into sagital cryosections (8,12, and 17 m thick, respectively). For other ( 1 ). To defi ne the role of PCSK9 in brain, we fi rst focused on its expression during embryonic development. In situ hybridization revealed that, at E12.5, PCSK9 was expressed only in the liver, small intestine, and telencephalon ( Fig.  1A ), suggesting an early developmental role of PCSK9 in these tissues. In telencephalon, PCSK9 was expressed in the frontal cortex (FCx) ( Fig. 1A ). The expression of the LDLR was also assessed in adjacent sections and was found to colocalize with all PCSK9 expression sites (arrows in Fig.  1A ). Furthermore, immunofl uorescence analysis, under nonpermeabilizing conditions that refl ect cell surface expression, showed that LDLR protein levels were higher in Pcsk9 Ϫ / Ϫ telencephalon than in WT ones ( Fig. 1B ). LDLR quantitation by Western blot in 10 WT and 13 Pcsk9 Ϫ / Ϫ extracts confi rmed this observation, and revealed a ‫ف‬ 2.7fold increase in Pcsk9 Ϫ / Ϫ compared with WT telencephalons ( Fig. 1C ). Nissl staining revealed that the telencephalon organization was not grossly altered in Pcsk9 Ϫ / Ϫ mice ( Fig.  1D ), suggesting that at E12.5 the regulation of LDLR by PCSK9 was not critical for tissue integrity.

PCSK9 enhances the degradation of the LDLR in mouse cerebellum at P7
We previously showed that PCSK9 was expressed in the cerebellum at P7 ( 1 ). As assessed by in situ hybridization, PCSK9 mRNA was expressed in the external granular layer (EGL), where it colocalized with the LDLR mRNA ( Fig.  2A ). As in E12.5 telencephalon, cell surface LDLR labeling was higher in Pcsk9 Ϫ / Ϫ external granular cells versus WT ones ( Fig. 2B ). Western blot analysis of fi ve cerebella for each genotype showed that the ‫ف‬ 2.5-fold increase was signifi cant ( Fig. 2C ). Altogether, these data demonstrate that, at P7, cerebellum PCSK9 locally downregulates the protein levels of the LDLR.
We next determined whether the lack of PCSK9 can alter the development of the cerebellum. Nissl staining revealed a normal cerebellum organization in Pcsk9 Ϫ / Ϫ mice, with no gross modifi cation of the EGL, molecular layer (ML), or internal granular cell layer (IGL) ( Fig. 2D ). This was confi rmed by the staining of Purkinje cells with calbindin (supplemental
BrdU incorporation into DNA was visualized using a specifi c antibody (Roche Diagnostics). Brain cryosections were dried and then fi xed 5 min in methanol at Ϫ 20°C, then incubated for 45 min in 2N HCl at 37°C and neutralized in borate buffer for 10 min. Brain slices were incubated overnight with a mouse BrdU antibody (1:100), detected with a biotin-labeled secondary antibody (PerkinElmer), and revealed using the Vectastain kit (Vector Laboratories) and DAB substrate (Zymed Laboratories). BrdU-positive cells in the dentate gyrus were counted on four brain slices regularly spaced for at least fi ve mice for each group. BrdU counts following ischemic stroke were expressed as percentages of counts in the nonlesioned side (fi xed at 100%).

Statistical analysis
Quantitations are defi ned as mean ± SEM. The statistical signifi cance of differences between groups was evaluated by Student-Newman-Keuls test or one-way ANOVA (ANOVA test) followed by a Dunnett test. A difference between experimental groups was considered statistically signifi cant at P < 0.05.

