Implication of the proprotein convertase NARC-1/PCSK9 in thedevelopment of the nervous system

Steve Poirier,* Annik Prat,* Edwige Marcinkiewicz,* Joanne Paquin,�Babykumari P. Chitramuthu,� David Baranowski,� Benoit Cadieux,�Hugh P. J. Bennett� and Nabil G. Seidah*

*Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, Montreal, Quebec, Canada

�Laboratoire de Neuroendocrinologie Developpementale, Departement de Chimie et de Biochimie, Universite du Quebec a Montreal,

Montreal, Quebec, Canada

�Endocrine Laboratory, Royal Victoria Hospital and McGill University Health Center Research Institute, Montreal, Quebec, Canada

Abstract

Neural apoptosis-regulated convertase-1/proprotein conver-

tase subtilisin-kexin like-9 (NARC-1/PCSK9) is a proprotein

convertase recently described to play a major role in choles-

terol homeostasis through enhanced degradation of the low-

density lipoprotein receptor (LDLR) and possibly in neural

development. Herein, we investigated the potential involve-

ment of this proteinase in the development of the CNS using

mouse embryonal pluripotent P19 cells and the zebrafish as

models. Time course quantitative RT–PCR analyses were

performed following retinoic acid (RA)-induced neuroectoder-

mal differentiation of P19 cells. Accordingly, the mRNA levels

of NARC-1/PCSK9 peaked at day 2 of differentiation and fell

off thereafter. In contrast, the expression of the proprotein

convertases subtilisin kexin isozyme 1/site 1 protease and

Furin was unaffected by RA, whereas that of PC5/6 and PC2

increased within and/or after the first 4 days of the differenti-

ation period respectively. This pattern was not affected by the

cholesterogenic transcription factor sterol regulatory element-

binding protein-2, which normally up-regulates NARC-1/

PCSK9 mRNA levels in liver. Furthermore, in P19 cells, RA

treatment did not affect the protein level of the endogenous

LDLR. This agrees with the unique expression pattern of

NARC-1/PCSK9 in the rodent CNS, including the cerebellum,

where the LDLR is not significantly expressed. Whole-mount

in situ hybridization revealed that the pattern of expression of

zebrafish NARC-1/PCSK9 is similar to that of mouse both in

the CNS and periphery. Specific knockdown of zebrafish

NARC-1/PCSK9 mRNA resulted in a general disorganization

of cerebellar neurons and loss of hindbrain–midbrain bound-

aries, leading to embryonic death at � 96 h after fertilization.

These data support a novel role for NARC-1/PCSK9 in CNS

development, distinct from that in cholesterogenic organs

such as liver.

Keywords: cholesterol, neural apoptosis-regulated conver-

tase-1/ proprotein convertase subtilisin-kexin like-9, neuro-

genesis, P19 cells, proprotein convertase, zebrafish.

J. Neurochem. (2006) 98, 838–850.

Received September 17, 2005; revised manuscript received March 17,2006; accepted March 23, 2006.Address correspondence and reprint requests to Nabil G. Seidah,

Clinical Research Institute of Montreal, 110 Pine Avenue West, Mon-treal, QC, H2W 1R7, Canada. E-mail: seidahn@ircm.qc.caAbbreviations used: a-MEM, modified minimum Eagle’s medium;

Ctrl, control; dpf, days post-fertilization; DRG, dorsal root ganglia; ECL,enhanced chemiluminescence; FBS, fetal bovine serum; GFAP, glialfibrillary acidic protein; HMG-CoA, 3-hydroxy-3-methylglutarylco-enzyme A; HMGCR, HMG-CoA reductase; hpf, hours post-fertil-ization; HRP, horseradish peroxidase; ISH, in situ hybridization; LDLR,low-density lipoprotein receptor; mAb, monoclonal antibody; mm,

mismatch; MO, morpholino oligonucleotide; NARC-1/PCSK9, neuralapoptosis-regulated convertase-1/proprotein convertase subtilisin-kexinlike-9; NeuN, neuronal nuclei; (n)SREBP, (nuclear) sterol regulatoryelement-binding protein; P1, day 1 after birth; PACE, paired basic aminoacid cleaving enzyme; PBS, phosphate-buffered saline; PBST, PBScontaining 0.1% Tween-20; PC, proprotein convertase; QPCR, quanti-tative RT–PCR; RA, retinoic acid; RXR, retinoid X receptor; SDS,sodium dodecyl sulfate; SG, spinal ganglia; SKI-1/S1P, subtilisin kexinisozyme 1/site 1 protease; TBST, Tris-buffered saline containingcontaining 0.1% Tween-20; TN, trigeminal nerve; WT, wild type; zf,zebrafish.

Journal of Neurochemistry, 2006, 98, 838–850 doi:10.1111/j.1471-4159.2006.03928.x

838 Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850� 2006 The Authors

The mammalian proprotein convertases (PCs) constitute afamily of nine serine proteinases related to bacterial subtilisin.These include the seven basic amino acid- specific conver-tases known as PC1/3, PC2, Furin, PC4, PACE4, PC5/6 andPC7 (Seidah and Chretien 1999; Zhou et al. 1999), and twoproteinases that cleave at non-basic residues, namely subtil-isin kexin isozyme 1/site 1 protease (SKI-1/S1P) (Sakai et al.1998b; Seidah et al. 1999) and neural apoptosis-regulatedconvertase-1/proprotein convertase subtilisin-kexin like-9(NARC-1/PCSK9) (Seidah et al. 2003; Benjannet et al.2004). These proteinases are implicated in the limitedproteolysis of precursors of secretory proteins that participatein several biological functions, such as development, repro-duction and immune response, and in numerous pathologies,including neurological disorders, infectious diseases, cancerand dyslipidemias (Decroly et al. 1997; Mbikay et al. 1997;Roebroek et al. 1998; Denis et al. 2000; Benjannet et al.2001; Lenz et al. 2001; Khatib et al. 2002; Seidah and Prat2002; Thomas 2002; Abifadel et al. 2003).

The PC NARC-1 (Seidah et al. 2003; Benjannet et al.2004), also known as PCSK9 (Abifadel et al. 2003), wasrecently characterized and shown to be highly expressed inliver and small intestine. Human point mutations within theNARC-1/PCSK9 coding sequencewere reported to be directlyassociated with either familial hypercholesterolemia in manycountries (Abifadel et al. 2003; Benjannet et al. 2004; Timmset al. 2004;Allard et al. 2005;Attie andSeidah 2005; Pisciottaet al. 2005) or familial hypocholesterolemia in black AfricanandAmerican populations (Cohen et al. 2005;Kotowski et al.2006), probably resulting from gain and loss of functionrespectively (Attie and Seidah 2005). So far, the only knownfunction of NARC-1/PCSK9 is its ability to enhance thedegradation of the low density lipoprotein receptor (LDLR)within an acidic compartment (Benjannet et al. 2004;Maxwelland Breslow 2004; Park et al. 2004), through an unknownmechanism (Attie and Seidah 2005).

Twogroupshave independentlyusedmicroarray technologyto study mRNAs that are regulated by cholesterol or by thesterol regulatory element-binding proteins (SREBPs), andobserved that hepatic rodent NARC-1/PCSK9 expression isup-regulated in SREBP-2 transgenicmice (Horton et al. 2003)and down-regulated by diet-induced excess circulatingcholesterol (Maxwell et al. 2003). Recent data from our groupdemonstrated that NARC-1/PCSK9 mRNA levels areup-regulated in hepatocyte-derived human HepG2 cells andprimary hepatocytes by the statins, which are 3-hydroxy-3-methylglutaryl co-enzyme (HMG-CoA) reductase inhibitors,but not by the liver X receptor agonist 22-hydroxycholesterol,nor by the retinoid X receptor agonist 9-cis-retinoic acid(Dubuc et al. 2004). This process can be reversed by meval-onate treatment, in agreement with the implied role of SREBP-2 in the up-regulation ofNARC-1/PCSK9 (Dubuc et al. 2004).In contrast to that ofNARC-1/PCSK9, the expressionofSKI-1/S1P is not sensitive to cholesterol, even though it is a key

protease, which, together with the metalloprotease site-2protease (S2P), activates the formation of the nuclear form ofSREBPs (Sakai et al. 1998a; Brown and Goldstein 1999).

During development and in adulthood, NARC-1/PCSK9 isexpressed in liver and small intestine, two regeneratingorgans implicated in cholesterol metabolism. By comparison,NARC-1/PCSK9 is only transiently expressed in corticalcells of the kidney and in specific brain regions where activeneurogenesis takes place. These include the telencephalon,rostral extension of the olfactory epithelium and thecerebellum (Seidah et al. 2003). Whether expression in thelatter organs is also controlled by cholesterol is unknown.Overexpression of NARC-1/PCSK9 cDNA in primaryneurons isolated from telencephalon at embryonic day 12,suggested that this enzyme can enhance neurogenesis ofprogenitor brain telencephalic cells (Seidah et al. 2003). Inorder to understand the neural function of NARC-1/PCSK9,we undertook an analysis of its expression in pluripotentmouse P19 embryonal carcinoma cells, which are well suitedfor studies related to neuroectodermal cell development andmaturation. These cells can be efficiently induced todifferentiate in culture into neurons and astroglia by combi-ning cell aggregation and brief treatment with all trans-retinoic acid (RA) (Rudnicki et al. 1989; McBurney 1993).

