Remodeling of ventricular conduction pathways in healed canine infarct border zones

Remodeling of ventricular conduction pathways

in healed canine infarct border zones.

R A Luke, J E Saffitz

J Clin Invest.

1991;

87(5)

:1594-1602.

https://doi.org/10.1172/JCI115173

.

Remodeling of myocyte interconnections may be an important determinant of ventricular

tachycardia in regions bordering healed infarcts. We used quantitative electron microscopy

to characterize the distribution of gap junctions in 10 canine left ventricles 3-10 wk after

coronary occlusion. In three normal canine left ventricles analyzed ultrastructurally,

myocardial gap junctions were distributed anisotropically; gap junction profile length was

significantly greater in the transverse than in longitudinal planes of section. In infarct border

zone tissues, the normal anisotropic distribution was completely abolished and fewer gap

junctions per unit intercalated disk length were observed. Analysis of individual gap junction

profile length distributions revealed selective disruption of the largest gap junctions that

collectively comprised only 9.6% of total junction profiles, but encompassed nearly 40% of

aggregate gap junction length in the transverse plane of section. Three-dimensional

reconstructions of myocyte interconnections by high resolution quantitative light microscopy

of serial sections demonstrated a reduction in the number of cells connected by intercalated

disks to a single myocyte from 11.2 +/- 1.0 in normal myocardium to 6.5 +/- 1.3 in border

zone tissues (P less than 0.001). Connections of cells in primarily side-to-side apposition

were reduced by 75%, whereas primarily end-to-end connections were reduced by only

22% (P less than 0.05). These alterations would disproportionately enhance axial resistivity

in the transverse direction, potentially contributing to development […]

Research Article

Remodeling of Ventricular

Conduction Pathways

in Healed

Canine Infarct Border

Zones

Robert A. Luke andJeffreyE.Saffitz

Cardiovascular Division and Department ofPathology, Washington University School ofMedicine, St. Louis, MO

Abstract

Remodeling ofmyocyteinterconnectionsmaybeanimportant

determinant of ventricular tachycardia in regions bordering healed infarcts. We used quantitative electron microscopyto

characterize the distribution ofgapjunctions in 10 canine left

ventricles 3-10wk after coronaryocclusion. In three normal

canine left ventricles analyzed ultrastructurally, myocardial

gapjunctions were distributed anisotropically; gapjunction profile lengthwassignificantlygreaterin thetransversethan in longitudinal planes of section. In infarct borderzonetissues,

the normalanisotropicdistributionwas completely abolished

and fewergapjunctionsperunit intercalated disk lengthwere

observed.Analysis of individualgapjunction profile length

dis-tributions revealed selective disruption of the largestgap

junc-tions that collectively comprised only 9.6% of total junction profiles, but encompassed nearly 40% ofaggregategapjunction

length in the transverse plane of section. Three-dimensional reconstructions ofmyocyteinterconnections by high resolution quantitative light microscopy ofserial sectionsdemonstrateda

reductioninthe number of cells connected by intercalated disks

toa singlemyocyte from 11.2±1.0 in normal myocardium to

6.5±1.3 in borderzonetissues(P<0.001).Connections of cells in primarily side-to-side apposition were reduced by 75%, whereas primarily end-to-end connectionswerereducedby only

22% (P < 0.05). These alterations would disproportionately

enhance axialresistivityinthe transversedirection, potentially contributingtodevelopmentof reentrantarrhythmias. (J.Clin.

Invest. 1991.87:1594-1602.) Key words: Ventricular

tachycar-dia*gapjunctions*electronmicroscopy*fibrosis*intercalated

disks

Introduction

The pathogenesis ofreentrantarrhythmias requires both

unidi-rectional block and slow conduction within thereentrant cir-cuit(1).Whileaconsiderablebodyof evidence hasimplicated

alterations of active membranepropertiesinthedevelopment

ofconduction abnormalities, morerecent observations have

indicated that changesinpassive myocardial resistivity likely playaroleaswell(2-8). Slow, heterogeneouselectrical

propaga-Address correspondencetoDr.Jeffrey E. Saffitz,Department

ofPathol-ogy,Box 8118, Washington University School of Medicine,660 So. Euclid Ave., St. Louis, MO 63310.

Receivedfor publication8 May 1990 andin revisedform28

No-vember 1990.

tion and complex fractionated electrograms have been ob-served in fibrotic myocardium with normal cellular active membrane

properties, which

has led to

speculation

that

alter-ations of

specific

patterns

of

myocyte interconnections in

re-gions bordering healed infarcts may be an important compo-nent

of

the

anatomic

substrate

of

reentrant

ventricular

tachy-cardia(3, 5, 6, 8-14).

Spach and Dolber (15) have

demonstrated

lossof electrical

coupling of atrial

myocytes

oriented

in

side-to-side apposition

due to

progressive interstitial fibrosis associated with aging.

Others,

such as Dillon et al. (5), have characterized reentrant

circuits

in healed

infarct

borderzonesand havesuggested that

alterations

in

side-to-side connections similar

to those

de-scribed

in the

aged

atria may be

important determinants

of

arrhythmogenesis. The

present

study is

an

extension

of

the work

of

Spach and Dolber and was

designed

torigorously

quantify

the

three-dimensional distribution of

myocyte

inter-connections in selected

regions

of the healed infarct border

zones

known

to be

sites of slow heterogeneous

propagation

critical

inreentrant

arrhythmogenesis.

We have recently

characterized

the

three-dimensional

structure

and distribution of

gap

junctions

and patterns

of

cel-lular

interconnection

in normal

canine left ventricular

myo-cardium

using light

andelectron

microscopic

morphometry (16, 17). We observed an

extraordinary

level

of

structural

com-plexity.