RESULTS
PCSK9 enhances the degradation of the LDLR in mouse telencephalon at E12. 5 We previously demonstrated that during development the expression of PCSK9 is tissue-and time-dependent PCSK9 is upregulated in the dentate gyrus following an ischemic stroke To test whether PCSK9 is regulated following a traumatic brain injury, as observed after partial hepatectomy ( 6 ), we examined by in situ hybridization the kinetic of its expression in mouse brain following a tMCAO. After 6 h of reperfusion, PCSK9 mRNA was still not detectable in brain areas other than the RE-OP. In contrast, 24 h and 72 h postreperfusion, PCSK9 transcripts were upregulated in the ipsilateral lesioned side of the dentate gyrus, and much less in the contralateral nonlesioned side, and disappeared one week following tMCAO, without signifi cantly affecting the expression of PCSK9 in the RE-OP ( Fig. 4 ). This is the fi rst demonstrated that the absence of PCSK9 does not affect LDLR protein levels in either the RE-OP or Ob ( Fig. 3B ). Note that the LDLR is poorly expressed in these structures, as a load of 200 g protein per lane was required to detect it by Western blot. This may explain why we were not able to sense any immunofl uorescence signal in these structures (data not shown). Nissl staining also revealed that the lack of PCSK9 did not affect the organization of the layers in the RE-OP ( Fig. 3C ) and Ob ( Fig. 3D ). In addition, Western blot analyses of the cell proliferation (PCNA) and cell differentiation (Tuj-1, NeuN, and GFAP) markers revealed no changes in their protein levels in Pcsk9 Ϫ / Ϫ mice (supplemental Fig. III).  dentate gyrus under our tMCAO conditions ( Fig. 5C ). In addition, hippocampal neurons were healthy as demonstrated by the absence of Fluoro-Jade® B labeling in hippocampus (data not shown). The mRNA expression of clusterin, previously shown to be upregulated and to have a neuroprotective role after permanent middle cerebral artery occlusion ( 27 ), was also not modulated in Pcsk9 Ϫ / Ϫ dentate gyrus (supplemental Fig. IV). Interestingly, although hippocampus LDLR protein levels were reduced in WT and Pcsk9 Ϫ / Ϫ mice on the lesioned side of the brain after the ischemic stroke, the decrease in LDLR levels was attenuated by 50% in Pcsk9 Ϫ / Ϫ mice ( Fig. 6A ). Note that the LDLR protein levels in the nonlesioned side of the hippocampus were similar to those of shamoperated control mice ( Fig. 6B ). Altogether, our results evidence that PCSK9 mRNA levels are upregulated following a traumatic injury in vivo.
Because tMCAO impacts on behavior in a manner related to the extent of lesion, we measured the lesion volume and mouse behavior following tMCAO ( 20 ). The data showed that mouse behavior (scale 2, corresponding to mild consistent curling, with >50% attempts to curl to the contralateral side, see Methods) and lesion volumes of Pcsk9 Ϫ / Ϫ mice ( ‫ف‬ 70% compared with the nonlesioned side) did not differ from those of WT mice ( Fig. 5A , B ).
Following tMCAO, PCSK9 was not expressed in the infarct and penumbra areas, suggesting that PCSK9 does not play a signifi cant role in neuronal death. In contrast, it was expressed after ischemic stroke in the dentate gyrus ( Fig. 4 ), a brain area in which neurogenesis takes place ( 26 ). Because Pcsk9 Ϫ / Ϫ mice exhibited impaired hepatocyte proliferation following partial hepatectomy ( 6 ), we tested the impact of PCSK9 on neurogenesis in hippocampus following ischemic stroke. Although cell proliferation, based on BrdU incorporation, tends to increase on the stroke side, in WT and Pcsk9 Ϫ / Ϫ mice, our data failed to reveal any significant impact of PCSK9 on the de novo neurogenesis in the Ϫ / Ϫ mice. Bar = 1 mm.

DISCUSSION
The strikingly lower levels of LDL-cholesterol ( Ϫ 40% or Ϫ 85%) found in individuals carrying loss-of-function mutations on one or both PCSK9 alleles generated a strong interest from researchers and pharmaceutical industries to develop a PCSK9 inhibitor/silencer for the treatment of dyslipidemias ( 8,9 ). The injection of PCSK9-specifi c monoclonal antibodies ( 30 ) or antisense oligonucleotides in mice ( 31,32 ) and cynomolgus monkeys ( 33 ) provided the proof of principle that PCSK9 inhibition/silencing is a promising therapeutic approach for dyslipidemias. It was thus critical to defi ne the clinical phenotypes of the lack of PCSK9 to fully understand its physiological roles in various organs. From this perspective, the impact of the lack of PCSK9 on brain recovery following ischemic stroke was also of interest for future patients who would be treated with a PCSK9 inhibitor/silencer.
We herein demonstrated that, in mouse brain, PCSK9 is co-expressed with the LDLR in the telencephalon at E12.5, cerebellum at P7, and RE-OP at adulthood ( Figs. 1-3 ). The low LDLR protein levels in the RE-OP and olfactory bulb are unchanged in adult Pcsk9 Ϫ / Ϫ mice compared with WT mice, suggesting that in adult mice PCSK9 does not promote the degradation of the LDLR in these brain areas ( Fig. 3 ). Our data are consistent with those reported for adult transgenic mice overexpressing PCSK9 in the liver ( 34 ), in which no LDLR regulation by ectopic PCSK9 in the adult hippocampus and cortex was observed ( 1,6 ).
