Hepatic versus Pulmonary Uptake of Particles Injected into the Portal Circulation in Sheep Endotoxin Escapes Hepatic Clearance Causing Pulmonary lntlammatlcn'?
MALCOLM M. DECAMP, ANGELINE E. WARNER, RAMON M. MOLINA, and JOSEPH D. BRAIN
Introduction
Resident macro phages in contact with circulating blood make up an important part of the body's host defenses. These cells of the mononuclear phagocyte system (MPS) playa central role in keeping the blood free of denatured proteins, foreign particles, injured or senescent cells, bacteria, endotoxin, and products of intravascular coagulation (1).The liver and spleen have traditionally been viewed as the primary organs with sinusoidal macrophages responsible for clearance of particulate materials from the bloodstream (2-5). This generalization has important exceptions. We have shown that sheep, calves, goats, and cats have a large population of avidly phagocytic pulmonary intravascular macrophages (PIMs) residing in their lung capillaries (6-8). PIMs have also been described in pigs (9). In these species, the lungs compromise a prominent compartment of the MPS, and the pulmonary uptake of pathogenic particulates (e.g., gram-negative bacteria or endotoxin) by PIMs is associated with evidence of their activation and rapid onset oflocal tissue injury (10-12). We believe that PIMs playa pivotal role in this injury through release of oxygen radicals and inflammatory mediators, and that PIM phagocytic activity may contribute to the considerable pulmonary sensitivity of certain species (especially, e.g., sheep and pigs) to circulating bacteria or endotoxin. These species have become the most popular experimental animal models of the human adult respiratory distress syndrome (ARDS), a syndrome of inflammatory lung injury that often followsgram-negative septicemia or endotoxemia. Experimentally, endotoxin administered for organ uptake studies is usually injected via a peripheral vein. The intestinal tract, however, is a major source of 224
SUMMARY Removal of circulating partiCUlates (bacteria, cell debris, endotOXin) Is accomplished in most species by macrophages resident in the liver and spleen. We have shown that sheep and other species have phagocytic macrophages resident In their pulmonary capillaries. Moreover, these pulmonary Intravascular macrophages accomplish the bulk of uptake of Injected tracer partiCles, bacteria, or endotoxin (LPS). Because bacteria or LPS of Intestinal origin enter the portal circulation, they would first encounter hepatic mononuclear phagocytes. We sought to determine the extent to which particulates Injected Into the portal circulation of sheep would be taken up by liver or by lung macrophages. Sheep (four per group) were injected via a mesenteric vein with radlolabeled gold colloid, magnetic Iron oxide particles, live Pseudomonas aeruginosa, or "51E. coli endotoxin. For each, the uptake pattern was determined 1 h after injection. Lung and liver were also fixed to determine the cells responsible for uptake and SUbsequent Inflammatory changes. Wefound that for circulating gold colloid, Iron oxide particles, or bacteria, hepatic uptake predominated, and Kupffer cells were responsible. After hepatic uptake of bacteria, Inflammatory changes were confined to the liver. In contrast, nearly 50% of endotoxin escaped hepatic clearance and was subsequently removed by the lungs. We then saw Inflammatory changes In both lungs and liver. ThUS, hepatic macrophages are active In species with pulmonary Intravascular macrophages, partially sparing the lungs from uptake and acute Inflammation. Endotoxin, however, may elude hepatic uptake, be sequestered in the lungs, and Initiate Inflammation there. AM REV RESPIR DIS 1992; 146:224-231
endotoxin in vivo. Gut-derived lipopolysaccharide (LPS) can pass to the general circulation via direct absorption into portal blood, by translocation to intestinal lymphatics, or by escape across the intestinal wallinto the peritoneal space (13). Liver phagocytes are thought to protect the heart, lungs, and systemiccirculation from the deleterious effects of such pathogens derived from the intestinallumen. Many patients who develop ARDS have no evident pulmonary focus of infection to explain the genesis of lung injury, and failure of hepatic MPS removal of portal bacteria or LPS originating from the gastrointestinal tract has been suspected (14). These pathogens could then circulate, become localized in mononuclear phagocytes, and cause injury to distant organs, including the lungs. The studies described here had two purposes. First, we sought to determine to what extent the pulmonary uptake of particles and pathogens wehad observed in the sheep after jugular injection would occur if these agents entered through the
portal vein. Then, hepatic macrophages would have first access to the test substances. Second, we wanted to study whether the site of pathogen uptake determined the site of early tissue injury, or whether macrophage activation at one site can cause release of inflammatory cytokines that mediate distant tissue injury. Therefore, we injected particles directly into a mesenteric vein in sheep so that particles were carried to the portal vein and through hepatic sinusoids past Kupffer cells. This is in contrast to our previous studies, in which particles (Received in original form June 3, 1991 and in revised form January 22, 1992) I From the Respiratory BiologyProgram, Harvard School of Public Health, Boston, Massachusetts. 2 Supported by Grants HL-07118,HL-31029,and HL-0l670 from the National Institutes of Health. 3 Correspondence and requests for reprints should be addressed to Dr. Angeline E. Warner, Respiratory Biology Program, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115.