In this study low levels of NARC-1/PCSK9 mRNA weredetected in naive P19 cells. Upon neuroectodermal inductionof P19 cells by RA, NARC-1/PCSK9 transcripts wereinduced approximately 7-fold, with a maximum at day 2 ofthe RA treatment, followed by repression to levels belowthose of naive P19 cells. These findings suggested that early,transient induction of NARC-1/PCSK9 mRNA may beneeded to modulate neurogenesis/gliogenesis, reminiscent ofits expression in telencephalon (Seidah et al. 2003). How-ever, in contrast to findings in the liver, NARC-1/PCSK9mRNA did not seem to be significantly regulated by SREBP-2 in P19 cells and its up-regulation did not affect the proteinlevel of the endogenous LDLR.

We further extended our data into the realm of a wholeanimal, using the model zebrafish (Danio rerio). As inmouse, zebrafish (zf)-NARC-1/PCSK9 is well expressed inthe developing liver and intestine, as well as in corticalneurons and cerebellar granules where continued neurogen-esis occurs. Specific knockdown of zf-NARC-1/PCSK9mRNA in zebrafish embryos using morpholino antisenseoligonucleotide technology led to the loss of CNS structures,such as the midbrain and hindbrain, and culminated inlethality at � 3–4 days after fertilization.

Materials and methods

Cell culture and neuroectodermal differentiation

Undifferentiated P19 embryonal carcinoma cells were maintained at

37�C under a humidified atmosphere of 5% CO2 in modified

NARC-1/PCSK9 and neural development 839

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850

minimum Eagle’s medium (a-MEM) (Invitrogen-Gibco, Burlington,

ON, Canada) supplemented with 10% fetal bovine serum (FBS)

(Cansera, Etobicoke, ON, Canada), penicillin (50 U/mL) and

streptomycin (50 lg/mL; Gibco), and passaged every 48 h. To

allow aggregate formation, 1 · 106 cells were cultured in 100-mm2

bacterial-grade dishes (Fisher, Nepean, ON, Canada) in a-MEM

supplemented with 5% FBS and 5% donor bovine serum (Cansera).

The neuroectodermal differentiation of these cells was then induced

by a 4-day treatment (in quadruplicate, n ¼ 4) or not (control, n ¼3) with 500 nM all-trans retinoic acid (RA) (Sigma-Aldrich,

Oakville, ON, Canada), with renewal of the medium after 2 days.

Aggregates were then dissociated into single cells with 0.025%

trypsin)1 mM EDTA (Invitrogen-Gibco) and cells were replated at

1 : 4 in the absence of RA on gelatinized-coated tissues culture

dishes in 10 mL Neurobasal medium containing B27 supplement,

Glutamax and antibiotics (all products from Invitrogen-Gibco). The

medium was replenished after 3 days and cells were maintained

until day 10. On selected days, cells were washed three times in

phosphate-buffered saline (PBS) and pelleted for further analysis.

Immunocytochemistry of P19 cells

For immunocytochemistry, we used monoclonal antibodies (mAb), at

a dilution of 1 : 50, against either neuronal nuclei (NeuN) (Chemicon

International, Temecula, CA, USA) or glial fibrillary acidic protein

(GFAP) (Sigma-Aldrich), as markers of neurons and glia respectively.

Accordingly, naive P19 cells or those cultivated up to day 10 with or

without RA induction were successively fixed for 15 min in 4%

paraformaldehyde/PBS, permeabilized with 0.1% Triton X-100/PBS,

incubated with 150 mM glycine/PBS for 5 min, blocked with 1%

bovine serum albumin/PBS (Sigma-Aldrich) for 30 min, and then

incubated overnight with each mAb at 4�C. Immunoreactivity was

revealed with goat anti-mouse Alexa Fluor�555 (Invitrogen-Molecu-

lar Probes, Burlington, ON, Canada) at 1 : 200 dilution in PBS for

45 min, increasing and then stabilizing the fluorophorewith a solution

of 5% Dabco by weight (Sigma-Aldrich) in 90% glycerol/10% PBS.

Immunofluorescence analyses were performed with a Zeiss Axiovert

S100 tv microscope (Zeiss Axiovert, Toronto, ON, Canada).

RNA preparation and cDNA synthesis

Total RNA was isolated using TRIzol reagent (Invitrogen, Burling-

ton, ON, Canada) from undifferentiated P19, P19 aggregates and

P19-derived neurons according to the recommendations of the

manufacturer. Total RNA integrity was verified by 1% ethidium

bromide-stained agarose gel according to predominant ribosomal

RNA bands. Nucleic acid purification was measured by A260/A280

and values of 1.6–2.0 were considered to indicate pure preparations.

Typically, 250 ng total RNA was used for cDNA generation in a

total volume of 20 ll using SuperScript II reverse transcriptase,

25 lg/mL oligo(dT)12)18, 0.5 mM 2¢-deoxynucleoside 5¢-triphos-phates and 40 U RNaseOUT; all products were used according to

the manufacturer’s instructions (Invitrogen).

Quantitative RT–PCR (QPCR)

All primers (Invitrogen) were designed using Primer3 software (http://

frodo.wi.mit.edu/cgi-bin/primer3/primer3_http://www.cgi) to pro-

duce amplicons that overlapped exonic splicing junctions, to avoid

genomicDNAamplification (Table 1). The primers were submitted to

BLAST databases to verify their specificity. Optimization was

evaluated using the MX4000 multiplex quantitative PCR (QPCR)

instrument and software (Stratagene, La Jolla, CA, USA). Optimal

primer concentrations and cDNA standard curves were obtained for

each target gene. All samples were submitted to two independent PCR

reactions: one for the normalizing ribosomal protein S16 and the other

for the gene of interest, each in triplicate as reported by Dubuc et al.(2004). Each reaction was in a final volume of 25 ll using QuantiTecSYBR green PCR master mix from Qiagen (Mississauga, ON,

Canada) in a thermal profile of an activation step (15 min at 95�C)followed by 40 cycles of 30 s at 94�C, 30 s at 58�C and 30 s at 72�C.Relative mRNA levels for each sample were quantified using the Ct

approach (fluorescence threshold), normalized with respect to S16

expression as an endogenous RNA standard, and calibrated by setting

the control (day 0) at 1. The data shown correspond to representative

experiments, in which P19 cells were treated (n ¼ 4) or not treated

(n ¼ 3) with RA.

Western blot analysis

Cells were washed three times in PBS and lysed in RIPA buffer

[50 mM Tris/HCl, pH 8.0, 1% (v/v) Nonidet P40, 0.5% sodium

deoxycholate, 150 mM NaCl and 0.1% (v/v) sodium dodecyl sulfate

(SDS)] with a Complete Protease Inhibitor Cocktail (Roche Applied

Science, Laval, QC, Canada). Proteins were separated by SDS–poly-

acrylamide gel electrophoresis (8% gels) and blotted on to HyBond

Table 1 Primers used for QPCR

mRNA assessed Forward primer Reverse primer

hs14 (NM_001025071) GGCAGACCGAGATGAATCCTCA CAGGTCCAGGGGTCTTGGTCC

hPCSK9 (NM_174936) ATCCACGCTTCCTGCTGC CACGGTCACCTGCTCCTG

mS16 (NM_013647) AGGAGCGATTTGCTGGTGTGG GCTACCAGGGCCTTTGAGATG

mPCSK9 (NM_153565) TGCAAAATCAAGGAGCATGGG CAGGGAGCACATTGCATCC

mPCSK5/6AB (XM_129214) ACTCTTCAGAGGGTGGCTA GCTGGAACAGTTCTTGAATC

mPCSK2 (NM_008792) TGACAAGTGGCCTTTCAT ATCAGGGTCCATTCCTTC

mSKI-1 (NM_019709) GCCCTCAAGTGAGACCTTTG GTCCCACCTCCTGGTTGTAG

mFurin (NM_011046) CATGACTACTCTGCTGATGG GAACGAGAGTGAACTTGGTC

mLdlr (NM_010700) GTATGAGGTTCCTGTCCATC CCTCTGTGGTCTTCTGGTAG

mHMGCR(NM_008255) TCAGAAGTCACATGGTTCAC TTGCATGTTAGTCCTTGAGA

mSREBP-2 (NM_033218) GTTCTGGAGACCATGGAG AAACAAATCAGGGAACTCTC

Accession numbers of the chosen QPCR forward and reverse primers are included.

