INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

CIRCULATING BLOOD is thought not to clot in healthy individuals for two principal reasons. First, blood is not normally exposed to concentrations of tissue factor (TF) sufficient to sustain blood coagulation. Second, any traces of activated clotting factors that might be generated, except for factor VIIa (24), are rapidly inactivated and cleared as blood flows through the microvasculature. Mechanisms regulating coagulation in interstitial and other slow-moving extravascular fluids are less well understood. TF apoprotein has been demonstrated by immunostaining on many extravascular cell types (11, 14). Nevertheless, extravascular clotting leading to generation of cross-linked fibrin has been reported to require perturbation of microvascular permeability (12). It has been postulated from data taken from human synovial fluid that a very low concentration of the high-molecular-weight coagulation cofactors, factor VIII and factor V, limits coagulation in extravascular fluids (7).

The composition of lymph from a given body region is thought to reflect the composition of its interstitial fluid. Whereas considerable data are available on coagulation factor levels in thoracic duct lymph (3, 4, 8, 15, 27), very few data are available on the levels of hemostatic factors in limb lymph. Fifty years ago, Brinkhous and Walker (5) reported that in dogs the average prothrombin concentration was 51% of plasma concentration in thoracic duct lymph but only 8% of plasma concentration in femoral lymph. This is not surprising, since lymph from the liver, which is the site of synthesis of most hemostatic factors, contributes significantly to thoracic duct lymph (10). Only scanty additional data on hemostatic factors in limb lymph have been reported (6, 18). Therefore, we have measured levels of a number of procoagulant and anticoagulant hemostatic factors in rabbit limb lymph. The data provide a basis for an increased understanding of mechanisms regulating coagulation in extravascular fluids.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Obtaining and handling of lymph samples. Lymph samples were collected from specific pathogen-free New Zealand White rabbits (6 female, 4 male; mean weight 2,294 g). The samples were then processed at the University of Arizona and sent to the University of California, San Diego (UCSD) for measurement of hemostatic factors. Both institutions approved the animal protocol. The rabbits, anesthetized by an intramuscular injection of 5 parts ketamine (100 mg/ml) plus 1 part acepromazine (10 mg/ml), were placed supine on an electric heat barrier, and the abdomen was shaved and exposed by a midline excision. A retroperitoneal lymphatic collector, which was cephalad to the joining of the lower limb lumbar lymphatics but below the cisterna chyli, was identified and isolated by a dissection technique that avoided drainage areas from which the lymph was to be collected. After a small incision was made with a needle, we inserted a polyethylene tube (PE-50, Intramedic). This tubing had been treated with Sigmacote (Sigma) and rinsed with distilled water and then with a buffered citrate anticoagulant (0.06 M sodium citrate, 0.04 M citric acid). A drop of tissue adhesive (Nexaband) was placed at the point of insertion of the cannula to glue it in place and close the incision of the vessel. This technique permitted collection of limb lymph samples free of lymph from gut lymphatic collectors.

Lymph was allowed to drain briefly to flush the tubing and was then collected in 250-µl aliquots into 1.5-ml siliconized, plastic microcentrifuge tubes containing 50 µl of the anticoagulant to yield a final sample volume of 300 µl. The tubes were kept on ice/ice water during the time of collection, which varied from 15 to 60 min. Two to seven aliquots were collected from each rabbit. A 5-µl subsample of each aliquot was removed to measure total protein by a hand-held clinical refractometer (T. S. meter; VWR Scientific). After centrifugation for 7 min at 50 g to remove cells, the samples were frozen for transport.

Blood, obtained from the abdominal vena cava during the time of lymph collection, was added to the citrate anticoagulant in a plastic tube in a 9:1 ratio. Plasma was frozen and sent with the lymph samples to UCSD. After sample collections, the deeply anesthetized rabbits were euthanized.

