In the blood plasma of humans and rats, ceruloplasmin is the major copper-binding protein and ferroxidase, accounting for 70% of the copper present in the plasma, with the rest binding primarily to albumin and a macroglobulin. Systematic studies with fresh plasma were carried out to compare what occurs in the mouse. C57BL6 mice had half as much copper and pPD (p-phenylene diamine) oxidase activity as humans and rats, 20–40% as much ferroxidase activity as humans (determined using three different assays) and less inhibition by azide. Plasma from ceruloplasmin knockout mice had no pPD oxidase activity, but retained >50% ferroxidase activity (which was not as affected by azide). Modelling of mouse ceruloplasmin against the known X-ray structure of human ceruloplasmin indicated subtle but potentially significant changes in the pPD- and azide-binding sites. Purification and in-gel assays after native PAGE confirmed that mouse ceruloplasmin had ferroxidase activity but revealed an additional ferroxidase in ceruloplasmin knockout mouse plasma, which is also seen in size-exclusion chromatography. In the wild-type mouse, the ‘ceruloplasmin’ peak contained ∼55% of the total copper, but ceruloplasmin knockout plasma exposed a major additional peak (180 kDa) which co-eluted with ferroxidase activity. Two other ferroxidases (700 and 2000 Da) were also detected in mouse and human plasma. Mammalian blood thus contains copper components and ferroxidases not reported previously.
- azide inhibition
- ferroxidase assay
- p-phenylene diamine (pPD) oxidase activity
Previous studies of human and rat plasma and serum fractionated by size-exclusion chromatography showed that there are at least three copper-binding proteins [1–5]. These are ceruloplasmin, comprising approx. 70% of the total plasma copper; albumin, with approx. 15%; and a third larger component named transcuprein, more recently identified as the macroglobulin α1-inhibitor-3 in rodents and α2-macroglobulin in humans , with approx. 10%. All three of these proteins have multiple physiological functions, including copper transport. Thus albumin is an amino acid store and carrier not only of free fatty acids and bilirubin, but also of copper, which binds to a high-affinity site at the N-terminus . Macroglobulins were first known for their ability to trap and inactivate proteases, but they also bind unrelated proteins (like β1-microglobulin) , as well as zinc and copper [2,6,8]. Ceruloplasmin delivers copper to tissues [2,3,9,10], neutralizes radicals [2,3] and catalyses oxidation reactions involving molecular oxygen and various natural and synthetic amines [including pPD (p-phenylene diamine) and o-dianisidine] [2,11], as well as Fe(II) and apparently also nitrogen oxide . The ferro-oxidase activity of ceruloplasmin was first discovered by Frieden , leading to its designation as ferroxidase I, in contrast with a second larger ferroxidase (ferroxidase II), which was later reported by the same group [14,15]. Loss of ferroxidase I activity (through genetics or copper deprivation) has since then been linked to the accumulation of excess iron in the liver, spleen, pancreas and basal ganglia in humans, pigs and other experimental animals [2,16–18], as well as ceruloplasmin knockout mice [19,20]. The theory has been that iron leaves cells in the form of Fe(II) with the help of a transporter, like ferroportin [21,22], but that it must be oxidized to Fe(III) (at the cell surface) by ceruloplasmin (or hephaestin) [22,23] before it can bind to its blood transport protein, transferrin, for distribution to cells of the bone marrow and elsewhere [2,13]. Thus, in the absence of active ceruloplasmin, the exit of iron from storage cells (or those active in erythrocyte turnover) might be slowed, leading to iron accumulation. In support of this idea, it was shown that infusion of active ceruloplasmin into the circulation of animals or perfused livers of copper-deficient dogs or pigs (where iron had accumulated) resulted in immediate release of iron into the medium [2,16,17,24]. However, it should be noted that, in the absence of active ceruloplasmin, the rate of tissue iron accumulation is very low. Only after several decades in the human  and many months in the rodent  does evidence of iron overload occur. This is despite a large daily flux of iron into and out of hepatocytes and other cells from erythrocyte turnover alone (20–24 mg of iron daily in the human adult). Moreover, transferrin itself has some ferroxidase activity . Thus ceruloplasmin is a relatively minor player in causing iron accumulation, and/or the intracellular ferroxidase hephaestin might substitute for when ceruloplasmin is lacking, and/or ceruloplasmin has another as yet unrecognized role in the process of iron flux out of cells.