840 S. Poirier et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850� 2006 The Authors

nitrocellulose membranes (Amersham Biosciences, Piscataway, NJ,

USA). The blots were incubated for 1 h in TBS-T (50 mM Tris-HCl,

pH 7.5, 150 mM NaCl, 0.1% Tween-20) containing 10% non-fat dry

milk. PC2 (C-terminal, against amino-acid (Cter) 529–637; Benjan-

net et al. 1993), LDLR (Abcam, Cambridge, MA, USA), neural

adhesion L1 (Kalus et al. 2003) or b-actin (Sigma-Aldrich)

antibodies were diluted 1 : 1000, 1 : 2500, 1 : 2000 and 1 : 1000

respectively in the same buffer. After 2 h incubation at room

temperature (22�C), membranes were washed and reincubated for

1 h with either donkey anti-rabbit–horseradish peroxidase (HRP) for

PC2 and b-actin detection (1 : 5000; Amersham Biosciences) or

rabbit anti-chicken-HRP for LDLR detection (1 : 5000; Abcam).

Blots were probed using enhanced chemiluminescence (ECL) using

an ECL plus kit (Amersham Biosciences).

In situ hybridization (ISH) in mouse

For ISH, mouse sense and antisense cRNA probes coding for mouse

NARC-1/PCSK9 (nucleotides 1197–2090) (Seidah et al. 2003) andmouse LDLR (nucleotides 1800–2565; accession no. BC019207)

were labeled with [35S]UTP and [35S]CTP (1250 Ci/mmol; Amer-

sham, Oakville, ON, Canada), to obtain high specific activities of

� 1000 Ci/mmol. whole C1 mouse cryosections (8–10 lm obtained

at day 1 after birth (P1) from unperfused mice were fixed for 1 h in

4% formaldehyde and hybridized overnight at 55�C as described

previously (Marcinkiewicz et al. 1999; Seidah et al. 2003). For

autoradiography, the sections were dipped in photographic emulsion

(NTB-2; Kodak Ile des Soeurs, Verdun, Quebec, Canada), exposed

for 6–12 days, developed in D19 solution (Kodak), and stained with

hematoxylin.

ISH in zebrafish

In order to prevent the appearance of melanin pigmentation, embryos

at approximately 18–24 h post-fertilization (hpf) were grown in egg

water supplementedwith 0.003%of the tyrosinase inhibitor 1-phenyl-

2-thiourea (phenylthiocarbamide; Sigma). Staged embryos were

manually dechorionated and fixed for 2 h at room temperature or

overnight at 4�C in 4% paraformaldehyde/PBS. After several washes

in PBS, embryos were stored in 100% methanol until needed.

For riboprobe synthesis, a 1323-bp PCR fragment of the zf-NARC-

1/PCSK9 cDNA comprising the initiator methionine (nucleotides

388–1710) was cloned into pCR 2.1-TOPO, excised with EcoRI andsubcloned into pBluescript II. For antisense and sense riboprobe

synthesis, T3 and T7 RNA polymerases were used after linearization

of the plasmid with SmaI and HindIII respectively. At all stages

examined, both sense and antisense probes were analyzed. Whole-

mount ISH with digoxygenin-labeled RNA probes and antibody

staining were done essentially according to Schulte-Merker et al.(1992) and Thisse et al. (1993) at a hybridization temperature of

70�C. Stained whole-mount embryos were mounted in glycerol and

visualized under a Leica MZFLIII stereomicroscope (Leica Micro-

systems, Richmond Hill, ON, Canada). Images were taken with a

Leica DC350F camera (Leica Camera AG, Solms, Germany).

Production of zf-NARC-1/PCSK9 antibody and western blotting

A 14mer peptide (S144IPWNLQRVLQNK156C) corresponding to

the predicted N-terminal 144–156 sequence of mature zf-NARC-1/

PCSK9 following cleavage of the propeptide (Fig. S3) was

synthesized with an additional cysteine residue at the C-terminus

using solid-phase chemistry (Peptide Synthesis Facility of the

Sheldon Biotechnology Center, McGill University, Montreal, QC,

Canada). The peptide was coupled through the C-terminal cysteine

residue to keyhole limpet hemocyanin and a polyclonal antibody

was raised in rabbit by immunization with the conjugate. The

antiserum was purified on a peptide affinity column before use.

Single zebrafish embryos at 24 hpf were solubilized in 30 ll 2 ·Laemmli buffer (125 mM Tris, pH 6.8, 20% glycerol, 4% SDS, 0.2%

bromophenol blue, 0.01% 2-mercaptoethanol), heated at 96�C for

10 min, and centrifuged in a microfuge at 13 000 g for 15 min.

Proteins were separated by SDS–polyacrylamide gel electrophoresis

(10% gels) and blotted on to HyBond-C Extra nitrocellulose

membranes (Amersham). Blots were blocked with 0.5% membrane

blocking agent (Amersham) overnight at 4�C and incubated with

primary antiserum against zf-NARC-1/PCSK9, which was diluted to

1 : 4000 in 0.25% membrane-blocking agent (Amersham)/PBST

(0.01 MKH2PO4, 0.1 MNa2HPO4, 1.4 MNaCl, 0.03 MKCl, pH 7.4,

containing 0.1% Tween-20) for 1 h at room temperature, followed by

three washes in PBST. The membrane was then incubated with

donkey anti-rabbit–HRP secondary antibody (Amersham) diluted to

1 : 4000 in PBST, followed by six washes with PBST. The blot was

developed by ECL, with monitoring for chemiluminescence accord-

ing the manufacturer’s instructions, and developed using Hyperfilm

(Amersham). The membrane was then stripped by incubating for

30 min at 60�C in stripping buffer (65 mM Tris pH 6.7, 2% SDS) and

probed for actin using anti-actin antibody (Sigma) diluted at 1 : 500

for 1 h at room temperature. After three washes with PBST, the

membrane was incubated with goat anti-mouse–HRP (Calbiochem,

La Jolla, CA, USA) secondary antibody at 1 : 4000 dilution, for 1 h at

RT. The blotwas developed for chemiluminescence as outlined above.

Maintenance of zebrafish and morpholino oligonucleotide (MO)

microinjection

Adult zebrafish were obtained from Scientific Hatcheries (Huntington

Beach, CA,USA) andmaintainedwithin a controlled light–dark cycle

at 28.5�C (Westerfield 2000). Embryos were developed under

identical conditions in water containing 0.006% Instant Ocean salts.

MOswere obtained fromGene Tools, Inc. (Philomath, OR, USA) and

diluted to 5 ng/nL in Danieaux buffer [58 mM NaCl, 0.7 mM KCl,

0.4 mM MgSO4, 0.6 mM Ca(NO3)2, 5.0 mM Hepes pH 7.6] contain-

ing 0.1% phenol red (Nasevicius and Ekker 2000). Approximately

2 nL (10 ng MO) was injected into the yolk of one- to two-cell stage

embryos using a PLI-100 microinjection system (Harvard Apparatus,

St Laurent, QC, Canada). Phenotypic observation and documentation

were accomplished using a LeicaDC300F digital camera connected to

a Leica MZFLIII stereomicroscope. Morpholino sequences were:

MO1, 5¢-GACGCTTCTCATTCTCTGTGCTTTC-3¢; five base-pair

mismatch (MO1-mm), 5¢-GAGGCTTGTCATTGTCTCTGGTTTC-3¢; and negative scramble control (Ctrl), 5¢-CCTCTTACCTCAGTT-ACAATTTATA-3¢.

Results

Neuroectodermal differentiation of P19 cells and the

neuroendocrine convertase PC2

We used the pluripotent P19 cell line as a model to studywhether NARC-1/PCSK9 was associated with neuroecto-

NARC-1/PCSK9 and neural development 841

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850

dermal induction and/or differentiation. Accordingly, P19cells were induced to differentiate into neuroectodermalderivatives following aggregation in the presence of 500 nMRA (days 0–4). Upon dissociation of the aggregates, theinduced cells were cultured for another 6 days (days 5–10) asadherent monolayers in the absence of RA, resulting inneurite outgrowth (already visible at day 7; not shown)extending from neuronal cells (see arrows in left panels ofFig. 1). In order to assess the extent of neuronal (NeuNpositive) versus glial (GFAP positive) differentiation inducedby RA, we labeled the cells with mAbs against thesemarkers. At day 10 after RA treatment > 95% cells expressedNeuN, whereas < 5% expressed GFAP (Fig. 1), both beinginduced by RA treatment (Fig. 1 and Fig. S1). Therefore, weconclude that the RA-treated cell population at day 10 ismostly (> 95%) composed of neuronal cells.

