Three different sampling techniques were compared in untreated control mice: tail cut, heart puncture and Vena cava puncture. Tables 1 and 2 summarize the results.

Table 1 shows a comparison of results from measurements of PT and aPTT in samples obtained either by Vena cava puncture or by tail cut. Citrate dilutions of 1:5 and 1:10 were used as indicated. None of the samples obtained by tail cut were evaluable because they coagulated prior to the test. As expected, the clotting times for 1:5 and 1:10 citrate dilutions differed slightly. The samples diluted with citrate at a 1:5 citrate:blood ratio showed a high standard deviation, and some samples were not evaluable at all. This was most likely due to a relative lack of Ca⁺⁺ ions. Adjusting the Ca++ ion concentration might have circumvented this problem. However, due to the high sample volume required for PT and PTT measurements, plasma samples could not be analyzed simultaneously for PT/PTT and for the panel of SF, ATIII, TAT, WBC and PLT. Therefore it was not necessary to use the 1:5 citrate:blood ratio, when analyzing PT and PTT. Similarly, when we analyzed samples for thrombin-antithrombin complexes or soluble fibrin, samples obtained by tail cut were either not evaluable or showed systemic coagulation activation (data not shown). In conclusion, tail cut seems to be an unsuitable blood sampling method for coagulation parameter analysis.

In Table 2, a comparison of results from soluble fibrin (SF) determinations in Vena cava puncture samples and heart puncture samples at different citrate:blood ratios are shown. Only samples obtained by Vena cava puncture at a 1:5 citrate:blood ratio were negative for SF, as expected in untreated mice. Since assays for TAT, WBC and PLT could also be successfully performed using this sampling technique (data not shown), Vena cava puncture at a citrate:blood dilution factor of 1:5 was our preferred sampling method (with the possible exception of PT and aPTT).

First we induced SF in vitro by treatment of murine or human plasma with thrombin, and verified that SF formation had taken place using Western Blotting (Figure 1A).

Two functional assays developed to measure SF in human samples were then adapted for the use in mice:

The erythrocyte agglutination assay is an easily performed test kit, which measures SF in a semiquantitative fashion. When we carried it out according to the manufacturers instructions, but using murine samples, we observed agglutination reactions in our negative control samples (normal mouse plasma). This was most likely due to an unspecific bridging factor. However, after varying assay conditions (dilution of plasma, reagent volumes and reagent conditions) we were able to eliminate the unspecific agglutination. Using the optimized protocol as described in the methods part, the positive controls, i.e. thrombin-treated samples, which had been shown by Western Blotting to contain SF, showed an agglutination reaction, which was not observed in negative control samples (untreated normal mouse plasma). This is demonstrated in Figure 1B. We then treated plasma samples with various doses of thrombin, resulting in corresponding amounts of SF. Figure 1C illustrates that the agglutination reaction was visibly stronger in the presence of higher amounts SF, yielding a semiquantitative evaluation.

The chromogenic assay is more difficult and time consuming. It originally was developed for automated measurements, but we established a manual procedure which measured 12 samples per test. Positive controls for the standard curve and negative controls were generated as described in the Methods section. The standard curve was linear for normal mouse plasma standards treated with 15 mU/ml to 50 mU/ml thrombin. Thus, within this range, sample values could be quantitated in arbitrary units corresponding to the amount of activating thrombin (Figure 1D). A direct conversion into μg/ml SF was not possible, because human plasma reacted differently than mouse plasma (data not shown), and therefore the human standards provided in the kit could not be used in the murine setting.

The detection limits were comparable for the agglutination assay and the chromogenic assay (Figure 1C, D). Positive controls of the chromogenic assay test kit were also positive in the agglutination test. When both tests were performed with the same plasma samples, their results correlated with each other and with a third test, the TAT assay (Table 3): In mice injected with 20 μg LPS i.p., 4/5 mice became positive in all three assays. When 300 U of hirudin was injected in addition to 20 μg LPS, all but one mouse in the Berichrom FM test, were negative for SF or TAT in all three assays. Thus we confirmed the thrombin specificity of the signals observed in the three assays.

