Probing platelet factor 4 alpha-granule targeting

ORIGINAL ARTICLE

Probing platelet factor 4 a-granule targeting

V. BR IQUET-LAUGIER ,* 1 C. LAVENU-BOMBLED ,� 1 A. SCHMITT ,� M. LEBOEUF ,� G. UZAN,�A. DUBART-KUPPERSCHMITT� and J . - P . ROSA**U348 INSERM, IFR6-Circulation, Hopital Lariboisiere, Paris; �Department of Hematology, Hopital de Port-Royal, Maternite de Port-Royal,

Institut Cochin, Paris; �U506 INSERM, Hopital Paul Brousse, Villejuif, France

To cite this article: Briquet-Laugier V, Lavenu-Bombled C, Schmitt A, Leboeuf M, Uzan G, Dubart-Kupperschmitt A, Rosa J-P. Probing platelet

factor 4 a-granule targeting. J Thromb Haemost 2004; 2: 2231–40.

Summary. The storage mechanism of endogenous secretory

proteins inmegakaryocyte a-granules is poorly understood.We

have elected to study the granule storage of platelet factor 4

(PF4), a well-known platelet a-granule protein. The reporter

protein green fluorescent protein (GFP), PF4, or PF4 fused to

GFP (PF4-GFP), were transfected in the well-characterized

mouse pituitary AtT20 cell line, and in the megakaryocytic

leukemic DAMI cell line. These proteins were also transduced

using a lentiviral vector, in human CD34+ cells differentiated

into megakaryocytes in vitro. Intracellular localization of

expressed proteins, and colocalization studies were achieved

by laser scanning confocal microscopy and immuno-electron-

microscopy. In preliminary experiments, GFP, a non-secretory

protein (no signal peptide), localized in the cytoplasm, while

PF4-GFP colocalized with adrenocorticotropin hormone

(ACTH)-containing granules in AtT20 cells. In the megakary-

ocytic DAMI cell line and in human megakaryocytes differen-

tiated in vitro, PF4-GFP localized in a-granules along with the

alpha granular protein von Willebrand factor (VWF). The

signal peptide of PF4 was not sufficient to specify a-granulestorageofPF4, sincewhenPF4 signal peptidewas fused toGFP

(SP4-GFP), GFP was not stored into granules in spite of its

efficient translocation to the ER-Golgi constitutive secretory

pathway. We conclude that the PF4 storage pathway in

a-granules is not a default pathway, but rather a regular granulestorage pathway probably requiring specific sorting mecha-

nisms. In addition PF4-GFP appears as an appropriate probe

withwhich to analyze a-granule biogenesis and its alterations inthe congenital defect gray platelet syndrome.

Keywords: green fluorescent protein intracellular targeting,

megakaryocyte alpha granule, von Willebrand factor.

Introduction

Regulated secretion, i.e. the pathway leading to granule storage

of secretory proteins and their release from cells upon specific

stimuli, is central to a number of functions in vascular biology:

release of growth factors or chemokines by inflammatory cells,

of a-granule proteins by platelets, or of von Willebrand factor

(VWF) from Weibel–Palade bodies in endothelial cells. Secre-

tory proteins follow the secretory pathway and are thus

sequentially translocated into the ER via a signal peptide-

dependent machinery [1], followed by export to the Golgi for

maturation [2], till they reach the last compartment, the trans-

Golgi network (TGN) where they undergo sorting [3,4].

Secretory proteins not stored away are constitutively released

into the medium, hence the �constitutive secretory pathway�.Other secretory proteins are sorted and stored in specific

compartments: for example hydrolases are sorted to lysosomes,

through a mannose-6-phosphate receptor-dependent mechan-

ism [5]. Non-lysosomal secretory proteins are sorted to storage

granules through a distinct pathway not involving the

mannose-6-phosphate receptor, but the mechanism of which

is still not fully understood. Several models are proposed,

including a mechanism based on a sorting receptor [6],

targeting specifically secretory proteins to storage granules, or

a �retention model� in which all proteins exit the TGN in

immature granules, from which non-granular proteins are

�passively� excluded [4,7].

In the megakaryocytic-platelet lineage, secretory proteins are

stored in a-granules and lysosomes. The a-granule contains

adhesive proteins (such as VWF, fibronectin…), growth factors

(such as PDGF, VEGF…), or chemokines [such as the platelet

factor 4 (PF4), the b-thromboglobulin] [8]. a-granules also

contain proteins not synthesized by megakaryocytes such as

IgGs, albumin or fibrinogen, which are endocytosed from

plasma, strongly suggesting that a-granules result from fusion

of endosomes and TGN-derived vesicles [9]. The storage

mechanism of a-granule proteins is unknown. For example, it

is not known whether a-granule storage is a megakaryocytic

default pathway for any secretory protein, or if it is a sorting

Correspondence: J.-P. Rosa, U348 INSERM, Hopital Lariboisiere, 41

boulevard de la Chapelle, 75475 PARIS Cedex 10, France.

Tel.: +33 15 320 3788; fax: +33 14 99 58579; e-mail: rosa@arib.

inserm.fr

Received 12 September 2003, accepted 11 August 2004

1These authors contributed equally to this work.

Journal of Thrombosis and Haemostasis, 2: 2231–2240

� 2004 International Society on Thrombosis and Haemostasis

pathway for specific secretory proteins. Hence the molecular

basis of the gray platelet syndrome, a hereditary bleeding

disorder characterized by an absence of megakaryocyte

a-granule endogenous proteins [10,11], is unknown. It has

been proposed that the gray platelet syndrome stems from

alteration of the delivery machinery (sorting receptor?) of

endogenous proteins to the a-granule [10], though no experi-

mental support was provided. Clearly a better understanding of

secretory granule storage biogenesis, which would undoubtedly

help understand the molecular basis of the gray platelet

syndrome, is awaited in platelet biology.

In this paper we report the first assessment of the trafficking

of the a-granular protein PF4. We have chosen PF4 as a

paradigm because of its small size (70 amino acid-long mature

sequence), and its known 3D structure [12], allowing future

structure–function predictions. GFP [13] was chosen as a

reporter protein, for ease of detection (fluorescence) and as it is

non-secretory per se (cytosolic protein, without signal peptide).