To follow the neural differentiation process, we also usedthe neuroendocrine PC2 (Seidah et al. 1990; Scopsi et al.1995; Jeannotte et al. 1997) as a marker. This convertase isresponsible for the processing of numerous precursors ofpolypeptide hormones within secretory granules, such as pro-opiomelanocortin (Benjannet et al. 1991), proenkephalin(Breslin et al. 1993), prosomatostatin (Galanopoulou et al.1995) and prodynorphin (Berman et al. 2000). Its precursorproPC2 (�74 kDa) is autocatalytically processed withinsecretory granules into mature PC2 (�67 kDa) (Benjannetet al. 1993). Thus, both the mRNA expression level of PC2(QPCR; Fig. 2a) and the extent of its zymogen processing(western blotting; Fig. 2b) were expected to increase uponneuroectodermal differentiation. Indeed, treatment with RAresulted in a � 350-fold up-regulation of PC2 mRNA at day10 (compared with the control at day 0), whereas in theabsence of RA only a � 30-fold increase was observed(Fig. 2a). Western blot analysis confirmed these data; bothproPC2 and PC2 were up-regulated and the PC2/proPC2ratio reached a maximum (‡ 0.55) at days 8–10 (Fig. 2b),

Fig. 1 Immunocytochemistry of RA-treated

P19 cells at day 10. The type of mAb used

(NeuN or GFAP) is shown (right panels).

The left panels show the cells under visible

light. Arrows point to neural extensions.

(a)

(b)

Fig. 2 PC2 as a neuronal marker. Undifferentiated P19 cells (day 0),

RA-treated P19 aggregates (days 1–4) and P19 derived-neurons

(days 5–10) were collected every 24 h. (a) Relative PC2 mRNA

expression analyzed by QPCR. PC2 expression was normalized with

respect to that of ribosomal protein S16 and calibrated with that found

in undifferentiated cells (day 0). The histogram shown is representa-

tive of three and four independent differentiations on untreated and

treated cells respectively. Each QPCR experiment was done in tripli-

cate; values are mean ± SEM. (b) Immunoblot analyses of PC2 pro-

tein and b-actin marker during neuroectodermal differentiation of P19

cells. (c) Results of the ImageQuant quantitation of the PC2 and

proPC2 arbitrary protein levels.

842 S. Poirier et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850� 2006 The Authors

which is in accord with a more favorable neuroendocrineenvironment for the autocatalytic zymogen processing ofproPC2 into PC2 at these late stages. A similar up-regulationof neurofilament M protein was observed (not shown).

Uniqueness of the NARC-1/PCSK9 expression profile

during neuroectodermal differentiation of P19 cells

The mRNA expression profile of NARC-1/PCSK9 was verydifferent from that of PC2 (Fig. 3). A maximum expressionlevel (� 7-fold) was attained at day 2 during RA exposure,and then abruptly decreased to reach levels � 3–5-fold belowthose of undifferentiated P19 cells from day 3 onwards up today 10. This repression after the aggregation period was alsoobserved in the absence of RA treatment, suggesting that itmay be related to P19 cell handling and/or culture conditions.RA and not cell aggregation caused the transientup-regulation of NARC-1/PCSK9 because aggregation inthe absence of RA coincided with the convertase down-regulation at day 2 (Fig. 3).

Neuroectodermal cell derivatives are also obtained when a4-day RA treatment is applied to cell monolayers instead ofaggregates, but the proportion of neurons in the differentiatedpopulations is decreased, whereas fibroblast-like cells areabundant (Laplante et al. 2004). For comparison, RA wasincubated with either adherent or suspended P19 cells for4 days, and then the cells were replated at the end of day 4and cultured until day 10. Morphological analysis at day 10revealed � 50% less neuritic extensions in the adherent cellprotocol compared with the suspension/aggregate one.Furthermore, PC2 mRNA up-regulation was less pronouncedwith the adherent P19 cells (�150-fold increase; not shown).Nevertheless, when the 4-day RA treatment was applied toadherent P19 cells, the peak of NARC-1/PCSK9 mRNAexpression was observed at day 2 (not shown), similar to thatnoted during RA treatment of suspended P19 aggregates.Thus, RA can induce the expression of NARC-1/PCSK9 inboth adherent and suspended P19 cells. Unfortunately, whena polyclonal antibody raised against mouse NARC-1/PCSK9was used (N. Nassouri and N. G. Seidah, unpublished data),western blots or biosynthetic analyses were not sensitiveenough for the detection of endogenous NARC-1/PCSK9protein expression in P19 cells before and after RA treatment(not shown). Therefore, it was not possible to define the fateof NARC-1/PCSK9-positive cells that were induced by RAtreatment, or to determine whether they would differentiateinto glia or neurons at days 6–10. In contrast to observationsin P19 cells, the up-regulation of NARC-1/PCSK9 by RAtreatment was not detected in adherent HepG2 cells (Fig. S2),indicating either that the RA effect is indirect or that P19cells contain an extra factor(s) that is absent from HepG2cells.

In a similar fashion, we analyzed the mRNA profile of theregulated PC, PC5/6, and those of the constitutivelyexpressed ubiquitous convertases Furin and SKI-1/S1P

(Seidah et al. 2003). Furin and SKI-1/S1P did not exhibitsignificant quantitative variations in their profiles throughoutthe differentiation programme, whereas PC5/6 showed asimilar profile to that of PC2, increasing gradually andreaching a maximum at day 10 (� 14-fold). Interestingly, weconsistently observed the presence of a peak of PC5/6expression (� 4-fold increase) at day 2.

Fig. 3 QPCR profiles of NARC-1/PCSK9, PC5/6, Furin and SKI-1/

S1P during neuroectodermal differentiation of P19 cells. Each sample

was normalized with respect to ribosomal protein S16 expression

and calibrated with undifferentiated cells (day 0). Values are mean ±

SEM.

NARC-1/PCSK9 and neural development 843

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850

Analysis of the cholesterogenic genes encoding

HMG-CoA reductase and LDLR

Because NARC-1/PCSK9 was shown to be down-regulatedby cholesterol in the liver (Maxwell et al. 2003), probablythrough the decreased activation of SREBP-2, it was ofinterest to test whether the observed NARC-1/PCSK9up-regulation in P19 cells following RA treatment (Fig. 3)was concomitant with that of cholesterogenic genes. How-ever, mRNA levels of SREBP-2 were not significantlyaffected by the RA treatment (not shown). Furthermore, theexpression profile of HMG-CoA reductase, a key limitingenzyme involved in cholesterol synthesis that is stronglyup-regulated by active SREBP-2 (Horton et al. 2002), wasvery different from that of NARC-1/PCSK9 (Fig. 4a). Itreached a plateau at days 9–10, but did not significantly peakat day 2, the time at which NARC-1/PCSK9 mRNA levelswere maximal (Fig. 3). This indicated that in P19 cells the

RA-induced up-regulation of NARC-1/PCSK9 mRNA at day2 most probably involves another regulatory mechanism.

NARC-1/PCSK9 is known to enhance the degradationof the LDLR (Maxwell and Breslow 2004), via an indirectmechanism (Benjannet et al. 2004) and in a cell- andtissue-specific manner (Park et al. 2004). As expected fromthe data obtained for HMG-CoA reductase, QPCR analysisrevealed that mRNA levels of the LDLR were notsignificantly altered by RA treatment (not shown). Inaddition, the peak of NARC-1/PCSK9 expression at day 2did not affect LDLR protein levels, as analyzed by westernblotting (done in triplicate at days 0–4; Fig. 4b). NARC-1/PCSK9 may thus fulfill other function(s) in these cells.Accordingly, comparative ISH analysis of NARC-1/PCSK9and LDLR mRNAs in a whole mouse at P1 (Fig. 4c)revealed that, apart from liver and intestine where bothmRNAs were abundant and co-localized, their expressionwas unique. Thus, the LDLR was widely distributed, withnotable hotspots of expression in the thymus, teeth, spinaland dorsal root ganglia and trigeminal nerve. In contrast,NARC-1/PCSK9 was detected in few tissues other thanliver and intestine, including cerebellar neurons (Seidahet al. 2003) where LDLR expression was not prominent(Fig. 4c).

Cloning and developmental expression of NARC-1/

PCSK9 in zebrafish

Several attempts to knockdown the NARC-1/PCSK9 mRNAin P19 cells using various short interfering RNAs andtransfection conditions failed to decrease efficiently the levelof endogenous NARC-1/PCSK9 mRNA at day 2, the time ofmaximal expression, even though we obtained a � 60%reduction at day 0 (not shown). Accordingly, we decided totest the implications of NARC-1/PCSK9 in the developmentof the nervous system in vivo using antisense MO knock-down approaches in the zebrafish. Alignment of human androdent NARC-1/PCSK9 cDNAs with the zebrafish genomeallowed us to identify a zebrafish ortholog using a combi-nation of Ensembl and University of California, Santa Cruz(UCSC) annotations. The Ensembl gene from nucleotides162 004–191 952 coding for zf-NARC-1/PCSK9 can befound at http://www.ensembl.org/Danio_rerio/geneview?gene¼ENSDARG00000037996;db¼core. The UCSC zebra-fish 5606-bp annotation was found on chromosome 7 (http://genome.ucsc.edu/cgi-bin/hgTracks?hgsid¼60551137&hgt.right1¼+%3E+&position¼chr7%3A14772303-14775302).Thus, a single NARC-1/PCSK9 gene was detected in thezebrafish genome, which exhibited 68% protein identitywithin the catalytic subunit of its mouse ortholog. Thecomplete assembled sequence of zf-NARC-1/PCSK9, to-gether with its predicted signal peptide and post-translationalmodifications, are shown in Fig. S3.