In addition to copper binding to ceruloplasmin, albumin and macroglobulins (transcuprein), human and rat studies indicate that trace amounts are associated with a variety of other known copper-containing proteins. These include the enzymes extracellular copper/zinc superoxide dismutase and amine oxidase, as well as metallothionein and histidine-rich glycoprotein , all of which are found in the blood plasma. Compared with humans and rats, the blood plasma of mice has been much less studied. Results published previously indicated that there might be significant differences: the total copper content of adult mouse plasma [26–28] and its pPD and o-dianisidine oxidase activities [26,29] might be half or less than those in humans and rats. None of the reports showed direct comparative measurements of these parameters between samples from mice, rats and humans; they contained values from different mouse strains, and were mostly performed using serum, which could be missing copper contributed by macroglobulin (S. Pizzo, personal communication). Since we were embarking on copper transport studies involving knockout and transgenic mice, we felt it essential to make a systematic study of the copper components and oxidase activities of mouse plasma versus those of the human and rat. The results confirmed the presence of significant differences, particularly in the quantity and activity of ceruloplasmin, and led to the discovery of an additional plasma copper component as well as additional ferroxidases.
Animals and plasma samples
Breeding pairs of ceruloplasmin knockout mice were obtained from Dr Z. Leah Harris (Department of Anesthesiology and Critical Care Medicine, Johns Hopkins University, Baltimore, MD, U.S.A.). Heterozygous ceruloplasmin (Cp+/−) mice were mated to obtain aceruloplasminaemic (Cp−/−) and corresponding wild-type (Cp+/+) offspring. The genetic identities of the offspring were determined by amplification of tail DNA by PCR. Adult male Sprague–Dawley rats were obtained from Simonson Laboratories (Gilroy, CA, U.S.A.). All animal procedures were carried out in accordance with the United States NIH guidelines [Guide for the Care and Use of Laboratory Animals (1985), DHEW Publication no. (NIH) 85–23: Office of Science and Health Reports, DRR/NIH, Bethesda, MD, U.S.A.] and approved by the university institutional animal care and use committee. To prevent suffering, animals were killed by exsanguination from the vena cava which bled into the thoracic cavity upon pneumothorax while under pentobarbital anaesthesia (40–50 mg/kg) and treatment with heparin, as described previously . Plasma was obtained by centrifugation at 4000 g for 10 min and was either used fresh or frozen at −20 °C in aliquots. Samples of normal adult human plasma and serum were obtained from volunteers by venosection at the University Health Center. The procedures for obtaining human blood samples was carried out in accordance with the Declaration of Helsinki (2000) of the World Medical Association, after approval of the informed consent-based protocol by the IRB (institutional review board) of the university.
Plasma samples were analysed for copper concentration by furnace atomic absorption spectroscopy as described previously . Alternatively, we used ICP-MS (inductively coupled plasma-MS), either linked directly to our size-exclusion HPLC outflow or for total copper quantification, as described previously , with individual copper isotopes being quantified using a PerkinElmer 6100 DRC ICP-MS from standard curves of copper generated from dilutions of a multi-element standard with known isotopic ratios of 63Cu/65Cu (SCP Science, Champlain, NY, U.S.A.).
