ArticlePDF Available

Human growth hormone stimulates proliferation of human retinal microvascular endothelial cell in vitro

Authors:
  • Molecular Research Center Inc., USA

Abstract and Figures

Growth hormone (GH) has been implicated in the pathogenesis of proliferative diabetic retinopathy. We sought to determine whether this could be mediated by an effect of GH on proliferation of endothelial cells, and, for this purpose, established long-term cultures of human retinal microvascular endothelial cells (hREC) from normal postmortem human eyes. High-purity (greater than 95%) hREC preparations were selected for experiments, based on immunofluorescence with acetylated low density lipoprotein (LDL) and anti-factor VIII-related antigen. Growth requirements for these cells were complex, including serum for maintenance at slow growth rates and additional mitogens for more rapid proliferation. Exposure of hREC to physiologic doses of human GH (hGH) resulted in 100% greater cell number vs. control (P less than 0.01) but could be elicited only in the presence of serum. When differing serum conditions were compared, hGH stimulated [3H]thymidine incorporation up to 1.6- to 2.2-fold under each condition and increased DNA content significantly in the presence of human, horse, and fetal calf serum. Depending on the culture conditions used, the threshold hGH concentration for significant stimulation of hREC proliferation was 0.4-4 micrograms/liter. In contrast, proliferation of human umbilical vein endothelial cells was not significantly altered by hGH added to concentrations as high as 200 micrograms/liter. In summary, hREC respond to physiologic concentrations of hGH in vitro with enhanced proliferation. This specific effect of GH on retinal microvascular endothelial cells supports the hypothesis of a role for GH in endothelial cell biology.
Content may be subject to copyright.
Proc.
Natl.
Acad.
Sci.
USA
Vol.
88,
pp.
617-621,
January
1991
Medical
Sciences
Human
growth
hormone
stimulates
proliferation
of
human
retinal
microvascular
endothelial
cells
in
vitro
(somatotropin/diabetic
retinopathy/neovascularization)
ZBIGNIEW
RYMASZEWSKI*t,
ROBERT
M.
COHEN*t,
AND
PIOTR
CHOMCZYNSKI*
Departments
of
*Medicine
(Endocrinology/Metabolism)
and
tOphthalmology,
University
of
Cincinnati
College
of
Medicine,
Cincinnati,
OH
45267-0547
Communicated
by
Donald
F.
Steiner,
September
2S,
1990
(received
for
review
May
15,
1990)
ABSTRACT
Growth
hormone
(GH)
has
been
implicated
in
the
pathogenesis
of
proliferative
diabetic
retinopathy.
We
sought
to
determine
whether
this
could
be
mediated
by
an
effect
of
GH
on
proliferation
of
endothelial
cells,
and,
for
this
purpose,
established
long-term
cultures
of
human
retinal
mi-
crovascular
endothelial
cells
(hREC)
from
normal
postmortem
human
eyes.
High-purity
(>95%)
hREC
preparations
were
selected
for
experiments,
based
on
immunofluorescence
with
acetylated
low
density
lipoprotein
(LDL)
and
anti-factor
VIII-
related
antigen.
Growth
requirements
for
these
cells
were
complex,
including
serum
for
maintenance
at
slow
growth
rates
and
additional
mitogens
for
more
rapid
proliferation.
Expo-
sure
of
hREC
to
physiologic
doses
of
human
GH
(hGH)
resulted
in
100%
greater
cell
number
vs.
control
(P
<
0.01)
but
could
be
elicited
only
in
the
presence
of
serum.
When
differing
serum
conditions
were
compared,
hGH
stimulated
[3H]thymidine
incorporation
up
to
1.6-
to
2.2-fold
under
each
condition
and
increased
DNA
content
significantly
in
the
presence
of
human,
horse,
and
fetal
calf
serum.
Depending
on
the
culture
condi-
tions
used,
the
threshold
hGH
concentration
for
significant
stimulation
of
hREC
proliferation
was
0.4-4
pg/liter.
In
contrast,
proliferation
of
human
umbilical
vein
endothelial
cells
was
not
significantly
altered
by
hGH
added
to
concentra-
tions
as
high
as
200
pzg/liter.
In
summary,
hREC
respond
to
physiologic
concentrations
of
hGH
in
vitro
with
enhanced
proliferation.
