INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

SEPSIS and systemic inflammatory response syndrome induce profound changes in the circulatory system and, in particular, the microcirculation (12). The precise mechanisms by which the microvasculature in different tissues is injured, however, have not been elucidated. In addition to changes in microvascular properties, including vascular reactivity (35), leukocyte-endothelial cell adhesion, and vascular leakage (27), an important characteristic of microvascular injury during the course of sepsis is the development of microcirculatory derangements. These have been reported as losses of capillary density (CD) (24), indicating a loss of surface area for gas exchange, a maldistribution of blood flow, and increased flow heterogeneity (18). The implications are that microvascular dysfunction during sepsis contributes to the somewhat paradoxical decrease in oxygen extraction (26) by compromising oxygen delivery and to tissue injury by preventing adequate tissue oxygenation (13).

Factors suggested to contribute to microcirculatory dysfunction include changes in blood rheology arising from decreased erythrocyte (1, 33) and leukocyte deformability (3), red blood cell (RBC) aggregation (2), and coagulation mechanisms (21) as well as systemic hypotension caused by increased nitric oxide (NO) production from the induction of inducible NO synthase (iNOS) (14). Interestingly, NO may play a multifactorial role in the development of microvascular injury, because it has been reported to modulate both RBC deformability (31) and signal transduction pathways in the RBC (23). RBC from both septic patients and animal models (1, 22) are known to exhibit decreased deformability, although the functional significance of this on capillary perfusion is unknown.

In this novel acute rat model of normotensive sepsis after cecal ligation and perforation (CLP), we utilized intravital video microscopy to study functional CD in the extensor digitorum longus (EDL) skeletal muscle and concurrently measured systemic erythrocyte membrane deformability () and plasma nitrite/nitrate (NO_x) and assessed NO accumulation in RBC. We hypothesized that preventing decreases in RBC deformability would improve capillary blood flow distribution, as measured by functional CD. To test this hypothesis, we used aminoguanidine (AG), an in vivo inhibitor of iNOS (34), which in pilot studies has demonstrated the dual properties of inhibiting NO production and preventing decreased RBC deformability. We report, for the first time, that preventing the decrease in erythrocyte deformability during acute sepsis is a factor in preventing the decrease of functional CD in skeletal muscle and that the mechanism appears to be NO mediated.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals. Nonfasting adult male Sprague-Dawley rats were randomly divided into the following four treatment groups: sham/saline, sham/AG treatment, CLP/saline, and CLP/AG treatment. An additional two groups of animals, naive and CLP, were used to assess lipid peroxidation in the experimental model. Experimental protocols were approved by the University of Western Ontario Council on Animal Care.

Acute murine model of fecal peritonitis-septicemia (CLP). Anesthetized rats (4% halothane-balance oxygen, Fluotec 3 vaporizer, Cyprane) were fitted with a mask, and inspired gas was set to 1% halothane-30% oxygen-balance nitrogen (flow rate 3.3 l/min). Oxygen levels were monitored by an oxygen sensor (Critikon Oxychek). Animal temperature (37°C) was maintained by a heat lamp and monitored by a rectal probe (Simpson 383). The right carotid artery was canulated [polyethylene (PE)-10/PE-50, Clay Adams] to enable monitoring of mean arterial pressure (MAP) and heart rate (HR) (Digi-Med blood pressure analyzer and Baxter flow transducers). The left jugular vein was cannulated (Bio-Sil tubing, Silmed) for intravenous fluid resuscitation [1 ml · 100 g¹ · h¹ of 0.9% sodium chloride (Baxter) by Harvard Apparatus syringe pump] and drug delivery. Animals were tracheotomized and mechanically ventilated (model 683, Harvard rodent ventilator). Halothane concentration, respiratory rate (78-85 breaths/min), and tidal volume (1.6-2.2 ml) were adjusted to maintain blood gases within the spontaneously breathing range (PO₂ 85-100 mmHg, PCO₂ 35-45 mmHg, pH 7.3-7.4). Animals underwent either 1) CLP or 2) sham laparotomy. For CLP, the cecum was exposed through a midline incision with the use of sterile surgical techniques, ligated with a 3-0 silk suture, and perforated (0.5-cm incision). Fecal material was then expressed into the peritoneal cavity, inducing acute fecal peritonitis-septicemia, and the midline incision was sutured with 3-0 silk. Sham laparotomy consisted of exposing and manipulating the cecum.

