Animal Model
The experimental protocol was approved by the local animal rights
protection authorities and followed the National Institutes of Health
guidelines for the care and use of laboratory animals. Male Sprague-Dawley
rats (250 to 300 g) were anesthetized with an intraperitoneal injection of 55
mg/kg of 6% pentobarbital sodium (Narcoren; Merial, Hallbergmoos, Germany).
Immediately prior to the injury, the animals received either 10 mg/kg of
parecoxib sodium (Dynastat; Pharmacia GmbH, Erlangen) (the parecoxib-0h group)
or an equal volume (0.5 mL) of saline solution (the saline group)
intravenously by the caudal vein. In an additional series of animals, 10 mg/kg
of parecoxib was administered intravenously two hours after the injury (the
parecoxib-2h group). Anesthetized animals that did not receive soft-tissue
trauma or therapy served as a time-matched control group (the sham group).
According to the analysis establishing the number of replications needed to
detect a given true difference between
means12, a total of
seven animals was included in each experimental group.
By means of a pneumatically driven and computer-controlled impact device, a
standardized soft-tissue injury was induced on the lateral compartment of the
left hindlimb, simulating high-velocity trauma of the lower extremity. The
nature and kinetics of tissue injury with use of this impact device have been
described by our group in detail
previously13. The
controlled-impact technique was initially developed as a model of standardized
traumatic brain injury in rats, reproducing the pathophysiological and
morphological responses of severe closed-head injury found in
humans14,15.
The controlled-impact device consists of a compressed nitrogen gas source, an
adjustable impactor, a displacement transducer, and a personal
computer-assisted interface for data transmission and analysis of
time-displacement parameters of the impact. The impact parameters that we
selected were an impact velocity of 7 m/s, a deformation depth of 11 mm, and
an impact duration of 100 ms with an impactor diameter of 10 mm. The left
hindlimb was placed in a plastic mold (Technovit; Kulzer, Wertheim, Germany)
shaped like the hindlimb to guarantee optimal energy transmission to the
tissue by avoiding hindlimb movement during the impact.
At eighteen hours after trauma induction, the animals were anesthetized
again and placed on a heating pad to maintain body temperature at 37°C.
Following a tracheotomy, the animals were mechanically ventilated (tidal
volume of 1 mL/100 g of body weight and fifty breaths per minute). Catheters
(PE-50; Portex, Hythe, Kent, United Kingdom) were placed in the right carotid
artery and the left jugular vein for continuous monitoring of central
hemodynamics (Sirecust; Siemens, Erlangen, Germany).
The left extensor digitorum longus muscle was microsurgically prepared to
allow direct access for in vivo high-resolution multifluorescence microscopy.
The preparation technique was first described by Tyml and
Budreau16 and was
modified for in vivo microscopy by our
group13. During
preparation, tissues were superfused with 37°C warm physiological saline
solution to prevent drying. After final exposure of the extensor digitorum
longus muscle, the tissue was covered with a cover glass.
After obtaining baseline recordings (the inclusion criteria included a mean
arterial blood pressure of 100 to 110 mm Hg, a hematocrit of 45% to 50%, a
pCO2 of 35 to 40 mm Hg, and a pH of 7.35 to 7.45) and a
twenty-minute stabilization period after completion of the exposure, in vivo
microscopy of the extensor digitorum longus muscle was performed. At the end
of the experiments, the animals were killed by exsanguination. Muscle tissue
was sampled for Western blot protein analysis, histology, and
immunohistochemistry.
Platelet Preparation
Resting platelets were isolated with use of the Sepharose column as
described
previously17. Blood
was drawn from healthy human volunteers (twenty-five to forty years old)
without history of disease or anticoagulant therapy, after they had provided
informed consent. After centrifugation, platelet-rich plasma was layered on a
prepared Sepharose column (Amersham Pharmacia Biotech, Uppsala, Sweden).
Isolated platelets were stained with
2',7'-bis(2-carboxyethyl)-5-(and-6-)-carboxy-fluorescein
acetoxymethyl ester (BCECF; Molecular Probes, Eugene, Oregon), passed again
through the Sepharose column, and were diluted in phosphate-buffered saline
solution to a final concentration of 1 ×108 cells
mL-1. Previous work in rodent models has shown that there is no
difference in the use of human platelets when compared with syngeneic
platelets17.