We demonstrate that PF4 fused to GFP (PF4-GFP) follows

the same pathway than PF4 alone, and is stored into granules

in ACTH-producing pituitary AtT20 mouse cells, in the

megakaryocytic DAMI cell line and in human megakaryo-

cytes. Finally we demonstrate by confocal microscopy and

immunoelectron microscopy, that targeting of PF4-GFP is

a-granule-specific in primary megakaryocytes, based on (a)

colocalization with VWF in morphologically identified

a granules, (b) absence of colocalization with the lysosomal

marker Lamp-1, (c) efficient secretion of PF4-GFP upon

thrombin stimulation. Importantly, we show thatGFP fused to

the PF4 signal peptide enters the secretory pathway but is not

stored in a-granules, demonstrating that PF4 signal peptide is

not sufficient to specify PF4 a-granule storage. Thus, our studysuggests that a-granule storage is likely to be a specific sorting

pathway in megakaryocytes. We believe that our approach

may prove useful to the understanding of both the mechanism

of a-granule storage, and the gray platelet syndrome defect.

Methods

Materials

DMEM (Dulbecco’s modified Eagle’s medium) and RPMI

1640, both GLUTAMAX I 4500 mg L)1D-glucose, FCS,

penicillin–streptomycin,MEM(Invitrogen,Gibco-BRL,Cergy

Pontoise, France), RM B00 serum-free medium (RTM,

Tourcoing, France); AtT20 cells (American Type Cell Culture,

Manassas, VA, USA); recombinant human SCF (stem cell

factor, kindly provided by Amgen, Neuilly/Seine, France), and

(TPO thrombopoietin, PromoCell, Heidelberg, Germany);

MiniMacs CD34 magnetic cell isolation kit (Miltenyi-Biotec,

Paris, France); transfection reagent ExGen 500 (Euromedex,

Souffelweyersheim, France); phenylmethylsulfonyl fluoride,

leupeptin, aprotinin, protein A-sepharose beads, thrombin,

gelatin and phoibol 12-mynistate 13-acetate (PMA) (Sigma-

Aldrich, Saint-Quentin Fallavier, France); polyvinylidene

difluoride membranes, bicinchoninic acid reagent (Pierce,

Rockford, IL, USA); chemiluminescent substrate (Bio-Rad,

Hercules, CA, USA); affinity-purified polyclonal rabbit

antihuman PF4 IgG (Stago, Paris, France); Texas Red� dye-

conjugated donkey antirabbit IgG or Texas Red� dye-conju-

gated goat anti-mouse IgG (Jackson Immunoresearch, West

Grove, PA); Alexa�Fluor 488 goat antirabbit IgG, and mouse

monoclonal GFP antibody (Molecular Probes, Inc. Leiden, the

Netherlands); rabbit anti-ACTH (Serotec, Raleigh, NC,USA);

mouse monoclonal anti-Lamp-1 antibody (Becton Dickinson,

San Diego, CA, USA); mix of six mouse monoclonal VWF

antibodies (kind gift of D. Baruch); rabbit polyclonal VWF

antibodies (Dako, Glostrup, Denmark); pure GFP protein,

expression vector pCMV/EGFP-N1, and polyclonal anti-GFP

(Clontech, Palo Alto, CA); peroxidase-conjugated donkey

antirabbit IgG and photographic Hyperfilm-ECL (Amersham

Pharmacia Biotech Europe GmbH, Orsay, France); IgG

coupled to gold particles (British Biocell International, Cardiff,

United Kingdom); ExpandTM High Fidelity PCR system

(Roche-Boehringer, Meylan, France); confocal laser scanning

microscope Leica TCS 4D (Leica microsystems, Rueil-

Malmaison, France); and electron microscope CM 10 (Philips,

Eindhoven, the Netherlands).

PF4 and PF4-GFP constructs A partial human PF4 cDNA

clone was obtained from M. Poncz (Philadelphia, PA). This

cDNA lacked the 5¢-untranslated and the 5¢-most 41 bp

encoding the N-terminal part of the signal peptide sequence

of PF4. Full-length PF4 was reconstituted (see Fig. 1A) using

the following two complementary primers: A, coding:

5�-CCGCAGCATGAGCTCCGCAGCAGGGTTCTGTG

CCTCACGCCCCGGTTT-3�; B, complementary: 5�-CAG

CAAACCGGGGCGTGAGGCACAGAACCCTGCTGC

GGAGCTCATGCTGCGGTAC-3�. These were designed

with the purpose of reconstituting the PF4 upstream

sequence, which essentially overlaps the PF4 signal peptide

(open box, Fig. 1A). The dimer exhibits a KpnI-compatible

5¢ recessed end and an 3¢ cohesive end, complementary with

the 5¢ end of the rest of the PF4 cDNA excised from

M13mp18 by FokI, [a class IIS restriction endonuclease,

which leaves a four nt-long cohesive overhang downstream

from its DNA binding site (5¢GGATG)] [14]. Silent

substitutions (bold nt) were introduced to minimize secon-

dary structures. Oligonucleotides A and B were annealed

together in the presence of FokI-digested PF4 cDNA.

PF4DSTOP, i.e. full-length PF4 deleted from the stop

codon, was reconstituted by PCR using primers C and D,

complementary to the 5¢- and the 3¢- ends of PF4, which

created KpnI and BamHI sites, respectively. After Kpn1/

BamH1 double digestion, and gel purification, PF4DSTOP

was subcloned into the multiple cloning site of pCMV/

EGFP-N1 in-frame with GFP. The cloned construct was

sequenced in both directions.

PF4 signal peptide-GFP chimera (SP4-GFP) (Fig. 1).

SP4-GFP corresponds to the PF4 signal peptide (its cleavage

site being determined by a specific algorithm, see below) fused

to the GFP cDNA deleted of its ATG start codon. SP4-GFP

2232 V. Briquet-Laugier et al


was generated by PCR. Two primers corresponding to the

PF4 signal peptide sequence were designed: E (coding),

5�-ATGAGCTCCGCAGCAGGGTTCTGTGCCTCACG

CCCCGGTTTGCTGTTCCTGGGGTTGCTGCTCCTG

CCACTTGTG-3�; F (complementary), 5�-CTCCTCGCC

CTTGCTCACCCCATCTTCTTCAGCTTCAGCGCTG

GCGAAGGCGACCACAAGTGGCAGGAGCAG-3�.