225
PARTICLE UPTAKE FROM THE PORTAL CIRCULATION IN SHEEP
Fig. 1. Injection site and circulatory pattern. Inert particles (radiolabeled gold colloid and iron oxide), Pseudomonas, or radiolabeled endotoxin were injected into a mesenteric vein and carried through liver sinusoids past Kupffer cells before passing through the systemic circulation. In contrast, particulates injected into the jugular vein would be carried directly to the pulmonarycirculation.
injected via the jugular vein were carried directly to the pulmonary circulation (figure 1). Methods
Animals Twelve mature female sheep (mean body weight, 40 kg) were obtained from a local breeder. They were housed separately and given food and water ad libitum. All were grossly free of disease at the time of postmortem examination. Gold Colloid Radiolabeled gold colloid ('98Au, 20-nm particle diameter) was purchased from Amersham Corp. (Arlington Heights, IL). The dose (0.001 mg/kg body weight in 4 to 5 ml sterile pyrogen-free saline) was equivalent to that used in our previous studies in calves and sheep (7, 8). The '98Au colloid content of blood and tissue samples was determined in a Packard gamma counter (Model 2001; Packard Instrument Co., Downers Grove, IL). The specific activity of the gold colloid was 7 mCi/mg. Magnetic Iron Oxide Magnetic iron oxide (gamrna-Fe.Os) was generated by combustion of iron pentacarbonyl vapors. These particles are O.5-llmdiameter clusters of 0.05- to G.l-um subunits. Their mass can be quantified by magnetic field measurement, and they are readily visible with light and electron microscopy. Sheep
beled with 125lodine was provided by Dr. David Morrison (University of Kansas, Kansas City, KS). Lyophilized LPS was reconstituted with sterile distilled water, diluted to single-dose aliquots in sterile, pyrogen-free saline, and stored at - 70° C. The preparation and iodination of this compound have been described (16). The specific activity was approximately 2 u Ci/ug, Sheep received I ug/kg of LPS (total volume, 0.5 ml); 1251 LPS content of blood and organ samples was determined using a Packard model 2001 gamma counter. Systemic ...----..' Circulotion
were infused with 5 mg/kg 0 f a sonicated suspension of gamma-Fe.O, in 200 ml of sterile pyrogen-free saline over 10min. Magnetic iron content of tissue samples was determined as previously described (15). Briefly, the frozen samples were subjected to a brief pulse (10 us) of a 0.1 Tesla magnetic field. This aligned all the magnetic directions of the particles within the sample and produced a detectable field proportional to the total mass of particles within the tissue. This remanent field was quantitated using a Forster fluxgate magnetometer.
Gram-negative Bacteria Pseudomonas aeruginosa strain P220 (courtesy of Dr. J. Pennington) was grown to log phase overnight in trypticase soy broth. The bacteria were washed twice in sterile, pyrogenfree saline and resuspended to an optical density (620 nm) of 1.2 in saline. Each animal received 5 ml of this bacterial suspension (3 x 10' cfu/kg). The bacterial content was confirmed using a trypticase soy agar pour-plate technique after appropriate tenfold dilutions of duplicate aliquots in sterile, distilled water. Plates were incubated overnight at 37° C, and colonies were counted the following day on a Quebec colony counter. Parallel diluted aliquots of washed Pseudomonas were incubated for 30 min in sheep blood or serum at 37° C to evaluate bactericidal activity in the circulating blood. Endotoxin Escherichia coli strain 0111:B4 LPS radiola-
Particle and Pathogen Protocol A 14-gauge, 5.7-cm jugular catheter was placed in each animaL Sheep were heavily sedated intravenously with diazepam (total dose, 10 mg) and xylazine (total dose, 2 mg) and maintained in sternal recumbancy, breathing spontaneously without intubation. The right flank was infiltrated with I % lidocaine and prepared for sterile paracostallaparotomy. Supplemental sedation was provided throughout the experiment based on vital signs and animal responsiveness. A dorsoventral flank incision was made 10 cm caudal to the costal margin to allow isolation and cannulation of a mesenteric vein draining a loop of small boweL Animals (n = 4) received 0.001 mg/kg of .98Au colloid via the mesenteric venous catheter' followed 30 min later by an infusion of magnetic iron oxide (5 mg/kg in 200 ml of 0.9% saline). We have shown that these two inert particles are rapidly cleared and well tolerated in sheep, calves, goats, and rats (7, 8). A second group of four sheep received 5 ml of a live, washed suspension of P. aeruginosa (mean dose, 3 x 10' cfu/kg) via the mesenteric venous catheter. A third group of four sheep were injected with I ug/kg of [12 5 1]E. coli endotoxin in 0.5 ml of pyrogen-free saline. Duplicate (Pseudomonas experiments) and triplicate ('98Au and 125 1 LPS experiments) l-ml jugular venous blood samples were drawn prior to and at serial intervals for I h after the injection. These blood samples were analyzed in a gamma counter (' 98Au, [USI]LPS) or by quantitative culture (Pseudomonas) in order to determine the kinetics of vascular particle clearance. Circulating half-life for particles and endotoxin was calculated using the injected dose divided by the estimated blood volume (7% total body weight) as the blood concentration at the zero time point. Data from the initial phase of clearance (zero to 5 min) were used to generate a linear regression from which the halflife was calculated. Sixty minutes after the injection of gold colloid, Pseudomonas, or endotoxin, the animals were humanely killed by intravenously administered barbiturate overdose, and the liver, kidneys, heart, spleen, right lung, and samples of bone marrow and skeletal muscle were collected and weighed. Between five and 10 samples of each organ or tissue were assayed for particle uptake. Gamma emission counting of fresh tissue was used to measure
226
DECAMP, WARNER, MOLINA, AND BRAIN
100.00
..