This
specific
effect
of
GH
on
retinal
microvas-
cular
endothelial
cells
supports
the
hypothesis
of
a
role
for
GH
in
endothelial
cell
biology.
Mechanisms
underlying
the
development
of
proliferative
diabetic
retinopathy
(PDR)
remain
poorly
understood
(1-3).
The
observation
that
pituitary
ablation
may
reverse
or
slow
the
progression
of
PDR,
first
made
>35
years
ago,
has
raised
the
question
of
a
role
for
a
pituitary-derived
or
-dependent
factor
in
this
disease
process
(4-8).
As
a
consequence
of
this
and
of
abnormalities
in
both
the
quantity
and
pattern
of
secretion
of
growth
hormone
(GH)
in
poorly
controlled
diabetes
(9-12),
GH
has
been
suggested
as
a
possible
medi-
ator
of
the
effect
of
pituitary
ablation
on
PDR,
potentially
via
insulin-like
growth
factors
(IGFs)
(2,
13).
However,
there
is
increasing
awareness
that
GH
acts
not
only
via
systemic
IGFs
but
also
directly
without
systemic
mediation
at
a
greater
number
of
target
tissues
than
has
been
recognized
(14,
15).
Therefore,
we
sought
to
determine
whether
GH
has
effects
directly
on
microvascular
endothelial
cells
(EC)
derived
from
the
retina.
Use
of
cells
differing
in
either
species
or
tissue
of
origin
is
a
potentially
confounding
variable
(16-23),
while
studies
on
human
microvascular
EC
have
been
limited
(24-
27).
We
now
report
the
establishment
of
long-term
cultures
of
such
cells
from
the
postmortem
human
retina
(hREC)
and
demonstrate
that
these
cells
respond
to
physiologic
concen-
trations
of
biosynthetic
human
growth
hormone
(hGH)
with
enhanced
proliferation.
MATERIALS
AND
METHODS
Reagents
were
obtained
from
the
following
sources:
fetal
calf
serum
(FCS),
horse
serum
(HS),
trypsin,
medium
199
(M199)
with
25
mM
Hepes
buffer,
and
bovine
serum
albumin
from
GIBCO;
EC
growth
supplement,
human
fibronectin,
colla-
genase
type
II-S
from
Sigma;
acetylated
low-density
lipopro-
tein
(LDL)
labeled
with
1,1'-dioctadecyl-1-(3,3,3',3'-tetra-
methyl)indocarbocyanine
perchlorate
(Dil-acetylated
LDL)
from
Biomedical
Technologies
(Stoughton,
MA);
Mito+
se-
rum
extender
from
Collaborative
Research;
rabbit
antiserum
to
human
factor
VIII
from
Calbiochem.
Biosynthetic
hGH
was
provided
by
Genentech,
San
Francisco;
pituitary
corti-
cotropin
(ACTH),
by
the
National
Hormone
and
Pituitary
Program;
and
biosynthetic
IGF-1
(Amgen
Biologicals),
by
S.
Chernausek.
Human
serum
(HuS)
was
obtained
from
healthy
male
volunteers,
ages
21-40.
The
endogenous
concentration
of
hGH
in
the
human
sera
used
was
determined
by
hGH
RIA
and
reported
with
each
individual
experiment,
and
GH-like
immunoreactivity
in
FCS
was
determined
by
using
an
ovine
GH
RIA
(28).
hREC
Isolation
and
Culture.
Postmortem
human
eyes
were
obtained
from
the
Cincinnati
Eye
Bank
for
Sight
Restoration
and
from
the
National
Disease
Research
Interchange
(Phil-
adelphia).
The
donors
were
previously
healthy
victims
of
trauma,
ages
2-37
years
of
both
sexes.
Exclusion
criteria
included
sepsis,
malignancy,
and
systemic
diseases.
Cells
were
isolated
and
established
in
culture
by
techniques
mod-
ified
from
previously
described
methods
(19,
20,
24,
26,
29).
Briefly,
an
incision
was
made
at
the
ora
serrata
under
sterile
conditions,
and
the
vitreous
was
removed.
The
neurovascu-
lar
retina
was
separated
from
the
optic
nerve,
transferred
to
a
tube
containing
chilled
M199
with
25
mM
Hepes
and
briskly
stirred
in
five
changes
of
medium.