Materials. Rabbit brain thromboplastin, bovine serum albumin (BSA), and tris(hydroxymethyl)aminomethane (Tris) were from Sigma Chemical (St. Louis, MO). Recombinant human thromboplastin (DADE Innovin), bovine thrombin, and a human fibrinogen reference standard were from Baxter Diagnostics (Deerfield, IL). Recombinant truncated human TF was a gift from Dr. James Morrissey of the Oklahoma Medical Research Foundation. Purified rabbit brain TF apoprotein was reconstituted into phospholipid vesicles as described earlier (33). An activated partial thromboplastin reagent was from Organon Teknika (Durham, NC). Rabbit factor X-deficient plasma, rabbit prothrombin-deficient plasma, and barium-adsorbed ox and rabbit plasma were prepared as described earlier (32, 33). Human factor V immunodepleted plasma was from American Bioproducts (Parsippany, NJ). Human factor Xa was prepared by activating factor X with Russell's viper venom and stored as described elsewhere (2). Recombinant human factor VIIa was from Novo/Nordisk (Copenhagen, Denmark). Chromozyme X and Chromozyme TH were from Boehringer Mannheim (Indianapolis, IN). Immulon 4 enzyme-linked immunosorbent assay (ELISA) plates were from Dynatech (Chantilly, VA), and Vectastain ABC kit was from Vector Laboratories (Burlingame, CA). Peroxidase substrate kit ABTS was from Bio-Rad (Hercules, CA). Biotin-N-hydroxysuccinamide ester was from Pierce (Rockford, IL). Horseradish peroxidase was from Calbiochem (La Jolla, CA).

Antibodies. Goat anti-rabbit antithrombin (AT) immunoglobulin (IgG) and goat anti-rabbit tissue factor pathway inhibitor (TFPI) IgG were prepared as previously described (23). A purified rabbit fibrinogen provided by Dr. Russell Doolittle of the University of California, San Diego, was used to raise goat anti-rabbit fibrinogen antiserum. A monoclonal antibody that reacted with both native and activated molecule of rabbit factor X was prepared as described elsewhere (33). Two monoclonal antibodies to human factor VIII (ESH-4 and ESH-8) that cross-reacted with rabbit factor VIII were obtained from Diagnostica Stago (Parsippany, NJ). A polyclonal IgG against human von Willebrand factor that cross-reacted with rabbit von Willebrand factor was a gift from Dr. Zaverio Ruggeri of The Scripps Research Institute, La Jolla, CA.

Pooled rabbit reference plasma. Two batches of pooled rabbit plasma, assigned a value of 100% activity and antigen, were prepared from five and six healthy young rabbits. They were used to prepare reference curves for all assays except the activity assay for fibrinogen.

One-stage clotting assays. Clotting times were determined at 37°C with an ST4 semiautomated coagulation instrument (Diagnostica Stago). Except for fibrinogen, test samples were diluted in 50 mmol/l Tris · HCl and 150 mmol/l NaCl, pH 7.5 (TBS), containing 1 mg/ml BSA (TBS-BSA).

Fibrinogen was determined by the method of Clauss (9). Test plasma was diluted 1:10 in Veronal buffer (28.4 mmol/l barbital sodium and 0.125 mol/l NaCl, pH 7.35). The two batches of reference plasma yielded fibrinogen values of 225 and 229 mg/dl when measured against the human fibrinogen standard of the assay. Therefore, a value of 227 mg/dl was considered equivalent to 1 U/ml of rabbit fibrinogen activity and antigen.

Prothrombin and factor X were determined in thromboplastin-based systems containing specific deficient substrate plasmas fortified with barium-adsorbed rabbit or bovine plasma to assure independence of test results from the factor V content of a test sample. In both assays, 50 µl of a 1:1 mixture of specific immunodepleted rabbit plasma and barium-adsorbed rabbit plasma, 50 µl of a dilution of test sample, and 50 µl of rabbit thromboplastin were incubated for 3 min before clotting was initiated with 50 µl of 35 mmol/l CaCl₂. Reference curves were prepared from serial dilutions of reference plasma between 1:20 and 1:800. Test samples were diluted 1:10 or 1:20 for lymph and 1:100 and 1:200 for plasma.