80.00
li
70.00
:::J
>
~
PSEUDOMONAS
Fig. 2. Clearance of gOldcolloid, endotoxin, and Pseudomonas bacteria injected into the portal circulation. Clearance was biphasic, with a rapid and nearly complete initial phase lasting 5 to 10 min and a much slower phase over the remainder of the hour after injection. At 60 min, clearances for the three test particles ranged from 92.8 ± 0.8 to 100% of the injected dose. n = 4.
ENDOTOXIN
60.00
:s
50.00
'0 C
..
40.00
iii Q.
20.00
u
GOLD COLLOID
90.00
30.00
10.00 0.00 0
10
20
30
40
50
60
Minutes Post-injection
19BAu or [12OIjLPS content. Tissue from animals that received magneticiron oxidewas frozen and assayed for quantitative iron uptake in a magnetometer. Samples from animals that received bacteriawerehomogenized in sterile, distilled water, and serial tenfold dilutions of the homogenates wereplated on trypticase soy agar and counted after overnight incubation at 37° C. The mean content per gram for each organ wasmultipliedby the organweightto calculate the total contentin that organ. Organ distribution results were expressedas percent of the total recovered dose for eachmaterial injected. Tissue weightsnot measured were estimated as a percentage of total body weight (bone marrow, 3070; skeletal muscle, 45%; peripheral blood, 7%) as described previously (8). Response to Particle and Pathogen Injection Rectaltemperature,pulse,and respiratoryrate weremeasured in all sheep before and 15,30, 45, and 60 min after particle injection. Total peripheral whiteblood cellcounts weremade on duplicate blood samplesdrawn from each animal at 0,5, 10, 15,30, and 60 min postinjection. Differential counts were made on stained blood smears at the same time points to follow changes in circulating granulocyte and monocyte populations. Preparation of Tissue for Microscopy After one mainstem bronchus was clamped off, the remaining lung was fixed in situ via the trachea with potassium-phosphatebuffered 2.5% glutaraldehyde (pH, 7.4; osmolarity, 340 mOsm/L) at 20 to 25 em H 2 0 pressure.The opposite lung wasremovedand used for quantitation of particleuptake (gamma counting, magnetometricquantitation, or quantitative culture). The fixed lung was excised and immersedin the same fixativeovernight. Random blocks of tissue 5 or 1 mmwerethen taken from all lobes and processed for light and electron microscopy. Cubes of liver tissue 1 em on a side were perfused intraparenchymallywith heparinized saline followed by 2.5% glutaraldehyde and cut into 2- or O.5-mm cubes for processing. Finally, larger samples of both liver and lung were fixedby immersionin 10%buffered formalin.
Tissue processing for light microscopy included dehydration in graded alcohols and embeddingin JB-4 glycol methacrylate(Polysciences, Warrington, PA).Sections 1to 21lm thick were cut with glass knives and stained with methylene blue and basic fuchsin. Formalin-fixed tissue was embedded in paraffin, and 7-llm sections were stained with hematoxylin-eosin. Tissue blocks werepostfixedin 1% osmium tetroxide and stained en b/ocwith 0.5% uranylacetate for electronmicroscopy. After dehydration in graded alcohols, tissue was embedded in epoxy resin (Epox; Polysciences). Sections (80 nm thickness) were cut on a Sorval MT 6000 ultramicrotome (Sorval, Newtown, CT), stained, and examined in a Phillips 300 electron microscope. Results
Vascular Clearance All sheep cleared the test substances rapidly from the bloodstream. Clearance included a rapid initial phase lasting 5 to 10 min and a much slower phase over the balance of the hour after injection. After 10 min, 92.5 ± 0.7 to 99.9 ± 0.01% of gold colloid, bacteria, and LPS were cleared from the peripheral blood. This brought the absolute mean numbers of circulating Pseudomonas from 4.7 X 105 cfu/ml at injection to 4.5 cfu/ml at 10 min and the mean endotoxin concentration from 13.1 ng/ml at injection to 0.98 ng/ml at 10 min. After 60 min, clearances for the three test particles measured varied from 92.8 ± 0.8 to 100% of the injected dose (figure 2). Particle size affected both the speed and efficiency of clearance. Pseudomonas organisms (mean length, 1.5 urn) were initially cleared with a half-life of 0.7 min as compared with 1.2 min for colloidal gold (20-nm particle diameter) or 1.0 min for endotoxin (20 to 120 nm estimated aggregate size) (17). The differences between these clearance half-lives are statistically significant (ANOVA p < 0.01). Moreover, no residual bacteria were
circulating at 60 min whereas less than 1% of the colloidal gold and 8% of the iodinated LPS were still detected. The 8% residual label for [125I]LPS circulating at 60 min probably also reflects endotoxin degradation and/or radionuclide dissociation, as detectable activity was found in bile, urine, and thyroid tissue.