Forexample, the

typical canine ventricular

myocyte has on

its surface

- 100 gap

junctions

located

in

specific

do-mains

of

the

intercalated

disk and

having specific sizes, shapes,

and

orientations ( 16).

Asubset

of

these

junctions

includes

ex-ceptionally long (up

to

8

,tm in length), ribbon-shaped

gap

junctions

oriented with their long

axes transverse tothe myo-cyte

longitudinal axis,

an arrangement that

is

likely

to

mini-mize

shear

forces

on

stiff

patches

of

densely

packed

channels

connecting adjacent cells.

These

junctions

arelocated in

inter-plicate regions of the intercalated disk.

The

remaining

gap

junctions

are

small

discoid

structures

located within

the

plicate

segment

of

the

intercalated disk.

Representative

electron

mi-crographs of long

interplicate

and smaller

plicate

segment gap

junctions

areshown in

Fig.

1.

We have also shown thatan

individual

cell

is connected

by

intercalated

disks

to - 10 other cells in

variable degrees

of

side-to-side

and end-to-end

packing

(

16).

Inthe present

study,

we

applied

a

similar

quantitative

structural

approach

to

charac-terize

the

effects of interstitial fibrosis

on

intercellular

junctions

in

epicardial regions bordering

healed

canine infarcts.

The

re-sults

indicate

that

accumulation

of interstitial collagen

coin-cides

with the

disruption

and

remodeling

of intercellular

junc-tions. Long transversely oriented

gap

junctions appeared

tobe

especially

vulnerable andwere

essentially

absent

from infarct

border zone

myocardium.

These structural

derangements

would enhance the

anisotropy of cellular connections

and

dis-proportionately increase

the

axial

resistivity encountered by

a

wavefront

propagatedtransverse tothe

long fiber

axis.

1594 R. A.Luke andJ. E.Saffitz

J.Clin. Invest.

©TheAmerican Society for Clinical Investigation,Inc.

0021-9738/91/05/1594/09 $2.00

Figure1.Electronmicrographs of gap junction profiles in normal canine epicardial tissue. A long,ribbon-shaped gapjunction(between

arrows) is readilyapparent in thetransverse

planeof section (toppanel). Small discoid gap

junctionsinplicatesegments(open arrow)and short axis views of the long, ribbon-like junctions in interplicate segments(closedarrows)are

shown in the longitudinal plane of section

(bot-tompanel).Bar, 1.0 !m in eachpanel.

Methods

Tissuepreparation for electron microscopic morphometry. Tissue was

obtainedfrom 10 adult mongrel dogs of either sex that had previously beensubjected to coronary occlusion and from 3 adult mongrel control dogs.Surgical ligation of the left anterior descending coronary artery wasperformed as previously described(18). After convalescence under approvedveterinary care for 3-10 wk, the animals were reanesthetized withsodium thiopental (20 mg/kg i.v.) and the hearts were excised as previously described (17). Tissue samples -1 x 1 X 0.2 cm in size weredissected from left ventricular epicardial regions overlying trans-muralinfarcts in the experimental animals and from corresponding left ventricular regions of control animals. The tissue was immersed in modified Karnovsky's fixative (2.0% glutaraldehyde and 1.0% parafor-maldehyde in 0.08 M sodium cacodylate buffer containing 2 mM cal-cium chloride, pH 7.4) and fixedat

40C

for at least12h.

10-20tissue blocks measuring 1 X 1 X0.5 mmwere cut from each epicardial tissue sample.Adissecting microscope was used to deter-mine fiber orientation and to aid in selection ofblocks composed of an interspersion of myocytes and collagen. Individual blocks were pro-cessedfor electron microscopy by postfixation for 15 min in 1.0% os-miumtetroxide, dehydration in ethanol, and embedment in Spurr's resin. Thick sections (0.5

Mm)

ofeach block were examined light

micro-scopicallytoverifythat myocytes had been cut in planes either parallel

orperpendicularto thelongitudinal fiber axis and to determine the structuralcompositionof the sample.

Aspectrumof structural features ranging from 100% collagenous

scartobundles of normal appearing myocytes separated from other bundlesby collagenous septa or even entirely normal myocardium was observed in - 150 blocksexamined from epicardial infarct border

zones. Inview of the wide variability in the structural composition of border zone regions and the expectation that the corresponding effects

onintercellular connections would range from total disruption in fully scarredregionstonegligible alterations within normal appearing

myo-cardium, these regions werenotsampled randomly. Rather, we se-lectedspecific samples for analysis that were composed ofcardiac myo-cytesseparated bydiffuse interstitial fibrosis with maintenance of nor-mal fiber anisotropic orientation. Fibrosis in these samples was

exclusively interstitial and did not appear to have been formed by

re-placement ofnecrotic myocytes.

Becausegapjunction membranesare orientedanisotropically in normalcanineventricularmyocardium, intercellular junctions had to

beanalyzedmorphometricallyin three planes of sectionorthogonalto

thelongitudinal fiber axis.Innormaltissue, thiswasaccomplished by sectioning independent tissue samples in each orthogonal plane, as

reportedpreviously (16). Intheinfarct border zone, however, struc-turalheterogeneity made this approach impractical. Thus, sections of individual border zone tissue blocks were analyzed morphometrically in three planes orthogonal to the long fiber axis by first cutting sections in onelongitudinal plane (e.g., paralleltothelong fiber axis and paral-lel to theepicardialsurface), thenreorienting the block in the

micro-tomeby 900topermitsectioning in the other longitudinal plane (paral-lel tothelongfiber axis andperpendiculartotheepicardial surface), and finally reorienting the block asecond time to prepare sections transversetothelong fiber axis.

