Protein Z (PZ), a vitamin K-dependent plasma protein, dramatically enhances inhibition of coagulation factor Xa by the PZ-dependent protease inhibitor (ZPI) serpinA10 [1]. ZPI also directly inhibits FXIa [2]. The fact that PZ and ZPI knockout mice show enhanced responses in models of induced thrombosis supports a physiologically relevant role for the PZ–ZPI system in the regulation of coagulation [3,4].

The broad range of plasma PZ levels has led to the suggestion that the inflammatory response might affect PZ expression [5]. Potentially consistent with this proposition, several studies in which plasma samples were obtained near the time of stroke reported high levels of PZ, whereas others that used plasma samples obtained during convalescence found the opposite. Two studies investigating the association between inflammation and PZ levels, however, have produced conflicting results [6,7]. Here, murine models of the acute-phase response and the antiphospholipid syndrome (APS) are used to better define the relationship between PZ and ZPI levels and inflammation.

Subcutaneous injection of turpentine with the production of an aseptic abscess is a model of the acute-phase response induced by local inflammation [8]. As previously reported for wild-type mice in this model, PZ knockout mice and ZPI knockout mice injected subcutaneously with turpentine responded with significant weight loss (Fig. 1A), a dramatic increase in serum amyloid A (SAA) level (Fig. 1B), a drop in albumin level, and an increase in fibrinogen level (data not shown). Plasma ZPI levels significantly increased in response to turpentine in both wild-type mice and PZ knockout mice, with maximal levels occurring around day 2 (Fig. 1C). Plasma PZ levels significantly increased in response to turpentine in wild-type mice, with maximal levels occurring on day 4, but there was no effect of turpentine on PZ levels in ZPI knockout mice (Fig. 1D).


