supplemental figure 1. - media.nature.com · 2 supplemental figure 1. characterization of the heg1...

1

Supplemental Figure 1.

Nature Medicine: doi:10.1038/nm.1918

2

Supplemental Figure 1. Characterization of the Heg1 gene and predicted HEG1

amino acid sequence. The mouse Heg gene contains 17 exons and is expressed as three

alternatively spliced mRNAs: a full length transcript that encodes a type I transmembrane

receptor with a highly conserved 110 amino acid cytoplasmic tail, a near full length

transcript without exons 3 and 4, and a secreted splice variant encoding the signal peptide

and the first four exons. (a) A schematic of the murine Heg1 locus on chromosome 16 is

shown. Red lines indicate alternative splice variants detected using RT-PCR in mouse

embryos. A cryptic splice site is present in exon 4 that drives splicing to the full length

receptor mRNA. Gray boxes indicate 3’ untranslated regions for the truncated and full

length Heg1 transcripts. (b) Alignment of the predicted mouse and zebrafish HEG1

amino acid sequences. The signal peptide (SP) is indicated by a green line. The

transmembrane domain (TM) is indicated by a red line. The point of truncation for the

HEG1ΔC receptor is marked by a blue line. Note the high level of conservation of the

HEG1 intracellular carboxy tail.


3


Supplemental Figure 2. Expression of Heg1 in the developing cerebral mouse

vasculature. (a–d) Digoxigenin in situ hybridization of E14.5 mouse embryos

demonstrates that Heg1 is expressed in the vasculature of the developing brain in a

pattern similar to that of the endothelial marker Pecam1 (asterisks indicate neural tube

expression of Heg1; images courtesy of http://www.genepaint.org/).


4


Supplemental Figure 3. Expression of Ccm2 in the developing mouse vasculature.

Ccm2 expression is detected in the neural tube but not the endocardium or endothelium of

developing mice. (a,b) Radioactive in situ hybridization for Ccm2 in E9.5 embryos fails

to show vascular expression. (c–f) Wholemount X-gal staining of Ccm2lacZ/+ E10.5

embryos (c,e) is compared with that of Tie2-Cre; ROSA26R E10.5 embryos (d,f) in

which lacZ is expressed in developing endothelial cells (asterisks indicate endocardial

expression). E10.5 Ccm2lacZ/+ animals revealed strong X-gal staining in the neural tube,

but no staining above background in the developing heart or vessels. Prolonged X-gal

incubation of E10.5 Ccm2lacZ/+ but not control littermate embryos exhibited weak

generalized staining, a result suggestive of ubiquitious low-level Ccm2 expression (data

not shown). h, heart; nt, neural tube; ao, aorta.


5


Supplemental Figure 4. Generation of HEG1-deficient mice. (a) Heg1 gene targeting

strategy. Shown is the vector used to target exon 1 of the Heg1 gene. (b) Identification

of targeted ES cell clones by Southern blot analysis using the 3’ probe indicated

following Nsi1 digestion of genomic DNA. (c) Characterization of loss of Heg mRNA in

Heg1+/– and Heg1–/– embryos. Real-time RT-PCR was performed using primers to

amplify the parts of the Heg1 mRNA encoded by the exons indicated in whole E14

embryos. Note the approximately 50% reduction of all parts of the Heg1 mRNA in

Heg1+/– animals, and the complete loss of the Heg1 mRNA in Heg1–/– animals. (d)

Characterization of Heg1 mRNA expression in Heg1+/+, Heg1+/– and Heg1–/– littermate

embryos using radioactive in situ hybridization with a probe encompassing the terminal 5

exons.


6


Supplemental Figure 5. HEG1-deficient hearts have normal numbers of

proliferating cardiomyocytes. Antibodies to Ki67 were used to identify proliferating

cardiomyocytes in sections of Heg1+/+ and Heg1–/– littermate E16 embryos. (a)

Quantitation of the percentage of cardiomyocytes that express Ki67 (N = 1000–1200

cells analyzed in five separate sections). P > 0.05. (b) A representative Ki67 stain of

Heg1+/+ and Heg1–/– embryo hearts is shown.