Atotalof10 blocks ofborder zone tissue (1 from each of 10 animals subjectedtocoronaryocclusion) andatotalof 5 blocksfrom 3 control animals (2 blocks from each of 2 control animals and 1 block from 1

controlanimal)wereanalyzed. Ultrathin sectionswereprepared from eachorientation and collectedon200-meshcopper-rubidiumgrids after fiber orientation had been verified with lightmicroscopyof

0.5-gmthick sections. Ultrathin sectionswerestained with

uranyl

acetate

and leadcitrate, examined withanelectronmicroscope (model EM-200;Philips Electronic Instruments, Mahweh, NJ), andphotographed

with Eastman 5302finegrain release positive 35-mm film (Eastman KodakCo., Rochester, NY).

Quantitativemorphometricanalysis ofelectron

micrographs.

At

leastfivetestregionswererandomly selected for ultrastructural analy-sis from each group of ultrathin sectionscutin eachorientation of each tissue blockaspreviouslydescribed(16).Individualtestregionswere

photographedat afinal printmagnificationofx5,000. Then,all por-tions of eachtest areacontainingintercalateddisks and gapjunction

profileswererephotographed for furtheranalysisat afinalprint

magni-fication of x24,000.Atotalof 432highpowermicrographsof control tissue and 875micrographsof fibrotic tissuewereanalyzed.

eachhigh power micrograph usinganelectronic stylus and digitizing tablet (HoustonSystems Inc.,Austin,TX). Datawereexpressedasgap junction profile length per 100 ,um intercalated disk length, or gap junction profile length per unit myocyte area. The percentages of total section areaoccupied by cardiac myocytes and by interstitial fibrosis weremeasured inx5,000 micrographs of each individual test area by placing apoint array (1.0 cminterpoint distance) over each 8 X 10 inch micrograph and counting the number of points located over interstitial andnoninterstitial structures ( 19).

Inaddition tomeasuring gap junction profile lengths, individual gapjunctions were categorized as being located in either plicate or interplicate regions of the intercalated disk. Plicate region gap junc-tionsweredefinedasthoselocated within theinterdigitating adhesive processesof the disk, whileinterplicate junctions were located in junc-tional membranes between plicate segments.

Light microscopicanalysis ofpatternsofintercellular connection. Additional tissue blocks were analyzed with light microscopy to deter-mine the number of cardiac myocytes to whichanindividual myocyte of normal and borderzonetissuewasconnectedby intercalateddisks and tocharacterize therelativeside-to-side or end-to-end orientation of these interconnections. Blocks measuring -1 X 1 X 0.5 mmwere dissected fromepicardialregionsoverlying healed transmural infarcts andfrom corresponding regions of controltissue, postfixedfor 15 min in 1.0% osmium tetroxide containing 1.0% potassium ferrocyanide, dehydrated inethanol, and embedded inSpurr'sresin. The addition of

potassium ferrocyanideenhanced sarcolemmalstaining,thereby facili-tating identification of cell borders andspecifically junctional

mem-braneatthelight microscopic level of resolution ( 16).Acellular inter-connectionwastabulatedonly afterunequivocalidentification ofan

intercalated disk between the cells. Itwasassumed that the presenceof

one or moreintercalated disksconnectingtwocellsatthelight

micro-scopic levelof resolution indicated electrical interconnection atgap

junctions.

Aseries of 40consecutivesections,each - 1.5uiminthickness,was

cutinaplanelongitudinal tothe longfiber axisfrom each of four blocksof borderzonetissue and four blocksof normalmyocardium.

Sectionswerestained withtoluidine blue and each sectionwas

photo-graphedatafinalprint

magnification

of 1,000.Five individual in-dexcardiac myocyte profileswere randomlyselected from the 20th sectionof eachsetof 40 serialsections,and theseindex cells and all

neighboringmyocytestowhich theywereconnectedby intercalated disks weredelineated and countedby identifying, in their

entirety,

profiles of these cells in serial sections. Inadditiontodeterminingthe numberof cells connectedtothefive index cells in eachsetof serial

I

II

XT~~

1a=

lV LI

IJ

Figure 2.Diagramof types of intercellular connections used in three-dimensional reconstructions.

sections, thespatial distribution of each interconnectionwas catego-rized on the basis of its relative side-to-side or end-to-end orientation accordingtothefollowing system: Type I,>75%ofthe lateral borders of thetwocellsoverlapped; Type II, 25-75% lateral border overlap; Type III,<25%lateral border overlap and<50%end-to-end overlap; TypeIV,<25%lateral borderoverlap and>50%end-to-end overlap. Schematic diagrams of these types of connectionsare illustrated in Fig. 2.

Statisticalanalysis.Datawereexpressedasmeans±SD. Morpho-metric data were analyzed by nonparaMorpho-metric Wilcoxon tests (20)

ex-ceptasnoted.Histograms of gap junction profile lengthswere

exam-ined byanalysis of covariance (20). Because of the skewed lognormal distributions, analysis of covariancewasperformedfollowing logarith-mictransformation of the data with the log values as the dependent variables. Independent variables included the logof the bin, the group, and theinteraction between thetwo.Thefirsttwobins in each histo-gramwereeliminated asoutliers that violated the overall pattern ofthe distribution.Accordingly, the statisticalanalysis doesnotapplytothe two smallestbins of gap junctionprofile length. The data were also testedwith x2 analysis of the percentage of profiles above and belowan

arbitrary cut off of 3.0

Am.

x2analysiswasalso usedtotestthe dichoto-mousdistribution of gap junctions inplicate and interplicate regions of intercalated disks (20). NonparametricWilcoxontests wereusedto test

differences in each connection type between normalregionsand border

zoneregions (shown in Table I). Thesignificanceof the percent reduc-tions ofspecificintercellular connection typeswastestedusing analysis

of covarianceon aranktransformation of the data.

IP

NO.02-v

32

'

₂₈

.

4-.0

.c

24

?