The
retinas
then
were
minced,
vigorously
passed
by
syringe
through
an
18-gauge
needle,
and
the
capillaries
were
separated
by
sieving
through
nylon
mesh
of
sequentially
smaller
pore
size
(200-
to
50-,um
pores).
Retinal
capillaries
retained
by
the
last
nylon
screen
were
partially
digested
with
collagenase
(10
,ug/liter
in
M199
containing
1%
bovine
serum
albumin)
for
0.5-1.5
hr.
Small
fragments
of
capillaries
were
suspended
in
growth
medium
(M199
containing
20%
HS,
EC
growth
supplement
at
20
mg/liter,
heparin
at
90
mg/liter,
2
mM
L-glutamine,
and
Abbreviations:
Dil-acetylated
LDL,
acetylated
LDL
labeled
with
1,1'-dioctadecyl-1-(3,3,3',3'-tetramethyl)indocarbocyanine
perchlo-
rate;
FCS,
fetal
calf
serum;
GH,
growth
hormone;
hGH,
human
GH;
EC,
endothelial
cell(s);
hREC,
human
retinal
microvascular
endo-
thelial
cell(s);
hUVEC,
human
umbilical
vein
endothelial
cell(s);
HS,
horse
serum;
hGH,
human
growth
hormone;
HuS,
human
serum;
IGF,
insulin-like
growth
factor;
LDL,
low
density
lipoprotein;
M199,
medium
199;
PDR,
proliferative
diabetic
retinopathy;
ACTH,
corti-
cotropin.
tTo
whom
reprint
requests
should
be
addressed.
617
The
publication
costs
of
this
article
were
defrayed
in
part
by
page
charge
payment.
This
article
must
therefore
be
hereby
marked
"advertisement"
in
accordance
with
18
U.S.C.
§1734
solely
to
indicate
this
fact.
618
Medical
Sciences:
Rymaszewski
et
al.
mitogen
serum
extender
at
1
ml/liter).
The
cell
suspension
was
then
plated
in
dishes
coated
with
0.5%
gelatin
and
fibronectin
(1
Ag/cm2).
This
medium
was
used
for
all
proce-
dures
except
as
otherwise
noted.
Cellular
outgrowth
from
small
microvascular
fragments
began
on
days
3-4
and
consisted
of
colonies
of
both
EC
and
pericytes
and,
commonly,
a
mixture
of
both
cell
types.
To
obtain
more
homogeneous
preparations,
we
initially
used
techniques
previously
applied
to
retinal
EC
isolation,
includ-
ing
(i)
"weeding"
of
unwanted
colonies
(30)
and
(ii)
Percoll
density
gradient
centrifugation
(22,
26).
However,
to
obtain
a
sufficient
yield
of
hREC,
it
was
necessary
to
use
fluores-
cence-activated
cell
sorting
(Becton
Dickinson
model
440)
after
labeling
cells
with
DiI-acetylated
LDL
(at
10
,ug/liter
of
culture
medium
for
12
hr;
ref.
31).
Cells
were
maintained
in
growth
medium
until
confluent.
For
subculture,
cells
were
trypsinized
(0.25%)
and
then
split
1:3-4.
The
purity
of
each
culture
was
monitored
frequently
by
morphology
based
on
light
microscopy
and
by
fluorescent
staining
with
DiI-
acetylated
LDL
and
anti-factor
VIII-related
antigen.
EC
from
human
umbilical
vein
(hUVEC)
were
isolated
by
es-
tablished
techniques
(32)
and
used
as
a
positive
control
for
both
staining
procedures.
Measurement
of
Cell
Number
and
Proliferation.
All
exper-
iments
were
performed
with
cells
that
were
between
passages
4
and
7
at
an
initial
density
of
1.3-1.8
x
104
cells
per
cm2.
Plating
medium
was
M199
containing
2
mM
L-glutamine
and
20%
HS
or
FCS.
The
day
after
plating,
the
medium
and
unattached
cells
were
removed,
the
cells
were
washed
once
with
M199,
and
incubation
of
the
cells
was
initiated
with
medium
containing
serum
and
hormones
for
4-9
days
as
described
below.
Hormone
concentrations
shown
in
the
figures
represent
the
concentration
added
(in
,.g/liter;
to
convert
to
pM,
multiply
by
50)
and
are
not
corrected
for
hormone
contributions
from
added
serum.