7


Supplemental Figure 6. Characterization of Ccm2lacz/lacz embryos. (a) Loss of 3’

Ccm2 mRNA in Ccm2lacz/lacz embryos. Ccm2 and Heg1 expression were measured in E9

embryo littermates from Heg1+/-;Ccm2lacZ/+ intercrosses. Ccm2 mRNA was measured by

real-time RT-PCR using primers in exon 10. (b) Ccm2 lacZ/lacZ embryos fail to establish a

patent blood vascular network. Transverse sections of E9 Ccm2 lacZ/lacZ embryos at three

levels are shown. H-E staining reveals the absence of blood-filled dorsal aortae (DA),

cardinal veins (CV) and branchial arch arteries (BAA) of normal caliber in Ccm2 lacZ/lacZ

embryos (below). Anti-Flk1 staining of adjacent sections at the level of the first two

branchial arch arteries is shown (middle). Flk1+ endothelial cells are present at the sites

of the dorsal aortae, cardinal veins and branchial arch arteries (arrows) in Ccm2 lacZ/lacZ

embryos but these cells do not form vessels of normal caliber with patent lumens. Sinus

venosus, SV. Scale bars, 50 µm.


8



9

Supplemental Figure 7. Endothelial cells lacking HEG1-CCM signaling exhibit

defects in tube formation but not endothelial vacuolization. (a–d) CCM2-deficient

endothelial cells exhibit a cell autonomous defect in lumen formation. (a) CCM2 mRNA

levels were reduced by >85% in HUVEC expressing CCM2-specific but not scrambled

(control) shRNA. (b) CCM2-deficient HUVEC migrate normally in response to VEGF

in a modified Boyden chamber assay. No significant differences were observed (P values

≥ 0.1). (c) Control shRNA expressing endothelial cells formed tubes with visible lumens

(black arrows), while CCM2-deficient endothelial cells formed cords that often lacked

detectable lumens (red arrows). GFP is co-expressed with lentiviral shRNAs. (d)

CCM2-deficient endothelial cells form more sprouts but fewer lumens. The number of

sprouts per 20 beads and number of visible lumens per sprout for HUVEC stably

expressing CCM2 shRNA or control shRNA are shown. N = 100 beads for each group;

data shown are representative of four independent experiments. (e) Endothelial vacuole

formation is preserved in zebrafish embryos lacking heg or ccm2. Shown are the

intersegmental vessels of 24-26 hpf fli1a:EGFP-cdc42 transgenic zebrafish embryos

treated with scrambled morpholino (control) or morpholinos to block expression of heg

or ccm2. The GFP-cdc42 fusion protein outlines the endothelial cell vacuoles that form

and fuse during lumen formation (white arrows). (f) The intersegmental vessels of

zebrafish embryos lacking heg or ccm2 are patent. Red fluorescent quantum dots injected

into the dorsal aorta of 24-26 hpf fli1a:EGFP-cdc42 morphant embryos reveals patent

intersegmental vessels lacking heg or ccm2. Note the presence of red quantum dots

within the green endothelial lumens (white arrows) of all embryos. b, bead. Scale bars in

c, 5 µm; in e and f, 20 µm.


10



11

Supplemental Figure 8. HEG1 and CCM2 are required to form normal endothelial

junctions in vivo. (a–c) Endothelial junction length is reduced in the endocardium of

ccm2 morphant zebrafish (mean length of 1314 ± 247 nm for ccm2 morphant junctions

versus 1857 ± 357 nm for control morphant junctions, P = 0.03). (a) The mean and

standard deviation of junction lengths for control and ccm2 morphant endothelial cells are

shown, divided into terciles (1st, shortest third of junctions in each group; 2nd, middle

third of junctions in each group; 3rd, longest third of junctions in each group). N = 88

control junctions and 107 ccm2 morphant junctions. (b) The percent of endothelial

junctions that were less than 1,000 nm (black), 1,000-2,500 nm (grey), and over 2,500 nm