In Vivo Fluorescence Microscopy
After intravenous injection of fluorescein-isothiocyanate (FITC)-labeled
dextran (15 mg/kg of body weight) (Sigma, Deisenhofen, Germany) and rhodamine
6G (0.15 mg/kg of body weight) (Sigma), in vivo microscopy was performed with
use of a microscope (E600-FN; Nikon, Tokyo, Japan) equipped with a 100-W
mercury lamp and filter sets for blue (excitation of 465 to 495 nm and
emission of >505 nm), green (510 to 560 nm and >575 nm, respectively),
and ultraviolet (340 to 380 nm and >400 nm, respectively) epi-illumination.
BCECF-stained platelets were injected intra-arterially (1 ×
108 platelets per time-point over thirty seconds) and were allowed
a period of 120 seconds to recirculate until the analysis of their
intravascular adhesion with use of blue light
epi-illumination17.
With a physiological platelet concentration in rats of approximately 600
× 103 platelets/µL in whole blood and a total blood volume
of 8 mL/100 g of body
weight18, the
labeled fraction was about 1% of all circulating platelets. By use of
water-immersion objectives (Nikon) at ×20/0.75 W and ×40/0.80 W,
final magnifications of 306 and 630 times were achieved. Images were recorded
by means of a charge-coupled video camera (FK 6990-IQ-S; Pieper, Schwerte,
Germany) and transferred to a Super-VHS video system for subsequent off-line
analysis.
Microcirculatory Analysis
For quantitative off-line analysis, a computer-assisted microcirculation
image-analysis system (version 7.4, CapImage; Zeintl, Heidelberg, Germany) was
used. As previously
described13,19,20,
functional capillary density was defined as the total length of red
blood-cell-perfused capillaries per observation area in cm/cm2. To
assess leukocyte-endothelial cell interaction in postcapillary venules, flow
behavior of leukocytes was analyzed with respect to free floating, rolling,
and adherent leukocytes. Rolling leukocytes were defined as those cells moving
along the vessel wall at a velocity of <40% of that of leukocytes at the
centerline and were expressed as a percentage of the total leukocyte flux.
Venular leukocyte adherence was defined as the number of leukocytes not moving
or detaching from the endothelial lining of the venule wall during an
observation period of twenty seconds. Assuming cylindrical microvessel
geometry, leukocyte adherence was expressed as nonmoving cells per endothelial
surface (n/mm2), calculated from the diameter and length of the
vessel segment analyzed. In postcapillary venules, centerline red blood-cell
velocity was determined with use of the line-shift method (CapImage). Platelet
adhesion was analyzed within ten observation fields of skeletal muscle tissue
and was given as the number of adherent thrombocytes per square millimeter.
Reduced nicotinamide adenine dinucleotide (NADH) fluorescence of skeletal
muscle tissue was densitometrically assessed after two seconds of ultraviolet
epi-illumination by computer-assisted gray-level
determination21. To
avoid interference of gray levels with microvascular structures, analysis was
strictly limited to the intercapillary space.
Laboratory Analysis
Arterial blood samples were withdrawn for analysis of blood gases (Rapidlab
348; Bayer Vital, Fernwald, Germany) and blood cell count with use of a
Coulter Counter (AcTdiff; Coulter, Hamburg, Germany).
Western Blot Analysis
For Western blot analysis of COX isoforms, traumatized extensor digitorum
longus muscle tissue was homogenized in lysis buffer (10 mM Tris, pH 7.5, 10
mM NaCl, 0.1 mM EDTA, 0.5% Triton-X 100, 0.02% NaN3, and 0.2 mM
phenylmethylsulphonylfluoride (PMSF); prior to use, the buffer received a
protease inhibitor cocktail [1:100 v/v; Sigma]), incubated for thirty minutes
on ice and centrifuged for fifteen minutes at 10 g. The soluble whole protein
fraction was saved for subsequent analysis. Protein concentrations were
determined with use of the bicinchoninic acid (BCA) protein assay (Sigma) with
bovine serum albumin as a standard. Twenty micrograms of protein per lane were
separated discontinuously on sodium dodecyl sulfate polyacrylamide gels (12%)
and were transferred to a polyvinyldifluoride membrane (Immobilon-P;
Millipore, Eschborn, Germany). After blockade of nonspecific binding sites,
membranes were incubated for two hours at room temperature with goat
polyclonal anti-COX-1 and goat polyclonal anti-COX-2 antibodies (each 1:200;
Santa Cruz Biotechnology, Heidelberg, Germany) followed by
peroxidase-conjugated donkey polyclonal anti-goat IgG (1:40000; Santa Cruz
Biotechnology) as a secondary antibody. Ponceau-S staining of membranes served
to check for equal loading of lanes.