Site of Particulate Uptake Almost all injected particles were found in organs that make up the MPS. Hepatic uptake predominated for gold colloid, iron oxide, and Pseudomonas (figure 3). The degree of hepatic uptake correlated with particle size. The liver content of the larger Pseudomonas organisms represented 95.6 ± 1.7% of the recovered inoculum, hepatic iron oxide content was 81.9 ± 5.8070, whereas liver uptake accounted for only 70.9 ± 5.1% of the recovered dose of the smaller gold colloid. Pulmonary uptake accounted for only 4.3 ± 1.7% of the bacteria, 3.8 ± 1.5% of the iron oxide, and 22.8 ± 3.8% of the gold dose. In contrast to these particles, only 36.7 ± 3.5 % of iodinated LPS was localized in the liver, and the largest fraction of recovered LPS (49.6 ± 4.0%) was in the lungs (figure 3). The total recovery of each of the injected particles was nearly complete, with the exception of live Pseudomonas organisms. We recovered 86.7, 98.4, and essentially 100% of the calculated injected doses of gold colloid, iron oxide, and LPS, respectively. Our net yield for the live bacteria from all sampled tissues accounted for only 17.3% of the injected dose. The low yield may reflect some tissue-associated killing of cleared bacteria during the 60-min protocol and was similar to our results with bacterial injection into sheep via the jugular vein (11). In vitro studies showed no bactericidal effects when live Pseudomonas organisms at the calculated dose per milliliter of circulating blood were incubated in whole blood or serum for 30 min at 37° C. Electron microscopic examination of tissue from all sheep confirmed the uptake of ultrastructurally visible particles (iron oxide and Pseudomonas) within phagosomes of Kupffer cells in the liver (figure 4A and B). Because of the small percentage of pulmonary uptake of these particles after portal venous injection and the small fraction of tissue that can be examined ultrastructurally, we did not identify the cellular location of particles retained in the lungs. We found no evidence for uptake of
227
PARTICLE UPTAKE FROM THE PORTAL CIRCULATION IN SHEEP
particulates by circulating phagocytes. At 60 min after injection we found no radioactivity C9 8 A u, [125I]LPS) or colonyforming ability (Pseudomonas) associated with any of the formed elements of blood.
Physiologic Changes and White Blood Cell Kinetics There were no significant changes in rectal temperature, pulse, or respiratory rate in sheep that received mesenteric venous injections of inert particles (gold colloid, iron oxide). During the 1 h after Pseudomonas infusions, however, sheep demonstrated an average temperature increase of 0.5 0 C, an increase in heart rate of 15010, and an increase in respiratory rate of 24%. Similar findings were seen in the endotoxin group, with a mean rise in rectal temperature of 0.7 0 C, and 15 and 30% increases in pulse and respiratory rate, respectively. The circulating leukocyte response of sheep after injections of Pseudomonas and LPS are shown in figure 5A and B. Both elicited dramatic changes in peripheralleukocyte numbers during the 60 min postinjection, whereasthe administration of gold colloid and iron oxide caused only a 5.8% decrease in total white cell counts. Both bacteria and LPS were largely cleared from the bloodstream before appreciable declines in neutrophil numbers, further suggesting that neutrophils were not primarily involved in particle clearance. Whereas LPS caused a slow, steady decline in total circulating leukocytes, especially neutrophils, Pseudomonas infusion resulted in a slight neutrophilia for 5 to 15 min, which was not statistically significant. This neutrophilia was accompanied by a concomitant decrease in circulating lymphocyte numbers. After this, total leukocyte and neutrophil numbers decreased to below 50% of control values. Morphologic Changes There were no visible pathologic changes in the livers or lungs of sheep that received gold colloid and iron oxide. After portal bacteremia, however, hepatic sinusoids showed accumulations of neutrophils in a periportal distribution (figure 6A). Some Kupffer cellsappeared activated as evidenced by increased erythrophagocytosis. There werelipid accumulations in many periportal hepatocytes, a finding also seen after bacteremia in rats where hepatic uptake predominated (11). Sheep lungs after portal bacteremia showed capillary congestion. However, only occasional neutrophils were
100.00 f/l
Fig. 3. Distribution of recovered particles injected into the portal circulation of sheep. Nearly all injected particles were found in the liver, lungs, blood, and the remaining organs and tissues ('Other includes spleen, bone marrow, kidney, and cardiac and skeletal muscle). Hepatic uptake predominated for gold colloid, iron oxide, and Pseudomonas, but the largest fraction of recovered endotoxin was lou nd in the lungs. n = 4.