We measured factor VII by incubating 50 µl of a 1:1 dilution of human hereditary factor VII-deficient plasma and barium-adsorbed bovine plasma with 50 µl of a dilution of test sample for 3 min before initiating clotting with 100 µl of Innovin. Reference curves for factor VII were prepared from dilutions of reference plasma between 1:5 and 1:60. Test samples were diluted 1:4 for lymph and 1:20 for plasma. Factor VIIa was determined with the use of a recombinant truncated human TF as described previously (31).

Factor V was measured by incubating 50 µl of human immunodepleted factor V-deficient plasma, 50 µl of test sample, and 50 µl of rabbit thromboplastin for 3 min before clotting with 50 µl of 35 mmol/l CaCl₂. Reference curves were prepared from dilutions of reference plasma between 1:50 and 1:1,600. Test samples varied from undiluted to a 1:20 dilution for lymph and were diluted 1:200 and 1:400 for plasma.

Factor VIII was measured by incubating 50 µl of hereditary human factor VIII-deficient plasma, 50 µl of a dilution of test sample, and 50 µl of an activated partial thromboplastin reagent for 5 min before clotting with 50 µl of 35 mmol/l CaCl₂. Reference curves were prepared from dilutions of reference plasma between 1:5 and 1:80. Lymph samples were assayed undiluted and plasma samples were assayed at a 1:20 dilution.

Factor XI was measured as performed for factor VIII except that human hereditary factor XI-deficient plasma was used as the substrate. Lymph samples were assayed at a 1:3 dilution and plasma samples at a 1:10 and 1:20 dilution.

Antithrombin activity, antigen, and antithrombin-Xa complexes. Antithrombin activity, antigen, and antithrombin-Xa complexes were measured as described earlier (23). Lymph samples were diluted 1:50 and 1:100 to measure activity, 1:1,000 and 1:2,000 to measure antigen, and 1:10 to measure antithrombin-Xa complexes. Plasma samples were diluted 1:200 and 1:400 to measure activity, 1:2,500 and 1:5,000 to measure antigen, and 1:10 to measure antithrombin-Xa complexes.

TFPI activity and antigen. TFPI activity was measured in a two-stage capacity assay (30), and TFPI antigen was measured by ELISA (23). Lymph samples were diluted 1:25 and 1:50 to measure activity and 1:10 to measure antigen. Plasma samples were diluted 1:100 and 1:200 to measure activity and 1:25 and 1:50 to measure antigen.

Factor VIII antigen. ELISA plate wells were coated overnight at 4°C with 100 µl of 5 µg/ml of monoclonal antibody ESH-8 in 0.1 mol/l bicarbonate sodium buffer, pH 9.6. Wells were blocked with 3% BSA in TBS for 4 h at 37°C and washed four times with 15 mmol/l phosphate-buffered saline containing 0.05% Tween (PBS-Tween). Then, 50 µl of test sample diluted in PBS-Tween containing 0.3% BSA and 50 µl of 5 µg/ml of horseradish peroxidase-conjugated monoclonal antibody ESH-4 diluted in PBS-Tween containing 0.5 mol/l NaCl were added. After incubation for 3 h at 37°C, the wells were washed five times with PBS-Tween, 100 µl of ABTS peroxidase substrate were added, and the initial rate of color development was recorded continuously with an ELISA reader (as mOD/min). Lymph samples were assayed at a 1:30 or 1:50 dilution and plasma samples were assayed at a 1:300 dilution.