Protein expression was visualized by means of luminol-enhanced
chemiluminescence (ECL Plus; Amersham Pharmacia Biotech, Freiburg, Germany)
and exposure of the membrane to a blue light-sensitive autoradiography film
(Kodak BioMax Light Film; Kodak Industrie, Chalon-sur-Saone, France). Signals
were assessed densitometrically (Gel-Dokumentations-Systeme E.A.S.Y. Win32;
Herolab GmbH, Wiesloch, Germany).
COX-1 and COX-2 protein expression was assessed in the two parecoxib groups
and the saline-solution group at eighteen hours after trauma induction, and it
was assessed at eighteen hours in the sham-treated animals of the control
group. The kinetics of COX-1 and COX-2 expression were investigated with use
of protein extraction from extensor digitorum longus tissue obtained from
additional animals (five animals per time-point) that were killed at four,
eight, twelve, and eighteen hours after the induction of trauma. Extensor
digitorum longus muscle from animals without induction of trauma was harvested
at either zero hours (the 0h-sham group) or eighteen hours (the 18h-sham
group) and served as control tissue.
Histology and Immunohistochemistry
At the end of each experiment (eighteen hours after trauma), extensor
digitorum longus muscle tissue was fixed in 4% phosphate-buffered formalin for
two to three days and was then embedded in paraffin. From the
paraffin-embedded tissue blocks, 4-µm sections were cut and stained with
hematoxylin-eosin for histological analysis. To assess the temporal profile of
trauma-associated COX-2 expression, immunohistochemistry was performed in
extensor digitorum longus muscle tissue of additional animals (four animals
per time-point), which were killed at four, eight, twelve, and eighteen hours
after trauma. Extensor digitorum longus muscle tissue from animals without
trauma induction was harvested at either zero hours (the 0h-sham group) or
eighteen hours (the 18h-sham group). COX-2 was detected by means of a goat
polyclonal anti-COX-2 antibody (1:1000; Santa Cruz Biotechnology), followed by
a donkey polyclonal anti-goat antibody (Santa Cruz Biotechnology) and
counterstained with new fuchsin (Fuchsin Substrate-Chromogen System; Dako,
Carpinteria, California) and haemalaun. Quantitative analysis was performed by
counting the number of COX-2 positive cells in fifty consecutive high-power
fields (400 times magnification).
Statistical Analysis
The results were given as the mean and the standard error of the mean.
After proving the assumption of normality, comparisons between the
experimental groups were performed by one-way analysis of variance, followed
by the appropriate post hoc multiple comparison procedure, including
Bonferroni correction (SigmaStat; Jandel, San Rafael, California). To assess
the correlations between different microcirculatory parameters, Pearson
product moment correlation analysis was used. Significance was set at p <
0.05.
Western Blot Analysis of COX-1 and COX-2 Expression in Skeletal
Muscle Tissue
As illustrated by Western blot analysis of rat skeletal muscle
tissue, controlled impact device-induced closed soft-tissue injury caused a
transient increase in COX-2 protein expression with peak levels at eight hours
and twelve hours after trauma induction, followed by a recovery to almost
control levels at eighteen hours (Fig. 1,
A and B). In line with this, animals assessed at
eighteen hours after trauma did not differ with respect to COX-2 expression
compared with sham-operated control animals regardless of being given saline
solution or parecoxib before or after the injury (data not shown). Trauma
induction was not only accompanied by an upregulation of the inducible isoform
but also caused a pattern of COX-1 protein expression comparable with that
seen for COX-2 protein (Fig. 1, C
and D).
COX-2 Immunohistochemistry of Skeletal Muscle Tissue
In parallel with the kinetics of COX-2 protein expression, cells in the
perivascular connective tissue of injured muscle showed marked
immunoreactivity for COX-2 (Fig. 2,
A and B). Quantitative analysis of
COX-2-expressing cells revealed an increase from four to twelve hours after
trauma, with a threefold increase at eight hours
(Fig. 2, C),
decreasing to essentially baseline at eighteen hours.
Systemic Hemodynamics
Animals in the four experimental groups did not differ with respect to mean
blood pressure or heart rate (see Appendix). Moreover, no differences were
detected in hemoglobin, hematocrit, leukocyte count, and electrolytes among
the groups. Of interest is the observation that animals treated with saline
solution revealed a marked thrombocytopenia compared with those treated with
parecoxib and those in the sham group (see Appendix).