These primers are partially complementary (underlined

overlapping sequences), and F is complementary to the 18most

5¢ nt (sequence in bold characters in primer D) of GFP cDNA

except for the ATG start codon of GFP, which was skipped.

The full-length SP4-GFP cDNA was generated by SOEing

PCR [15] using these 2 primers and GFP cDNA as template,

and primers G, forward, 5�-GGGGTACCGCAG

CATGAGCTCCGCAGCAGGG-3� and H, reverse, 5�-GCT

CTAGATTACTTGTACAGCTCGTC-3�, for amplification.

These primers created an 5¢ Kpn I site and an 3¢ XbaI site,

which allowed subcloning into the Clontech pCMV vector.

Signal peptide cleavage site prediction Signal peptide

cleavage site of PF4 is located between Ala 30 and Glu 31.

The prediction, derived from an algorithm accessible at http://

www.cbs.dtu.dk/services/SignalP [16], gave a signal peptide

cleavage site between Ala 32-Glu 33. The signal peptide

sequence was extended up to Gly 37 to preserve the sequence

environment of the signal peptidase cleavage site.

Lentiviral vectors The TRIPDU3-EF1a/GFP lentiviral

vector encoding EGFP has already been described [17].

Briefly, this vector leads to efficient integration of the

transgene within the host cell genome, even in non-dividing

cells, due to a specific sequence allowing vector cDNA import

through the nuclear membrane. In addition, it allows efficient

transgene expression under the control of an internal

eukaryotic promoter (here the EF1a ubiquitous promoter).

TRIPDU3-EF1a/PF4-GFP vector was generated as follows:

PF4-GFP cDNA was amplified by PCR using the respective

forward and reverse primers 5¢-CTAGGTTCTAGACCA

CCGGTCGCCACCATGAGCTCCGCAGCAGGG-3¢ and

5�-TGGACGAGCTGTACAAGTAACTCGAGACCTAG-3¢.These generate amplified fragments exhibiting a Xba I and a

Xho I sites at their 5¢- and 3¢-ends (Fig. 1). The PCR was

performed for four cycles (annealing temperature, 54 �C), andthen for 25 cycles (annealing temperature, 60 �C) using either

PF4-GFP or SP4-GFP constructs as templates. The PCR

fragments were then digested by XbaI and Xho I and subcloned

directionally into the TRIPDU3-EF1a vector, resulting in

TRIPDU3-EF1a/PF4-GFP and TRIPDU3-EF1a/SP4-GFP

vectors, respectively. Vector particles were produced in 293T

cells by transient calcium phosphate cotransfection of the

lentiviral vector plasmids, an encapsidation plasmid lacking all

accessory HIV-1 proteins, and a VSV-G envelope protein

expression-plasmid as previously described [17]. cDNAs

inserted within the constructs were sequenced in both

directions.

Cell lines AtT20 cells (derived from mouse pituitary) were

grown in T175 flasks in 20 mL of DMEM GLUTAMAX I

4500 mg L)1 D-glucose pH 7.8 supplemented with 10% fetal

calf serum and penicillin–streptomycin. DAMI cells were

grown in suspension in T75 flasks in 35 mL of RPMI

GLUTAMAX 1000 mg L)1 D-glucose supplemented with

10% fetal calf serum and penicillin–streptomycin.

KpnlA

B

C

Kpnl

C Fokl

DBamHl

GFP

GFP

GFP

GFP GFP

GFP GFP M.W.(kDa)

M.W.(kDa)

PF4

PF4 PF4 PF4

1 2 3 4 5

1 2 3 4 5

46

30

21.5

46

30

21.5

14.3

6.5

anti-GFP blot

anti-PF4 blot

PF4

Not l

B

A

F

G

E

pure

Plts moc

kGFP

IP

IP

PF4-GFP

PF4-GFP

GFP

PF4

GFPGFP PF4-

GFP

PF4-

GFP

moc

k

HXba l

SP4

SP4

SP4

Fokl

PF4

PF4∆stop

PF4

PF4-GFP

SP4-GFP

Fig. 1. (A) Schematic diagram of the design of PF4, PF4-GFP and SP4-

GFP constructs. PCR oligonucleotides C, D, G and H are shown as

arrows. Oligonucleotides A, B, E and F used as templates are indicated as

bars. Restriction sites cited in the text are shown. The open box corres-

ponds to PF4 signal peptide. Box with vertical stripes at the C-terminus of

PF4 indicates the position of the stop codon and the 3¢ untranslatedsequence, deleted in PF4Dstop construct. (B) and (C) Expression of the

PF4-GFP fusion protein in AtT20 cells by immunoprecipitation and

Western blotting. Mock-, GFP-, or PF4-GFP-transfected AtT20 cells and

platelets were homogenized in sample lysis buffer, and subjected to

immunoprecipitation (IP) with anti-GFP (GFP) or anti-PF4 antibody

(PF4), followed byWestern blotting. Membranes were probed with rabbit

anti-GFP (anti-GFP blot, B) or anti-PF4 (anti-PF4 blot, C) followed by

horseradish peroxidase-conjugated donkey antirabbit IgG. Lanes 1: pure

GFP in (B); PF4 immunoprecipitated from platelet lysate in (C); lanes 2

and lanes 4: anti-PF4 immunoprecipitation of mock-transfected AtT20

cells (negative control), and of PF4-GFP-transfected AtT20 cells,

respectively; lanes 3 and lanes 5: anti-GFP immunoprecipitation on

GFP- and PF4-GFP-transfected cells, respectively. Molecular weight

standards (MW) are shown on the right. Bands corresponding to

PF4-GFP, GFP and PF4 are indicated by arrows.

Megakaryocyte granule targeting 2233


DNA transfection

AtT20 andDAMI cells were in exponential growth phase prior

to transfection. They were transiently or stably transfected with

plasmid constructs using the poly-imine transfection reagent

ExGen 500. For transient expression, DNA (2 or 3 lg,respectively) was added to 20 000 AtT20 cells freshly seeded

onto Laboratory-Tek slides, or 100 000 DAMI cells, in

suspension. Transient transfection efficiencies were 30% for

AtT20 cells and 5–10% for DAMI cells. Cells were then

processed for immunofluorescence microscopy 48 or 72 h after

transfection, respectively. For stable transfection, plasmid

constructs were first linearized by restriction enzyme digestion,

transfected at the same DNA/cell ratio than for transient

transfection, and then selected in G418 at 0.5–1.0 mg mL)1.