Western blots of TFPI. Two hundred to three hundred microliters of plasma or lymph were passed over a 1-ml Hi-Trap protein G column (Pharmacia Biotech, Piscataway, NJ) that had been equilibrated with 20 mmol/l sodium phosphate, pH 7.4. The pass-through volume was collected in 0.5 ml fractions, and fractions containing protein were pooled to yield samples devoid of rabbit immunoglobulin. Fifty microliters of 5.3 mg/ml goat anti-rabbit TFPI IgG were then added per milliliter of pooled sample and incubated for 2 h at 4°C. Then, 50 µl of protein A agarose (Pierce) per milliliter of sample were added, and the incubation was continued at 4°C for an additional 2 h. The protein A agarose beads were separated by centrifugation and washed twice with TBS. Pellets were then suspended in loading gel buffer containing 5% 2-mercaptoethanol and heated at 90°C for 30 min. The beads were removed by centrifugation. The supernate was subjected to 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transblotted onto a polyvinylidene difluoride membrane at 10 V overnight. The membrane was blocked with 5% milk in TBS containing 0.05% Tween (TBS-Tween) for 2 h at room temperature, washed in TBS-Tween, and incubated for 2 h at room temperature with 3 µg/ml of goat anti-rabbit TFPI IgG diluted in TBS-Tween containing 0.5% milk and 0.6 mol/l of NaCl. After we washed the membrane with TBS-Tween, we immersed the membrane for 30 min at room temperature in reagents A and B (2 drops each in 15 ml of TBS-Tween) of an ABC kit. The membrane was then washed with TBS-Tween, exposed to ECL (Amersham) for 1 min as per the manufacturer's instructions, and developed immediately.

Western blots of fibrinogen. Blots of fibrinogen were obtained by subjecting plasma and lymph samples to 10% reduced SDS-PAGE followed by transblotting as described above. Membranes were blocked with 3% BSA in TBS, washed with TBS-Tween, and incubated with 8 µg/ml of goat anti-rabbit fibrinogen diluted in TBS-Tween containing 0.5% milk. After we washed the membrane with TBS-Tween, the membrane was incubated with reagents A and B of the ABC kit prepared as described above. After we made four additional washings with TBS-Tween, the membrane was developed with 3,3'-diaminobenzidine substrate.

Western blots of von Willebrand factor. Blots of von Willebrand factor were obtained by subjecting plasma and lymph samples to horizontal electrophoresis carried out on a 12 × 26 cm glass plate. The running gel was 1.4% agarose and 0.1% SDS in 375 mmol/l Tris, pH 8.8. The stacking gel was 0.8% agarose and 0.1% SDS in 125 mmol/l Tris, pH 6.8. The electrophoresis buffer was 50 mmol/l Tris, 384 mmol/l glycine, and 0.1% SDS, pH 8.3. Plasma and lymph samples were diluted in sample buffer (10 mmol/l Tris, 1 mmol/l EDTA, 2% SDS, and 8 mol/l urea) and heated for 15-20 min at 60°C. After overnight electrophoresis at 30 V and 15°C, the gel was soaked in buffer (80 mmol/l phosphate and 0.1 m/l NaCl, pH 7.2) and transblotted overnight by paper compression onto a polyvinylidene difluoride membrane. The membrane was blocked with 4% milk in TBS-Tween, incubated with 50 µg/ml of goat anti-human von Willebrand factor IgG diluted in the blocking buffer, washed in TBS-Tween, and incubated with Vector's anti-goat IgG diluted 1:1,000 in blocking buffer. After 2 h the membrane was rinsed, washed in TBS-Tween, incubated with ABC solution, and developed with 3,3'-diaminobenzidine substrate as described in Western blots of fibrinogen.