(white) in the indicated groups is shown. (c) Representative low magnification (far left)

and high magnification images of endothelial cells are shown. Arrowheads indicate

endothelial junction limits in each image. Note the presence of a blood cell (bc) in the

space between the endothelial cells (ec) and myocardial cells (myo) in ccm2 morphant

but not control hearts. (d–f) Constricted dorsal aortae in E9 Ccm2lacZ/lacZ embryos also

exhibit shortened endothelial junctions (mean length of 1340 ± 91 nm for Ccm2lacz/lacz

junctions versus 1830 ± 251 nm for Ccm2+/+ junctions, P = 0.03). (d) The mean and

standard deviation of endothelial cell junction lengths from Ccm2+/+ and Ccm2lacZ/lacZ

embryo aortae are shown, divided into terciles (as defined in a). N = 128 Ccm2+/+

junctions and 107 Ccm2lacZ/lacZ junctions. (e) The percent of endothelial junctions that

were less than 1,000 nm (black), 1,000-2,500 nm (grey), and over 2,500 nm (white) in the

indicated groups is shown. (f) Representative low magnification (far left) and high

magnification images of endothelial cells are shown. Note the absence of blood cells in

the Ccm2lacZ/lacZ dorsal aorta.


12



13

Supplemental Figure 9. HEG1 or CCM2 deficiency does not reduce expression of

endothelial junctional proteins. (a) Immunostaining for VE-cadherin (green) in 48 hpf

kdr:EGFP (red) transgenic zebrafish embryos treated with control morpholinos or

morpholinos directed against heg or ccm2 is shown. (b) Immunostaining for the

lymphatic endothelial marker LYVE1 and junctional proteins beta-catenin and claudin-5

in serial sections of dilated lymphatic vessels from a neonatal HEG1-deficient mouse is

shown. (c) Immunostaining for beta-catenin in CCM2-deficient and control HUVEC

monolayers. (d) Immunoblotting for VE-cadherin, beta-catenin and GAPDH was

performed using cell lysate from control and CCM2-deficient HUVEC. Scale bars in a,

20 µm; in b and c, 100 µm.


14


Supplemental Figure 10. HEG1-YFP is localized at cell-cell junctions. HEG1-YFP

and CCM2-CFP expressing CHO cells and HUVEC stained for beta-catenin were

visualized using confocal microscopy. Each horizontal set of images is derived from a

single 0.48 µm thick optical section. (a) HEG1-YFP but not CCM2-CFP co-localizes

with beta-catenin in CHO cells. (b) HEG1-YFP but not CCM2-CFP co-localizes with

beta-catenin in endothelial cells.


15


Supplemental Figure 11. FLAG-HEG1 and FLAG-HEG1ΔC receptors are

expressed at equivalent levels on the surface of live HEK293T cells. Boxed regions

indicate the % of live (propidium iodide-negative) FLAG+ cells.


16


Supplemental Figure 12. ccm2 but not ccm2 L197R rescues the big heart phenotype

of ccm2-deficient zebrafish embryos. Wild type zebrafish embryos were co-injected

with a control morpholino or a splice morpholino to block expression of endogenous

ccm2 and cRNAs to express either wild type ccm2 or ccm2 L197R. (a) Expression of

ccm2 but not ccm2 L197R rescues the development of the dilated heart and pericardial

edema conferred by the ccm2 morpholino. (b) Quantitation of the rescue shown in a. N

= 198 wild type embryos, 1198 ccm2 morphants, 327 ccm2 morphants plus ccm2 cRNA,

170 ccm2 morphants plus ccm2 L197R cRNA; P < .0001 for rescue with ccm2 cRNA and

P = 0.7 for rescue with ccm2 L197R cRNA. Scale bars, 100 µm.


17

Supplemental Movies 1–3. Endothelial vacuole formation is preserved in zebrafish

embryos lacking heg or ccm2. Shown are the intersegmental vessels of 24–26 hpf

fli1a:EGFP-cdc42 transgenic zebrafish embryos following injection of scrambled

morpholino (control, Supp. Movie 1) or morpholinos to block expression of heg (Supp.

Movie 2) or ccm2 (Supp. Movie 3). The GFP-cdc42 fusion protein is expressed in

endothelial cells and outlines the vacuoles that form and fuse during lumen formation.

Supplemental Movies 4–6. The intersegmental vessels of zebrafish embryos lacking

heg or ccm2 are patent. Red fluorescent quantum dots were injected into the dorsal

aorta of 24–26 hpf fli1a:EGFP-cdc42 transgenic zebrafish embryos following injection

of scrambled morpholino (control, Supp. Movie 4) or morpholinos to block expression of

heg (Supp. Movie 5) or ccm2 (Supp. Movie 6). The presence of red fluorescent dots in

the intersegmental vessels indicates vessel patency.