Drug dose and delivery. AG hemisulfate (Sigma) was administered via venous catheter at 60 mg · kg¹ · h¹. Pilot studies using 20, 40, and 60 mg · kg¹ · h¹ in CLP animals have shown that a dose of 60 mg · kg¹ · h¹ maintained plasma NO_x levels at baseline over a 6-h period. Drug delivery was started 2 h post-CLP or sham laparotomy to coincide with the upregulation of iNOS known to occur within 2-3 h after injury (5).

EDL muscle preparation. The right hindlimb EDL muscle was blunt dissected, exposed by cutting the distal tendon, and reflected along its length to provide a microscopic ventral surface view (Leitz Metallux 3). Moderate tension was applied to restore the approximate in situ muscle length. The muscle was bathed in warmed (37°C) 0.9% saline, covered in Saran Wrap (which acted as an oxygen barrier), and slightly compressed by an 18 × 18-mm coverslip (VWR Scientific). The animal and the microvascular preparation were then allowed to stabilize for 30 min.

Intravital video microscopy of EDL muscle. The EDL muscle was transilluminated by a 100-W mercury vapour light source, and its surface microcirculation was viewed through a ×20 (Leitz) long working distance objective, numerical aperture 0.4, with an effective magnification on the monitor of ×510. High-contrast images were achieved with a 431-nm band-pass filter. Video images of the microcirculation were obtained with a silicon-intensified target (SIT) camera (Dage-MTI-STI-66), viewed on a monitor (Panasonic WV-5410) via a closed-circuit video system, and recorded (Panasonic 7300 SVHS VCR) on SVHS videotape for offline analysis. At 15- to 20-min intervals, from 3.5 to 5.75 h post-CLP/laparotomy, random 5-min video recordings were made. In total, seven to nine random fields were recorded per experiment. Fields were first selected at 2 h post-CLP/laparotomy and then filmed later at 3.5-5.75 h post-CLP/laparotomy. Images were focused to maximize the number of sharply visible capillaries. If large arterioles or venules were present in the field of view, the muscle was repositioned to maximize the number of clearly focused capillaries. To confirm that in a CLP model of sepsis leukocytes do not accumulate in the postcapillary venules of the EDL microcirculation and obstruct capillary flow, as reported by Piper et al. (25), fields of view containing postcapillary venules were examined for rolling and adherent leukocytes.

Functional CD analysis. CD in the EDL was determined with the use of the method described by Piper et al. (24). Videotapes were analyzed by a single observer in a blinded fashion. A transparency with three staggered test or reference lines (150 µm), drawn perpendicular to the muscle fibers, was placed over the video monitor screen. Capillaries that intersected the test lines were scored over a 30-s time interval as being either 1) continuous; 2) intermittent, in which case capillary flow either reversed or came to full arrest at least once; or 3) stopped. Only capillaries containing RBC were counted. Empty capillaries were difficult to identify and considered to be nonfunctional with respect to RBC flux. Functional CD values [CD values in continuous (CD_cont), intermittent (CD_int), or stopped capillaries (CD_stop)] are expressed as the number of capillaries per millimeter test line (caps/mm).

Blood samples. Heparinized arterial blood (500 µl) was collected at 0, 3, and 6 h post-CLP/laparotomy. The blood volume was replaced at 0 and 3 h with physiological saline (Baxter). Ten microliters of blood was diluted into 5 ml of PBS (Sigma) for erythrocyte membrane displacement measurements, and the balance was centrifuged at 1,000 g for 5 min at 4°C. Plasma samples were stored at 20°C for later NO_x analysis. At 6 h, final blood gas determinations were made (ABL Radiometer), and the automated complete blood cell count and lactate analysis were obtained (model STKS 908, Coulter Electronics, Burlington, Ontario, and Synchron system, Beckman, Coulter, CA, respectively). At 6 h, 1 ml of arterial blood from three sham, CLP, and CLP + AG animals was directly transferred to a 125 × 4-mm, 2-mm inner diameter (ID) suprasil quartz electron paramagnetic resonance (EPR) tube (Wildmar Glass), sealed with a rubber septum, and immediately frozen in liquid nitrogen for later EPR analysis of hemoglobin-bound NO.