20

.a

8

₁₆

0

.o

12

0

8

0 C

0,o

Normal

Inforct Border Zones

(EM)

(EM)

(LM)

Figure 3. Relativeproportionsoftest areasoccupied byinterstitial space(allnonmyocytestructures)in normal(n= 15)andborder

zoneregions(n=_{30) analyzed}with electronmicroscopy

(EM)

and in borderzonetissues (n=_{4) analyzed}withlightmicroscopy(LM)

of serial sections. Interstitial space in normalsamples

analyzed

with lightmicroscopywas5.4±0.6%of total sectionarea(P=NS

com-pared with normalsamplesanalyzedultrastructurally).

1596 R. A.LukeandJ.E.

Saffitz

IF 0 i A

---13

ri

Results

Electron

microscopic morphometry of

intercellular connections. Therelativeamountsof total sectionareaoccupied by

intersti-tial spaceincontrol and infarct borderzone testregionsare

shown in Fig. 3. In normal tissues, the interstitium occupied - 6% ofsection _{area, a} valuesimilartothose reported in previous morphometric studies of canine ventricular

myocar-dium (21). In border zone regions, interstitial space

encom-passed - 22%oftotal sectional _areain samples analyzed by electronmicroscopy and - 15%insamples analyzed by light microscopy. Expansion of the interstitialspace wasdueto

ac-cumulation of bundles of collagen fibrils,asshownina

repre-sentative electron micrograph (Fig. 4).

Results of morphometric analysis ofgapjunction size and

distribution in normal and infarct borderzoneregionsare

shown inFigs.5and6. Becausenosignificant differenceswere

observedinanyparameteranalyzedinthelongitudinal planes

perpendicular

and

parallel

tothe

epicardial

surface, data from thesetwoplaneswerepooled andarereferredtoinFigs. 5-7as

the

longitudinal

planes

of

section. As shown in

Fig.

5, the

an-isotropic distribution of

gap

junction profile

length

is

readily

apparent in normal

canine myocardium.

Whether expressed as gap

junction profile

length per

unit

myocyte area or per

unit

intercalated

disk

length, values observed

in the

transverse planes

of section

were

significantly

greater than those

in

the

longitudinal

planes. As

demonstrated previously,

the

anisotro-pic

distribution

is a

consequence of

long,

ribbon-shaped

gap

junctions,

located

in

interplicate

segments,

and

having their

long axes

oriented

transverse to the long

fiber axis.

Thus, as shown

in

Fig. 6,

the average

individual

gap

junction profile

length in normal ventricular

myocardium

wasgreater

in

the transverse than

in

the

longitudinal planes of section (1.35

vs. 0.87 um,

respectively).

However, the number

of

gap

junction

profiles

wasthe same in both

orientations

(12.9 vs.

13.2

pro-files/ 100

,gm

intercalated

disk length

in

longitudinal

and trans-verseplanes of section,

respectively

[Fig. 6]).

Inmarked contrast to the pattern observed

in

normal

myo-cardium,

the

anisotropic distribution of

gap

junction

length

was

significantly

altered in the fibrotic

myocardium

of

the

in-Figure 4. Electron micrograph of a typical epicardial border zone region cut in a longitudinal plane of section. Interstitial bundles of collagen (C)

farct

border zone

(Fig.

5).

Compared with

normal

tissue,

there was a significant reduction in gap junction profile length in both

longitudinal

and transverse planes

of section,

but

this

dec-rement was

far

greater in the transverse plane, such that gap junction length per unit myocyte area or intercalated disk length became

equivalent

inboth orientations. The reduction inaggregate gapjunction profile length in border zone tissues was the result of there being fewer individual profiles per unit disk

length

in both

longitudinal

and transverse orientations (Fig. 6), although the

difference

in the transverse planefailed to

achieve statistical

significance

due to a

relatively

large standard

deviation. Nevertheless, these

data

indicate

that the

increase

of

interstitial fibrosis coincided with

a

disruption of cell junctions

with fewer individual gap

junction profiles

per

unit

interca-lateddisk length. Moreover, the average length

of

an

individual

gap

junction

inborderzone

samples

was

significantly

reduced in the transverse plane but not in the

longitudinal

planes

(Fig.

6).

These

observations

suggest that

long,

ribbon-shaped

gap

junction profiles,

known to have

their long

dimensions

ori-ented

transverse to the

longitudinal

fiber

axis,

may be

selec-tively

disrupted

in

fibrotic

tissue.

Histograms

of

the

distribution of

individual

gap

junction

profile lengths

in the transverse

and

longitudinal planes

of

sec-tion of

normal and

infarct

border zone

samples

areshown in

Fig.

7. The greater aggregate and mean

individual

gap

junc-20

0 x)< 16

EE

= v 12

0' 0 C

8

O 8

0 ₄ 0 @ 4

CD,

0

20

0.0' _.- c

E

tL ,,, 16

O. 4

0 6 12

o- 0

c- c 8

._ c

-f

c o tL

O 4

rPO.001h1

-NS _I

E-NS---Normal Border Zones

Transverse Longitudind Transverse Longtudinal

Normal BorderZones

Tran e Longitudinal Transve Longtudinal Figure5.Total gapjunctionprofilelength per Mm2 myocytearea(top

panel)and per 100Mmintercalated disk length(bottom panel)

mea-sured intransverseandlongitudinalplanes of section in normaland

borderzonetest areas.

.- 2.0- _P<0.001

ZL

-r-FPO.002>-i

1.6

a'~~~~~~~~N

CL

0

0.

C

0.2

.4

-0)

Normal BorderZones

Transverse Longitudinal Transverse Longitudinal

I P=0.12

20_-NS

0= 16

-NS-i

C_P.

0

12

E

4

0.

Normal BorderZones

Tranverse fLngiludinal TransVo Lgudia

Figure 6. Mean gapjunctionprofile length

(,m)

(toppanel), and the numberof individual gap junctionprofiles per 100Amintercalated disk length(bottom panel)intransverseandlongitudinalplanes of section innormal andborderzone test areas.

tional

profile lengths

observed

in

the transverse

compared with

the

longitudinal

planes

in

normal

tissue (demonstrated

in

Figs.