Based on this sequence information, we analyzed byRT–PCR the developmental expression of zf-NARC-1/

(a)

(b)

(c)

Fig. 4 Cholesterol-independent role of NARC-1/PCSK9 in P19 cells.

(a) QPCR analysis of the relative HMG-CoA reductase (HMGCR)

mRNA levels. (b) Immunoblot analysis of LDLR levels on days 0–4

in RA-treated P19 cells. The migration positions of the LDLR

(� 160 kDa) and the internal standard b-actin (� 42 kDa) are shown.

(c) Comparative ISH of mouse NARC-1/PCSK9 and LDLR at P1. SG,

spinal ganglia; DRG, dorsal root ganglia; TN, trigeminal nerve.

844 S. Poirier et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850� 2006 The Authors

PCSK9 (Fig. 5a). The data showed that zf-NARC-1/PCSK9 transcripts were detectable at the three-somitestage, coinciding with the onset of cell fate acquisitionwithin proneural domains (neurogenesis occurs in thethree-to-five somite stage) (Westerfield 2000; Korzh et al.2001; Lossi and Merighi 2003). Using ISH with a 1323-base zf-NARC-1/PCSK9-specific cRNA probe (Fig. S3),we analyzed the expression of this convertase at the six-somite stage (Fig. 5b) and 4–5 days post-fertilization (dpf)periods (Fig. 5c). At 12 hpf (six somites) zf-NARC-1/PCSK9 was ubiquitously expressed throughout the epi-blast, with higher levels within the presumptive notochord.At 4–5 dpf the enzyme was highly expressed in liver andintestine, as in mouse and rat (Seidah et al. 2003). Withinthe neural network, it was found in centers of continuedneurogenesis such as cortical and intracranial neurons andcerebellar granule cell precursors (Fig. 5c). Thus, NARC-1/PCSK9 is expressed in specialized neurogenic centers inboth zebrafish and mammals.

Knockdown of NARC-1/PCSK9 in zebrafish

We next used the technology of MO-based translationinhibition of NARC-1/PCSK9 using the zebrafish as a model

of vertebrate development. From the deduced assembledsequence of zf-NARC-1/PCSK9 mRNA, we identified aunique sequence comprising the initiator methionine whichwas used to synthesize an antisense MO (MO1) (Fig. S3). Inorder to avoid non-specific gene targeting, we performedBlast analyses of zebrafish genomic sequences and confirmedthe uniqueness of the chosen MO1. Furthermore, to test thespecificity of the MOs we generated an affinity-purifiedpolyclonal antibody against the N-terminal 144–156 peptidesequence of the catalytic subunit of zf-NARC-1/PCSK9(sequence S144IPWNLQRVLQNK156; Fig. S3). Using thisantibody we performed western blots of single embryos andshowed that injection of MO1 reduced the protein level ofendogenous zf-NARC-1/PCSK9 by � 60%, whereas theother two control MOs, including an MO1 5-bp mismatch(MO1-mm) and a standard scramble control (Scramble Ctrl),had no effect (Fig. 6a). Injection of MO1 in zebrafish eggsresulted in CNS degeneration with high penetrance (Fig. 6b),compared with findings for the MO1-mm and Scramblecontrol MOs, which yielded much decreased phenotypicpenetrance or normal CNS development respectively(Fig. 6b). At 24 hpf a small cleft, presumed to be thehindbrain, represented the only discernable CNS architecturewithin morphant embryos (Figs 6c and d). By 48 hpfabnormal neurogenesis continued, creating multiple neuralchambers in an anterior–posterior orientation (Figs 6e and f).These defects culminated in lethality between 48 and 96 hpf(Figs 6g and h). These data suggest that NARC-1/PCSK9 isan essential gene in zebrafish.

Discussion

Initial studies showed that NARC-1/PCSK9 is transientlyexpressed during embryonic development in neurogeniccenters such as those in the telencephalon and cerebellum,and that it is no longer expressed in mature CNS neurons ofrodents (Seidah et al. 2003). We undertook a study of thepossible role of this convertase in neuroectodermal differen-tiation using P19 cells and zebrafish as models. The datarevealed that upon RA induction of P19 cells leading toneurons and glial cells, NARC-1/PCSK9 mRNA levels peakat day 2 and fall off thereafter, a profile quite different fromthat of the other convertases (Fig. 3). This suggests thatNARC-1/PCSK9 may have a unique role at the onset of theneuronal/neuroectodermal differentiation process. Interest-ingly, a minor but statistically significant peak of PC5/6mRNA at day 2 was also observed, reminiscent of theup-regulation of both NARC-1/PCSK9 and PC5/6 duringliver regeneration following partial hepatectomy in rat(Seidah et al. 2003). Thus, it may well be that NARC-1/PCSK9 and PC5/6 have complementary and/or additive rolesin the predifferentiation period.

We next attempted to elucidate the mechanism behind theup-regulation of NARC-1/PCSK9 at day 2 in P19 cells. This

(a)

(b)

(c)

Fig. 5 Developmental expression of zf-NARC-1/PCSK9. (a) RT–PCR

analysis of NARC-1/PCSK9 expression in zebrafish during gastrula-

tion/segmentation. Actin was used as an internal standard. In situ

hybridization of zf-NARC-1/PCSK9 (b) 12 hpf at the six-somite stage,

and (c) at 4 and 5 dpf.

NARC-1/PCSK9 and neural development 845

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850

effect was only observed following RA treatment of eitheradherent cells or aggregated cells in suspension, but not inuntreated controls. Accordingly, we deduced that the aggre-gation step is not responsible for the observed up-regulation,but rather that RA is the inducing agent. In the nucleus, theRA signal is transduced by binding to a heterodimeric pair ofretinoid receptors, retinoic acid receptor/retinoid X receptor(Means and Gudas 1995). Because HepG2 cells, whichexpress both receptors (Denson et al. 2000), do not up-regulate the NARC-1/PCSK9 transcripts when treated withRA (Fig. S2), it is likely that the effect of RA in P19 cells isindirect, possibly involving activation of intervening proteinsor transcription factors (Wei et al. 2002), which may then up-regulate NARC-1/PCSK9 mRNA levels. Furthermore, noapparent RA-responsive element could be identified in theproximal � 2 kb of the mouse and human PCSK9 promoters(not shown).

Because NARC-1/PCSK9 transcription is highlyup-regulated by SREBP-2 (Horton et al. 2003; Dubuc et al.2004), we next turned our attention to the effect of RA onSREBP-2 transcription, and found that it was unchangedduring the differentiation programme (not shown). However,because it is mostly the nuclear nSREBP-2 protein that is theactive ingredient, it is plausible that the activation of theGolgi form of SREBP-2 into its nuclear form nSREBP-2 bythe concerted action of SKI-1/S1P and S2P (Brown et al.2000) may be limiting. Our data revealed that the transcrip-tion of SKI-1/S1P was not regulated by RA in P19 cells(Fig. 3). Furthermore, the mRNA levels of the limiting HMG-CoA reductase, which critically depends on nSREBP-2activity, only increased from days 5–10, indicating thatnSREBP-2 is not the key NARC-1/PCSK9 transcriptionfactor at day 2 (Fig. 3).We cannot at the present time eliminatethe possibility that the peak observed at day 2 may be due to an

(a) (b)

(c)

(e)

(f)

(g)

(h)

(d)

Fig. 6 Validation of morpholino-based

translation inhibition and phenotypic con-

sequences of zf-NARC-1/PCSK9 knock-

down. MOs directed against the starter

methionine of zf-NARC-1/PCSK9 (MO1), its

5-bp mismatch (MO1-mm) and scramble

control (Scramble Ctrl) at 10 ng each were

injected into one- to two-cell stage embryos

and overall morphology assessed at 24 hpf.

(a) Knockdown of zf-NARC-1/PCSK9 pro-

tein following morpholino injection in vivo.