Erythrocyte membrane deformability. The mechanical properties of individual RBC were assessed with the use of the micropipette aspiration technique, which measures the displacement of an individual RBC membrane () into a micropipette under a fixed negative pressure. The distance of penetration of the RBC membrane was defined by Leblond (20) as surface deformability and a measure of erythrocyte deformability. In this study, on average, 25 different RBC membrane displacement measurements were made at each time point.  data are reported here in two ways. First, as mean RBC membrane displacement values () at each time point, and, second, as pooled normalized values. For the latter,  values at 3 and 6 h were normalized (/₀, where ₀ is the baseline mean displacement value for a given experiment) to control for differences between micropipettes. /₀ data from each group were then pooled over all experiments and presented as frequency distributions centered about the baseline value, which was arbitrarily set to zero.

NO_x analysis by chemiluminescence. Plasma NO_x was measured by first chemically reducing NO_x to NO gas (0.05 M VCl₃ in 1 M HCl at 92°C) and then detecting the NO gas using a Sievers 270B NO analyzer (Sievers NOA, Sievers; Boulder, CO). A Chromatopak (Shimadzu) recorder was used to integrate the analyzer output signals. The instrument was calibrated against known concentrations of sodium nitrate.

EPR spectroscopy. The presence of NO in arterial blood bound to hemoglobin (16) was detected qualitatively using a Brucker ESP-300 spectrometer (Bruker Instruments). EPR spectra were obtained at 77 K operating at 9.42 GHz with a power of 20 mW and a modulation frequency of 100 kHz. Each spectrum was the result of signal averaging over two scans.

Lipid peroxidation: malondialdehyde and related aldehydes. The extent of oxidant damage during acute sepsis was estimated in CLP animals (n = 4) relative to naive (n = 4) controls by measuring the levels of thiobarbaturic acid-reactive substances (TBARS) in both plasma and RBC membrane fractions according to Buege and Aust (6). Briefly, 750 µl of plasma or packed RBC were mixed with 1,500 µl of 15% trichloroacetic acid, 0.375% thiobarbaturic acid (TBA), and 0.25 N hydrochloric acid (Sigma) and then boiled for 15 min. TBARS levels were estimated from the differences in absorbance at 531 and 600 nm using an extinction coefficient of 1.56 × 105 mol/cm (32) for the malondialdehyde-TBA complex. Results were expressed as nanomoles per milliliter of packed RBC and micromolars of plasma.

Statistical analysis. All values are expressed as means ± SE unless otherwise stated. For all tests of significance, P values < 0.05 were considered statistically significant. A 2 × 2 factorial experimental design composed of two factors (injury and drug treatment), each at two levels [CLP and sham (injury) and AG and saline (drug treatment)], was employed in this study. Repeated measures analysis of variance (ANOVA) was done with the SAS/PROC GLM procedure (SAS Institute, Cary, NC) on animal blood pressure and HR, plasma NO_x level, and erythrocyte deformability. Two-way ANOVA was used to assess end-point physiological data and the CV of CD_perfused. Multivariate ANOVA was used to assess overall differences in CD_cont, CD_int, and CD_stop, and ANOVA was used to assess total functional CD (CD_total). A Tukey-Kramer multiple comparison test for all pairwise differences was performed to identify specific pairwise group mean differences. Frequency distributions from pooled erythrocyte deformability and functional CD data were not all normally distributed in all groups, so Kruskall-Wallis ANOVA on ranks followed by Dunn's method was used to make multiple comparisons versus the sham group. A t-test was used to assess the difference in TBARS levels between naive and CLP animals.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals. Thirty rats weighing between 180-260 g were used in this study in the following groups: sham (n = 7), CLP (n = 8), CLP + AG (n = 8), and sham + AG (n = 7). An additional eight rats, naive (n = 4) and CLP (n = 4), were used to assess the extent of lipid peroxidation in the CLP animals.