5

and

6)

are seen

in

Fig.

7 to

be

due to a

subset of

long

gap

junctions (arbitrarily defined

as

having

a

profile length

>3.0

Am),

apparent

only

in thetransverse

plane

of section. In the

"normal transverse" histogram,

17

of 178

gap

junction profiles

were >

3.0 Mm

in length. Although only 9.6%

of

the total num-ber

of profiles,

these

junctions comprised 37.5% of

the aggre-gate gap

junctional length

in the transverse

plane of section.

Becausethe "short" axes

of these long,

ribbon-shaped

gap

junctions

are

oriented parallel

tothe

longitudinal

fiber

axis,

few

long

profiles

areobserved in normal

tissue

cut

in

the

longitu-dinal

planes of

section

(Fig. 1).

Incontrast tothe transverse

profile length distribution of

normal

tissue, infarct

border zone

samples showed

a marked

paucity of long

gap

junction profiles

in the transverse

plane

(Fig.

7,

bottom

left).

Only

2

of

244 (<

1%) profiles

in

the

"border

zone transverse"

histogram

had a

length

>3.0 um,

compared with

9.6% in normal

tissues (P

<

0.001).

Thus,

loss

of

the

anisotropic

junction distribution and reduction of the

average gap

junction profile length

in the transverse

plane of

infarct

border zone

samples (shown

in

Figs.

5and

6)

weredue to

selective

disruption

of

the

longest

gap

junction

profiles.

Inthe

longitudinal planes of

section,

no

significant

differ-ences were found in the gap junction length distribution

of

35-NORMALLONGITUDINAL (n=₂₂₀₎

30-

25-IfL fr lnm

Dnn

I I

I I

1.0 2.0 3.0 4.0 5.0

BORDER ZONE TRANSVERSi

0

J

hi

0

Ci

z

41

0

6.0 7.0 8.0 9.2

I.-U

z

E(n=244) a.

4 0

Un.

0

z

1.0 2D 3.0 4.0 5.0 6.0 7.0

GAP JUNCTION MEMBRANE PROFILE LENGTH (Hem)

8.0

BORDER ZONE LONGITUDINAL (n=235)

1.0 2.0 3.0 4.0 5.O 6.0 70 8.0

GAP JUNCTION MEMBRANEPROFILE LENGTH (em)

Figure 7.Histograms ofgapjunction profile lengthdistributions intransverseandlongitudinal planesof section in normal and borderzonetest

areas.Using analysisofcovariance,significantdifferenceswerefound between normal longitudinalandnormaltransverseplanes (P=0.032),

reflectingthe normalanisotropicdistribution ofgapjunctions,and between normaltransverseand borderzonetransverseplanes (P<0.0001), indicatingdisruption and loss of long, ribbon-shapedgapjunctions. Nosignificantdifferenceswereobserved between normallongitudinaland borderzonelongitudinal,orbetween borderzonetransverseandborderzonelongitudinal.

normal and borderzone tissues (Fig. 7), consistent with the

observation that the average individual gapjunction profile

lengthseeninthelongitudinal planes didnotchangeinfibrotic

tissue(Fig. 6).Because the shortaxes(longitudinal dimensions)

of the long, transversely oriented, ribbon-shaped junctions have been shown previouslytohavea meanlength of -1 um,

selectivedisruption of thesejunctionsinfibrotictissue would

notresultinadecrease in theaverageprofilelength observed in

longitudinal planes, but would cause reduction in aggregate

profile lengthand thetotal number ofgapjunction profilesper

unitintercalateddisk,asshowninFigs. 5 and 6.

Thelong, ribbon-shapedgapjunctions of normal

ventricu-larmyocardium have been shown previously toreside exclu-sivelyininterplicatesegments,whereas smallerdiscoid shaped junctionsarelocatedwithininterdigitatingadhesive elements

of the plicate segmentof the intercalated disk (16). Selective disruptionofthese long, interplicatesegmentgapjunctions in

infarct borderzone tissue would beexpected todecrease the proportion oftotal gapjunctions residing ininterplicate

seg-ments.Accordingly, thepercentagesof totalgapjunction

pro-fileslocatedininterplicatesegmentsof normaltissues and

in-farct border zonesamples wereassessed in bothlongitudinal

andtransverseplanesof section. Innormaltissues, 66% of all

gapjunction profileswerefound in interplicatesegments,while

theremaininggapjunction profileswerelocated within plicate

segments. In fibroticsamples,theproportionof totalgap

junc-tionprofilesidentified ininterplicatesegments decreased signif-icantly to 52% (P <0.001). These data are concordant with observations shown in Fig. 6 that long transversely oriented

gapjunctionareselectivelylost in infarct borderzonesamples.

Light

microscopic

morphometry of intercellular

connec-tions. Whereas results of electron microscopic morphometry

showed disruption of intercellularjunctionsin infarct border

zonetissues,these observations didnotindicate whether fewer cells wereactuallyinterconnected to oneanotherorwhether fewer connections existed betweenanygivencellpairwithouta

reductioninthetotal number of interconnected cells. In

addi-tion, because the ultrastructural analysis didnotindicate the relativeextent towhich side-to-side and end-to-end cell

inter-connectionsweredisruptedininfarct borderzones,light

micro-scopicmorphometrywasperformedtoanswerthesequestions. Fig. 8 shows representative fields of normal and borderzone

myocardiumstainedtoenhance cell borders andpermit three-dimensional reconstruction of cellular interconnectionsin

se-rial sections. Cellular interconnectionswerescoredonlywhen

a clearly identifiable intercalated disk was observed between

the cells.