Statistical analysis. A two-tailed paired t-test was used to assess the significance of paired measurements in individual animals.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Five to seven 300-µl lymph samples were obtained from seven rabbits and two to four 300-µl lymph samples from three rabbits. All samples were clear without gross evidence of red blood cell contamination. Total protein was measured in each sample and did not increase over the course of the collections. The mean value was calculated for each animal. Mean values for the 10 rabbits were the following: total lymph protein concentration, 1.9 g/dl (SD = 0.4 g/dl, range 1.5-2.6 g/dl); total plasma protein concentration, 4.1 g/dl (SD = 0.2 g/dl, range 4.0-4.6 g/dl); lymph-to-plasma total protein ratio, 0.43. This last value was virtually identical to that measured under basal conditions in canine limb lymph (6). Moreover, the mean lymph protein concentration was similar to a reported mean total protein concentration of 2.1 g/dl (SD 0.4 g/dl, n = 9) obtained by others from a lymph vessel from the foot of normal human volunteers (13). Therefore, except for one rabbit with data identified by arrows in Figs. 1, 4, and 7, it is assumed that our data reflect physiological concentrations of hemostatic factors in the interstitial fluids of the rabbit's hindlimb.

More than one lymph sample had to be used to measure all hemostatic factors in a given animal. If a given hemostatic factor was measured in more than one lymph sample from the same animal, values were averaged. Thawed lymph samples from two animals contained fibrin strands or grains. Data from these rabbits are reported separately at the end of RESULTS. Mean data for lymph hemostatic factors from the remaining eight rabbits, expressed as percentage of a pooled rabbit reference plasma and also as lymph-to-plasma ratios, are summarized in Table 1.

Fibrinogen, factor VIII, and factor V. These hemostatic factors have two common properties: 1) molecular mass exceeding 300,000 Da, and 2) sensitivity to proteolysis by only traces of thrombin, which can markedly affect the measurement of their concentration and/or activity. The distributions of the data for these factors are shown in Fig. 1.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Scattergram depicting concentrations in limb lymph of fibrinogen (Fib), factor VIII (FVIII), von Willebrand factor (vWF), and factor V measured as activity, antigen (Ag), or both and expressed as a percentage of the concentration in pooled rabbit reference plasma. Each point represents data from a single animal. Arrows identify data points from a single animal in which high values for multiple factors were obtained (see text). Horizontal lines identify mean values.

Fibrinogen in lymph was measured both as functional activity and as antigen with concordant results. Mean concentrations were the following: for activity, 48 mg/dl (range 23-95 mg/dl) and for antigen, 45 mg/dl (range 16-90 mg/dl). Western blots with anti-rabbit fibrinogen IgG were performed on lymph, on plasma as a negative control, and on plasma treated with exogenous plasmin as a positive control (Fig. 2). Bands corresponding to plasmin-induced fibrinogen degradation products were not seen in the lymph (lane 3, Fig. 2). A doublet with the mobility of fibronectin was present in the antigen used to immunize the goat (lane 4, Fig. 2) and in both the plasma (lanes 1 and 2 of Fig. 2) and the lymph (lane 3, Fig. 2).

View larger version (55K): [in this window] [in a new window]   Fig. 2.   Binding on Western blots of goat anti-rabbit fibrinogen immunoglobulin G (IgG) to rabbit lymph and plasma from a study animal. Lanes are the following: 1, plasmin-treated plasma as a positive control for presence of lower molecular weight fibrin degradation products; 2, untreated plasma; 3, untreated lymph; 4, purified rabbit fibrinogen used as antigen to immunize the goat. Arrowheads denote, from top to bottom, molecular mass standards of 97, 66, 45, 31, and 22 kDa.

Factor VIII in lymph was measured as activity in six rabbits and as antigen in all eight rabbits (Fig. 1). For the six rabbits in which both were measured, the mean activity level of 7% was lower than the mean antigen level of 15%, and the mean difference between paired measurements in individual animals was significant (P < 0.003). In contrast, in five animals in which reliable values for plasma factor VIII activity and antigen were obtained, the mean concentration as activity (87%) and as antigen (82%) were concordant. (In a sixth animal, a plasma factor VIII activity of 315% with an antigen level of 87% indicated in vitro activation, and these data were excluded from the above calculation).

Values for factor V activity measured in lymph and plasma from all eight rabbits were the following: for lymph from 1 to 19% (Fig. 1) and for plasma from 53 to 129% of the reference plasma.