Cell
proliferation
was
measured
by:
(i)
cell
counting
(Nebauer
hemocytome-
ter);
(ii)
determination
of
cellular
DNA
content
using
a
modified
microfluorometric
method
with
Hoechst
fluoro-
chrome
33258
(33,
34);
and
(iii)
determination
of
[3H]thymi-
dine
(80
Ci/mmol;
0.2
,uCi/ml
of
medium;
1
Ci
=
37
GBq)
incorporation
into
DNA
after
a
12-hr
incubation
(35).
Data
Analysis.
Data
are
expressed
as
means
±
SEM.
Data
were
analyzed
by
two-way
analysis
of
variance
(ANOVA)
with
Duncan's
new
modified
range
test
(see
Fig.
3)
or
by
one-way
ANOVA
(Figs.
4-6)
with
Dunnet's
modified
t
method
for
comparison
of
multiple
groups
to
a
single
control
or
by
unpaired
t
test
(effect
of
glucose
on
hREC
prolifera-
tion).
RESULTS
hREC
Isolation
and
Culture.
In
preliminary
experiments
to
establish
a
cell
isolation
procedure
and
culture
conditions
based
on
methods
reported
for
culture
of
EC
from
both
human
and
nonhuman
microvessels,
we
observed
a
low
rate
of
successful
culture
(<10%o
of
attempted
isolations)
as
assessed
by
serial
propagation
of
high-purity
EC
cultures.
Similarly
to
bovine
retinal
pericytes
(36),
human
retinal
pericytes
strongly
inhibited
proliferation
of
hREC
in
our
system.
In
mixed
culture,
acetylated
LDL-staining
cells
ceased
to
proliferate
at
40-50%
of
confluence,
whereas
in
pure
EC
culture,
proliferation
continued
until
confluence
had
been
achieved.
Purification
using
cell
sorting
based
on
label-
ing
of
EC
with
the
fluorescent
probe
Dil-acetylated
LDL
provided
a
higher
purity
of
viable
cells
than
did
the
alterna-
tives
and
was
used
in
subsequent
isolations.
In
four
sortings,
the
yield
of
Dil-acetylated
LDL-stained
cells
ranged
between
9%
and
20%
of
the
total
sorted.
This
method
allowed
long-
term
culture
of
high-purity
preparations
of
hREC
(>95%
cells
stained
positive
for
acetylated
LDL
and
factor
VIII)
from
four
donors;
two
of
these
preparations
have
been
maintained
thus
far
for
11
passages.
Even
at
that
stage,
hREC
grew
in
a
monolayer
with
cobblestone-like
morphology
and
continued
to
exhibit
cytoplasmic
perinuclear
granular
fluorescent
stain-
ing
with
both
Dil-acetylated
LDL
and
anti-factor
VIII-related
antigen
(Fig.
1).
hUVEC
stained
with
anti-factor
VIII-related
antigen
served
as
a
positive
control,
whereas
human
peri-
cytes
and
fibroblasts
failed
to
fluoresce
(not
shown).
hGH
and
hREC
Proliferation:
Serum
Requirement
for
an
Effect
of
hGH.
Initially,
the
proliferative
response
of
hREC
to
different
concentrations
of
hGH
was
measured
by
cell
count-
ing
after
incubation
in
growth
medium
from
which
serum
had
been
omitted,
in
comparison
with
growth
medium
modified
to
2.5%
HS.
In
the
absence
of
serum,
hGH
did
not
exert
a
mitogenic
effect
on
hREC
(Fig.
2).
However,
in
the
presence
of
2.5%
HS,
hGH
at
as
low
a
concentration
as
1.2
,tg/liter
(0.06
nM)
resulted
in
increased
cell
number
by
almost
2-fold
more
than
in
the
absence
of
GH
during
a
4-day
incubation.
Comparison
of
the
Effect
of
hGH
on
[3HfThymidine
Incor-
poration
and
DNA
Content.
A
significant
mitogenic
effect
on
FIG.
1.
Fluorescent
staining
of
representative
preparations
of
hREC
(after
passage
10).
(Upper)
hREC
labeled
with
dil-acetylated
LDL.
(x800.)
(Lower)
hREC
labeled
with
rabbit
anti-human
VIII-
related
antigen
and
fluorescein
coupled
to
goat
anti-rabbit
IgG
(Calbiochem).
(x380.)
Proc.