Microvascular Perfusion in Closed Soft-Tissue Injury
Closed soft-tissue trauma caused a substantial impairment in nutritive
capillary perfusion, with mean value (and standard error of the mean) of 296
± 30 cm/cm2, whereas treatment with parecoxib, even when
applied two hours after trauma, restored nutritive capillary perfusion up to
almost physiological baseline values, 474 ± 11 cm/cm2 for
the sham group, 434 ± 15 cm/cm2 for the parecoxib-0h group,
and 399 ± 8 cm/cm2 for the parecoxib-2h group, as observed
in the sham control animals (Fig.
3). In line with nutritive perfusion failure, high NADH
autofluorescence of skeletal muscle tissue was observed after closed
soft-tissue injury in saline solution-treated animals, indicating pronounced
tissue hypoxia (mean, 100 ± 4 aU compared with 68 ± 5 aU for the
sham group). In contrast, animals treated with parecoxib exhibited a marked
reduction of tissue NADH (73 ± 2 aU for the parecoxib-0h group and 74
± 1 aU for the parecoxib-2h group), which is in line with the
concomitant improvement in capillary perfusion. Regression analysis revealed a
significant inverse correlation between functional capillary density and NADH
autofluorescence of skeletal muscle tissue with a regression coefficient of r
= -0.68 (p < 0.05) (Fig.
4).
Capillaries were found to be significantly widened in traumatized skeletal
muscle (mean, 5.7 ± 0.2 µm) compared with nontraumatized sham
controls (mean, 4.8 ± 0.1 µm) (p < 0.05). Capillary diameter was
reduced to a mean of 5.2 ± 0.2 µm in the animals that received
parecoxib prior to trauma. Parecoxib given after the injury was capable of
completely restoring capillary diameter (mean, 4.9 ± 0.1 µm in the
parecoxib-2h group compared with a mean of 5.7 ± 0.2 µm in the
saline-solution group; p < 0.05). Venular red blood-cell velocity was
slightly decreased in the saline-solution-treated animals and the
parecoxib-treated animals, but it was not significantly different from that in
the sham control animals (data not shown) (analysis of variance, p =
0.11).
Inflammatory Cell Response in Closed Soft-Tissue Injury
Soft-tissue trauma was characterized by an inflammatory cell response with
significant (p < 0.05) increases in leukocytes, both rolling along
(threefold) and firmly attaching to the venular endothelium (eightfold).
Parecoxib, given either prior to trauma or two hours after injury, effectively
limited the inflammatory response with low numbers of leukocytes interacting
with the venular endothelium, which was comparable with the numbers observed
in sham control animals (Table
I).
Following trauma, the saline solution-treated animals revealed enhanced
intravascular thrombocyte accumulation, which was absent in the animals
treated with parecoxib before or after injury
(Table I). Intramuscular
accumulation of leukocytes was significantly (p < 0.001) correlated with
that of thrombocytes (r = 0.62).
These results show that both preinjury and postinjury application of
the selective COX-2 inhibitor parecoxib is effective in treating
trauma-induced microcirculatory disturbances with almost complete restoration
to normal by eighteen hours after the trauma in this animal model. These
observations, together with the upregulation of COX-2 protein expression after
skeletal muscle trauma, suggest a role for COX-2 in the response to
soft-tissue injury.
Closed Soft-Tissue Injury
This model of closed soft-tissue injury mimics the characteristics observed
clinically in patients experiencing high-energy trauma and demonstrates a
marked decrease in capillary perfusion as well as a marked increase in
leukocyte-endothelial cell interaction and microvascular
permeability13.
Besides direct tissue destruction caused by the impact itself, tissue damage
results from traumatically induced inflammatory reactions, with tissue hypoxia
being the most likely trigger. In line with this, injured muscle revealed a
marked increase in tissue NADH autofluorescence, indicating pronounced tissue
hypoxia. In general, the interruption of oxidative phosphorylation due to an
inadequate oxygen supply is reflected by an increase in NADH
levels22. NADH
fluorimetry allows noninvasive investigation of organ metabolism, reflecting
alterations in oxidative
phosphorylation23.
So far, enhanced NADH fluorescence caused by hypoxia has been monitored in rat
liver tissue in vivo upon postischemic reperfusion and hemorrhagic
shock21,24.
We now demonstrate the strong inverse correlation between capillary perfusion
and NADH fluorescence within traumatized skeletal muscle tissue. The
insufficient oxygen supply can be attributed mainly to trauma-induced
perfusion shutdown of individual capillaries, but it may in addition be due to
capillary vasomotor
dysfunction25, as
indicated in the present study by extreme dilatation of these microvascular
segments. Although the mechanisms of capillary diameter control in skeletal
muscle tissue are incompletely known, it is reasonable to speculate that
pericytes, endothelial cells, and the endothelin-nitric oxide system also
control vascular diameter in muscular tissue, as has been shown for hepatic
and pancreatic
tissue26,27.