18

Supplemental Table 1.

E10.5 (129/B6)

E14.5 (129/B6)

E18.5 (129/B6)

P1 (129/B6)

P21 (129/B6)

P1 (B6)

P21 (B6)

Heg1+/+ 8 10 19 130 164 17 12 Heg1+/– 21 25 27 237 324 25 23 Heg1–/– 7

(1 dead) 11

(2 dead) 10

(6 dead) 90

(20%) P=.009

66 (12%)

P=.0001

9 (18%)

4 (10%)

Supplemental Table 1. Offspring of Heg1+/– X Heg1+/– crosses. Shown are the

offspring of Heg1+/– intercrosses on mixed SV129:C57Bl/6 and >95% C57Bl/6

backgrounds. The expected % of Heg1–/– offspring is 25%. P values were calculated

using Chi Square analysis.

Supplemental Table 2.

Heg1: Heg1+/+ Heg1+/–

Heg1–/– Heg1+/+ Heg1+/–

Heg1–/– Heg1+/+ Heg1+/–

Heg1–/–

Ccm2: Ccm2+/+ Ccm2+/+ Ccm2+/lacZ Ccm2+/lacZ Ccm2lacZ/lacZ Ccm2 lacZ/lacZ E8.5 25 6 39 13 19 5 P1 31 8 60 0 0 0

% observed % expected

30% 19%

9% 6%

61% 38%

0% 12%

P=.0001

0% 19%

P=.0001

0% 6%

P=.0001

Supplemental Table 2. Offspring of Heg1+/–;Ccm2+/lacZ X Heg1+/–;Ccm2+/lacZ crosses. P

values shown were calculated using Chi Square analysis.


19

Supplemental Methods

Immunostaining. For whole-mount X-gal staining, embryos were fixed for 1 h in 2%

paraformaldehyde, rinsed in PBS for 2 h and stained with X-gal staining solution for 2-24

h. For fluorescent immunocytochemistry, HEG1-YFP and CCM2-CFP were transfected

into CHO cells with Lipofectamine LTX (Invitrogen) and into HUVEC with FugeneHD

(Roche) or nucleofection (Amaxa), according to manufacturer’s instructions. At 10-12 h,

cells were washed once in PBS, fixed in 4% paraformaldehyde for 15 min, washed 3 x 5

min in PBS, permeabilized with PBSX (PBS + 0.1% Triton X-100) for 10 min, and

washed 3 x 5 min in PBS. Cells were blocked with 2% BSA in PBS for 15 min and then

incubated with primary antibodies overnight at 4 °C. Cells were washed 5 x 10 min in

PBS, reblocked with 2% BSA in PBS for 15 min, and then incubated with secondary

antibodies for 1 h at room temperature. Finally, cells were washed 3 x 10 min with PBS

and mounted with ProLong Gold (Invitrogen). All steps were performed at 25 °C unless

otherwise specified. The following antibodies were used: mouse monoclonal antibody to

beta-catenin, 1:1000 (BD Transduction); Alexa 568-conjugated goat antibody to mouse

IgG (Invitrogen), 1:1000. Images were taken with a Nikon Eclipse 80i or a Leica SP2

confocal microscope (0.48 µm slice thickness, 63X objective) and processed in ImageJ.

For zebrafish whole mount immunofluorescence, 48 hpf embryos were fixed in 4%

paraformaldehyde for 2 h at room temperature. Embryos were washed 2 x 5 min in

PBST (PBS + 0.1% Tween) and 3 x 1 h in PBSTX 0.5% (PBS + 0.1% Tween + 0.5%

Triton X). Embryos were blocked in PBSTX (PBS + 0.1% Tween + 0.1% Triton X) +

10% BSA + 1% NGS for 2h and then incubated with primary antibodies (in PBSTX +


20

1% BSA + 0.1% NGS) overnight at 4 °C. Embryos were washed 6 x 1h in PBSTX + 1%

BSA + 0.1% NGS and then incubated with the secondary antibody (in PBSTX + 1%

BSA + 0.1% NGS) overnight at 4 °C. Embryos were finally washed several times in

PBST. All steps were performed at RT except for antibody incubations. The following

antibodies were used: mouse monoclonal antibody to human ZO-1, 1:200 (Zymed);

rabbit polyclonal antibody to zebrafish CDH5 {Blum, 2008 #129}, 1:200; Alexa 568-

conjugated goat antibody to rabbit IgG (Invitrogen), 1:1000; Alexa 568-conjugated goat

antibody to mouse IgG (Invitrogen), 1:1000. Images were taken with a Leica SP1

confocal microscope.