SF was also measured in another model of systemic coagulation activation: 2 U of thrombin were injected i.v. into mice, and 7 out of 9 mice showed positive test results. The remaining two mice probably failed to develop a systemic coagulation activation because thrombin initiated a strong local coagulation reaction in the tail vein, consuming the vast majority of thrombin at the injection site.

Comparisons between untreated mice and mice treated with 0.9% NaCl-solution (clinical grade) revealed that i.v. injections can induce the formation of SF. Figure 2 demonstrates that SF became detectable in 32 % of plasma samples from mice after a slow i.v. injection of 0.4 ml physiological NaCl-solution in comparison with 0% in untreated mice. LPS-treated mice served as a positive control. This demonstrates that i.v. injection of an inert solution unrelated to the coagulation system can result in systemic coagulation activation, likely by mechanically damaging the venous wall. The agglutination test for SF was sensitive enough to detect this reaction.

As illustrated in Figure 3, all test parameters changed over time, after injection of LPS. SF became positive at early time points, TAT and free ATIII at later time points (6 h and 24 h after LPS injection). The counts for WBC and PLT showed a pattern, which was previously described for humans 28, i.e. WBC counts fell initially, followed by an increase above normal, and PLT counts were reduced. PT rose after 24 hours, but not in all mice, therefore standard deviations were high. aPTT was elevated as early as 2 h post injection and continued to rise. Schistocyte counts changed only slightly (from 3 to 6 fragments per field after LPS injection and from 4 to 9 fragments per field after tumor necrosis factor-alpha injection, data not shown). In conclusion, all parameters tested here are suited for detecting a systemic coagulation activation, with schistocyte counts being the least sensitive.

We measured each of the laboratory parameters in several subcutaneously growing tumor models and compared the values to those of non-tumor bearing mice (Figure 4). In the models tested, we did not find any significant changes in laboratory parameters compared to control mice, except for WBC and PLT. WBC counts were elevated in the lymphoma model and PLT counts both in the myeloma and the lymphoma model.

This argues against a procoagulant status in these models, but would have to be evaluated separately for other tumor models.

The selection of tests was based on the following criteria: Sensitivity of the test, suitability for measuring systemic coagulation activation, sample volume required and the feasability for testing a large number of samples. The required sample volume is even more critical in studies with diseased mice, because in our experience the volume of blood that can be obtained from sick mice is lower than that from healthy mice.

In consequence, the measurement of PT and aPTT was not included in the panel because i) a total of at least 100 μl plasma (about 200 μl of blood) is needed for these two tests; ii) it is well accepted that PT and aPTT are global markers and not specific for systemic coagulation activation 40; and iii) a comparison of aPTT results between different laboratories is difficult 41.

Schistocyte counts are very labour intensive, somewhat subjective, and did not prove to be very sensitive. Therefore they were not included in the panel.

In contrast, SF and TAT indicate a systemic coagulation activation and the required sample volumes are low (5 μl and 20 μl of plasma, respectively). ATIII measures a different parameter (consumption) with a different kinetics and also requires a low sample volume (5 μl of plasma). WBC and PLT, which can both be performed with a total sample volume of 10 μl blood, have a known pattern of change in DIC.

In conclusion, we propose SF, TAT, ATIII, WBC and PLT as a useful panel of tests for systemic coagulation activation.

We then tested the impact of a VTA, which selectively induces coagulation activation in tumor vasculature 5, on coagulation parameters at different time points after injection into tumor bearing mice. The results are illustrated in Figure 5 and demonstrate that there was no indication of systemic cogulation activation after treatment with that particular VTA.

Three different techniques were used: Vena cava puncture, heart puncture and tail cutting. If not otherwise specified, blood was drawn from the Vena cava.