Natl.
Acad.
Sci.
USA
88
(1991)
Proc.
Natl.
Acad.
Sci.
USA
88
(1991)
619
30
0
0
0
.2
12
12
E
GH,
pg/liter
FIG.
2.
Effect
of
serum
on
biosynthetic
hGH
stimulation
of
hREC
proliferation.
Cells
in
passage
7
from
donor
1
were
plated
at
a
density
of
1.8
x
104
per
cm2
(4
x
104
per
well)
and
exposed
to
hGH
for
4
days.
At
that
time,
cell
number
was
then
determined
on
all
samples
in
triplicate.
o,
No
serum
added;
u,
2.5%
HS
added;
a,
P
<
0.05
vs.
control
(no
serum);
b,
P
<
0.01
vs.
control
(no
serum);
c,
P
<
0.01
vs.
absent
GH
(2.5%
HS).
hREC
was
demonstrable
(by
both
DNA
content
and
[3H]thy-
midine
incorporation)
at
the
lowest
hGH
concentration
tested,
1.2
,ug/liter
(0.06
nM),
under
each
of
several
serum
conditions
tested
(growth
medium
in
which
HuS
or
HS
was
substituted
to
either
2.5%
or
10o
concentration);
data
are
shown
only
from
incubations
in
2.5%
HS
(Fig.
3).
Since
the
concentration
of
hGH
in
the
HuS
used
was
0.2
,g/liter,
the
endogenous
hGH
contribution
to
the
final
hGH
concentration
when
serum
was
added
to
a
final
concentration
of
2.5%
or
10%
was
negligible
(0.005
and
0.02
,ug/liter,
respectively).
Threshold
for
hGH
Stimulation
of
hREC
Proliferation.
Because
the
proliferative
response
appeared
to
be
very
sensitive
to
hGH,
dose-response
studies
on
the
proliferation
of
hREC
were
subsequently
conducted
extending
the
hGH
concentration
range
downward
(0.1-200
pmg/liter;
Fig.
4)
under
conditions
specifically
selected
to
provide
slower
baseline
proliferation.
This
was
achieved
by
using
M199
containing
either
20%o
FCS
or
15%
pooled
HuS
and
2
mM
L-glutamine
with
omission
of
the
other
growth
supplements
(heparin,
EC
growth
supplement,
and
mitogen
serum
extend-
er).
Serum
conditions
were
selected
for
these
experiments
as
those
necessary
to
maintain
cell
attachment
in
the
absence
of
0
0_
_
2C0
4-
0-
2-
.CxC
_
o
:i,2-
_c
v
_E
cm
C
**
**
**
**
1
10
hGH,
,g/liter
-300
0
.Eip
C
z
0-
4-
200
Z
0
z
a
a,
is
C
010
4._
0)
-
0
0
Ir
T
T
/
T_,5K\*l
I
.1
0
1
10
100
hGH,
ug/liter
FIG.
4.
Effect
of
low
concentrations
of
biosynthetic
hGH
on
hREC
proliferation.
Cells
in
passage
7
from donor
2
were
plated
at
a
density
of
1.5
x
104/cm2
(corresponding
to
310
ng
of
DNA
per
well)
and
incubated
in
the
presence
of
20%
FCS
for
4
days
with
omission
of
other
growth
factors
(see
text),
at
which
time
they
were
harvested
and
the
DNA
content
was
determined
(all
samples
in
quadruplicate).
*,
P
<
0.01
vs.
control.
the
supplements.
GH-like
immunoreactivity
measured
64
,ug/liter
in
the
FCS
(providing
an
endogenous
bovine
GH
concentration
of
13
ttg/liter
from
20%o
FCS),
while
that
in
the
pooled
HuS
measured
0.5
,ug/liter
(adding
0.075
pg/liter
to
the
hGH
concentration
in
the
culture
well).
In
the
presence
of
20%o
FCS,
the
mean
DNA
content
per
well
increased
even
at
the
lowest
added
hGH
concentration
tested,
with
the
effect
becoming
statistically
significant
at
0.4
ug/liter.
Indeed,
the
effect
at
0.4
,ug/liter
of
hGH
was
indistinguishable
from
that
at
all
higher
concentrations
up
to
200
pug/liter.