CII CJ
L-
Q.
as
70.00 60.00
'C
50.00
CII
40.00 30.00
~
>
o
CJ CII
0:
o~
•
90.00 80.00
Pseudomonas
CJ
Iron Oxide
~
Gold Colloid
D
Endotoxin
20.00 10.00 0.00 Liver
Lungs
Blood
Other
Organ
Fig. 4. Site of particle uptake in the liver. A. One hour after injection of iron oxide. the electron-dense iron particles (arrowhead) were within phagolysosomes of Kupffer cells (K). No particles were found within the sinusoids, endothelial cells, or white blood cells. B. After injection of live Pseudomonas organisms, engulfed bacteria (arrow. head) were found only within Kupffer cell (I
MALCOLM M. DECAMP, ANGELINE E. WARNER, RAMON M. MOLINA, and JOSEPH D. BRAIN
Introduction
Resident macro phages in contact with circulating blood make up an important part of the body's host defenses. These cells of the mononuclear phagocyte system (MPS) playa central role in keeping the blood free of denatured proteins, foreign particles, injured or senescent cells, bacteria, endotoxin, and products of intravascular coagulation (1).The liver and spleen have traditionally been viewed as the primary organs with sinusoidal macrophages responsible for clearance of particulate materials from the bloodstream (2-5). This generalization has important exceptions. We have shown that sheep, calves, goats, and cats have a large population of avidly phagocytic pulmonary intravascular macrophages (PIMs) residing in their lung capillaries (6-8). PIMs have also been described in pigs (9). In these species, the lungs compromise a prominent compartment of the MPS, and the pulmonary uptake of pathogenic particulates (e.g., gram-negative bacteria or endotoxin) by PIMs is associated with evidence of their activation and rapid onset oflocal tissue injury (10-12). We believe that PIMs playa pivotal role in this injury through release of oxygen radicals and inflammatory mediators, and that PIM phagocytic activity may contribute to the considerable pulmonary sensitivity of certain species (especially, e.g., sheep and pigs) to circulating bacteria or endotoxin. These species have become the most popular experimental animal models of the human adult respiratory distress syndrome (ARDS), a syndrome of inflammatory lung injury that often followsgram-negative septicemia or endotoxemia. Experimentally, endotoxin administered for organ uptake studies is usually injected via a peripheral vein. The intestinal tract, however, is a major source of 224
SUMMARY Removal of circulating partiCUlates (bacteria, cell debris, endotOXin) Is accomplished in most species by macrophages resident in the liver and spleen. We have shown that sheep and other species have phagocytic macrophages resident In their pulmonary capillaries. Moreover, these pulmonary Intravascular macrophages accomplish the bulk of uptake of Injected tracer partiCles, bacteria, or endotoxin (LPS). Because bacteria or LPS of Intestinal origin enter the portal circulation, they would first encounter hepatic mononuclear phagocytes. We sought to determine the extent to which particulates Injected Into the portal circulation of sheep would be taken up by liver or by lung macrophages. Sheep (four per group) were injected via a mesenteric vein with radlolabeled gold colloid, magnetic Iron oxide particles, live Pseudomonas aeruginosa, or "51E. coli endotoxin. For each, the uptake pattern was determined 1 h after injection. Lung and liver were also fixed to determine the cells responsible for uptake and SUbsequent Inflammatory changes. Wefound that for circulating gold colloid, Iron oxide particles, or bacteria, hepatic uptake predominated, and Kupffer cells were responsible. After hepatic uptake of bacteria, Inflammatory changes were confined to the liver. In contrast, nearly 50% of endotoxin escaped hepatic clearance and was subsequently removed by the lungs. We then saw Inflammatory changes In both lungs and liver. ThUS, hepatic macrophages are active In species with pulmonary Intravascular macrophages, partially sparing the lungs from uptake and acute Inflammation. Endotoxin, however, may elude hepatic uptake, be sequestered in the lungs, and Initiate Inflammation there. AM REV RESPIR DIS 1992; 146:224-231
endotoxin in vivo. Gut-derived lipopolysaccharide (LPS) can pass to the general circulation via direct absorption into portal blood, by translocation to intestinal lymphatics, or by escape across the intestinal wallinto the peritoneal space (13). Liver phagocytes are thought to protect the heart, lungs, and systemiccirculation from the deleterious effects of such pathogens derived from the intestinallumen. Many patients who develop ARDS have no evident pulmonary focus of infection to explain the genesis of lung injury, and failure of hepatic MPS removal of portal bacteria or LPS originating from the gastrointestinal tract has been suspected (14). These pathogens could then circulate, become localized in mononuclear phagocytes, and cause injury to distant organs, including the lungs. The studies described here had two purposes. First, we sought to determine to what extent the pulmonary uptake of particles and pathogens wehad observed in the sheep after jugular injection would occur if these agents entered through the
portal vein. Then, hepatic macrophages would have first access to the test substances. Second, we wanted to study whether the site of pathogen uptake determined the site of early tissue injury, or whether macrophage activation at one site can cause release of inflammatory cytokines that mediate distant tissue injury. Therefore, we injected particles directly into a mesenteric vein in sheep so that particles were carried to the portal vein and through hepatic sinusoids past Kupffer cells. This is in contrast to our previous studies, in which particles (Received in original form June 3, 1991 and in revised form January 22, 1992) I From the Respiratory BiologyProgram, Harvard School of Public Health, Boston, Massachusetts. 2 Supported by Grants HL-07118,HL-31029,and HL-0l670 from the National Institutes of Health. 3 Correspondence and requests for reprints should be addressed to Dr. Angeline E. Warner, Respiratory Biology Program, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115.