The highest lymph values for fibrinogen, factor V, and factor VIII (denoted by arrows in Fig. 1) were from the same animal. This rabbit's total lymph protein concentration of 2.6 g/dl exceeded the mean total protein concentration in the lymph of the other nine rabbits (1.77 g/dl SD = 0.31 g/dl) by >2 SD.

Von Willebrand factor. A low concentration of von Willebrand factor antigen was measurable by ELISA in the lymph of all eight animals: mean, 5%, and range 2-10% of our reference plasma. It was present on Western blots primarily as lower molecular weight multimers, but higher molecular weight multimers were also seen in several samples (e.g., lanes 1, 4, and 5, Fig. 3).

View larger version (95K): [in this window] [in a new window]   Fig. 3.   Binding on Western blots of a goat anti-human von Willebrand factor IgG to illustrate multimeric forms of von Willebrand factor present in rabbit plasma and lymph. Lanes 1-6 are lymph samples from 6 study animals. Lane 7 is pooled reference rabbit plasma.

Prothrombin, factor VII, factor X, and factor XI. Prothrombin, factor X, and factor VII are vitamin K-dependent proteins with molecular masses between 50,000 and 72,000 Da. Their mean lymph activity levels are given in Table 1, and the distribution of values is shown in the scattergram of Fig. 4. The highest value for factor VII activity, denoted by the arrow in Fig. 4, came from the same animal with the highest values for fibrinogen, factor VIII, and factor V denoted by the arrows in Fig. 1. Lymph prothrombin and factor X activities were not measured in this animal.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Scattergram depicting concentrations in limb lymph of prothrombin (FII), factor VII (FVII), factor X (FX), and factor XI (FXI) activity expressed as a percentage of concentration in pooled rabbit reference plasma. Symbols and lines are as described in the legend of Fig. 1.

For the remaining seven rabbits, the mean level of factor VII in lymph was lower than the mean level of prothrombin or factor X. The P values for mean differences among activity levels of prothrombin, factor X, and factor VII were the following: for prothrombin versus factor X, P = 0.622; for prothrombin versus factor VII, P = 0.002; for factor X versus factor VII, P = 0.004.

Factor VIIa levels were measured in lymph samples from three rabbits and were between 0.03 and 0.06 ng/ml. These values approached the limits of detection of the assay. Plasma factor VIIa levels in these rabbits were between 0.12 and 0.18 ng/ml.

Factor XI activity was measured in lymph from four animals (Table 1, Fig. 4).

Antithrombin. Antithrombin was measured as both activity and antigen in lymph and plasma from the eight rabbits (Table 1). The mean antithrombin activity in the lymph was 32% (range 17-49%), and the mean antithrombin antigen was 44% (range 24-75%) of our pooled reference plasma. In the lymph, the difference in individual rabbits between antithrombin measured as activity and antigen was significant, P = 0.04. In the plasma, the difference in individual rabbit between the two measurements was not significant, P = 0.20.

Antithrombin-factor Xa complexes were also measured in these eight animals and expressed as percentage of the maximum complexes formed when pooled rabbit reference plasma was clotted in vitro with Russell's viper venom in the presence of heparin (23). In the lymph, the mean value for these complexes was 0.84% (range 0.38-1.25%). In the plasma, the mean value for these complexes was 0.40% (range 0.29-0.48%). The P value for the mean of the differences for the concentration of complexes in the lymph and in the plasma of the individual rabbits was significant, P = 0.008.

Tissue factor pathway inhibitor. TFPI was measured as activity and as antigen in lymph and plasma in all eight animals. In the plasma, the values for activity and antigen were concordant (Table 1, Fig. 5). In contrast, in lymph the mean TFPI activity (37%) differed from the mean TFPI antigen level (80%), and the difference between paired activity and antigen levels in individual animals was highly significant, P < 0.0001. Western blots of lymph samples from two rabbits yielded no evidence of TFPI degradation products (Fig. 6).