As
an
indica-
tion
of
the
specificity
of
the
effect
of
hGH,
the
effect
of
ACTH
at
a
supraphysiologic
concentration
was
tested
during
the
same
experiment;
1
ILM
ACTH
did
not
alter
hREC
prolifer-
ation
(410
±
10
ng
of
DNA
per
well
in
ACTH-treated
cells
vs.
430
±
20
in
controls).
When
15%
HuS
was
substituted
under
the
same
otherwise
deprived
conditions,
mean
DNA
content
per
well
increased
in
hREC
cultures
beginning
at
1
jtg
of
hGH
per
liter,
with
the
effect
becoming
significant
at
4
,ug/liter
(Fig.
5).
There
was
only
minimal
further
effect
upon
addition
of
up
to
200
,ug/liter
of
hGH.
A
similar
effect
on
proliferation
was
observed
when
IGF-1
was
added
at
200
ug/liter.
In
contrast,
hUVEC
that
had
been
maintained
in
culture
for
a
'a
z
.P
oI
c,
8
:
r
-
100
100
1000
FIG.
3.
Comparison
of
the
effect
of
biosynthetic
hGH
on
hREC
proliferation
measured
by
both
[3H]thymidine
incorporation
and
DNA
content.
Cells
in
passage
6
from
donor
1
were
plated
at
a
density
of
1.3
x
104
per
cm2
(corresponding
to
270
ng
of
DNA
per
well)
and
incubated
for
6
days
in
2.5%
HS.
During
the
last
12
hr,
medium
containing
[3H]thymidine
(0.2
,uCi/ml)
was
added.
Cells
were
then
harvested
to
measure
both
[3H]thymidine
incorporation
(left
ordinate)
and
DNA
content
(right
ordinate).
All
samples
were
analyzed
in
triplicate
except
for
duplicate
determinations
of
[3H]thy-
midine
incorporation
at
400
pg
of
GH
per
liter.
*,
P
<
0.05
vs.
control;
**,
P
<
0.01
vs.
control
(analysis
of
variance).
**
TT
I
I
-?I
T
A
I
0
.1
1
10
hGH,
Ag/liter
100
IGF-1
FIG.
5.
Comparison
of
hGH
and
IGF-1
effects
on
hREC
and
hUVEC
proliferation.
hREC
(e)
in
passage
7
from
donor
2
were
plated
at
a
density
of
1.1
x
104
per
cm2
(corresponding
to
220
ng
of
DNA
per
well),
and
hUVEC
(o)
in
passage
7
were
plated
at
a
density
of
1.9
x
104/cm2
(390
ng
of
DNA
per
well);
both
were
incubated
in
the
presence
of
15%
pooled
HuS
for
9
days
with
omission
of
other
growth
factors
(see
text).
Cells
were
harvested,
and
DNA
content
was
determined
in
quadruplicate.
*,
P
<
0.01
vs.
control.
Medical
Sciences:
Rymaszewski
et
al.
620
Medical
Sciences:
Rymaszewski
et
al.
similar
number
of
passages
and
grown
under
the
same
conditions
failed
to
show
a
significant
response
to
either
hGH
or
IGF-1.
Glucose
and
hREC
Proliferation.
To
assess
the
potential
for
synergy
between
hGH
and
glucose
in
the
diabetic
milieu,
the
effect
of
25
mM
glucose
was
compared
with
that
of
5
mM
glucose
on
hREC
DNA
content
after
a
9-day
incubation
(data
pooled
from
two
experiments
in
which
cells
were
plated
at
a
density
of
3000
per
cm2;
n
=
10).
At
5
mM
glucose,
the
final
DNA
content
per
well
was
3.3
±
0.2
times
the
initial
DNA
content,
whereas
at
25
mM
glucose,
the
final
DNA
content
was
1.8
+
0.2
times
the
initial
content
(P
<
0.002).
Because
glucose
and
hGH
had
opposite
rather
than
similar
effects,
potential
interactions
have
not
been
pursued
further.
DISCUSSION
There
are
at
least
two
lines
of
evidence
to
support
a
role
for
a
pituitary-derived
or
pituitary-dependent
factor
in
the de-
velopment
of
PDR
and
a
third
line
of
evidence
to
specifically
implicate
hGH:
(i)
the
effect
of
pituitary
ablation
on
PDR
(4-6),
(ii)
the
relationship
of
development
of
retinopathy
to
puberty
or
adolescence
or
both,
and
(iii)
findings
in
ateliotic
dwarfs.