However, in the present study we provide the fi rst demonstration that endogenous PCSK9 regulates the levels of LDLR during mouse brain development and following ischemic stroke. Total LDLR protein levels were indeed у 2.5-fold higher in the telencephalon at E12.5 and in the cerebellum at P7 of Pcsk9 Ϫ / Ϫ embryos or newborns ( Figs.  2, 3 ). Furthermore, they decreased 2-fold less in the Pcsk9 Ϫ / Ϫ -lesioned dentate gyrus compared with that in WT mice ( Fig. 6A ). Consistently, cell surface LDLR levels during brain development were increased, suggesting that, as in liver, PCSK9 enhances LDLR internalization and degradation in the analyzed developing or lesioned brain areas. Although annexin A2 has been shown to be an inhibitor of extra-hepatic PCSK9 ( 35 ), in Pcsk9 Ϫ / Ϫ brain during development and at adulthood, the protein levels of annexin A2 were unchanged compared with those of WT mice (supplemental Fig. V). However, we cannot exclude that, at adulthood, the lack of LDLR regulation by PCSK9 in WT mice is due to a possible modulation of PCSK9 binding to LDLR by annexin A2, especially because LDLR levels are particularly low in the RE-OP.
The absence of V5 immunoreactivity in the CSF of transgenic mice overexpressing V5-tagged PCSK9 in liver (supplemental Fig. VI) demonstrated that circulating PCSK9 does not cross the blood-brain barrier (BBB). This may also explain the absence of brain LDLR regulation by suggest that, even though LDLR levels were downregulated following an ischemic stroke, PCSK9 still promotes its degradation.

Downregulation of the protein levels of apoE in Pcsk9 ؊ / ؊ mice during development
To determine the role of the regulation of brain LDLR by PCSK9 and because apoE is the main apolipoprotein in brain that can bind LDLR, we quantifi ed the protein levels of apoE during development, at adulthood, and following ischemic stroke. Western blot analyses showed that the ‫ف‬ 34 kDa untruncated apoE protein levels were ‫ف‬ 25% lower in Pcsk9 Ϫ / Ϫ mice in the telencephalon at E12.5 and in the cerebellum at P7 ( Fig. 7A , B ), but not at adulthood in the RE-OP, olfactory bulb, and cerebrospinal fl uid (CSF) ( Fig. 7C, D ). Following ischemic stroke, the ‫ف‬ 34 kDa apoE protein levels increased ‫ف‬ 1.6-fold in the lesioned dentate gyrus compared with the nonlesioned side ( Fig. 7E ), consistent with the upregulation of apoE in infarcted cortex and striatum ( 28 ). However, no signifi cant difference in the levels of upregulated apoE protein levels between WT and Pcsk9 Ϫ / Ϫ mice was observed. Note that the apoE protein levels of sham-operated mice were similar to those of nonlesioned dentate gyrus (data not shown). In addition to the ‫ف‬ 34 kDa apoE form, another form running with an apparent molecular weight of ‫ف‬ 28 kDa is present at all stages except at P7, as previously reported ( 29 ). The protein levels of this apoE form are also not modulated in Pcsk9 Ϫ / Ϫ mice. Altogether, these data suggest that adulthood, apoE ( Fig. 7C, D ) and LDLR ( Fig. 3B ) protein levels are not regulated in the brain of Pcsk9 Ϫ / Ϫ mice. Following transient ischemic stroke, in spite of a diminution of LDLR by ‫ف‬ 2-fold less in Pcsk9 Ϫ / Ϫ mice compared with WT mice, apoE protein levels similarly increased in WT and Pcsk9 Ϫ / Ϫ lesioned mice compared with the nonlesioned side ( Fig. 7E ), suggesting that the ‫ف‬ 2-fold higher levels of LDLR in the lesioned hippocampus of Pcsk9 Ϫ / Ϫ mice is not suffi cient to enhance the general degradation of apoE in brain. However, during brain development, PCSK9 induces a downregulation of the protein levels of the nontruncated apoE form ( ‫ف‬ 25%) ( Fig. 7A, B ), which may be due to the increase of cell surface LDLR and consequently to an increase in apoE uptake and degradation by lysosomes. This is consistent with the upregulation of apoE protein levels reported in CSF and cortex of Ldlr Ϫ / Ϫ mice ( 46 ), and the downregulation by 50-90% of apoE levels in the brain of transgenic mice overexpressing LDLR ( 40 ). The downregulation or degradation of apoE levels by PCSK9, probably through LDLR upregulation, is not enough to induce clear physiological consequences, since during development and following transient ischemic stroke we did not observe any brain morphology alteration or modulation of