View larger version (14K): [in this window] [in a new window]   Fig. 5.   Scattergram depicting concentrations in limb lymph and in plasma of tissue factor pathway inhibitor (TFPI) activity (Act) and antigen (Ag) expressed as a percentage of concentration in pooled reference plasma. Symbols and lines are as described in legend of Fig. 1.

View larger version (84K): [in this window] [in a new window]   Fig. 6.   Binding on Western blots of goat anti-rabbit TFPI IgG to purified rabbit plasma TFPI (lane 1), concentrated pooled rabbit reference plasma (lane 2) and concentrated lymph from two study rabbits (lanes 3 and 4). Arrowheads denote, from top to bottom, molecular mass standards of 97, 66, 45, and 31 kDa.

Lymph-to-plasma concentration ratios. In addition to the above data presented as percentage of our pooled rabbit reference plasma, a lymph-to-plasma ratio was calculated for each hemostatic factor in each animal. The ratios obtained in individual animals were then averaged to obtain a mean lymph-to-plasma ratio for each factor (Table 1). The distribution of the individual ratios is shown in Fig. 7.

View larger version (14K): [in this window] [in a new window]   Fig. 7.   Scattergram depicting hemostatic ratios in individual animals of concentrations in lymph to concentrations in plasma determined as percentage of concentration in pooled rabbit reference plasma. Each data point represents a ratio from a single animal and horizontal lines denote mean lymph-to-plasma ratio for all animals.

Hemostatic factors in lymph samples containing fibrin. As mentioned earlier, thawed lymph samples from two animals contained fibrin grains or strands. After removal of the fibrin by centrifugation, fibrinogen levels in the samples were 6 and 7 mg/dl measured as activity and 14 and 8 mg/dl measured as antigen. Factor VIII activity was 5% in one animal and below detection in the other. Factor V activity was 4% in both animals. Prothrombin, factor VII, factor X, antithrombin, and TFPI activities were within the ranges observed in the lymph samples from the eight animals without evidence of visible fibrin in thawed samples.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In this study we provide substantial data, not previously available, on hemostatic factor levels in peripheral limb lymph. The lymph had passed through popliteal, femoral, and parailiac lymph nodes. However, under our experimental conditions in which pathogen-free rabbits had been housed under a controlled environment, the passage of lymph through the microvasculature of these unstimulated modes is thought to have affected only minimally the macromolecular composition of the lymph. The observation that mean total lymph protein content in our samples was similar to values reported in canine limb lymph (6) and in lymph collected from the foot of normal human volunteers (13) indirectly supports this assumption. Therefore, we believe that the values for lymph hemostatic factors reported here reflect the composition of these factors in the interstitial fluid of rabbit hindlimb tissues.

Fibrinogen, factor V, and factor VIII are large proteins with molecular masses exceeding 300,000 Da. Olszewski and colleagues (19) have reported that concentrations of immunoglobulins and complement components in human limb lymph are inversely proportional to their molecular masses. Therefore, we were surprised to find a mean lymph fibrinogen level of almost 30% of the mean plasma level, which was substantially higher than the mean lymph factor V and factor VIII levels of <10% of their mean plasma levels (Table 1). This leads us to suspect that a mechanism other than capillary sieving, such as receptor-mediated transcytosis (25), is involved in the transport of fibrinogen across vascular endothelial cells.

Very low residual values for both fibrinogen activity and antigen were found in lymph samples from two rabbits in which fibrin was seen and removed after the samples were thawed. This finding plus the concordant mean values for fibrinogen activity (48 mg/dl) and antigen (45 mg/dl) of the other eight rabbits provide convincing indirect evidence that little if any nonclottable fibrinogen/fibrin was present in the limb lymph. Moreover, fibrin/fibrinogen degradation products were not detectable on Western blots of the lymph (Fig. 2). Our data are consistent with the report that only negligible amounts of extravascular cross-linked fibrin could be demonstrated in normal guinea pig tissues unless vascular permeability was perturbed (12). Therefore, we conclude that fibrin does not form under normal physiological conditions in either limb lymph or limb interstitial fluid, despite a substantial concentration of fibrinogen in these very slow moving fluids.