The
effect
of
pituitary
ablation
on
PDR
is
controver-
sial
but
is
supported
by
two
controlled
studies
and
a
recent
long-term
follow-up
(7,
8,
37).
The
latter
demonstrated
far
greater
reversal
of
proliferative
changes
and
preservation
of
vision
(unrelated
to
thyroxine,
glucocorticoid,
or
sex
steroid
replacement)
than
one
would
expect
in
relation
to
the
pro-
gressive
course
of
untreated
PDR,
as
documented,
for
ex-
ample,
in
the
control
group
of
the
Diabetic
Retinopathy
Study
(38).
Use
of
pituitary
ablation
has
been
documented
in
>900
patients
with
PDR
(39).
Given
that
the
mediator
of
the
effect
and
the
degree
of
hormonal
ablation
necessary
were
not
known
and
assays
for
pituitary
peptides
were
less
sensitive
than
those
today,
adequacy
of
ablation
of
specific
putative
mediator(s)
and
hence
whether
the
data
fairly
addressed
the
role
of
a
pituitary
factor
is
difficult
to
judge.
It
should
not
be
surprising
that
variable
results
were
obtained.
Subsequently,
the
prevalence
of
retinopathy
has
been
shown
in
a
6-
to
23-year-old
population
to
be
greater
in
those
older
than
15
years,
when
the
duration
of
diabetes
had
been
taken
into
account
(40).
This
further
supports
a
role
in
retinopathy
for
a
factor
present
in
both
sexes
that
is
related
to
either
growth
or
sexual
development,
most
likely
a
pituitary-derived
or
-dependent
factor.
Studies
in
ateliotic
dwarfs
demonstrated
a
high
frequency
of
diabetes.
Despite
this,
in
comparison
to
matched
control
patients
with
type
II
diabetes,
they
had
a
markedly
lower
frequency
of
retinopathy
and
did
not
have
the
skeletal
muscle
capillary
basement
membranes
thicken-
ing
found
in
patients
with
diabetes
(41).
There
are
numerous
observations
that
hGH
is
elevated
in
poorly
controlled
diabetes
mellitus,
which
focused
attention
on
the
role
of
the
hGH
axis
as
the
mediator
of
pituitary
ablation
in
PDR
(9-12,
42).
These
have
led
to
two
lines
of
investigation,
concerning
the
IGFs
and
alterations
in
neu-
ropharmacologic
control
of
hGH
secretion.
Plasma
and
vit-
reous
IGF-1
concentrations
have
been
reported
higher
in
diabetic
patients
with
PDR
than
in
those
without
(43,
44),
and
IGF-1
has
been
shown
to
alter
retinal
endothelial
cell
chemo-
taxis
and
secretion
of
plasminogen
activator
(25,
26).
There
is
controversy
whether
elevated
concentrations
of
IGF-1
reflect
increased
production
of
or
altered
vascular
perme-
ability
to
IGF-1
(15,
43,
44).
In
vivo
human
pharmacologic
studies
have
demonstrated
increased
responsiveness
of
pi-
tuitary
hGH
secretion
to
thyrotropin-releasing
hormone
(a
hypothalamic
peptide
that
ordinarily
does
not
stimulate
GH
secretion)
and
to
arginine
in
patients
with
retinopathy
(45,
46).
These
studies
provide
evidence
of
altered
control
of
hGH
secretion
without
addressing
whether
hGH
has
biologic
ef-
fects
that
may
modify
the
course
of
retinopathy.
It
is
in
this
context
that
the
question
arises
whether
pituitary
hormones,
specifically
GH,
have
direct
effects
on
the
endothelium.
The
present
studies
demonstrate
that
cells
from
the
human
retinal
endothelium
are
indeed
capable
of
responding
to
hGH
under
the
conditions
described.
Both
morphologic
and
functional
features
of
endothelium
were
preserved
under
the
conditions
of
study.
The
present
studies
do
not
distinguish
whether
the
increased
proliferation
rate
resulted
from
hGH
action
alone
or
whether
hGH
induced
the
secretion
of
another
mediator
by
endothelial
cells
(e.g.,
IGF-1
or
another
factor),
which
in
turn
induced
mitotic
activity.