225
PARTICLE UPTAKE FROM THE PORTAL CIRCULATION IN SHEEP
Fig. 1. Injection site and circulatory pattern. Inert particles (radiolabeled gold colloid and iron oxide), Pseudomonas, or radiolabeled endotoxin were injected into a mesenteric vein and carried through liver sinusoids past Kupffer cells before passing through the systemic circulation. In contrast, particulates injected into the jugular vein would be carried directly to the pulmonarycirculation.
injected via the jugular vein were carried directly to the pulmonary circulation (figure 1). Methods
Animals Twelve mature female sheep (mean body weight, 40 kg) were obtained from a local breeder. They were housed separately and given food and water ad libitum. All were grossly free of disease at the time of postmortem examination. Gold Colloid Radiolabeled gold colloid ('98Au, 20-nm particle diameter) was purchased from Amersham Corp. (Arlington Heights, IL). The dose (0.001 mg/kg body weight in 4 to 5 ml sterile pyrogen-free saline) was equivalent to that used in our previous studies in calves and sheep (7, 8). The '98Au colloid content of blood and tissue samples was determined in a Packard gamma counter (Model 2001; Packard Instrument Co., Downers Grove, IL). The specific activity of the gold colloid was 7 mCi/mg. Magnetic Iron Oxide Magnetic iron oxide (gamrna-Fe.Os) was generated by combustion of iron pentacarbonyl vapors. These particles are O.5-llmdiameter clusters of 0.05- to G.l-um subunits. Their mass can be quantified by magnetic field measurement, and they are readily visible with light and electron microscopy. Sheep
beled with 125lodine was provided by Dr. David Morrison (University of Kansas, Kansas City, KS). Lyophilized LPS was reconstituted with sterile distilled water, diluted to single-dose aliquots in sterile, pyrogen-free saline, and stored at - 70° C. The preparation and iodination of this compound have been described (16). The specific activity was approximately 2 u Ci/ug, Sheep received I ug/kg of LPS (total volume, 0.5 ml); 1251 LPS content of blood and organ samples was determined using a Packard model 2001 gamma counter. Systemic ...----..' Circulotion
were infused with 5 mg/kg 0 f a sonicated suspension of gamma-Fe.O, in 200 ml of sterile pyrogen-free saline over 10min. Magnetic iron content of tissue samples was determined as previously described (15). Briefly, the frozen samples were subjected to a brief pulse (10 us) of a 0.1 Tesla magnetic field. This aligned all the magnetic directions of the particles within the sample and produced a detectable field proportional to the total mass of particles within the tissue. This remanent field was quantitated using a Forster fluxgate magnetometer.
Gram-negative Bacteria Pseudomonas aeruginosa strain P220 (courtesy of Dr. J. Pennington) was grown to log phase overnight in trypticase soy broth. The bacteria were washed twice in sterile, pyrogenfree saline and resuspended to an optical density (620 nm) of 1.2 in saline. Each animal received 5 ml of this bacterial suspension (3 x 10' cfu/kg). The bacterial content was confirmed using a trypticase soy agar pour-plate technique after appropriate tenfold dilutions of duplicate aliquots in sterile, distilled water. Plates were incubated overnight at 37° C, and colonies were counted the following day on a Quebec colony counter. Parallel diluted aliquots of washed Pseudomonas were incubated for 30 min in sheep blood or serum at 37° C to evaluate bactericidal activity in the circulating blood. Endotoxin Escherichia coli strain 0111:B4 LPS radiola-
Particle and Pathogen Protocol A 14-gauge, 5.7-cm jugular catheter was placed in each animaL Sheep were heavily sedated intravenously with diazepam (total dose, 10 mg) and xylazine (total dose, 2 mg) and maintained in sternal recumbancy, breathing spontaneously without intubation. The right flank was infiltrated with I % lidocaine and prepared for sterile paracostallaparotomy. Supplemental sedation was provided throughout the experiment based on vital signs and animal responsiveness. A dorsoventral flank incision was made 10 cm caudal to the costal margin to allow isolation and cannulation of a mesenteric vein draining a loop of small boweL Animals (n = 4) received 0.001 mg/kg of .98Au colloid via the mesenteric venous catheter' followed 30 min later by an infusion of magnetic iron oxide (5 mg/kg in 200 ml of 0.9% saline). We have shown that these two inert particles are rapidly cleared and well tolerated in sheep, calves, goats, and rats (7, 8). A second group of four sheep received 5 ml of a live, washed suspension of P. aeruginosa (mean dose, 3 x 10' cfu/kg) via the mesenteric venous catheter. A third group of four sheep were injected with I ug/kg of [12 5 1]E. coli endotoxin in 0.5 ml of pyrogen-free saline. Duplicate (Pseudomonas experiments) and triplicate ('98Au and 125 1 LPS experiments) l-ml jugular venous blood samples were drawn prior to and at serial intervals for I h after the injection. These blood samples were analyzed in a gamma counter (' 98Au, [USI]LPS) or by quantitative culture (Pseudomonas) in order to determine the kinetics of vascular particle clearance. Circulating half-life for particles and endotoxin was calculated using the injected dose divided by the estimated blood volume (7% total body weight) as the blood concentration at the zero time point. Data from the initial phase of clearance (zero to 5 min) were used to generate a linear regression from which the halflife was calculated. Sixty minutes after the injection of gold colloid, Pseudomonas, or endotoxin, the animals were humanely killed by intravenously administered barbiturate overdose, and the liver, kidneys, heart, spleen, right lung, and samples of bone marrow and skeletal muscle were collected and weighed. Between five and 10 samples of each organ or tissue were assayed for particle uptake. Gamma emission counting of fresh tissue was used to measure
226
DECAMP, WARNER, MOLINA, AND BRAIN
100.00
..