Postoperative observations on the animals. Generalized purulent peritonitis was confirmed at post mortem in all CLP animals. The appearance of the cecum varied from being severely inflamed to gangrenous. Serosanguinous fluid was observed in the peritoneal cavity. In contrast, sham animals had no signs of peritonitis.

MAP and HR. MAP and HR profiles are shown in Fig. 1. The acute CLP procedure caused changes in MAP over time in CLP and CLP + AG animals (time × injury effect, P = 0.014). This was due to the transient increase in MAP from 1.5 to 3 h post-CLP. Beyond 3 h, no differences in MAP between sham and CLP animals were detected, indicating the septic injury was now normotensive. AG increased the MAP in CLP animals relative to nontreated CLP animals at 260 min post-CLP. Despite a trend toward increasing HR in both CLP and CLP + AG animals (time × injury effect, P = 0.108), no significant differences with sham animals were detected. AG had no effect on HR in either sham or CLP animals.

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Mean arterial pressure (MAP; A) and heart rate (HR; B) profiles. MAP and HR profiles are presented for clarity of the presentation as follows: sham, sham + aminoguanidine (AG), cecal ligation and perforation (CLP), and CLP + AG. All sham animals had stable MAP and HR over the experiment. Whereas MAP appeared to increase transiently in all CLP animals relative to sham animals, only the CLP + AG group was found to have significant differences (*P < 0.05, CLP + AG vs. sham animals). A difference in MAP between the CLP groups was detected at 260 min (+P < 0.05, CLP vs. CLP + AG animals). No significant differences in HR were detected among the groups.

Physiological data. A number of physiological changes were induced by the CLP injury (Table 1). AG had no effect in CLP animals on leukocyte count, hematocrit, hemoglobin, pH, arterial PO₂, or arterial PCO₂ (no injury × drug interactions). AG did, however, increase lactate levels in both CLP and sham animals.

Functional CD. During the 2.25-h observation period (3.5-5.75 h post-CLP/laparotomy), there was no detectable change in CD_total (Fig. 2) in the EDL muscle capillaries. There was, however, a significant 149% increase in CD_stop in CLP animals (CLP vs. sham, P < 0.05) (Fig. 2). We did not observe leukocytes occluding the postcapillary venules of random fields of view. AG decreased the CD_stop in CLP animals by 39% (CLP + AG vs. CLP, P <0.05) to a level that was not significantly different from sham (sham vs. CLP + AG, P = 0.598) and had no effect in sham + AG animals. No changes in intermittent capillary flow were detected among the groups. The change in capillary perfusion (CD_perfused = CD_cont + CD_int) between groups was manifested by a significant injury × drug interaction (P < 0.05). This was attributed to a 55.5% decrease in continuous flow capillaries in CLP animals (from 19.11 ± 5.5 to 8.49 ± 3.4 caps/mm, sham vs. CLP, P = 0.082) and to the effect of AG, which prevented any change in continuous flow in CLP + AG animals relative to sham (12.8 ± 11.3 vs. 19.11 ± 5.5 caps/mm, CLP + AG vs. sham, P = 0.371).

View larger version (31K): [in this window] [in a new window]   Fig. 2.   Functional capillary density (CD) 3.5-5.75 h post-CLP/laparotomy in extensor digitorum longus (EDL) muscle. CDtotal, the sum of continuous, intermittent, and stopped flow CD (CDcont + CDint + CDstop), remained the same in all groups. CDstop significantly increased in the CLP animals compared with sham animals (*P < 0.05). AG significantly decreased CDstop in CLP + AG animals compared with the CLP group (+P = 0.005, CLP + AG vs. CLP) to a level that was unchanged from the sham animals. The decrease in continuous flow in CLP animals was not significant (P = 0.082, CLP vs. sham); however, it was a contributing factor to the overall injury by drug interaction for capillary perfusion (P < 0.05).