The
finding
reported
here
is
consistent
with
those
of
other
investigators
in
the
last
several
years
that
certain
actions
of
GH
occur
directly
at
a
number
of
target
tissues
rather
than
as
a
consequence
of
systemic
IGF-1
secretion
(14,
15).
The
effect
of
hGH
on
hREC
proliferation
requires
the
presence
of
serum.
This
requirement
may
explain
the
lack
of
a
GH
effect
in
results
reported
by King
and
co-workers
(23)
studying
bovine
retinal
endothelial
cells
under
serum-free
conditions.
While
species
specificity
of
the
cells
cannot
be
excluded,
the
presence
of
serum
would
seem
the
more
likely
explanation.
The
nature
of
the
specific
component
of
serum
necessary
for
this
GH
effect
is
unclear
at
this
time.
The
effect
of
hGH
was
numerically
small
in
some
exper-
iments-e.g.,
a
30%
increase
when
DNA
content
was
used
as
the
measure
of
cell
number
(Fig.
3).
However,
this
corre-
sponded
to
a
simultaneously
measured
[3H]thymidine
incor-
poration
that
was
2.3
times
the
basal
rate
(Fig.
3),
and
in
other
experiments
a
2-fold
increase
over
4-9
days
in
either
cell
number
or
DNA
content
was
found.
Since
mitoses
are
infrequent
in
endothelium
(estimated
turnover
time
-10,000
days;
refs.
47
and
48)
and
the
exposure
to
these
concentra-
tions
of
GH
is
prolonged,
even
a
seemingly
small-magnitude
effect
in
a
brief
experiment
may
reflect
a
process
that
compounds
many
times
in
vivo
and
has
substantial
implica-
tions
for
disease
progression.
hREC
are
sensitive
in
a
dose-dependent
manner
to
exceed-
ingly
low
hGH
concentrations
in
vitro,
0.4-4
jtg/liter.
While
there
was
a
high
content
of
bovine
GH
immunoreactivity
in
the
experiments
conducted
in
FCS,
the
species
specificity
of
GHs
is
such
that
the
biological
activity
contributed
in
this
human
cell
system
would
be
expected
to
be
much
smaller.
The
fact
that
we
were
able
to
measure
a
statistically
signif-
icant
response
in
a
dose-dependent
manner
at
markedly
lower
hGH
concentrations
argues
for
a
relatively
small
biological
effect
of
bovine
GH
on
these
human
cells.
The
dose-response
was
shifted
to
the
left
in
FCS
compared
to
that
in
HuS,
suggesting
the
effect
of
another
component
of
one
or
both
sera
used,
either
a
stimulator
in
the
FCS
or
a
suppressor
in
the
HuS.
The
failure
of
EC
from
a
different
vascular
source
(the
umbilical
vein)
to
respond
to
hGH
both
supports
the
specificity
of
the
response
and
suggests
that
this
commonly
used
EC
experimental
model
will
not
be
an
appropriate
substitute
for
studying
these
interactions
in
human
disease.
To
our
knowledge,
effects
of
GH
on
other
vascular
prepa-
rations
in
vitro
have
only
been
shown
previously
with
non-
endothelial,
nonhuman
material-i.e.,
myomedial
cell
prolif-
eration
and
synthetic
function
in
rabbit
aortic
explants
(49,
50).
The
dose
dependence
observed
is
comparable
to
the
most
sensitive
responses
to
GH
we
are
aware
of
described
in
vitro
(51,
52).
This
GH
concentration
range
is
commonly
exceeded
in
vivo
not
only
in
disease
states,
diabetes
and
acromegaly,
but
also
in
healthy
people.
The
lowest
concentration
of
hGH
at
which
a
mitogenic
effect
is
seen,
0.4
jg/liter,
is
interme-
diate
between
the
levels
we
found
at
baseline
in
patients
with
poorly
controlled
diabetes
and
normal
control
subjects,
using
Proc.
Natl.
Acad
Sci.
USA
88
(1991)
Proc.
Natl.
Acad.
Sci.
USA
88
(1991)
621
a
highly
sensitive
hGH
RIA
(ref.
11;
unpublished
data).
The
in
vitro
conditions
selected
may
lack
factor(s)
present
in
vivo
that
restrain
the
ability
to
respond
to
hGH.
Such
factors
may
also
represent
functional
differences
in
the
retinal
microvas-
culature
between
individuals
with
and
without
diabetes
and
are
presently
under