80.00
li
70.00
:::J
>
~
PSEUDOMONAS
Fig. 2. Clearance of gOldcolloid, endotoxin, and Pseudomonas bacteria injected into the portal circulation. Clearance was biphasic, with a rapid and nearly complete initial phase lasting 5 to 10 min and a much slower phase over the remainder of the hour after injection. At 60 min, clearances for the three test particles ranged from 92.8 ± 0.8 to 100% of the injected dose. n = 4.
ENDOTOXIN
60.00
:s
50.00
'0 C
..
40.00
iii Q.
20.00
u
GOLD COLLOID
90.00
30.00
10.00 0.00 0
10
20
30
40
50
60
Minutes Post-injection
19BAu or [12OIjLPS content. Tissue from animals that received magneticiron oxidewas frozen and assayed for quantitative iron uptake in a magnetometer. Samples from animals that received bacteriawerehomogenized in sterile, distilled water, and serial tenfold dilutions of the homogenates wereplated on trypticase soy agar and counted after overnight incubation at 37° C. The mean content per gram for each organ wasmultipliedby the organweightto calculate the total contentin that organ. Organ distribution results were expressedas percent of the total recovered dose for eachmaterial injected. Tissue weightsnot measured were estimated as a percentage of total body weight (bone marrow, 3070; skeletal muscle, 45%; peripheral blood, 7%) as described previously (8). Response to Particle and Pathogen Injection Rectaltemperature,pulse,and respiratoryrate weremeasured in all sheep before and 15,30, 45, and 60 min after particle injection. Total peripheral whiteblood cellcounts weremade on duplicate blood samplesdrawn from each animal at 0,5, 10, 15,30, and 60 min postinjection. Differential counts were made on stained blood smears at the same time points to follow changes in circulating granulocyte and monocyte populations. Preparation of Tissue for Microscopy After one mainstem bronchus was clamped off, the remaining lung was fixed in situ via the trachea with potassium-phosphatebuffered 2.5% glutaraldehyde (pH, 7.4; osmolarity, 340 mOsm/L) at 20 to 25 em H 2 0 pressure.The opposite lung wasremovedand used for quantitation of particleuptake (gamma counting, magnetometricquantitation, or quantitative culture). The fixed lung was excised and immersedin the same fixativeovernight. Random blocks of tissue 5 or 1 mmwerethen taken from all lobes and processed for light and electron microscopy. Cubes of liver tissue 1 em on a side were perfused intraparenchymallywith heparinized saline followed by 2.5% glutaraldehyde and cut into 2- or O.5-mm cubes for processing. Finally, larger samples of both liver and lung were fixedby immersionin 10%buffered formalin.
Tissue processing for light microscopy included dehydration in graded alcohols and embeddingin JB-4 glycol methacrylate(Polysciences, Warrington, PA).Sections 1to 21lm thick were cut with glass knives and stained with methylene blue and basic fuchsin. Formalin-fixed tissue was embedded in paraffin, and 7-llm sections were stained with hematoxylin-eosin. Tissue blocks werepostfixedin 1% osmium tetroxide and stained en b/ocwith 0.5% uranylacetate for electronmicroscopy. After dehydration in graded alcohols, tissue was embedded in epoxy resin (Epox; Polysciences). Sections (80 nm thickness) were cut on a Sorval MT 6000 ultramicrotome (Sorval, Newtown, CT), stained, and examined in a Phillips 300 electron microscope. Results
Vascular Clearance All sheep cleared the test substances rapidly from the bloodstream. Clearance included a rapid initial phase lasting 5 to 10 min and a much slower phase over the balance of the hour after injection. After 10 min, 92.5 ± 0.7 to 99.9 ± 0.01% of gold colloid, bacteria, and LPS were cleared from the peripheral blood. This brought the absolute mean numbers of circulating Pseudomonas from 4.7 X 105 cfu/ml at injection to 4.5 cfu/ml at 10 min and the mean endotoxin concentration from 13.1 ng/ml at injection to 0.98 ng/ml at 10 min. After 60 min, clearances for the three test particles measured varied from 92.8 ± 0.8 to 100% of the injected dose (figure 2). Particle size affected both the speed and efficiency of clearance. Pseudomonas organisms (mean length, 1.5 urn) were initially cleared with a half-life of 0.7 min as compared with 1.2 min for colloidal gold (20-nm particle diameter) or 1.0 min for endotoxin (20 to 120 nm estimated aggregate size) (17). The differences between these clearance half-lives are statistically significant (ANOVA p < 0.01). Moreover, no residual bacteria were
circulating at 60 min whereas less than 1% of the colloidal gold and 8% of the iodinated LPS were still detected. The 8% residual label for [125I]LPS circulating at 60 min probably also reflects endotoxin degradation and/or radionuclide dissociation, as detectable activity was found in bile, urine, and thyroid tissue.