Collection and fractionation of hematopoietic CD34+

cells Cord blood (CB) cell samples were collected with the

informed consent of themothers, according to approved french

institutional guidelines. CD34+ cells were isolated using the

MiniMacs immuno-magnetic kit (Miltenyi Biotech, Gladbach,

Germany) as recommended by the manufacturer and trans-

duced immediately.

Transduction of CD34+ cells and megakaryocyte cell

culture We plated human CD34+ cell populations at 106

cells/mL in RM B00 serum-free medium (RTM, Tourcoing,

France) in the presence of recombinant human (rhu) SCF

(5 ng mL)1), and rhuTPO (20 ng mL)1) together with

lentiviral vector particles at a concentration of 2500 ng of

viral P24/mL [18]. After 24 h an identical amount of lentiviral

vector particles was added and the transduction was prolonged

up to 72 h. Cells were then washed, numbered and seeded at

5 · 104 cells per mL in fresh medium complemented as above,

in six-well plates onto sterile glass coverslips coated for 30 min

at 37 �C with 0.02% gelatin in CaCl2 phosphate-buffered

saline. After an overall 10-day culture, cells were treated for

immunofluorescence staining. At that stage, more than 90% of

the cells were CD41+/CD42b+megakaryocytes as assessed by

FACS analysis (not shown). In some experiments megak

aryocytes were assayed for granule secretion and challenged

with thrombin (1 U mL)1) in resuspension buffer containing

10 mM Hepes pH 7.4, 140 mmol L)1 NaCl, 3 mmol L)1 KCl,

0.5 mmol L)1 MgCl2, 5 mmol L)1 NaHCO3, 10 mmol L)1

glucose during 30 min, the optimal incubation time found in

preliminary experiments.

Immunoprecipitation

Cells were homogenized for 30 min on ice in sample lysis buffer

containing 1% Triton X-100, 10 mM Tris-HCl pH 8, 150 mM

NaCl, 1 mM phenylmethylsulfonyl fluoride, 2 lg mL)1

leupeptin and 5 lg per 10 mL aprotinin, and centrifuged at

12 000 · g for 5 min at 4 �C to remove nuclei and insoluble

debris. Total soluble cellular protein was quantified using

the bicinchoninic acid method and instructions from the

manufacturer. Fifty microliters of protein A-Sepharose beads

in 10 mM sodium phosphate, pH 7.2, 0.15 mol L)1 NaCl

(1 : 1, v:v) was added to lysate/antibody mixes (4 mg of total

proteins/2 lL antibody, preincubated overnight at 4 �C) andincubated for 1 h at 4 �C with gentle constant shaking. The

samples were centrifuged at 12 000 · g for 2–5 min, beads

were washed five times with sample lysis buffer, and finally

resuspended in 130 lL of loading buffer (2% SDS,

10 mmol L)1 HEPES, 20% glycerol, 5 mmol L)1 EDTA,

10 mmol L)1 dithiothreitol pH 6.8) for 10 min at 95 �C.Samples were then centrifuged for 5 min at 12 000 · g, and

60 lL of each supernatant were loaded onto two separate

polyacrylamide gels.

Western blot analysis

The samples were electrophoresed on 13% SDS/PAGE, and

electro-blotted onto membranes. Membranes were blocked for

1 h in Western buffer (20 mmol L)1 Tris-HCl, 0.15 mol L)1

NaCl, 0.1%Tween, pH 7.4), supplemented with 5% non-fat

milk. Membranes were incubated with antihuman PF4 or anti-

GFP [1 : 2000 and 1 : 1000 (v:v), respectively], overnight at

4 �Cwith gentle shaking.Membranes were rinsed four times in

water, and incubated in Western buffer for 5 min with gentle

shaking. This procedure was repeated five times. Membranes

were then incubated with peroxidase-conjugated donkey anti-

rabbit IgG (at a 1 : 10 000 dilution) in Western buffer for 1 h,

washed as above, and finally incubated with chemiluminescent

substrate for 5 min before exposing for 10–30 s to photogra-

phic film.

Immunofluorescence staining and confocal microscopy AtT20

cells were grown onto LabTeks, before immunofluorescence

processing, whereas CD34+ and DAMI cells were grown onto

sterile glass coverslips (in the presence of 100 mmol L)1 PMA

and 10 ng mL)1 of TPO for DAMI cells). Cells were washed

three times with 0.5 mmol L)1 CaCl2 PBS. Between each step

cells were also washed three times at room temperature. Cells

were fixed in 4% paraformaldehyde for 15 min at room

temperature; then permeabilized for 1 h at room temperature

with permeabilization buffer containing 20 mM HEPES,

300 mM sucrose, 50 mmol L)1 NaCl, 3 mmol L)1 MgCl2,

1% Triton X-100, pH 7.0. Cells were then incubated in the

presence of 5% (w/v) bovine serum albumin for 30 min at

37 �C. Cells were stained for 30 min at 37 �C with primary

antibodies diluted in 2.5% bovine serum albumin: rabbit anti-

ACTH (1 : 25, v:v), mouse antivon Willebrand factor (1 : 200

v/v); secondary antibodies were Texas red� dye-conjugated

donkey antirabbit IgGs or Texas red� dye-conjugated goat

antimouse IgGs. Cells were then analyzed using a Leica TCS

4D confocal laser scanning microscope.

Immunoelectron microscopy

For morphological examination, megakaryocytes were

postfixed in 1% osmic acid, dehydrated in ethanol, and



embedded in Epon by standardmethods. For immunolabeling,

megakaryocytes were fixed in 1.5% glutaraldehyde and

embedded in sucrose. Immunochemical reactions were per-

formed on thin sections collected on nickel grids. The antibody

specific for GFPwas used at 40 lg mL)1 to immunolabel PF4-

GFP and SP4-GFP; bound anti-GFP IgGs were detected by

affinity-purified goat antirabbit IgGs coupled to 10 nm gold

particles. Polyclonal rabbit antihuman antibody to VWF was

used at 100 lg mL)1, and bound anti-VWF IgGs detected by

affinity-purified goat anti rabbit IgG coupled to 5-nm gold

particles. IgG coupled to gold particles were used at 1 : 30

dilution. Ultrathin sections were observed on a Philips CM 10

electron microscope.