markers of cell proliferation, cell differentiation, synapses, or clusterin, a neuroprotective protein for ischemic stroke (supplemental Figs. I-IV). Thus, the physiological role of PCSK9 in the brain remains to be defi ned. Further investigations may be warranted as we cannot exclude from our study that the effect of PCSK9 on PCSK9 at adulthood. Because the BBB is more permissive at earlier developmental stages and the establishment of its integrity during embryogenesis, or at postnatal stages, has been controversial ( 36,37 ), the potential contribution of circulating PCSK9 in LDLR downregulation at E12.5 and P7 remains to be elucidated. However, the excellent colocalizations of PCSK9 and LDLR expression in the frontal cortex (E12.5) and external granular cell layer (P7) suggest a major contribution of local PCSK9. Furthermore, we showed by in situ hybridization that the expression level of other members of the PC family relevant to cholesterol regulation (Furin, PC5/6, and SKI-1/S1P) ( 38 ) do not change upon loss of PCSK9 expression in the liver, where the major expression of PCSK9 occurs (Ref. 1 and supplemental Fig. VII), suggesting that absence of PCSK9 does not infl uence the mRNA expression of these proprotein convertases (PCs). Consequently, the observed regulation of LDLR levels in Pcsk9 Ϫ / Ϫ brain is likely a consequence of the absence of PCSK9.
The physiological and pathological functions of LDLR in the nervous system remain unclear. Ldlr Ϫ / Ϫ mice have normal brain morphology but exhibit impaired learning and memory ( 39 ), and the impact of excess LDLR levels on general brain functions has not yet been reported, except in the case of Alzheimer's disease progression ( 40 ). However, it has been shown that LDLR has the highest affi nity for apoE in brain (41)(42)(43)(44). ApoE, a ‫ف‬ 34 kDa glycoprotein, is the major apolipoprotein in the brain ( 45 ), and it plays an important role in brain cholesterol metabolism. At mouse brain is more subtle, since cholesterol is involved in nerve conduction velocity through the formation of myelination and synaptogenesis ( 47 ). It also plausible that local hydroxy-cholesterol, rather than circulating cholesterol, is critical for CNS development ( 48 ). Injection of purifi ed PCSK9 or a recombinant adenovirus overexpressing PCSK9 in mice reduced the protein levels of the LDLR in liver, lung, kidney, and small intestine, but not in the adrenal glands ( 11,12 ). In addition, upon overexpression in kidney, the secreted plasma PCSK9 promotes LDLR degradation mostly in the liver and raises plasma LDL ( 49 ). Altogether, our data and those in the literature show that not all tissues respond equally to local or circulating PCSK9 and that in adult brain LDLR does not respond to circulating PCSK9. Because the gene ablation of PCSK9 in mice, which completely abolishes PCSK9 expression, does not affect brain organization or brain recovery following ischemic stroke, the silencing of PCSK9, either using monoclonal antibodies ( 30 ) or antisense therapies ( 33 ) that may partially inhibit PCSK9 expression, should not affect brain PCSK9 levels. However, we cannot exclude that low circulating LDL-cholesterol in treated adult patients with PCSK9-silencing approaches could affect brain LDLR levels via regulation of PCSK9 in the central nervous system ( 1 ), likely through the SREBP-2 ( 50 ) or HNF-1 ␣ pathways ( 51 ). Furthermore, rare human subjects exhibiting either PCSK9 loss-of-function mutations [e.g., R46L ( 3 ) and R434W ( 52 )] or gain-of-function mutations [e.g., and S127R ( 2 ) and D374Y ( 53 )], may during development also exhibit higher or lower levels of LDLR in the brain, respectively, resulting in subtle effects later on in adults.
In conclusion, the present results suggest that PCSK9 inhibition should not interfere with brain development and morphology or with brain recovery/damage after an ischemic stroke. They are consistent with humans carrying heterozygous loss-of-function mutations in PCSK9 who appear healthy and have normal a lifespan ( 54,55 ). Thus, PCSK9 inhibition is likely not to have overt deleterious effects on patients affected by a stroke event and would not be expected to cause major side effects in patients treated with a PCSK9 inhibitor or silencer for hypercholesterolemia.