Site of Particulate Uptake Almost all injected particles were found in organs that make up the MPS. Hepatic uptake predominated for gold colloid, iron oxide, and Pseudomonas (figure 3). The degree of hepatic uptake correlated with particle size. The liver content of the larger Pseudomonas organisms represented 95.6 ± 1.7% of the recovered inoculum, hepatic iron oxide content was 81.9 ± 5.8070, whereas liver uptake accounted for only 70.9 ± 5.1% of the recovered dose of the smaller gold colloid. Pulmonary uptake accounted for only 4.3 ± 1.7% of the bacteria, 3.8 ± 1.5% of the iron oxide, and 22.8 ± 3.8% of the gold dose. In contrast to these particles, only 36.7 ± 3.5 % of iodinated LPS was localized in the liver, and the largest fraction of recovered LPS (49.6 ± 4.0%) was in the lungs (figure 3). The total recovery of each of the injected particles was nearly complete, with the exception of live Pseudomonas organisms. We recovered 86.7, 98.4, and essentially 100% of the calculated injected doses of gold colloid, iron oxide, and LPS, respectively. Our net yield for the live bacteria from all sampled tissues accounted for only 17.3% of the injected dose. The low yield may reflect some tissue-associated killing of cleared bacteria during the 60-min protocol and was similar to our results with bacterial injection into sheep via the jugular vein (11). In vitro studies showed no bactericidal effects when live Pseudomonas organisms at the calculated dose per milliliter of circulating blood were incubated in whole blood or serum for 30 min at 37° C. Electron microscopic examination of tissue from all sheep confirmed the uptake of ultrastructurally visible particles (iron oxide and Pseudomonas) within phagosomes of Kupffer cells in the liver (figure 4A and B). Because of the small percentage of pulmonary uptake of these particles after portal venous injection and the small fraction of tissue that can be examined ultrastructurally, we did not identify the cellular location of particles retained in the lungs. We found no evidence for uptake of
227
PARTICLE UPTAKE FROM THE PORTAL CIRCULATION IN SHEEP
particulates by circulating phagocytes. At 60 min after injection we found no radioactivity C9 8 A u, [125I]LPS) or colonyforming ability (Pseudomonas) associated with any of the formed elements of blood.
Physiologic Changes and White Blood Cell Kinetics There were no significant changes in rectal temperature, pulse, or respiratory rate in sheep that received mesenteric venous injections of inert particles (gold colloid, iron oxide). During the 1 h after Pseudomonas infusions, however, sheep demonstrated an average temperature increase of 0.5 0 C, an increase in heart rate of 15010, and an increase in respiratory rate of 24%. Similar findings were seen in the endotoxin group, with a mean rise in rectal temperature of 0.7 0 C, and 15 and 30% increases in pulse and respiratory rate, respectively. The circulating leukocyte response of sheep after injections of Pseudomonas and LPS are shown in figure 5A and B. Both elicited dramatic changes in peripheralleukocyte numbers during the 60 min postinjection, whereasthe administration of gold colloid and iron oxide caused only a 5.8% decrease in total white cell counts. Both bacteria and LPS were largely cleared from the bloodstream before appreciable declines in neutrophil numbers, further suggesting that neutrophils were not primarily involved in particle clearance. Whereas LPS caused a slow, steady decline in total circulating leukocytes, especially neutrophils, Pseudomonas infusion resulted in a slight neutrophilia for 5 to 15 min, which was not statistically significant. This neutrophilia was accompanied by a concomitant decrease in circulating lymphocyte numbers. After this, total leukocyte and neutrophil numbers decreased to below 50% of control values. Morphologic Changes There were no visible pathologic changes in the livers or lungs of sheep that received gold colloid and iron oxide. After portal bacteremia, however, hepatic sinusoids showed accumulations of neutrophils in a periportal distribution (figure 6A). Some Kupffer cellsappeared activated as evidenced by increased erythrophagocytosis. There werelipid accumulations in many periportal hepatocytes, a finding also seen after bacteremia in rats where hepatic uptake predominated (11). Sheep lungs after portal bacteremia showed capillary congestion. However, only occasional neutrophils were
100.00 f/l
Fig. 3. Distribution of recovered particles injected into the portal circulation of sheep. Nearly all injected particles were found in the liver, lungs, blood, and the remaining organs and tissues ('Other includes spleen, bone marrow, kidney, and cardiac and skeletal muscle). Hepatic uptake predominated for gold colloid, iron oxide, and Pseudomonas, but the largest fraction of recovered endotoxin was lou nd in the lungs. n = 4.
CII CJ
L-
Q.
as
70.00 60.00
'C
50.00
CII
40.00 30.00
~
>
o
CJ CII
0:
o~
•
90.00 80.00
Pseudomonas
CJ
Iron Oxide
~
Gold Colloid
D
Endotoxin
20.00 10.00 0.00 Liver
Lungs
Blood
Other
Organ
Fig. 4. Site of particle uptake in the liver. A. One hour after injection of iron oxide. the electron-dense iron particles (arrowhead) were within phagolysosomes of Kupffer cells (K). No particles were found within the sinusoids, endothelial cells, or white blood cells. B. After injection of live Pseudomonas organisms, engulfed bacteria (arrow. head) were found only within Kupffer cell (I