Real-time PCR analysis. Total RNA was generated by using RNEasy (Qiagen). For

reverse transcriptase reactions, 0.5-1.5 ug total RNA and 100 ng random hexamers were

used to generate cDNA by using the First Strand cDNA Synthesis Kit (Invitrogen). Real-

time quantitative PCR was performed using a commercially available SYBR Green

master-mix reagent (Applied Biosystems). Sequences of real-time PCR primers are

available below.

Zebrafish studies. For angiography of intersegmental vessels, 0.1 µM Qtracker (655)

quantum dots (Invitrogen) were injected in the dorsal aorta of 24-26 hpf Tg(fli1a:EGFP-

cdc42)y48 zebrafish embryos. Embryos were embedded in a drop of 0.7% low melting

agarose containing 0.01% tricaine (Sigma) (pH 7.5) and images were acquired every 5

min using a Leica Sp1 confocal microscope (slice thickness 1 µm, 63X objective) at a

temperature of 28.5 °C. Movies were processed using Imaris (Bitplane) and ImageJ.


21

Cell culture. HUVEC (Cambrex) were maintained in EGM2 media (Lonza) and used

between passages 3 and 8. CHO and HEK293T cells were cultured in DMEM with 10%

FBS and 1% pen/strep.

Lentiviral shRNA. pGIPZ lentiviral vectors encoding either human CCM2-specific

shRNA (Open Biosystems Cat No. RHS4430-98485624) or scrambled shRNA were

packaged in HEK293Ts according to the manufacturer’s protocol (Open Biosystems),

concentrated with Centricon Plus-20 centrifugal filters (Millipore), and titered by GFP

positivity in HUVEC (Lonza). Passage 2 HUVEC were infected with lentivirus at an

MOI of 1-2 and GFPhigh HUVEC were sorted by flow cytometry (BD FACSVantage with

DiVa option). CCM2 knockdown was verified by real-time RT-PCR using two different

primer sets. Primer sequences are available in Supplemental Methods.

Endothelial assays. Endothelial tube formation was assessed using a fibrin bead assay,

as previously described{Nakatsu, 2003 #130}. Briefly, HUVEC were cultured with

dextran-coated Cytodex 3 beads (Amersham Pharmacia Biotech) at roughly 400

cells/bead. Approximately 100 HUVEC-coated beads were embedded in a fibrin clot in

one well of a 24-well tissue culture plate. 2 x 104 skin fibroblasts were seeded on top of

the clot. Assays were monitored for 10 days. Number of sprouts per bead and number of

lumens per sprout were quantitated on day 10.


22

Endothelial migration was assessed using a modified Boyden chamber assay, as

previously described {Shen, 2008 #134}. Briefly, the upper surface of a 6.5mm

Transwell insert (8.0µm pore size) (Corning) was coated with 0.1% gelatin. HUVEC

grown to 80% confluency were starved in EBM2 (Lonza) with 0.2% FBS overnight.

Cells were detached from the plate with 5mM EDTA, rinsed and resuspended in EBM2

with 0.2% FBS. 5x104 cells were added to the insert well and the lower chamber was

filled with 600µl of EBM2 with 0.2% FBS with or without 7.5ng/ml VEGF-165 (R&D

Systems). After 4 hours incubation at 37 °C, non-adherent cells were removed by

aspiration and non-migrated cells on the upper surface of the insert well were wiped off

using a PBS wetted cotton swab. The insert wells were stained with coomassie blue

staining buffer (0.05% coomassie blue in 30% methanol, 10% acetic acid) for 10 min and

rinsed with water. The insert well filter was cut off and mounted. The number of cells in

four 20x fields was counted for each filter. The relative migration was calculated by

normalizing the number of migrated cells in each condition to the number of migrated

cells in the condition of scrambled shRNA HUVEC without VEGF. The chemotactic

index was defined as the number of migrated cells with VEGF divided by the number of

migrated cells without VEGF. Values are the mean and standard deviations from three

independent experiments in triplicate.