For Vena cava or heart puncture, mice were anaesthetized with ether, the thoracical cavity was opened, and blood was drawn from the upper Vena cava or from the heart into a syringe containing sodium citrate (1/10 or 1/5 volume of 3.1% sodium citrate). For tail cutting, mice were warmed up under an infrared lamp (250 W) for one minute. The tail was cut with a scalpel, the first three drops of blood were not used, and then blood was collected into a heparin-coated centrifuge tube, which contained sodium citrate (see above). After removing blood for blood cell counts (WBC, PLT, schistocytes), citrated blood was centrifuged twice (500 g and 8000 g) to remove platelets. Platelet-poor plasma was analyzed immediately for soluble fibrin and the remainder of the samples were frozen at -80°C until further use.

Negative control mice were either untreated mice or mice injected with 0.9% NaCl-solution (clinical grade). In some experiments pooled plasma was used, which was prepared as follows: Blood from 20 mice was drawn by Vena cava puncture and platelet-poor plasma was obtained as described above. In rare cases the sampling technique could produce outliers. In order to exclude these artifacts from our negative control pool, each sample was tested for soluble fibrin and for thrombin-antithrombin complexes (method see below). The vast majority of samples were negative in these tests, and those samples were pooled. The pool was frozen in aliquots at -80°C.

As a positive control, we incubated either purified human fibrinogen or human and murine plasma samples with 0.02 mU/ml bovine thrombin (Dade Behring, Liederbach, Germany). To prove the formation of crosslinked fibrin and fibrin-dimers, Western blotting of digested plasma or fibrinogen samples was performed. Samples were separated by SDS-PAGE (sodium dodecyl-sulfate polyacrylamide-gel electrophoresis) followed by capillary heat transfer onto nitrocellulose membranes. Membranes were stained with a polyclonal sheep-anti-rabbit fibrin(ogen) antibody (Haemochrom, Essen, Germany), known to be crossreactive with mouse and human fibrin(ogen), followed by chemiluminescent detection.

SF in plasma was then measured with two functional assays: i) a modification of a commercially available agglutination assay (FM-test, Roche-Diagnostics, Mannheim, Germany) which measures the agglutination reaction of SF with fibrin monomer-coated erythrocytes and ii) a modification of a chromogenic assay (Berichrom FM assay, Dade Behring), which is based on the ability of SF to enhance the tissue plasminogen activator (tPA)-induced plasminogen conversion. For the modified agglutination test, 5 μl of citrated mouse plasma were diluted in 100 μl of Owren's buffer (28 mM sodium barbital in 125 mM sodium chloride solution, pH = 7.35), and mixed on a slide (provided in the kit) with 50 μl of the erythrocyte suspension. After incubation at 37°C for 10 minutes, samples were mixed again and the agglutination reaction compared with positive and negative control samples. Positive controls were mouse plasma samples treated with 0.02 U/ml thrombin (Dade Behring), negative controls were untreated normal mouse plasma samples. For the chromogenic assay, 250 μl of human mini-plasminogen solution (provided in the kit) were pipetted into the wells of a 96 well microtiter plate, and the plate was preheated to 37°C. A standard curve was generated by incubating murine plasma samples with different amounts (2 mU/ml – 100 mU/ml) of bovine thrombin. 5μl samples and controls were added to the plasminogen-solution, and 50 μl tPA reagent was pipetted to each well with a multichannel pipettor. Samples were incubated at 37°C for 5 minutes, then 25 μl of plasmin substrate was added with a multichannel pipettor and absorption at 405 nm (minus 650 nm) was measured kinetically in an ELISA reader. The tests were also carried out with thrombin-treated or untreated human plasma.

To verify that the test results were due to the action of thrombin, we added the specific thrombin inhibitor hirudin. Hirudin (Roche, Mannheim, Germany) was either mixed in vitro at a concentration of 1.4 U/ml with the positive control samples, or injected in addition to LPS in the in vivo studies (see below).