The number of cells connectedtoanindividualventricular myocyte wasreduced significantly from 11.2±1.0 in control tissueto 6.5±1.3 in infarct borderzone_{samples (P}<_0.001).

0

15-G. K0

-w

z

4

w

5-0

0 2-2 U0

0

25

0

20 I

.~~~~.

F

\aX

B

of-

r+ e M is i

tS,.s-

P.04w.

it

_|Y

Telin;Rndw

I~~~~~~~~~~;I

KE,~~~~~~~~4

e~vesawX++2Z...

., . .

*

~~~~~~~~~~~~~~~~'T

sij s

*Z'i' .. X~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.

Mow'..~~~t.

I le

i'''fno

P

$Ad I f

-e^(>.:.Of-"*tN4,yg...Hi.,k.,, . . . t : .,....I..

.;,..a.1

Figure 8. Lightmicrographsofrepresentativeserial sectionsof normal(top

panel)

and borderzone

(bottom

panel)

test

regions

showing

enhanced sarcolemmastainingtofacilitate three-dimensional reconstruction of patterns of cell

apposition. Examples

of intercalated disks

_connecting

cells in

primarily

side-to-side(vertical arrows)orend-to-end(horizontal arrows) appositionareindicated.Anintercellular connectionwasscored

only

whenaclearlyidentifiable diskwasobserved betweentwocells.X640.

The value

obtained for

normal

myocardium

is

in agreement

with reported

values

(16, 22).

Previously,

weobserved 9.1±2.2

connections

per normal

canine

ventricular

myocyte

(16),

a

value that

is

significantly different (P

<

0.01)

thanthe value

observed

in borderzone

samples

in the present

study.

The loss

of

cellular

interconnections in fibrotic tissue did

notoccurina

random

spatial

orientation,

butwas most

marked

in cells

hav-ing

primarily

side-to-side

apposition

andleast apparent in cells

1600 R. A.Luke andJ. E.

Saffitz

f,

1;-*1-W7, "'PUf!

TV

connected

primarily

end-to-end.

Asshown in Table

I, Type

I connections (cells having > 75% lateral border overlap) were diminished by

75%,

while

Type

IVconnections

(>

50% end-to-end overlap) were reduced by only 22% (P < 0.05). Interme-diate types of connections

(Types

II andIII) were diminished by 35-45%. Thus, connections between cells oriented

side-to-side

were more

severely disrupted

than connections in cells

joined

end-to-end.

Discussion

Results of this

quantitative

structural

study

indicate that mod-est

accumulation

of

collagen

in the healed canine infarct

border

zone that is sufficienttoincrease the interstitial space

from

a normal value of - 6% to 15-20% while maintaining myocyte

fiber anisotropy, is associated

with

disruption

ofinter-cellular

junctions

anda marked reduction in the number of

intercellular connections.

Electron

microscopic morphometry

revealed that the

largest

connexonarrays

in

the heart, arranged

in long,

ribbon-shaped transversely oriented

gap

junctions,

are

selectively disrupted.

Light microscopic

reconstruction of

cell-packing

patterns

indicated

that the total number

of cell-to-cell

contacts

is

substantially

reduced and

that

side-to-side

connec-tions

are

selectively affected.

Thesestructural

alterations

are

summarized

in

Fig.

9.

Structural

abnormalities

in

regions bordering healed

in-farcts

in the present

study varied from

extremes

of

dense

collag-enous scar to

relatively

normal

myocardium. Often,

this

range

of

structural

alteration

occurred over a

distance of only

1-2 mm.

Relatively

small areas

of

the border zone wereactually

composed of

the

interspersion of

myocytes and

interstitial

col-lagen

that

constituted the tissue sample analyzed in the

present

study. This

pattern

of

interstitial fibrosis with maintained

tis-sue

anisotropy

is

one

of

multiple potential

anatomic

substrates

of ventricular

tachycardia

that have been

identified in

areas

bordering

both human and

canine infarcts (5, 6, 8, 10-14,

23-25).

Although

the results

of

the present study

indicate

that border zone myocytes

contain fewer

gap

junction profiles

per

unit

intercalated disk length

and

that the

longest

profiles

(> 3.0

,gm)

are

selectively disrupted,

it isnotknown whether

only

the

longest

profiles

are

lost,

orwhether both

long

and

short

profiles

become

disrupted,

orwhether

long

profiles

become converted

toshorter

profiles.

Our

observations

suggest,

however, that

gap

junctional

membrane

is

reduced

by

agreater extent than over-all

intercalated diskjunctional membrane.

Because the

longest,

TableI.Distribution ofMyocyte Connection Types inNormal and BorderZone Areas

Connection Normal Border Percent type regions zone reduction

1 2.4±0.8 0.6±0.6*

75.0*

II 3.1±1.2 1.9±1.4§ 37.7

III 2.2±0.8 1.2±0.8§ 45.0

IV 3.6±0.9 2.8±0.81 22.0 Valuesindicate the number of

myocytes

connected to a single myo-cyteby each connection type.

*P<0.001.

P<0.05vs.percent reduction of TypeIVconnections.

P<0.005.

'PP<0.05vs.normalregions.

BORDER ZONELONGITUDINAL BORDER ZONE TRANSVERSE

Figure9.Diagram of the effects of interstitial collagen depositionon

gapjunctiondistribution andpatternsofintercellular connection. Gap junctionsareindicated by thickenedsegmentsof the junctional

membraneconnectingcells. Asignificant reduction intotalgap

junction membrane lengthoccurswith selective disruption oflong, ribbon-shaped interplicategapjunctions. Becausethelonggap