Results

Preliminary experiments demonstrate that PF4-GFP is tar-

geted to ACTH-containing granules in AtT20 cells. AtT20

cells, which are derived from mouse pituitary gland, exhibit

regulated secretory granules containing ACTH [19]. This cell

line has been extensively utilized to study trafficking to

intracellular storage organelles of membrane proteins, inclu-

ding the megakaryocyte membrane protein P-selectin [20], as

well as secretory proteins such as VWF [21,22]. AtT20 cells thus

provide a suitable model for studying intracellular storage of

megakaryocytic secretory proteins. Figure 1(B,C) show that

after transfection, AtT20 cells express GFP as a 27 kDa band

as detected by immunoprecipitation and Western blotting (see

Fig. 1B, lane 3). When transfected with PF4-GFP, AtT20 cells

expressed a 35 kDa polypeptide, consistent with the predicted

size of PF4-GFP, and that cross-reacted with both anti-GFP

and anti-PF4 antibodies (Fig. 1B,C, lanes 4 and 5). Identical

results were obtained with the megakaryocytic cell line DAMI

(not shown).

To check whether GFP fusion altered PF4 trafficking, as

well as to determine PF4-GFP intracellular localization

comparatively to GFP or PF4 alone, transfected AtT20 cells

were analyzed by immunofluorescence by laser scanning

confocal microscopy (Fig. 2). Figure 2(A) shows that, when

transfected alone, GFP yields a diffuse fluorescence distributed

throughout the cell body, consistent with the cytosolic local-

ization of a non-secretory protein. In contrast, PF4 displayed a

granular pattern (green in Fig. 2C), which matched ACTH-

containing granules (red in Fig. 2D), yielding yellow granules

upon image merging (Fig. 2E). This demonstrated colocaliza-

tion of PF4 with ACTH. PF4-GFP exhibited the same pattern

than PF4 alone, and colocalized with ACTH (Fig. 2F–H), thus

demonstrating that, when fused to PF4, GFP does not interfere

with granule storage.

Altogether, these results suggest that (i) PF4 follows the

regular granule storage pathway of secretory cells and (ii) its

granule targeting ability is not altered by the GFP fusion.

The signal peptide of PF4 is not sufficient to promote

granule storage. Because granule targeting determinants of

secretory proteins are still not fully understood, we wondered

whether PF4 granule targeting was the result of a specific

granule storage pathway, or of a default pathway for any

secretory protein entering, via its signal peptide, the ER and the

secretory pathway. To answer this question we engineered

GFP (a cytoplasmic protein, therefore non-secretory) so as to

enter the ER and the general secretory pathway by way of an

added signal peptide. To be able to directly compare the result

with granule targeting of PF4-GFP, we used the PF4 signal

peptide fused to GFP (seeMethods, SP4-GFP construct). SP4-

GFP distribution was strikingly different fromGFP, exhibiting

a marked localization to a juxta-nuclear compartment, most

likely the Golgi apparatus, and a faint reticular compartment

(Fig. 2I), clearly distinct from ACTH storage granules (no

yellow granules, Fig. 2J,K). This indicated that SP4-GFP

trafficked through the ER and the Golgi, as a consequence of

PF4 signal peptide addition, but was not stored in ACTH-

containing granules. These results indicate that the PF4 signal

peptide, though required for entry into the secretory pathway,

is not sufficient to promote entry into the ACTH granule

storage pathway.

PF4-GFP follows the granule storage pathway of VWF in

the megakaryocytic DAMI cell line and in human primary

megakaryocytes differentiated in vitro. We then assessed PF4

trafficking in amegakaryocytic context.We surveyed a number

of cell lines withmegakaryocytic potential for their ability to (a)

express VWF, (b) undergo efficient transfection, and (c) exhibit

granule storage. Of all the cell lines tested, CHRF-288, Meg01,

Fig. 2. PF4 and PF4-GFP, but not SP4-GFP, traffick to ACTH-con-

taining storage granules in AtT20 cells. GFP (A), PF4 (C,D,E), PF4-GFP

(F,G,H), and SP4-GFP (I,J,K) constructs were stably transfected inAtT20

cells. Cells were fixed, permeabilized, and incubated with a polyclonal anti-

ACTH (D,G,J), and anti-PF4 (C) antibody which was detected with

Alexa�Fluor 488-conjugated goat antirabbit IgG, and analyzed by con-

focal laser scanning microscopy. (A,F,I) GFP fluorescence (green); (B)

mock transfection; (C) anti-PF4 (green); (D,G,J) anti-ACTH (red); (E)

C/D merging; (H) F/G merging; (K) I/J merging. Yellow granules in (E)

and (H) demonstrate colocalization of PF4or PF4-GFPwithACTH.Note

the absence of colocalization of SP4-GFP does not colocalize with ACTH

in (K). The Golgi apparatus is indicated. Total magnification 1200 ·.



K562, CMK, HEL and DAMI, only the latter exhibited

significant VWF expression and granule storage. DAMI cells

were induced further into megakaryocytic differentiation by

5-day incubation with PMA and the megakaryocytic-specific

cytokine TPO (Briquet-Laugier et al., in press). We compared

the intracellular localization of transfected PF4-GFP to that of

endogenous VWF, in DAMI cells (Fig. 3A–C). In Fig. 3(A), a

wide field of PF4-GFP-transfected DAMI cells shows that

colocalization of PF4-GFP with VWF (cells exhibiting yellow

staining), is not a rare event in stably transfected DAMI cells.

Moreover PF4-GFP trafficked to a granular compartment

colocalizing at least partially with VWF-containing granules, as

evidenced by the presence of a number of yellow-stained

granules throughout the cytoplasm (Fig. 3B). The PF4-GFP

granular compartment is different from Golgi and lysosomes

since no colocalization between b-COPI, a specific Golgi

marker (not shown), or Lamp-1, a specific membrane marker

of lysosomes, and PF4-GFPwas observed (Fig. 3C). SP4-GFP

did not colocalize with VWF (Fig. 3D), confirming that, like in

AtT20 cells, the PF4 signal peptide alone does not specify

entry into the granule storage pathway in a megakaryocytic

context.