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Perfused CD (CDperfused) and CDstop frequency distributions. Frequency distributions of CDperfused and CDstop were generated from pooled data over all animal groups. Note the significant rightward shift of the CDstop distribution in CLP animals (B) compared with sham animals (A) (#P < 0.05) and the corresponding leftward shift of the CDperfused distribution compared with sham animals (#P < 0.05). AG restored CLP distributions (D) to those of sham animals and had no detectable effect on sham animals (C).

NO accumulation in RBC. The large signal between 3,300 and 3,400 G in the EPR spectrum of CLP animals (Fig. 4) indicated that NO had accumulated in the RBC as nitrosylhemoglobin (16) by 6 h post-CLP. AG prevented NO accumulation in CLP + AG animals because the EPR spectrum appeared unchanged from that of sham animals.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Electron paramagnetic resonance (EPR) spectra of arterial blood samples at 6 h post-CLP/laparotomy. Representative EPR spectra from sham, CLP, and CLP + AG animals are shown. The presence of nitric oxide (NO) within the red blood cells (RBC) was evident by the large signal between 3,300 and 3,400 G. This signal is characteristic of the six coordinated Hb -NO form of nitrosylhemoglobin. AG prevented the accumulation of NO in CLP + AG animals.

Time course: erythrocyte deformability and plasma NO_x levels. Significant changes in both and plasma NO_x levels occurred over time (NO_x × time and × time effects, P < 0.05, respectively) as the septic injury progressed (Fig. 5). In sham animals, and plasma NO_x levels remained stable over the experimental period. In CLP animals, at 3 h, had decreased by 5.2% from baseline with a concomitant 39% increase in plasma NO_x levels (P < 0.001) compared with sham animals. By 6 h, had decreased by 12.7% (P < 0.001) and plasma NO_x levels had increased by 254% (P < 0.001) compared with sham animals. The temporal relationship between NO overproduction and , expressed as percent change in each animal, is shown in Fig. 6. AG maintained plasma NO_x at baseline levels in the CLP + AG animals and prevented any change in over time.

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Time course profiles of changes in erythrocyte deformability and plasma nitrite/nitrate (NOx) levels. CLP animals showed significant decreases in mean RBC membrane displacement (; A), a measure of RBC deformability, by 6 h. Plasma NOx levels (B) were elevated by 3 and 6 h in CLP animals compared with sham animals (*P < 0.001). AG treatment started 2 h after injury maintained both plasma NOx and at baseline in CLP + AG animals and had no effect in sham + AG animals.

View larger version (20K): [in this window] [in a new window]   Fig. 6.   Dynamic changes in erythrocyte deformability relative to NOx overproduction. Scatter plots showing the percent changes in , a measure of RBC deformability, against the percent changes in NOx production in all animals over time are shown. A: no changes occurred in the sham animals with or without AG at 3 or six 6 h. B: in the CLP group, at 6 h, all NOx levels had increased by >100%. This was associated with a decrease in of >12%. AG restored NOx levels to baseline and prevented the loss of RBC deformability.

Erythrocyte deformability frequency distributions. Frequency distributions of normalized RBC membrane deformability data (/₀) are shown in Fig. 7. At 3 h, the CLP distribution (Fig. 7C) was similar to that of sham animals, which ranged to ±30% of baseline mean deformability. By 6 h, however, the CLP deformability distribution was found to be significantly different from the sham distribution (Fig. 7D) (CLP vs. sham animals, P < 0.05). A subpopulation of RBC (5.3%) with decreased deformability beyond the lower range of sham animals was detected (Fig. 7D, circled area). AG prevented this subpopulation from developing in CLP + AG animals.

View larger version (27K): [in this window] [in a new window]   Fig. 7.   Erythrocyte deformability frequency distributions. Frequency distributions were generated from normalized RBC deformability measurements (/0, where 0 is the baseline mean displacement value) that were pooled over all animals. At 3 h (A and C), no change in the range of the distribution was seen in either CLP, CLP + AG, or sham + AG animals relative to sham animals. By 6 h (B and D), however, the distribution in the CLP group was shifted to the left (#P < 0.05), and a distinct subpopulation of RBC (5.3%) with deformabilities below the sham range was detected (D, circled area). AG prevented the shift of the RBC distribution and the development of the subpopulation with markedly decreased deformability in CLP + AG animals.