junc-tionsareseeninthe short axisinlongitudinal planes of section, their

loss is observedas amodest reductioningapjunction profilenumber

withoutasignificantdecreaseinaverageprofile length. In the

trans-verseplaneof section,however,theirloss resultsin significant reduc-tion inmeangapjunction profile length, producinga moreuniform

distributionof shortgapjunction profiles. Reduced coupling is

ob-servedatthe cellular leveltooccurdisproportionately in cells

con-nected inside-to-side orientation.End-to-endmyocyteconnections

arerelatively spared.

transversely oriented gapjunctions are often located onthe

surfaces ofmyocytes atthe edges of intercalated disks (i.e., they

arecontiguous with nonjunctional sarcolemma) (16), they

wouldlikely be moresusceptiblethan shorterprofilesto

me-chanical forcesassociated withaccumulation ofinterstitial col-lagen. Plicatesegmentgapjunctions, embedded entirely within

theadhesivecomponents of intercalated disks and

concen-tratedatthe ends ofmyocytes,wouldlikely bemoreresistantto

suchforces. Thisinterpretationof the observations in the

pres-entstudyassumesthat disruption ofgapjunctions and

remodel-ing ofpatternsof intercellular connectionsarecaused by

me-chanicalforcesattributabletoaccumulationandorganization ofcollagen bundles oriented paralleltothe longmyocyteaxis. However, suchacausalrelationship hasnotbeenestablished. The significant reduction ofgapjunction surface density

delineated with ultrastructural morphometry inborderzone

tissuesindicates that thesemyocytesarerelatively uncoupled

and thatthetissue would have increased passive internal

resis-tanceincomparison with normal myocardium. This observa-tionisconsistent withthe reducedspaceconstantsmeasuredin infarctborderzonesby Spearetal. (2).Selective disruption of side-to-sideintercellularconnections demonstrated in three-di-mensionalreconstructions wouldappeartoenhancethe

anisot-ropy ofintercellular connections and disproportionately in-creaseresistancetopropagation inadirectiontransverse tothe

NORMAL LONGITUDINAL

A_~~I

_Bn

long myocyte axis. Whereas any single gap junction may sub-serve current transfer in either transverse or longitudinal direc-tion, and loss of gap junctions would be expected to increase internal resistance in both directions, disruption of side-to-side connections with relative sparing of end-to-end connections would increase the path length and number of intercellular junctions traversed by a wavefront moving in thetransverse

direction. These predicted electrophysiologic effectsare consis-tent with the observations of Dillon et al. (5) and Zuanettietal. (18), who analyzed patterns of propagation in the canine in-farctborder zone.

Results ofthe present study also provide insights into mech-anisms of fractionated electrograms that have been observedin

tissues prone to arrhythmias (26-28). In experimental canine

infarcts,

complex fractionated electrograms are not apparent during the early phase of infarct healing but occur and persist as the

necrotic myocardium

is resorbed and replaced by collage-nous scar

tissue

that also

affects

the spatial orientation and packing of surviving myocytes at the infarct edge (8, 11). Gardner et al. have suggested that alterations in intercellular

junctions accompanying

interstitial fibrosis were

responsible

for theappearance of fractionated electrograms, and for slow heterogenous

propagation

in

epicardial infarct

border regions

(1

1).

Before

the present study,

however,

no

quantitative

struc-tural data onmyocyte connections in infarct borderzoneswere

available

to

substantiate this conclusion. Our

results

provide

a more detailed

understanding of

the structural

basis

underlying

slowpropagation, conduction block, and fractionated electro-grams, all

of

which appear to be important determinants of

arrhythmogenesis.

Acknowledgments

Wegratefully acknowledgetheassistance ofDr.Giulio Zuanettiand Dr.PeterB.Corr who generously provided canine epicardialborder zonetissue.WethankDr.RobertH.Hoyt for helpful scientific

discus-sions,WilliamKraft andTimothy Tolley for technical assistance, Louise SchoelchandMargaretMcHughfor photographic assistance,

Dr.KennethSchechtman for statistical assistance,and SusanJohnson forpreparationof thetypescript.

This workwassupported by National Heart, Lung and Blood

Insti-tutegrantsHL-36773 andHL-17646,SCOR in Ischemic Heart

Dis-ease. Dr.Saffitz isanEstablishedInvestigatorof the American Heart Association.

References

1.Fozzard, H.A.,andM. F.Arnsdorf. 1986. Cardiac electrophysiology.In The Heart andCardiovascular System.H. A.Fozzard,E.Habert,R. B.Jennings,

A.M.Katz, andH. F.Morgan, editors.RavenPress, New York. 1-25. 2. Spear,J.F., E. L.Michelson, andE. N. Moore. 1983. Reducedspace constantin slowly conducting regions of chronically infarcted canine

myocar-dium. Circ.Res.53:176-185.

3. Spach,M.S.,P. C.Dolber,and J. F.Heidlage. 1988.Influence of the passive anisotropic propertiesondirectionaldifferences inpropagation following modification of the sodium conductance in human atrial muscle.Amodel of

reentry based onanisotropicdiscontinuous propagation. Circ.Res.62:811-832.