Western blot analysis of embryos injected

with 10 ng zf-NARC-1/PCSK9 or control

morpholinos using rabbit zf-NARC-1/

PCSK9 polyclonal antibody (upper panel)

and anti-mouse actin monoclonal antibody

as a loading control (lower panel). Protein

extracted from single 24 hpf embryo was

loaded on to each lane. WT, uninjected. (b)

CNS degeneration was observed with the

following frequencies: 50 of 54 (MO1), eight

of 48 (MO1-mm) and none of 50 (Scramble

Ctrl), expressed as percentages in the his-

togram. (c) WT and (d) MO1-injected

embryos at 24 hpf displayed defective

neurogenesis including absence of tectum

(red arrow), midbrain–hindbrain boundary

(arrowhead) and decreased hindbrain

architecture (black arrow and asterisk).

Comparison of (e) WT and (f) MO1-injected

embryos at 48 hpf indicated the presence of

further abnormalities in neural develop-

ment. (e–h) These defects culminated in

lethality between 48 and 96 hpf.

846 S. Poirier et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850� 2006 The Authors

enhanced specific stabilization of NARC-1/PCSK9 mRNA,possibly via specific interactions with its 5¢ untranslatedregion, as was the case for insulin, PC1 and PC2 mRNAs athigh glucose levels (Schuppin and Rhodes 1996), rather thandirect transcriptional activation. However, this mechanismwould have to be quite restrictive because mRNA levels of thecholesterogenic genes encodingHMG-CoA reductase, LDLR,SKI-1/S1P and SREBP-2 did not peak at day 2.

The only attributable function of NARC-1/PCSK9 isenhancement of degradation of the LDLR in liver andhepatocyte-derived cell lines (Benjannet et al. 2004; Max-well and Breslow 2004; Park et al. 2004; Rashid et al.2005). However, the mechanism behind this effect is notunderstood, as it does not seem to result from direct cleavageof the LDLR by NARC-1/PCSK9 in an acidic compartment(Attie and Seidah 2005). This suggests that in liver NARC-1/PCSK9 may activate and/or change the trafficking of anotherprotein(s) that is responsible for the degradation of theLDLR. Because the protein level of the LDLR in P19 cellstreated with RA did not change from days 0–4 (Fig. 4b),even though NARC-1/PCSK9 was up-regulated at day 2, thisindicates that the cellular conditions in P19 cells may not beappropriate for LDLR-enhanced degradation by NARC-1/PCSK9. This led us to conclude that, similar to findings inChinese hamster ovary cells (Park et al. 2004), the NARC-1/PCSK9-induced enhanced degradation of the LDLR is notefficient in P19 cells, and that NARC-1/PCSK9 may haveother function(s) in these cells. Indeed, comparative ISH ofNARC-1/PCSK9 and LDLR in consecutive sections of awhole mouse on the first day after birth (P1) revealed bothdistinct and overlapping expression patterns (Fig. 4c). Thus,both mRNAs were co-localized in liver, intestine and kidney,whereas the high level of expression of NARC-1/PCSK9 incerebellum did not coincide with that of the LDLR, which isgenerally not abundant in the CNS. Therefore, NARC-1/PCSK9 may have new functions in the nervous system,possibly related to neurogenesis, among others (Seidah et al.2003). This may involve the NARC-1/PCSK9-enhanceddegradation of other receptor types or proteins duringdevelopment of the cerebellum and telencephalon.

We decided to study such function(s) within the realm of awhole animal such as the zebrafish, which represents a wellstudied experimental model as it develops externally withvisual clarity. These qualities, together with antisense-basedtranslation inhibition techniques, allow visual phenotypicassessment of targeted gene knockdown. Mining of zebrafishgenomic databases allowed us to put together a completeprimary structure of zf-NARC-1/PCSK9 (Fig. S3). Thededuced sequence predicts a 667-amino acid proprotein witha 22-amino acid signal peptide (amino acids 1–22), and a121-amino acid prosegment (amino acids 23–143), resultingin a 525-amino acid mature protein. The zymogen activationsite predicted at SSIFAQ143flSIPWN is very similar to thatfound in human and mouse, with the only conservative

variation of zebrafish Ile140 to Val in mammals. Notably, wealso deduced the possible presence of an extra N-glycosy-lation site within the prosegment of zf-NARC-1/PCSK9(Asn75) that is not found in mammalian orthologs. Finally,the C-terminal Cys/His-rich domain of zf-NARC-1/PCSK9 issimilar to that found in the human and mouse sequences,with the presence of two CysCys-X6-Cys motifs.

We used MOs selectively to inhibit zf-NARC-1/PCSK9translation (Fig. 6), to create morphant phenotypes and todocument the resulting visible morphological defects. Basedon murine NARC-1/PCSK9 gene expression patterns (Seidahet al. 2003), it was predicted that defects would be associatedwith neural and/or hepatic/intestinal/renal organogenesis. Inthe zebrafish, neurogenesis begins postgastrulation (approxi-mately 10 hpf) with the formation of the neural plate, aconserved vertebrate structure. The neural plate then under-goes anterior–posterior patterning during segmentation, cre-ating visible regionalized structures by 24 hpf (Schier et al.1996). Hepatic cell fate specification from the presumptiveanterior endoderm begins at approximately 16–18 hpf beforeprimitive gut tube formation (Korzh et al. 2001). However,the liver only acquires a histologically distinct architecture at34 hpf (Wallace and Pack 2003). At 24 hpf, embryosinjected with MO1 displayed defective neurogenesis(Fig. 6d) as compared with wild type (WT) (Fig. 6c). Thisincluded absence of tectum (red arrow), midbrain–hindbrainboundary (arrow head), as well as decreased hindbrainarchitecture (black arrow and asterisk) (Figs 6c and d).Comparison of WT (Fig. 6e) and MO1-injected (Fig. 6f)embryos at 48 hpf indicated the presence of further abnor-malities in neural development. The observed overall neuraldegeneration was followed by regionalization at 48 hpf intoseveral chambers of unknown identity (Fig. 6f). The knock-down of zf-NARC-1/PCSK9 expression resulted in an� 60% decreased in protein levels, but not complete loss ofprotein (Fig. 6a). The specific phenotypes observed suggestthat in zebrafish neurogenesis is highly sensitive to alteredNARC-1/PCSK9 expression. This is reminiscent of theeffects observed in human NARC-1/PCSK9 heterozygotemissense mutations (50% loss of function), resulting infamilial hypocholesterolemia (Cohen et al. 2005; Kotowskiet al. 2006). It remains unclear at this time whether hepaticspecification or liver organogenesis is affected in thezf-NARC-1/PCSK9 morphants. Further investigation bymarker gene in situ and histological analysis is required.

The refractile (as opposed to transparent) appearance ofneural tissue in morphants is a hallmark of cell death,particularly associated with neural degeneration, and wasused initially to identify neuron survival mutants in large-scale zebrafish chemical mutagenesis screens (Furutani-Seikiet al. 1996). Although not visibly evident, apoptosis is anessential component of normal neurogenesis, wherebyneuronal pathways form via axonal extension in a ‘first-come-first-served’ basis, requiring redundant projections to

NARC-1/PCSK9 and neural development 847

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850

undergo programmed cell death (Lossi and Merighi 2003).The close association between differentiation and apoptosisduring neurogenesis may confer increased sensitivity tocellular alterations, particularly those affecting the cell cycleand metabolism (Furutani-Seiki et al. 1996). Interestingly,the known autosomal dominant natural human NARC-1/PCSK9 mutations resulting in hypercholesterolemia placethis gene within such a metabolic context (Abifadel et al.2003; Benjannet et al. 2004; Timms et al. 2004; Attie andSeidah 2005).

While this work was in progress, the complete knockout ofNARC-1/PCSK9 in mouse was reported (Rashid et al.2005). The latter work confirmed earlier studies, showingthat NARC-1/PCSK9 enhances degradation of the LDLR; itsloss would therefore result in higher hepatic LDLR levels(Benjannet et al. 2004; Maxwell and Breslow 2004; Parket al. 2004). Nevertheless, it was astonishing to discover thatin mouse the loss of NARC-1/PCSK9 expression does notresult in a lethal phenotype (Rashid et al. 2005), as is thecase in zebrafish (this work).

The association of NARC-1/PCSK9 with LDLR degrada-tion is insufficient to explain the zebrafish morphant defectsbecause LDLR–/– mice are viable (Ishibashi et al. 1993),suggesting that this PC embodies biological activities beyondLDLR homeostasis, as also suggested by its expression inbrain areas lacking the LDLR (Fig. 4c). Nevertheless,differences do exist between the mouse and zebrafish,including the absence of placenta and external embryonicgrowth in the fish. In this context, a recent study on Vephshowed that the knockdown of this transiently expressed CNSprotein in neurogenesis centers results in a lethal phenotype inzebrafish but not in Veph–/– mice (Muto et al. 2004). Deathin developing zebrafish lacking Veph transcripts seems tofollow defects in the midbrain–hindbrain boundary and oticvesicle formation. A similar phenotype was observed uponknockdown of zf-NARC-1/PCSK9, with defective neurogen-esis including loss of the tectum and midbrain–hindbrainboundary, as well as disorganized hindbrain architecture(Figs 6c and d). It remains to be seen whether NARC-1/PCSK9 has different functions in mammals and fish.Alternatively, the CNS of mammals may have developedmechanisms to compensate for the lack of NARC-1/PCSK9that are absent in teleosts. Detailed mapping of the functionaldomains of NARC-1/PCSK9 in zebrafish and mouse maylead to a better understanding of the various functions of thisfascinating convertase.