Finally, we examined PF4-GFP granule targeting in the

more physiological context of primary megakaryocytes,

obtained by in vitro differentiation of cord blood CD34+

progenitors differentiated in vitro. In Fig. 4(A), as expected,

GFP is localized in the cytosol of the CD41+ (amegakaryocyte

differentiation marker) cell population, while VWF is detected

in well-identified granules (Fig. 4B), no colocalization being

observed between the two proteins (Fig. 4C, absence of yellow

granules). In contrast, PF4-GFP and VWF colocalized in

granular structures (Fig. 4D–F). PF4-GFP does not colocalize

with lysosomes as evidenced by absence of colocalization with

Lamp-1 (Fig. 4G–I). Interestingly, as opposed to PF4-GFP,

and like in DAMI cells, SP4-GFP localizes in a reticular

compartment (Fig. 4J) very distinct from that of VWF

(Fig. 4K), exhibiting no significant colocalization (Fig. 4L).

These results were confirmed by electron microscopy ima-

ging. In Fig. 5, megakaryocytes exhibit morphological features

specific of the megakaryocytic lineage, including demarcation

membranes and granules some of which showing a dense core,

specific of mature alpha granules [9]. More importantly, PF4-

GFP-transduced megakaryocytes exhibit numerous dense core

storage granules reacting with both anti-VWF and anti-GFP

immunogold antibodies, highly suggesting that PF4-GFP

reached alpha granules (Fig. 5A,C). In contrast, colocalization

of VWF and SP4GFP appears very poor in SP4GFP-

transduced megakaryocytes (Fig. 5B,D). Altogether both our

confocal and EM studies demonstrate that like for other cell

types, addition of the PF4 signal peptide, though presumably

allowing its entry into the secretory pathway, does not target

GFP to the VWF granular compartment. Thus granular

targeting of PF4 in megacaryocytes does not rely on its sole

entry into the secretory pathway.

Altogether our data demonstrate that PF4-GFP targets to

the same granular compartment than VWF, and suggest that

the a-granule storage pathway of PF4, and presumably of

other secretory proteins, is not a default pathway due to entry

into the endoplasmic reticulum–Golgi pathway via the signal

peptide.

PF4-GFP is secretable upon thrombin cell stimulation. One

essential feature of a-granules (as well as any secretory granule)is exocytosis in response to megakaryocyte or platelet stimu-

lation. Attempts at triggering granule release from DAMI cells

with thrombin were unsuccessful. However, in human megak-

aryocytes challenged with thrombin (1 U mL)1), the release of

PF4-GFP-containing granules was efficient as evidenced by the

marked decrease in PF4-GFP fluorescence following activation

(Fig. 6A,B). The extent of secretion, as assessed by the ratio of

mean fluorescence of activated vs. nonactivated megakaryo-

cytes, was 83.7 ± 8.1% SEM (P < 0.0001) by image analysis

on a total of five fields. These results are confirmed by

immunoelectron microscopy with a nearly complete disap-

pearance of storage granules containing both GFP and VWF

immunogold staining after thrombin stimulation (Fig. 6C,D).

Thus PF4-GFP reaches a granular compartment in megak-

aryocytes capable of secretion upon stimulation, and which

thus belongs to the regulated secretory pathway.

Fig. 3. Colocalization of PF4-GFP with VWF in peripheral granules and

not lysosomes, in DAMI cells. PF4-GFP and SP4-GFP constructs were

stably or transiently transfected in DAMI cells. Cells were fixed, perme-

abilized, stained with monoclonal antibodies to VWF, or to Lamp-1, a

lysosome membrane marker, and detected with Texas red� dye-conju-

gated goat antimouse IgG. Co-immunofluorescence of GFP and Texas

red� dye was then assessed by confocal microscopy. (A) A wide field of

PF4-GFP-transfected DAMI cells at 400 · magnification. Three distinct

cells are shown in (B), (C) and (D). (B) PF4-GFP fluorescence, green; anti-

VWF, red, colocalization of both proteins in yellow. (C) PF4-GFP

fluorescence, green, anti-Lamp-1, red, absence of yellow indicates no

colocalization of PF4-GFP with lysosomes. (D) SP4-GFP fluorescence,

green, anti-VWF, red; diffuse green of SP4-GFP over VWF red granules

yields some yellow background, but note the absence of peripheral yellow

granules, compared with (B) suggesting absence of colocalization of

SP4-GFP with VWF. Total magnification is as indicated.



Discussion

Our goal is to understand the mechanism of biogenesis of

megakaryocyte storage a-granules. As a first step we have

seeked to assess the trafficking of a megakaryocyte a-granularprotein, PF4. We have thus fused PF4 to GFP, as a fluorescent

reporter and compared its trafficking to that of PF4 alone. We

find that whether fused toGFP or not, PF4 is always efficiently

sorted to true storage granules distinct from lysosomes, in the

non-hemopoietic ACTH-producing pituitary mouse cell line

AtT20, and the megakaryocytic cell line DAMI. We thus

conclude that GFP fused to PF4 does not interfere with its

trafficking properties, validating the PF4-GFP construct as a

proper tool for studying PF4 granule targeting.

One important question addressed here is: Is granule

storage in the megakaryocytic pathway a default pathway?

Answering this question would be of relevance to mega-

karyocyte biology, given the central role played by the

secretion of platelet secretory proteins in hemostasis, inflam-

matory or wound healing contexts. One way to address this

question is to use proteins known to follow the constitutive

secretory pathway and see whether they are stored in granules.