Von Willebrand antigen was present in a very low concentration in limb lymph, primarily as low molecular weight multimers (Figs. 1 and 3). However, higher molecular weight multimers were also visible on Western blots of some samples (Fig. 3). Von Willebrand factor secreted abluminally by vascular endothelium should have remained bound to extravascular matrix proteins (26, 28). Therefore, since von Willebrand factor has been demonstrated by immunostaining in lymphatic endothelial cells (20), we believe that the von Willebrand factor we measured in lymph was released from lymphatic endothelial cells.

Limb lymph contained substantial concentrations of two major protease inhibitors of coagulation, antithrombin and TFPI. Antithrombin activity was present at a mean level of ~40% of mean plasma activity, which is consistent with earlier measurements of antithrombin in canine limb lymph (6) and in human synovial fluid (7). TFPI activity was also present at ~40% of mean plasma activity. Measurements of plasma TFPI activity and antigen were concordant. In contrast, mean lymph TFPI antigen in lymph was approximately twice the mean lymph TFPI activity (Table 1 and Fig. 5). However, on Western blotting (Fig. 6), only a single band of TFPI was visible in lymph with the same mobility as plasma TFPI and purified rabbit TFPI (29). We cannot presently account for the difference between the measurements of TFPI and antigen in our lymph samples.

The activity of factor VII in limb lymph was lower than the activity of factor X and of prothrombin despite the similar molecular weights and other properties of these three proteins. It is possible that its lower lymph concentration reflects the binding of factor VII to TF apoprotein constituitively expressed on the surface membrane of extravascular cells (11, 14). Such TF-bound factor VII could be internalized or activated to factor VIIa-TF complexes with subsequent neutralization of factor VIIa activity by TFPI or antithrombin (21). The finding of low but measurable amounts of factor VIIa in limb lymph is consistent with this possible sequence of events.

Although both plasma and lymph contained only very low concentrations of antithrombin-factor Xa complexes, their concentration in lymph exceeded their concentration in plasma. This finding is compatible with a limited, basal factor VIIa-TF-catalyzed activation of factor X in interstitial fluid, lymph, or both.

Despite our precautions to prevent clotting during the collection of the lymph, fibrin was present in thawed lymph samples from two rabbits. The catalytic activity of traces of thrombin presumably formed during or after sample collection persisted despite levels of antithrombin activity in these samples comparable to activity present in the lymph samples of the other eight animals. This suggests to us that anticoagulantly active glycosaminoglycans on lymph endothelium that could markedly potentiate antithrombin activity may play an important role in preventing limb lymph from clotting in vivo.

We conclude that under physiological conditions at least four mechanisms are involved in preventing an extravascular VIIa-TF activation of factor X from progressing to the generation of fibrin in the interstitial fluid and lymph of peripheral tissues. The first is the unavailability of cell surface anionic phospholipids to support more than minimal VIIa-TF catalytic activity (1, 16, 17). The second is the inactivation of VIIa-TF catalytic activity by TFPI-factor Xa complexes (21). The third is the very low concentration of the cofactors, factor VIII and factor V. The fourth is the substantial level of antithrombin and the presumed presence of anticoagulantly active glycosaminoglycans both in extravascular matrix and on lymph endothelium needed for antithrombin to suppress factor Xa-catalyzed back activation of factor VII bound to TF (22), to inhibit and dissociate from TF any factor VIIa molecules that did form (21), and to neutralize rapidly traces of thrombin that might be generated.

We recognize that other factors not evaluated here, e.g., heparin cofactor II and protease nexins released from extravascular cells, could also contribute to the regulation of extravascular coagulation. Nevertheless, we believe that the data reported here add substantially to the previously very limited information (5, 6, 13) on the levels of hemostatic factors in peripheral lymph and their relationship to the regulation of extravascular coagulation.