Figure 1. Protein Z (PZ) and PZ-dependent protease inhibitor (ZPI) responses. (A–D) PZ and ZPI proteins in the acute-phase response. Wild-type (C57BL/6J; Jackson Laboratory, Bar Harbor, ME, USA) and PZ and ZPI knockout (KO) mice [4,5], 12–20 weeks of age, were injected with 100 μL of turpentine (or phosphate-buffered saline [PBS] as a control) subcutaneously into the hindquarters; mouse weights were monitored (A). At various times, retro-orbital blood samples were collected into heparinized capillary tubes, and plasma was prepared and analyzed by immunoassays for serum amyloid A (SAA) (B),with the E-90SAA kit (Immunology Consultants Laboratory, Newberg, OR, USA); for ZPI (C), with human factor XIa (Enzyme Research Laboratories, South Bend, IN, USA) as capture reagent, horseradish peroxidase-labeled polyclonal antibody raised to recombinant human ZPI, which cross-reacts with mouse ZPI as the detection reagent, and a standard curve generated with pooled C57BL/6J plasma [18]; and for PZ (D), with the Zymutest Protein Z ELISA kit RK031A (Aniara Corp., Mason, OH, USA), which cross-reacts with mouse PZ, and a standard curve generated with pooled C57BL/6J plasma. (E) PZ and ZPI messages in the acute-phase response. Livers, isolated from C57BL/6J mice injected subcutaneously with 100 μL of turpentine, were collected at the indicated times into RNAlater; RNA was extracted with RNeasy kits, and treated with RNase-free DNase as suggested (Qiagen, Valencia, CA, USA). First-strand DNA synthesis was performed with the SuperScript III First Strand cDNA synthesis kit and random hexamer primers (Invitrogen, Carslbad, CA, USA). Samples were analyzed with TaqMan gene expression assays by use of an Applied Biosystem StepOnePlus real-time PCR system and FAM-probe kits for murine glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Mm99999915.g1), PZ (Mm00482203.m1), ZPI (Mm00522856.m1), and fibrinogen (Mm00513575.m1) (Applied Biosystems, Foster City, CA, USA). PZ, ZPI and fibrinogen messages were normalized to GAPDH message. (A–E) The results in control, PBS-treated mice were similar for all mouse groups; therefore, the data were pooled and are represented as a single group (PBS). Shown for all datasets are means ± standard error (SE), with 5–20 animals per group. (F) Gel chromatography profiles of ZPI and PZ for various mouse plasmas. Heparinized mouse plasmas (0.5 mL), obtained from wild-type, PZ knockout and ZPI knockout mice, as indicated, were passed through two 10 × 300-mm Superdex  200 GL columns (GE Healthcare, Piscataway, NJ, USA), which were connected in series, equilibrated, and run with HS buffer (20 mmol L−1 Hepes, 100 mmol L−1 NaCl, pH 7.4). Half-milliliter fractions were assayed for PZ and ZPI by immunoassay, and quantified with purified recombinant murine PZ or ZPI (see below). (G) PZ and ZPI circulating half-life determinations. PZ knockout and PZ/ZPI double knockout mice were injected sub-orbitally with 50 μL of PBS with 0.1 mg mL−1 bovine serum albumin (BSA) containing 1 μg of [125I]PZ/PCtag recombinant mouse protein (2 × 104 c.p.m. μg–1) labeled with Na125I (M. P. Biomedical, Solon, OH, USA) using Iodogen tubes, and separated from free label with Zeba desalt spin columns (Pierce Biotechnology, Rockford, IL, USA). Blood samples were collected from the non-injected eye at the indicated times post-injection, and 10 μL of each blood sample was counted in an LKB  1275 mini gamma counter. For each animal, the initial time point count was assigned a value of 100, and the counts for each successive time point were normalized relative to this value. Shown are the normalized averages and standard deviations for the five animals in each group. Recombinant mouse PZ with a protein C tag (PCtag) was prepared as follows. Mouse PZ cDNA was cloned into pCDNA4.0 with the 36-base sequence coding for the PCtag epitope (EDQVDPRLIDGK) inserted between the PZ C-terminal codon and the stop codon. This plasmid was transfected into Trex293 cells; stable cells expressing PZ/PCtag were identified by ELISA, and expanded in DMEM with 10% fetal bovine serum, 50 units mL−1 penicillin, 50 μg mL−1 streptomycin, 300 μg mL−1 Zeocin, 5 μg mL−1 blasticitin, 1 μg mL−1 tetracycline, and 10 μg mL−1 vitamin K. When ∼ 80% confluence had been reached, the medium was replaced with serum-free Opti-MEM, which, other than serum, contained the same additives as the DMEM medium. Conditioned medium was harvested and replaced every other day. PZ/PCtag was purified from conditioned medium with an anti-PCtag antibody, as described by the manufacturer (Invitrogen). Purified protein was > 90% pure as judged by SDS-PAGE with Coomassie staining, and was quantified with the Bio-Rad protein assay (Bio-Rad, Hercules, CA, USA), with BSA as standard. Recombinant mouse ZPI was produced and isolated in the same fashion, except that the PCtag was placed at the mature N-terminus of the recombinant ZPI. (H–K) PZ and ZPI responses in a model of APS. F1 offspring (48 males and 39 females) of BXSB/MpJ males crossed with NZW/LacJ females (Jackson Laboratory) were monitored up to 28 weeks of age for survival (H); retro-orbital blood samples were collected into heparinized tubes for platelet counts (I); and plasma was prepared and analyzed with immunoassays for PZ (J) and ZPI (K). Each data analysis included a minimum of 20 samples. Data shown are means ± SE. WT, wild type.

Download figure to PowerPoint

Both PZ and ZPI are expressed in the liver, and RT-PCR performed on liver-derived mRNA showed that the level of ZPI, but not that of PZ, mRNA was increased substantially in response to turpentine; the level of fibrinogen mRNA, as a positive control for the acute-phase response, was also increased in response to turpentine (Fig. 1E). The increase in ZPI message and protein levels in wild-type mice in response to turpentine defines it as an acute-phase response protein. In contrast, the increase in PZ protein level was ZPI-dependent and not related to a change in PZ message, implicating a mechanism other than PZ gene induction. Administration of lipopolysaccharide (LPS) to mice mimics the acute-phase response to infection [9]. Relative to the turpentine model, the LPS (100 μg of intraperitoneal Escherichia coli serotype 0111:B4; Sigma, St Louis, MO, USA) model showed similar, although more transient and less robust, responses in weight loss, and SAA, ZPI and PZ levels (data not shown).