Fig. 4. PF4-GFP and VWF colocalize in megakaryocyte granules after differentiation in vitro. Transduced CD34+ cells were differentiated into

megakaryocytes in vitro (see Methods), fixed, permeabilized, and incubated with a mix of monoclonal antibodies to VWF. Cells were then incubated with

secondary Texas red� dye-conjugated goat antimouse IgG, before analysis by confocal immunofluorescencemicroscopy.Megakaryocytes were transduced

with either GFP (A,B,C), PF4-GFP (D–I), or SP4-GFP (J,K,L). Green fluorescence corresponds to GFP (A), PF4-GFP (D,G), SP4-GFP (J). Red

fluorescence is VWF (B,E,K) or Lamp-1, a lysosomal marker (H). (C), (F), (I) and (L) are merged images of GFP fluorescence (green)/VWF (red) except

for (I), GFP/Lamp-1. Note efficient colocalization of PF4-GFP with VWF as yellow granules in (F), no colocalization with Lamp-1 in (I), and virtually no

colocalization of SP4-GFPwith VWF in (L). The few yellow small vesicles seen in (L) correspond to diffusion of VWF red fluorescence over the green SP4-

GFP-containing vesicles and not to true colocalization, as evidenced by the completely different SP4-GFP and VWF patterns [compare (J) and (K)],

contrasting with the matching patterns of PF4-GFP and VWF [compare (F) and (L)]. Total magnification 1200 ·.



However it is known that proteins following a given pathway

in one cell type may very well be routed to another one in a

different cell-type. As the biology of secretion remains poorly

explored in megakaryocytes, there is no known constitutively

secreted megakaryocytic protein. One way to circumvent this

problem is to take a protein naive for secretion, namely a

cytoplasmic one, and to add it a signal peptide to promote ER

entry. We used the PF4 signal peptide to ensure that this

sequence did not provide by itself granule targeting informa-

tion. Using this latter strategy, with GFP as a cytoplasmic

secretion-naive protein, we show that PF4 signal peptide

added to GFP, allows entry into the endoplasmic reticulum–

Golgi pathway. Additionally, GFP was found constitutively

secreted, demonstrating that it followed the constitutive

secretion pathway (data not shown). However SP4-GFP did

not enter the granule storage pathway in megakaryocytes.

This suggests that granule targeting of PF4 is not the natural

consequence of its entry into the secretory pathway via its

signal peptide. To our knowledge, this is the first direct

evidence that storage of an a-granule protein requires further

trafficking than endoplasmic reticulum–Golgi entry. What is

the exact nature of the PF4 determinants required for sorting

and storage, and whether this is true of all a-granule proteinsremains to be determined.

Finally we found that in the physiological context of

megakaryocytes, the PF4-GFP granular compartment con-

taining VWF and the membrane protein P-selectin (data not

shown), and therefore identified as a-granules, underwent

exocytosis in response to thrombin activation, and therefore

belonged to the regulated secretory pathway. Interestingly PF4

granule targeting ability appears independent of the cell type,

since observed in pituitary cells as well as in megakaryocytes,

and in mouse (murine AtT20 cells) and human context. This

suggests that the mechanism at play is most likely either highly

preserved throughout evolution, or redundant, and therefore

physiologically relevant.

Fig. 5. Electron microscopy analysis of transduced megakaryocytes. EM

photographs of PF4-GFP-expressing megakaryocytes were obtained as

described in Methods. (A) EM of a megakaryocyte expressing PF4-GFP.

Demarcation membranes are noted DMS, while granules with dense cores

are labeled asA. Large gold particles correspond to anti-VWF, while small

gold particles correspond to anti-GFP and thus identify PF4-GFP. Note

that large and small gold particles colocalize in a number of a-granules. (B)EM of an SP4-GFP-expressing megakaryocyte. Like in (A), large particles

identify VWF, and small particles identify SP4-GFP. Note near absence of

colocalization of SP4-GFP with VWF. (C) and (D) are enlargements of agranules from PF4-GFP- and SP4-GFP-expressing megakaryocytes,

respectively. Magnifications are 38 000· for (A) and (B), and 136 000· for

(C) and (D).

Fig. 6. Thrombin induces secretion of PF4-GFP from megakaryocytes.

PF4-GFP-transduced CD34+ cells were differentiated into megakaryo-

cytes in vitro. Cells were challenged with resuspension buffer alone (see

Methods) (A,C) or buffer containing 1 U mL)1 thrombin as a secreta-

gogue (B,D), for 30 min at 37 �C. (A) and (B) shown are merged confocal

images of PF4-GFP/VWF. Note the decrease in the overall number of

storage granules as well as in yellow granules after thrombin stimulation

(B) compared with (A). Nuclei are stained in blue with DAPI. Total

magnification 1200 ·. (C) EM of non-stimulated PF4-GFP-expressing

megakaryocytes. Note the intense costaining for PF4-GFP (small gold

particles, small arrows) and VWF (large gold particles, large arrows)

within the same a granules. (d) EM of thrombin-stimulated PF4-GFP-

expressing megakaryocytes. Note the weak immunogold staining and the

numerous empty granules, consistent with the efficient corelease of PF4-

GFP and VWF. Total magnification 70 000 ·.



In fully mature megakaryocytes and platelets, PF4 and

VWF are found in a-granules [8], suggesting that they either

follow the same targeting pathway, or that their pathways

eventually merge into a-granules. In DAMI cells and mega-

karyocytes differentiated in vitro, we found evidence for

granules containing each protein separately along with yellow

granules containing both PF4 and VWF. Aside from an

artefact due to sensitivity differences between GFP and

fluorescent antibodies, this observation may be biologically

significant and reflect the a-granule maturation process. Our

data may be consistent with a-granules not budding directly

off the Golgi, but originating from a more complex phenom-

enon such as for example Golgi exit of VWF and PF4

separately in distinct vesicles undergoing fusion into mature

a-granules. Interestingly, the mechanism of storage of VWF

has been studied recently in AtT20 cells and shown to depend

upon a domain overlapping aminoacids 201–741 of the

VWF-propeptide [23,24]. While this has not been confirmed in

megakaryocytes yet, it is conceivable that the same determi-

nants are required for VWF storage in a-granules. Of note, noclear primary sequence homology between PF4 and the

201–741 VWF sequence could be demonstrated (not shown).

However in both cases, granule storage is clearly not

dependent upon signal peptide alone, strengthening the idea

that a-granule storage requires an additional pathway than

entry into the ER.