4.Spach,M.S.,W.T.Miller, III,D.B.Geselowitz,R.C.Barr, J.M.Kootsey, andE. A.Johnson. 1981. The discontinuousnatureofpropagationin normal canine cardiac muscle. Evidence forrecurrentdiscontinuitiesofintracellular

resis-tancethataffect the membranecurrents.Circ. Res.48:39-54.

5.Dillon,S. M.,M. A.Allessie,P.C.Ursell,andA.L.Wit. 1988. Influences of anisotropictissuestructure andreentrantcircuits in theepicardialborderzoneof subacute canine infarcts. Circ.Res.63:182-206.

6. Spear,J.F.,E. L.Michelson,and E. N. Moore. 1983. Cellularelectrophysio-logic characteristics of chronically infarctedmyocardium in dogssusceptibleto

sustained ventricular tachyarrhythmias. J.Am.Coll. Cardiol. 1:1099-1110.

7.Spach,M.S.,W. T. Miller,III,P. C. Dolber, J. M. Kootsey, J. R.Sommer, andC.E.Mosher,Jr.1982.Thefunctionalroleof structural complexitiesin the

propagation of depolarizationin the atriumofthe dog. Cardiacconduction

dis-turbancesdue todiscontinuities of effective axial resistivity.Circ. Res.

50:175-191.

8. Ursell,P.C.,P.I.Gardner, A.Albala,J. J.Fenoglio,Jr., and A. L. Wit. 1985.Structuralandelectrophysiologicalchanges in theepicardialborder zone of

caninemyocardial infarcts during infarct healing. Circ.Res.56:436-451. 9. Friedman,P. L., J. J.Fenoglio, andA. L. Wit. 1975. Timecoursefor

reversalof electrophysiologicalandultrastructural abnormalitiesin subendocar-dialPurkinje fibers surviving extensive myocardial infarction indogs. Circ. Res.

36:127-144.

10.Richards,D. A., G. J. Blake, J. F. Spear, and E. N. Moore. 1984.

Electro-physiologicsubstratefor ventricular tachycardia: correlationofpropertiesin vivo

and invitro.Circulation.69:369-381.

11. Gardner,P. I., P. C. Ursell, J. J.Fenoglio,and A. L. Wit. 1985.

Electro-physiologicandanatomic basis for fractionated electrograms recordedfrom

healedmyocardial infarcts. Circulation. 72:596-61 1.

12.Surawicz,B. 1989.Ventriculararrhythmias:whyisit so difficult to find a

pharmacologic cure. J. Am. Coll. Cardiol. 14:1401-1416.

13.Wit,A. L., M. A.Allessie,F.I.M.Bonke, W.Lammers,J.Smeets,and

J. J.Fenoglio. 1982. Electrophysiologic mappingtodetermine the mechanism of experimental ventricular tachycardia initiated by premature impulses.

Experi-mentalapproachandinitialresultsdemonstrating reentrant excitation.Am. J.

Cardiol. 49:166-185.

14.Wit,A. L. 1989. Anisotropic reentry:Amodel of arrhythmias thatmay

necessitateanewapproachtoantiarrhythmic drug development.InLethal

ar-rhythmias resulting from myocardial ischemia and infarction. M. R. Rosen and

Y.Palti, editors.KluwerAcademic Publishers, Boston. 199-213.

15.Spach,M.S., andP.C. Dolber.1986. Relating extracellular potentials and their derivatives toanisotropic propagationat amicroscopiclevel in human cardiacmuscle-evidence for electrical uncoupling of side-to-side fiber

connec-tionswithincreasing age. Circ. Res. 58:356-371.

16. Hoyt,R.H.,M.L.Cohen, andJ. E.Saffitz. 1989. Distribution and three-dimensionalstructureofintercellularjunctionsincaninemyocardium.Circ. Res. 64:563-574.

17.Luke,R.A., E.C.Beyer, R. H.Hoyt,and J. E.Saffitz. 1989.Quantitative analysisof intercellularconnections byimmunohistochemistryofthe cardiacgap

junction protein connexin43. Circ. Res. 65:1450-1457.

18.Zuanetti, G.,R. H.Hoyt, andP. B.Corr.1990.fl-Adrenergic-mediated

influencesonmicroscopic conduction in epicardial regionsoverlyinginfarcted myocardium. Circ. Res. 67:284-302.

19.Weibel,E. R.1969.Stereological principles for morphometry in electron microscopic cytology. Int. Rev.Cytol.26:235-299.

20. SASInstituteInc.1988.SAS/STATTMUser'sGuide. Release 6.03 edi-tion. SASInstitute Inc., Cary, NC.

21. Gerdes,A.M., andF. H. Kasten.1980.Morphometricstudy of endomyo-cardium andepimyocardiumof the left ventricle inadultdogs.Am. J. Anat.

159:389-394.

22.Phillips, S. J.,E. A.Pappas,M.Paulosky, and D. Meir-Levi. 1981. Cellular connectivitybetweenadjacentmammalian ventricular heart cells.Anat.Rec.

199:202A. (Abstr.)

23. DeBakker,J. M.T.,F.J.L. vanCapelle,M.J.Janse,A.A. M.Wilde,R.

Coronell,A. E.Becker,K. P.Dingemans,N. M. vanHemel, and R.N. W. Hauer.

1988. Reentryas a causeofventriculartachycardiainpatientswith chronic ischemicheartdisease:electrophysiologicandanatomiccorrelation.Circulation. 77:589-606.

24.Cardinal, R.,M.Vermeulsen,M.Shenasa,F.Roberge,P.Page,F.Helie, andP.Savard. 1988.Anisotropicconduction and functional dissociation of isch-emic tissueduringreentrantventriculartachycardiaincaninemyocardial infarc-tion. Circulainfarc-tion. 77:1162-1176.

25. Kramer,J.B.,J. E.Saffitz,F. X.Witkowski,andP. B.Corr. 1985. Intra-muralreentry as amechanismof ventriculartachycardia during evolvingcanine myocardialinfarction. Circ. Res. 56:736-754.

26.Wit,A.L., andM. E.Josephson. 1985. Fractionatedelectrogramsand continuouselectricalactivity:factorartifact.InCardiacElectrophysiologyand Arrhythmias.D. P.Zipes, andJ.Jalife,editors. GruneandStratton,NewYork. 343-351.

27.Waxman,H.L., andR.J.Sung.1980.Significanceoffragmented

ventricu-larelectrogramsobservedusingintracardiacrecording techniquesinman. Circu-lation. 62:1349-1356.

28.Spach,M.S.,R.C.Barr, E.A.Johnson,and J.M.Kootsey.1973.Cardiac extracellularpotentials. Analysisofcomplexwaveforms about thePurkinje net-worksin dogs. Circ.Res.33:465-473.