Plasma levels of PZ and ZPI appear to correlate in both humans and mice, and a PZ–ZPI complex has been identified in humans [4,10]. On size-exclusion chromatography of plasma from wild-type mice, all of the ZPI appeared to elute with PZ in a PZ–ZPI complex; ∼ 35% of the PZ coeluted with ZPI, and ∼ 65% eluted as free PZ (Fig. 1F). With recombinant mouse PZ and ZPI as standards in immunoassays, mouse PZ circulates at 15 ± 8 μg mL−1 (mean ± standard deviation [SD], n = 20) with a range from 8 to 22 μg mL−1, whereas mouse ZPI circulates at 5 ± 3 μg mL−1 (mean ± SD, n = 20) with a range from 3 to 9 μg mL−1. Thus, PZ and ZPI circulate as a complex in both human and mouse plasma. In humans, there is excess free ZPI [10], whereas in the mouse there is excess free PZ. Nevertheless, a reduction in PZ levels in either species, as exemplified by warfarin treatment in humans and by murine PZ deficiency, is associated with reduced plasma levels of ZPI, and murine ZPI deficiency is associated with reduced plasma levels of PZ [4,10].

As the increase in PZ level following turpentine administration was dependent on ZPI, and PZ–ZPI complexes circulate in mice, we tested whether PZ–ZPI complex formation affected the circulating half-lives of PZ and ZPI. In preliminary studies, size-exclusion chromatography of plasma taken 30 min after injection of 1 μg of labeled recombinant mouse PZ into a PZ/ZPI double knockout mouse demonstrated that > 90% of label eluted at a size consistent with free PZ. In contrast, plasma taken 30 min after injection of 1 μg of labeled PZ in a PZ knockout mouse demonstrated that > 90% of label eluted at a size consistent with a PZ–ZPI complex. Similarly, 1 μg of labeled recombinant ZPI injected into a wild-type mouse (which naturally contains excess free PZ; Fig. 1A) forms a complex with PZ, as indicated by its size-exclusion chromatography profile (data not shown). Subsequent studies showed PZ half-lives of ∼ 210 min in PZ/ZPI double knockout mice and ∼ 580 min in PZ knockout mice (Fig. 1G). In a similarly designed study evaluating ZPI, the ZPI half-life was ∼ 320 min in PZ knockout mice vs. ∼ 660 min in wild-type mice (with excess circulating PZ; Fig. 1G). Taken together, these results demonstrate that the formation of a PZ–ZPI complex extends the half-life of each protein relative to its free form.

APS is an autoimmune state that is associated with circulating, predominantly β2-glycoprotein I-dependent, antiphospholipid antibodies (anticardiolipin and lupus anticoagulant), thrombocytopenia, thrombosis, and fetal wastage. Reduced levels of PZ have been consistently reported in individuals with antiphospholipid antibodies, and low levels of PZ are associated with the thrombotic complications of APS [11–13]. ZPI antigen levels, however, are not reduced in individuals with APS [13]. Therefore, PZ and ZPI levels were evaluated in a mouse model of APS. Crosses of NZW females with BXSB males produced F1 males that, much more frequently than females, developed thrombocytopenia, vascular thrombosis, and increased mortality (Fig. 1H,I) [14,15]. A drop in plasma PZ protein levels occurred with disease progression (Fig. 1J), but plasma ZPI levels remained unchanged (Fig. 1K). Mean SAA levels did not change significantly over the 28-week course of the mouse experiment (data not shown), which is consistent with the low levels of SAA and the limited inflammatory response reported in humans with primary APS [16,17].

In summary, the murine models show ZPI, but not PZ, to be a typical acute-phase reactant. The increase in murine plasma PZ levels in the acute-phase models was dependent on ZPI and potentially attributable, in part, to prolongation of the PZ half-life when it circulates in complex with ZPI. With regard to the formation of the PZ–ZPI complex in plasma, however, ZPI is limiting in the mouse, but in excess in humans. Therefore, the degree to which an increase in plasma ZPI secondary to an acute-phase response would affect PZ–ZPI complexation and the circulating half-life of PZ in humans is not known. ZPI, of course, could also influence PZ levels through alternative mechanisms, e.g. by affecting PZ synthesis, secretion, proteolysis, or extra-plasma localization.

In contrast to the vigorous acute-phase response induced by subcutaneous turpentine, the NZW × BXSB F1 murine model of APS and human primary APS are associated with a muted inflammatory response, and ZPI levels are not increased. The murine APS model demonstrates an acquired reduction in PZ levels that mirrors that seen in human APS, despite the differing relative proportions of PZ and ZPI in mouse and human plasma. Why the typical correlation between PZ and ZPI plasma levels is not maintained in mouse and human APS is not clear.

Disclosure of Conflict of Interests

  1. Top of page
  2. Disclosure of Conflict of Interests
  3. References

This work was supported by the National Institute of Health (grant HL 60782; G. J. Broze). The other authors state that they have no conflict of interest.


  1. Top of page
  2. Disclosure of Conflict of Interests
  3. References