Flow cytometry. HEK293T cells were detached with PBS containing 10mM EDTA.

10^5 cells were rinsed and stained in Tyrodes buffer with 1ug of anti-FLAG M2 FITC

(Sigma) at 4 °C for 1 h. Cells were rinsed and resuspended in Tyrodes buffer containing

50µg/ml propidium iodine for flow cytometric analysis (BD FACSort).


23

Anti-CCM2 antibody production and characterization. Polyclonal anti-CCM2 #2055

was raised in rabbits against recombinant GST-CCM2 fusion protein containing the C-

terminal 55 amino acids of CCM2, coupled to keyhole-limpet hemocyanin. Polyclonal

anti-CCM2 #2055 antisera was validated by immunoblotting using purified GST-CCM2

and cell lysates from HEK293T cells transfected with FLAG-tagged full-length CCM2.

Primers Genotyping primers: Gene Primer name Primer sequence

Forward 5'-CGGCTCCCACAACTTTTGC-3' Reverse (wildtype) 5'-CTCACGCCTCTGGGGACAC-3'

Heg1

Reverse (mutant) 5'-CCAGGTGACGATGTATTTTTCG-3' Forward 5’-GAAGAGTTGTGCTCCCTGCT-3’ Reverse (wildtype) 5’-CATCCCTGTCTGGGAACCTA-3’

Ccm2{Plummer, 2006 #17}

Reverse (mutant) 5’-TCTAGGACAAGAGGGCGAGA-3’ Forward 5’-GAACCTGATGGACATGTTCAGGGA-3’ Cre Reverse 5’-CAGAGTCATCCTTAGCGCCGTAAA-3’ Forward 5'-TCAATCCGCCGTTTGTTCC-3' R26R LacZ Reverse 5’-ACCATTTTCAATCCGCACCTC-3’

Real-time RT-PCR primers: Gene product Primer name Primer sequence

Forward 5’-ATCGAACACTCCTCCCGGT-3’ mHeg1 Exon 1-2 Reverse 5’-TTCTTTGTGACCTGGAACTGCTGG-3’ Forward 5’-GGAGGAGTTACGCAGAGTCTTCATCT-3’ mHeg1 Exon 4-5 Reverse 5’-ATTGCACCATCTTCCGGAGGAGAA-3’ Forward 5’-GCGTGGCACTCATTGTTACCTGTT-3’ mHeg1 Exon 15-16 Reverse 5’-ACGTCTGTCATCTGCAGGAGGTTT-3’ Forward 5’-AGCCATTGAAATGCACGAGAACGG-3’ mHeg1 Exon 16-17 Reverse 5’-TCTCGTCGCTGATGAAAGATGGGT-3’ Forward 5’-ACCACCTCCACATCCACCATCAAT-3’ mCcm2 Reverse 5’-TCTGAAATCATGCGGTCCCACTC-3’ Forward 5’-CAAAATGGTGAAGGTCGGTGTGAAC-3’ mGapdh Reverse 3’-TTGATGTTAGTGGGGTCTCGCTCC-3’

hCCM2 set 5 Forward 5’-TTCCCTGAATCTGTGGATGTGGGT-3’ Reverse 3’-TGCAAACTGCTGGATCTCCTGTGA-3’ hCCM2 set 12 Forward 5’-ACACTGTGGTGTTGTCATTGCCTG-3’


24

Reverse 3’-TGGTGGGCTCTCTTTGCATTGTCT-3’ hGAPDH Forward 5’-GAAGGTGAAGGTCGGAGTC-3’ Reverse 3’-GAAGATGGTGATGGGATTTC-3’ Miscellaneous primers: Gene product Primer name Primer sequence

Forward 5'-GCAATACATCCAGGAGGTGCCTAC-3' Heg1 southern probe Reverse 5'-CCGTTATCTTTCGATGCCTCG-3' Forward 5’-TGTAAGCGGAAGAGTCCAGAATG-3’ Heg1 in situ probe #1 Reverse 5’-GTTGCGTTCAAGTTCAGGGTTC-3’ Forward 5’-GACAGCTTTGGTAGGCATCGTC-3’ Ccm2 in situ probe Reverse 5’-ATATGCTACAGGCATTGCTGAGC-3’


supplemental figure 1. - media.nature.com · 2 supplemental figure 1. characterization of the heg1...

Documents