RBC and plasma TBARS levels. As shown in Table 2, there was no significant difference in the TBARS level in either the RBC or plasma fraction between naive and CLP animals.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Summary. In this acute animal study, we demonstrated that acute normotensive sepsis caused a significant increase in the density of stopped flow capillaries in the EDL skeletal muscle, in a seemingly random manner, between 3 and 6 h after induction of fecal peritonitis. Over the same time period, systemic NO production was significantly increased, NO accumulated in the RBC, erythrocyte deformability decreased, and a subpopulation of erythrocytes with markedly decreased deformability developed. There was no evidence of lipid peroxidation in the CLP animals. Delayed administration of AG did not prevent the sequestration of leukocytes or changes in hematocrit that occurred in CLP animals; however, AG maintained NO levels at baseline and prevented the decrease in both RBC deformability and functional CD.

Acute murine model of fecal peritonitis-septicemia. This study utilized a normotensive model of acute sepsis, which is representative of acute clinical peritonitis and systemic inflammatory response syndrome and demonstrates the remote microvascular injury that is observed in 18- to 48-h models of sepsis (18, 24). Because the remote injury in this acute model occus within 6 h, it is possible to follow the time course of the injury from normal to severely impaired microvasculature. The severity of the acute injury was evident by extensive serosal gut inflammation and necrosis, a significant twofold increase in lactate level, and the decrease of remote microvascular capillary perfusion density. The progressive nature of the response to injury was manifested by the time-dependent changes in RBC deformability and plasma NO_x levels.

The degree and nature of capillary bed injury during sepsis. A number of different experimental models of sepsis looking at diverse remote capillary beds, including the rat diaphragm (4) and EDL skeletal muscle (18, 24, 25) and the canine small intestine (8), have all reported significant increases in the number of stopped flow capillaries. The loss of capillary perfusion, ranging from 12.8 to 40.9%, may represent a general microvascular response to sepsis and is consistent with the results of our study in the EDL muscle.

Erythrocyte deformability in sepsis. Sepsis has been shown to induce changes in RBC mechanical and membrane properties, including changes in membrane viscosity (33), decreases in the deformability index (19) and elongation index (1, 30), and increases in shape recovery time (1). In this study, by measuring  of individual RBC, we have identified, for the first time, a subset of circulating RBC comprising ~5% of the total population that have suffered decreases in deformability beyond the normal RBC deformability range (Fig. 6D). This subpopulation is consistent with reports of suspected small populations of RBC in both clinical and experimental sepsis having "extremely long" RBC transit times (1). Our measure of decreased deformability may have been an underestimate because trapped RBCs could not be measured. When AG was administered to CLP animals, this subpopulation of RBC failed to develop and CD_stop did not increase, suggesting that less deformable RBC did indeed contribute to vascular stasis.

A role for NO in decreased RBC deformability. Changes in RBC morphology and deformability have been associated with a number of factors including pH, reduced ATP concentration, oxidative stress (7), spectrin-hemoglobin complexation (29), and lipid membrane peroxidation (1, 22). Despite several reports of increased lipid peroxidation, an indicator of oxidant damage, in advanced septic patients (11, 17, 22) and in an 18-h experimental model of sepsis (1), we found no increase in lipid peroxidation in our acute 6-h model of sepsis. This suggests that, in the early stages of sepsis, endogenous antioxidants are capable of preventing lipid peroxidation and minimizing oxidant damage.

Change in plasma NO_x as a marker of remote injury in sepsis. Langenfeld et al. (19) reported that changes in the RBC deformability index were "more helpful than either WBC count or temperature changes in discriminating patients with or without infection." By 6 h in our acute studies, we found that CLP animals had an 87.5% incidence of at least a 12% decrease in and a 100% incidence of at least a twofold increase in plasma NO_x (Fig. 6). Each parameter had a specificity of 100%. This was consistent with Machiedo et al. (22), who reported a 96% incidence of decreased RBC deformability with a specificity of 87% in septic patients. On the basis of our findings, we suggest that twofold increases in plasma NO_x could be used as an index or marker of decreased RBC deformability and remote microvascular injury in fecal peritonitis-septicemia in rats.