Acknowledgements

We would especially like to thank Johanne Duhaime for excellent

assistance in identifying and concatenation of the zf-NARC-1/

PCSK9 cDNA sequences. We also wish to thank Ann Chamberland

for QPCR analyses, Marie-Claude Asselin for cell culture and Josee

Hamelin for cDNA cloning of mouse LDLR. The authors are also

indebted to all the members of Dr Seidah’s laboratory for their

constant advice and help. The secretarial assistance of Mrs Brigitte

Mary is greatly appreciated. This work was supported in part by the

Canadian Institutes of Health Research grants: MOP-36496, a

Canadian Chair no. 201652, MGP-44363, and by a group grant

MGC-64518.

Supplementary material

The following supplementary material is available for this paper

online

Fig. S1 Immunocytochemistry of naive and untreated P19 cells.

The type of mAb used (NeuN or GFAP) is shown for either naive

P19 cells at day 0 (D0), or untreated (– RA) P19 cells at day 10

(D10). Left panels show the cells under visible light.

Fig. S2 QPCR of NARC-1/PCSK9 in HepG2 cells. HepG2 cells

were not treated (gray bars) or treated with RA for 4 days (black

bars) and QPCR was performed in triplicate for each day.

Fig. S3 cDNA and protein sequence deduced for zf-NARC-1/

PCSK9. The deduced sequence was deduced from concatenation of

zebrafish genomic and expressed sequence tag data. The predicted

signal peptide (amino acids 1–22) is underlined, together with the

active site. Asp177, His217, Ser377 and the oxyanion hole Asn309 are

emphasized. The zymogen activation site SSIFAQ143flSIPWN is

very similar to the SSVFAQ152flSIPWN site in human NARC-1/

PCSK9 (Seidah et al. 2003; Benjannet et al. 2004). The antigen usedfor antibody production is in bold and underlined. The position of the

predicted two N-glycosylation sites at Asn75,397 as well as the Tyr26

sulfation site are shown as shaded amino acids. The two oligonucle-

otides (sense and antisense) that were used to obtain the 1323-base

zf-NARC-1/PCSK9 cRNA by RT–PCR are boxed. Finally, the MO

sequence around the initiator 388ATG used in the knockdown

experiment is shaded. The stop codon is represented by an asterisk.

This material is available as part of the online article from http://

www.blackwell-synergy.com

References

Abifadel M., Varret M., Rabes J. P. et al. (2003) Mutations in PCSK9cause autosomal dominant hypercholesterolemia. Nat. Genet. 34,154–156.

Allard D., Amsellem S., Abifadel M. et al. (2005) Novel mutations ofthe PCSK9 gene cause variable phenotype of autosomal dominanthypercholesterolemia. Hum. Mutat. 26, 497.

Attie A. D. and Seidah N. G. (2005) Dual regulation of the LDL recep-tor – some clarity and new questions. Cell Metab. 1, 290–292.

Benjannet S., Rondeau N., Day R., Chretien M. and Seidah N. G. (1991)PC1 and PC2 are proprotein convertases capable of cleaving pro-opiomelanocortin at distinct pairs of basic residues. Proc. NatlAcad. Sci. USA 88, 3564–3568.

Benjannet S., Rondeau N., Paquet L., Boudreault A., Lazure C., ChretienM. and Seidah N. G. (1993) Comparative biosynthesis, covalentpost-translational modifications and efficiency of prosegmentcleavage of the prohormone convertases PC1 and PC2: glycosy-lation, sulphation and identification of the intracellular site ofprosegment cleavage of PC1 and PC2. Biochem. J. 294, 735–743.

Benjannet S., Elagoz A., Wickham L. et al. (2001) Post-translationalprocessing of beta-secretase (beta-amyloid-converting enzyme) andits ectodomain shedding. The pro- and transmembrane/cytosolic

848 S. Poirier et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850� 2006 The Authors

domains affect its cellular activity and amyloid-beta production.J. Biol. Chem. 276, 10 879–10 887.

Benjannet S., Rhainds D., Essalmani R. et al. (2004) NARC-1/PCSK9and its natural mutants: zymogen cleavage and effects on the lowdensity lipoprotein (LDL) receptor and LDL cholesterol. J. Biol.Chem. 279, 48 865–48 875.

Berman Y., Mzhavia N., Polonskaia A., Furuta M., Steiner D. F., PintarJ. E. and Devi L. A. (2000) Defective prodynorphin processing inmice lacking prohormone convertase PC2. J. Neurochem. 75,1763–1770.

Breslin M. B., Lindberg I., Benjannet S., Mathis J. P., Lazure C. andSeidah N. G. (1993) Differential processing of proenkephalin byprohormone convertases 1(3) and 2 and furin. J. Biol. Chem. 268,27 084–27 093.

Brown M. S. and Goldstein J. L. (1999) A proteolytic pathway thatcontrols the cholesterol content of membranes, cells, and blood.Proc. Natl Acad. Sci. USA 96, 11 041–11 048.

Brown M. S., Ye J., Rawson R. B. and Goldstein J. L. (2000) Regulatedintramembrane proteolysis: a control mechanism conserved frombacteria to humans. Cell 100, 391–398.

Cohen J., Pertsemlidis A., Kotowski I. K., Graham R., Garcia C. K. andHobbs H. H. (2005) Low LDL cholesterol in individuals of Africandescent resulting from frequent nonsense mutations in PCSK9.Nat. Genet. 37, 161–165.

Decroly E., Benjannet S., Savaria D. and Seidah N. G. (1997) Com-parative functional role of PC7 and furin in the processing of theHIV envelope glycoprotein gp160. FEBS Lett. 405, 68–72.

Denis F., Shoukry N. H., Delcourt M., Thibodeau J., Labrecque N.,McGrath H., Munzer J. S., Seidah N. G. and Sekaly R. P. (2000)Alternative proteolytic processing of mouse mammary tumor virussuperantigens. J. Virol. 74, 3067–3073.

Denson L. A., Auld K. L., Schiek D. S., McClure M. H., MangelsdorfD. J. and Karpen S. J. (2000) Interleukin-1beta suppresses retinoidtransactivation of two hepatic transporter genes involved in bileformation. J. Biol. Chem. 275, 8835–8843.

Dubuc G., Chamberland A., Wassef H., Davignon J., Seidah N. G.,Bernier L. and Prat A. (2004) Statins upregulate PCSK9, the geneencoding the proprotein convertase neural apoptosis-regulatedconvertase-1 implicated in familial hypercholesterolemia. Arteri-oscler. Thromb. Vasc. Biol. 24, 1454–1459.

Furutani-Seiki M., Jiang Y. J., BrandM. et al. (1996) Neural degenerationmutants in the zebrafish, Danio rerio. Development 123, 229–239.

Galanopoulou A. S., Seidah N. G. and Patel Y. C. (1995) Heterologousprocessing of rat prosomatostatin to somatostatin-14 by PC2:requirement for secretory cell but not the secretion granule.Biochem. J. 311, 111–118.

Horton J. D., Goldstein J. L. and Brown M. S. (2002) SREBPs: acti-vators of the complete program of cholesterol and fatty acid syn-thesis in the liver. J. Clin. Invest. 109, 1125–1131.

Horton J. D., Shah N. A., Warrington J. A., Anderson N. N., Park S. W.,Brown M. S. and Goldstein J. L. (2003) Combined analysis ofoligonucleotide microarray data from transgenic and knockoutmice identifies direct SREBP target genes. Proc. Natl Acad. Sci.USA 100, 12 027–12 032.

Ishibashi S., Brown M. S., Goldstein J. L., Gerard R. D., Hammer R. E.and Herz J. (1993) Hypercholesterolemia in low density lipopro-tein receptor knockout mice and its reversal by adenovirus-medi-ated gene delivery. J. Clin. Invest. 92, 883–893.

Jeannotte R., Paquin J., Petit-Turcotte C. and Day R. (1997) ConvertasePC2 and the neuroendocrine polypeptide 7B2 are co-induced andprocessed during neuronal differentiation of P19 embryonal car-cinoma cells. DNA Cell Biol. 16, 1175–1187.

Kalus I., Schnegelsberg B., Seidah N. G., Kleene R. and Schachner M.(2003) The proprotein convertase PC5A and a metalloprotease are

involved in the proteolytic processing of the neural adhesionmolecule L1. J. Biol. Chem. 278, 10 381–10 388.