The results of this study extend our previous observations by demonstrating
that injured muscle shows both leukocytic and thrombocytic sequestration. In
this study, intraorgan thrombocyte accumulation after injury was severe enough
to be associated with systemic thrombocytopenia. It has been reported that
peripheral soft-tissue trauma causes massive intrapulmonary trapping of
thrombocytes28,29,
but, to our knowledge, sequestration of platelets within the injured tissue
itself has not been previously demonstrated. Since platelets generate an array
of proinflammatory mediators and oxygen radicals, they, like leukocytes, might
be regarded as both mediator and effector
cells30. This is
further supported by the fact that, in the present study, the accumulation of
thrombocytes strongly correlates with that of leukocytes.
COX-2 Inhibition in Closed Soft-Tissue Injury
Our data showing the strong ability of parecoxib to limit features of
microvascular and inflammatory tissue damage implicate COX-2 as a potential
mediator of soft-tissue injury. This view is underscored by the transient
upregulation of COX-2, as shown by the time-course studies with use of both
Western blot analysis and immunohistochemistry. Comparably, COX-2 has been
reported to be upregulated in surgery-associated paraspinal muscle injury, but
peak levels were not reached before three
days31.
Although the present study exclusively focused on the inhibition of the
inducible COX-2 isoform in soft-tissue injury, there is evidence that the
contribution of each isoform to the prevention or development of disease is
more complex than originally described. For example, one study has
demonstrated that COX-1 inhibition equals that of COX-2 in efficacy to
attenuate lipopolysaccharide-induced hepatic
injury32. That
study and other
reports33,34
have challenged the current paradigm of a selective role of COX-2 in the
inflammatory response of tissue to injury. The present observation that
soft-tissue trauma caused an increase in COX-1 protein as well as COX-2
suggests that COX-1 may be involved in the response of soft tissue to
injury.
Besides their proinflammatory activities, prostaglandins have been shown to
be beneficial in the resolution of tissue injury and inflammation. For
example, prostacyclin exerts potent antiaggregatory and antiadhesive
properties. Thus, inhibition of COX-2 as the predominant prostaglandin
endoperoxide synthase not only may be of benefit but also may result in
augmentation of the inflammatory response. Accordingly, it has been shown that
superfusion of mesenteric venules with the COX-2 inhibitor, celecoxib,
promotes leukocyte
adherence35.
Moreover, inhibition of COX-2 exacerbated inflammation-associated colonic
injury in colitis
models36 and in
both liver and bowel injury upon resuscitation from hemorrhagic
shock37. However,
COX-2 inhibitors have also been used successfully to reduce inflammation, as
evidenced by attenuation of
ischemic38 and
traumatic brain
injury39,
postischemic liver
injury40,41,
and pancreatitis-associated local and remote organ
injury42,43.
We now report that the acutely injured muscle benefits from immediate COX-2
inhibition, whether it is administered before or after trauma. The observation
that preinjury and postinjury administration of parecoxib provided nearly
identical efficacy suggests that COX-2 exerts its deleterious effects not
earlier than two hours after trauma in this animal model. This suggests that
there might be a therapeutic window to allow for successful postinjury
treatment.
Parecoxib-induced reduction of leukocyte response within the injured tissue
is presumably due to a reduction in endothelial ICAM (intercellular adhesion
molecule)-expression, as this has been shown for the COX-2 inhibitor meloxicam
in diabetic
retinopathy44.
Reduced leukocyte adherence in outflow venules of parecoxib-treated animals
may be associated with a lower resistance to flow due to a higher
cross-sectional
area45, which, in
turn, may indirectly support better capillary perfusion. Moreover, reduced
leukocyte adherence goes along with less blood
viscosity46, which
additionally improves microvascular flow conditions.
In summary, the protective effect of COX-2 inhibition implies that COX-2
contributes to trauma-induced soft-tissue injury. COX-1 could also be targeted
in order to limit tissue injury, but this approach is hampered by the multiple
side effects exerted by the commonly available nonselective COX inhibitors.
Thus, selective COX-2 inhibitors exhibiting low side effects may be of
superior therapeutic value in protecting the microcirculation and preserving
skeletal muscle from secondary inflammatory tissue damage following closed
soft-tissue injury.
A table presenting the hemodynamic and hematologic parameters measured in
all four study groups is available with the electronic versions of this
article, on our web site at
(go to
the article citation and click on "Supplementary Material") and on
our quarterly CD-ROM (call our subscription department, at 781-449-9780, to
order the CD-ROM). ?