For platelet and white blood cell counts, citrated blood was mixed with 1% (w/v) ammonium-oxalate (for platelet counts) or with 3% (v/v) acetic acid (for white blood cell counts) at a dilution of 1:20, and incubated on a shaker for 10 min. Platelets and white blood cells were then counted under a light microscope at 400 × magnification in a counting chamber (type Neubauer Improved). For schistocyte quantification, thin blood smears were prepared from full blood. Smears were dried and stained with the Pappenheim Method: Fixation in 100% (v/v) methanol, followed by 5 minutes staining in May-Grünwald solution (Sigma-Aldrich, St. Louis, MO) and 30 minutes in Giemsa solution (Sigma-Aldrich). Schistocytes were counted in three high power fields at 400 × magnification on a light microscope. Prothrombin time (PT) and activated partial thromboplastin time (aPTT) were automatically measured on a BCS coagulation analyzer (Dade Behring) using the Innovin kit (Dade Behring) for PT, the Actin FS kit (Dade Behring) for aPTT and normal human plasma (SHP, Dade Behring) as a standard. Thrombin-antithrombin complexes (TAT complexes) were measured using an ELISA test (Enzygnost-TAT, Dade Behring) according to the manufacturer's instructions, but using 20 μl of plasma samples and standards. This test is fully crossreactive with murine samples 13. Briefly, microtiterplates which were coated with an anti-thrombin antibody were incubated with samples and standards (provided in the kit), washed, incubated with a peroxidase-conjugated anti-Antithrombin III antibody, washed and detected with OPD (o-phenylene-diamine-dihydrochloride) substrate. Absorption was measured on an ELISA reader. Normal platelet poor mouse plasma was used as a negative control. Antithrombin III (ATIII) was measured using a functional antithrombin test (Coamatic-Antithrombin, Chromogenix, Milano, Italy) according to the manufacturer's instructions, using 5 μl plasma and standards per test at the recommended dilutions. Briefly, plasma samples and standards (provided in the kit) were incubated with an excess of coagulation factor Xa in the presence of heparin. The residual factor Xa was quantified with a chromogenic substrate specific for coagulation factor Xa and the decrease of absorption was measured on an ELISA reader. Normal platelet poor mouse plasma was used as a standard to define 100% activity.

Mice were injected either i.v. with 2 U bovine thrombin (Dade Behring) or i.p. with 20 μg or 50 μg lipopolysaccharides (LPS from E.coli serotype 055:B5, Sigma-Aldrich). For schistocyte counts, mice were injected with 20 μg LPS i.p. or with 100 μg tumor necrosis factor-alpha i.v. After specified time periods, citrated blood was drawn and analyzed for coagulation parameters as described above. In the cases where thrombin action was to be inhibited, 300 U hirudin were injected i.v. prior to 20 μg LPS.

Immunodeficient mice were injected s.c. with 1 × 10⁷ tumor cells. Tumor cell lines used were: F9 murine teratocarcinoma cells (ECACC, Salisbury, UK), Colo 677 human myeloma cells (DSMZ, Braunschweig, Germany), L540rec human Hodgkin's lymphoma cells 43, HT29 human colon carcinoma cells (ATCC/LGC, Middlesex, UK), LS174T human colon carcinoma cells (ATCC), IMR5 human neuroblastoma cells (kindly provided by Dr. Unwicker, University of Marburg, Germany). Mice were analyzed for coagulation parameters when tumors had reached a size of 150–500 μl. Differences between groups were analyzed for statistical significance with the Mann-Whitney U test for unpaired groups. The term "significant" was used, where p-values were less than 0.05.

Mice bearing Colo677 tumors were treated with a VTA developed in our laboratory. It consisted of a single chain variable fragment antibody (scFv) directed against murine vascular cellular adhesion molecule-1 (VCAM-1), which was expressed on tumor endothelial cells, and the extracellular domain of tissue factor. This fusion protein induces selective coagulation activation in the tumor vasculature 5. One single dose of 20 μg was administered i.v. Mice were sacrificed at specified timepoints after treatment, and analysis of coagulation parameters was performed as described above.