Khatib A. M., Siegfried G., Chretien M., Metrakos P. and Seidah N. G.(2002) Proprotein convertases in tumor progression andmalignancy:novel targets in cancer therapy. Am. J. Pathol. 160, 1921–1935.

Korzh S., Emelyanov A. and Korzh V. (2001) Developmental analysis ofceruloplasmin gene and liver formation in zebrafish. Mech. Dev.103, 137–139.

Kotowski I. K., Pertsemlidis A., Luke A., Cooper R. S., Vega G. L.,Cohen J. C. and Hobbs H. H. (2006) A spectrum of PCSK9 allelescontributes to plasma levels of low-density lipoprotein cholesterol.Am. J. Hum. Genet. 78, 410–422.

Laplante I., Beliveau R. and Paquin J. (2004) RhoA/ROCK and Cdc42regulate cell-cell contact and N-cadherin protein level duringneurodetermination of P19 embryonal stem cells. J. Neurobiol. 60,289–307.

Lenz O., ter Meulen J., Klenk H. D., Seidah N. G. and Garten W. (2001)The Lassa virus glycoprotein precursor GP-C is proteolyticallyprocessed by subtilase SKI-1/S1P. Proc. Natl Acad. Sci. USA 98,12 701–12 705.

Lossi L. and Merighi A. (2003) In vivo cellular and molecular mecha-nisms of neuronal apoptosis in the mammalian CNS. Prog. Neu-robiol. 69, 287–312.

Marcinkiewicz M., Marcinkiewicz J., Chen A., Leclaire F., Chretien M.and Richardson P. (1999) Nerve growth factor and proproteinconvertases furin and PC7 in transected sciatic nerves and in nervesegments cultured in conditioned media: their presence in Schwanncells, macrophages, and smooth muscle cells. J. Comp. Neurol.403, 471–485.

Maxwell K. N. and Breslow J. L. (2004) Adenoviral-mediatedexpression of Pcsk9 in mice results in a low-density lipoproteinreceptor knockout phenotype. Proc. Natl Acad. Sci. USA 101,7100–7105.

Maxwell K. N., Soccio R. E., Duncan E. M., Sehayek E. and BreslowJ. L. (2003) Novel putative SREBP and LXR target genes identi-fied by microarray analysis in liver of cholesterol-fed mice. J. LipidRes. 44, 2109–2119.

Mbikay M., Tadros H., Ishida N. et al. (1997) Impaired fertility in micedeficient for the testicular germ-cell protease PC4. Proc. Natl Acad.Sci. USA 94, 6842–6846.

McBurney M. W. (1993) P19 embryonal carcinoma cells. Int. J. Dev.Biol. 37, 135–140.

Means A. L. and Gudas L. J. (1995) The roles of retinoids in vertebratedevelopment. Annu. Rev. Biochem. 64, 201–233.

Muto E., Tabata Y., Taneda T., Aoki Y., Muto A., Arai K. and WatanabeS. (2004) Identification and characterization of Veph, a novel geneencoding a PH domain-containing protein expressed in the devel-oping central nervous system of vertebrates. Biochimie 86, 523–531.

Nasevicius A. and Ekker S. C. (2000) Effective targeted gene ‘knock-down’ in zebrafish. Nat. Genet. 26, 216–220.

Park S. W., Moon Y. A. and Horton J. D. (2004) Post-transcriptionalregulation of low density lipoprotein receptor protein by proproteinconvertase subtilisin/kexin type 9a in mouse liver. J. Biol. Chem.279, 50 630–50 638.

Pisciotta L., Oliva C. P., Cefalu A. B. et al. (2005) Additive effect ofmutations in LDLR and PCSK9 genes on the phenotype of familialhypercholesterolemia. Atherosclerosis 186, 443–450.

Rashid S., Curtis D. E., Garuti R., Anderson N. N., Bashmakov Y., HoY. K., Hammer R. E., Moon Y. A. and Horton J. D. (2005)Decreased plasma cholesterol and hypersensitivity to statins inmice lacking Pcsk9. Proc. Natl Acad. Sci. USA 102, 5374–5379.

Roebroek A. J., Umans L., Pauli I. G., Robertson E. J., van Leuven F.,Van de Ven W. J. and Constam D. B. (1998) Failure of ventral

NARC-1/PCSK9 and neural development 849

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850

closure and axial rotation in embryos lacking the proprotein con-vertase Furin. Development 125, 4863–4876.

Rudnicki M. A., Reuhl K. R. and McBurney M. W. (1989) Cell lineswith developmental potential restricted to mesodermal lineagesisolated from differentiating cultures of pluripotential P19 embry-onal carcinoma cells. Development 107, 361–372.

Sakai J., Nohturfft A., Goldstein J. L. and Brown M. S. (1998a) Clea-vage of sterol regulatory element-binding proteins (SREBPs) atsite-1 requires interaction with SREBP cleavage-activating protein.Evidence from in vivo competition studies. J. Biol. Chem. 273,5785–5793.

Sakai J., Rawson R. B., Espenshade P. J., Cheng D., Seegmiller A. C.,Goldstein J. L. and Brown M. S. (1998b) Molecular identificationof the sterol-regulated luminal protease that cleaves SREBPs andcontrols lipid composition of animal cells. Mol. Cell 2, 505–514.

Schier A. F., Neuhauss S. C., Harvey M. et al. (1996) Mutationsaffecting the development of the embryonic zebrafish brain.Development 123, 165–178.

Schulte-Merker S., Ho R. K., Herrmann B. G. and Nusslein-Volhard C.(1992) The protein product of the zebrafish homologue of themouse T gene is expressed in nuclei of the germ ring and thenotochord of the early embryo. Development 116, 1021–1032.

Schuppin G. T. and Rhodes C. J. (1996) Specific co-ordinated regulationof PC3 and PC2 gene expression with that of preproinsulin ininsulin-producing beta TC3 cells. Biochem. J. 313, 259–268.

Scopsi L., Gullo M., Rilke F., Martin S. and Steiner D. F. (1995) Pro-protein convertases (PC1/PC3 and PC2) in normal and neoplastichuman tissues: their use as markers of neuroendocrine differenti-ation. J. Clin. Endocrinol. Metab. 80, 294–301.

Seidah N. G. and Chretien M. (1999) Proprotein and prohormone con-vertases: a family of subtilases generating diverse bioactive poly-peptides. Brain Res. 848, 45–62.

Seidah N. G. and Prat A. (2002) Precursor convertases in the secretorypathway, cytosol and extracellular milieu. Essays Biochem. 38, 79–94.

Seidah N. G., Gaspar L., Mion P., Marcinkiewicz M., Mbikay M. andChretien M. (1990) cDNA sequence of two distinct pituitary pro-teins homologous to Kex2 and furin gene products: tissue-specificmRNAs encoding candidates for pro- hormone processing pro-teinases. DNA Cell Biol. 9, 789.

Seidah N. G., Mowla S. J., Hamelin J. et al. (1999) Mammalian sub-tilisin/kexin isozyme SKI-1: a widely expressed proprotein con-vertase with a unique cleavage specificity and cellular localization.Proc. Natl Acad. Sci. USA 96, 1321–1326.

Seidah N. G., Benjannet S., Wickham L., Marcinkiewicz J., JasminS. B., Stifani S., Basak A., Prat A. and Chretien M. (2003) Thesecretory proprotein convertase neural apoptosis-regulated con-vertase 1 (NARC-1): liver regeneration and neuronal differenti-ation. Proc. Natl Acad. Sci. USA 100, 928–933.

Thisse C., Thisse B., Schilling T. F. and Postlethwait J. H. (1993)Structure of the zebrafish snail 1 gene and its expression in wild-type, spadetail and no tail mutant embryos. Development 119,1203–1215.

Thomas G. (2002) Furin at the cutting edge: from protein traffic toembryogenesis and disease. Nat. Rev. Mol. Cell Biol. 3, 753–766.

Timms K. M., Wagner S., Samuels M. E., Forbey K., Goldfine H.,Jammulapati S., Skolnick M. H., Hopkins P. N., Hunt S. C. andShattuck D. M. (2004) A mutation in PCSK9 causing autosomal-dominant hypercholesterolemia in a Utah pedigree. Hum. Genet.114, 349–353.

Wallace K. N. and Pack M. (2003) Unique and conserved aspects of gutdevelopment in zebrafish. Dev. Biol. 255, 12–29.

Wei Y., Harris T. and Childs G. (2002) Global gene expression patternsduring neural differentiation of P19 embryonic carcinoma cells.Differentiation 70, 204–219.

Westerfield M. (2000) The Zebrafish Book. A Guide for the LaboratoryUse of Zebrafish (Danio Rerio). University of Oregon Press,Eugene.

Zhou A.,Webb G., Zhu X. and Steiner D. F. (1999) Proteolytic processingin the secretory pathway. J. Biol. Chem. 274, 20 745–20 748.

850 S. Poirier et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 838–850� 2006 The Authors