PF4-GFP may prove an interesting probe with which to

analyze the defect responsible for the absence of endogenous

a-granule proteins in the human bleeding disorder, gray

platelet syndrome. While it appears likely that the defect

involves one or several steps along the packaging of a-granuleproteins, its mechanism and therefore the gene involved remain

unknown [11]. Determining the actual fate of PF4-GFP after

its transduction within gray megakaryocytes, such as its

constitutive release into the medium, or its misrouting to

lysosomes, may provide a basis for a better understanding of

the molecular basis of the gray platelet syndrome.

Acknowledgements

We thank Jan Bayer, Jalila Chagraoui, and Adeline Lepage-

Noll for their help at the bench during the early phase of this

study; Faezeh Legrand and Philippe Mangeot who helped us

with lentiviral constructs; Gerard Geraud for his expert help in

confocal microscopy; Jean-MarcMasse for his expert technical

assistance with the immunoelectron microscopy studies; Brig-

itte Izac for her technical expertise in producing lentiviral

vectors. We wish to acknowledge Dr M. Poncz for providing

PF4 cDNA, and Dr D. Baruch for anti-VWF monoclonal

antibodies. We also wish to thank Dr C. Negrier (Hospices

Civils de Lyon, France) for helpful discussions.

This project was supported in part by grants from INSERM,

Association pour la Recherche sur le Cancer–ARC and

Association Francaise contre les Myopathies-AFM. C. L.-B.

is a recipient from an AFM fellowship.

References

1 Nicchitta CVA. Platform for compartmentalized protein synthesis:

proteintranslation and translocation in the ER. Curr Opin Cell Biol

2002; 14: 412–6.

2 Donaldson JG, Lippincott-Schwartz J. Sorting and signaling at the

Golgi complex. Cell 2000; 101: 693–6.

3 Allan BB, Balch WE. Protein sorting by directed maturation of Golgi

compartments. Science 1999; 285: 63–6.

4 Tooze SA, Martens GJ, Huttner WB. Secretory granule biogenesis:

rafting to the SNARE. Trends Cell Biol 2001; 11: 116–22.

5 Zhu Y, Doray B, Poussu A, Lehto VP, Kornfeld S. Binding of GGA2

to the lysosomal enzyme sorting motif of the mannose 6-phosphate

receptor. Science 2001; 292: 1716–8.

6 Cool DR, Normant E, Shen F, Chen HC, Pannell L, Zhang Y, Loh

YP. Carboxypeptidase E is a regulated secretory pathway sorting

receptor: genetic obliteration leads to endocrine disorders in Cpe (fat)

mice. Cell 1997; 88: 73–83.

7 Gerdes HH, Glombik MM. Signal-mediated sorting of chromo-

granins to secretory granules. Adv Exp Med Biol 2000; 482: 41–54.

8 Harrison P, Cramer EM. Platelet alpha-granules. Blood Rev 1993; 7:

52–62.

9 Cramer EM. Megakaryocyte structure and function. Curr Opin

Hematol 1999; 6: 354–61.

10 Rosa JP, George JN, Bainton DF, Nurden AT, Caen JP, McEver

RP. Gray platelet syndrome: demonstration of alpha granule mem-

branes that can fuse with the cell surface. J Clin Invest 1987; 80: 1138–

46.

11 Smith MP, Cramer EM, Savidge GF. Megakaryocytes and plate-

lets in alpha-granule disorders.Baillieres ClinHaematol 1997; 10: 125–

48.

12 Zhang X, Chen L, Bancroft DP, Lai CK, Maione TE. Crystal

structure of recombinant human platelet factor 4. Biochemistry 1994;

33: 8361–6.

13 Tsien RY. The green fluorescent protein. Annu Rev Biochem 1998;

67: 509–44.

14 Li L, Chandrasegaran S. Alteration of the cleavage distance of Fok I

restriction endonuclease by insertion mutagenesis. Proc Natl Acad Sci

USA 1998, 1993; 90: 2764–8.

15 Horton RM. In vitro recombination and mutagenesis of DNA.

SOEing together tailor-made genes. Methods Mol Biol 1997; 67:

141–9.

16 Nielsen H, Engelbrecht J, Brunak S, von Heijne G. Identification of

prokaryotic and eukaryotic signal peptides and prediction of their

cleavage sites. Protein Eng 1997; 10: 1–6.

17 Sirven A, Ravet E, Charneau P, Zennou V, Coulombel C,

Guetard D, Pflumio F, Dubart-Kupperschmitt A. Improvement

of long-term transgene expression in all hematopoietic cells by

the use of modified TRIP lentiviral vectors. Mol Ther 2001; 3:

438–48.

18 Amsellem S, Ravet E, Fichelson S, Pflumio F, Dubart-Kupperschmitt

A.Maximal lentivirus-mediated gene transfer and sustained transgene

expression in human hematopoietic primitive cells and their progeny.

Mol Ther 2002; 6: 673–7.

19 Richardson UI, Schonbrunn A. Inhibition of adrenocorticotropin

secretion by somatostatin in pituitary cells in culture. Endocrinology

1981; 108: 281–90.

20 Disdier M, Morrissey JH, Fugate RD, Bainton DF, McEver RP.

Cytoplasmic domain of P-selectin (CD62) contains the signal for

sorting into the regulated secretory pathway. Mol Biol Cell 1992; 3:

309–21.

21 Wagner DD, Saffaripour S, Bonfanti R, Sadler JE, Cramer EM,

Chapman B, Mayadas TN. Induction of specific storage organ-

elles by von Willebrand factor propolypeptide. Cell 1991; 64:

403–13.



22 Rosenberg JB, Foster PA, Kaufman RJ, Vokac EA, Moussalli M,

Kroner PA, Montgomery RR. Intracellular trafficking of factor VIII

to von Willebrand factor storage granules. J Clin Invest 1998; 101:

613–24.

23 Haberichter SL, Jozwiak MA, Rosenberg JB, Christopherson PA,

Montgomery RR. The von Willebrand factor propeptide (VWFpp)

traffics an unrelated protein to storage. Arterioscler Thromb Vasc Biol

2002; 22: 921–6.

24 Haberichter SL, Jacobi P, Montgomery RR. Critical independent

regions in the VWF propeptide and mature VWF that enable normal

VWF storage. Blood 2003; 101: 1384–91.



Documents

Probing platelet factor 4 alpha-granule targeting