Ferroxidase activity was measured using three procedures. The first procedure was adapted from that of Minotti and Ikeda-Saito  and monitored the formation of Fe(III) from FeSO4 (250 μM) in 0.2 M acetate buffer with 0.3 M NaCl (pH 6.0) spectrophotometrically at 310 nm for 3 min at 30 °C. The reaction was linear for at least 5 min, and the change in absorbance corresponded to the volume of enzyme assayed. The second ferroxidase assay which was adapted from that of Erel  followed the loss of Fe(II) [70 mM Fe(NH4)2SO4] in 0.37 M acetate buffer (pH 5.8) at 37 °C, using Ferene S (Sigma, St. Louis, MO, U.S.A.) at 600 nm over 4–6 min. Changes in absorbance of less than 1.0 per 4–6 min were linear over time and corresponded to the volume of enzyme assayed. Changes in absorbance due to autoxidation of Fe(II) were subtracted and averaged approx. 0.05 absorbance units per 4–6 min. Activity (nmol/min per ml of sample) was calculated based upon an nanomolar absorption coefficient (ε) of the Fe(II)–chromagen of 0.034 [A600/nmol of Fe(II)]. The third ferroxidase assay used was the classic one of Johnson et al. , which follows the production of holotransferrin from apotransferrin (55 μM) spectrophotometrically at 460 nm at 30 °C using Fe(II) [120 μM as Fe(NH4)2SO4·6H2O] in 200 mM acetate buffer (pH 6.0). Ferroxidase activities in a given plasma/serum sample were stable in samples stored frozen over several weeks or at 4 °C for several days. Oxidation of pPD was assayed as described previously [11,26], using a fresh 0.5% substrate solution and acetate buffer (pH 6.0) at 37 °C. Oxidation of o-dianisidine was assayed at 30 °C in pH 5.0 in acetate buffer, also as described previously [11,26]. In some cases, sodium azide was added to the assays at a final concentration of 1–2.5 mM. In-gel ferroxidase activity assays were carried out as described by Chen et al. , after native gel electrophoresis (7% gels), by incubating with 0.00784% Fe(NH4)2SO4·6H2O in acetate buffer (pH 6.0) at 37 °C, followed by exposure to 15 mM ferrozene to show the Fe(II) remaining in the background.
Purification of ceruloplasmin
This was carried out by DEAE–Sepharose chromatography in 100–300 mM potassium phosphate (pH 6.8) as described previously . Alternatively, ceruloplasmin was purified using a method adapted from the one-step procedure of Calabrese et al. , which involved modifying glutathione–Sepharose 4B with chloroethylamine after treatment with 10 M NaOH and epichlorohydrin .
Electrophoresis and immunoblotting
SDS/PAGE and native PAGE were carried out using 7% gels following standard procedures and were either stained with Coomassie Brilliant Blue Stain (Sigma) or transferred on to PVDF membranes (Bio-Rad, Hercules, CA, U.S.A.) by semi-dry transfer (Bio-Rad), in Tris/Triton-buffered saline as described previously . Blocking was performed with 5% (w/v) non-fat dried skimmed milk powder. A primary antibody raised in rabbits against human ceruloplasmin was obtained from Dako (Carpinteria, CA, U.S.A.). The secondary antibody used was alkaline phosphate-conjugated goat-anti-rabbit IgG (BioRad).
Computational modelling of mouse ceruloplasmin
Homology modelling of mouse ceruloplasmin was accomplished using ICM-Pro 3,4,5d (2.3) (Molsoft LLC, La Jolla, CA, U.S.A.). The sequence of mouse ceruloplasmin was obtained from the UniProtKB/Swiss-Prot database (primary accession number Q61147), aligned and threaded on to the sequence of human ceruloplasmin extracted from the structure file (PDB ID 1KCW) . Full optimization of the side chain rotomer space and hydrogen bonding was performed. Sequence-to-structure compatibility was tested by examining the semi-empirical force field interaction energy distribution implemented in ICM (2.5) and the fold energy calculated in PROSAII . Standard geometry-based checks verified that the structure model did not deviate significantly from the expected geometry target values [40–43].
This was performed at 4 °C with open 25 or 500 ml columns (1 cm×25 cm or 3.7 cm×48 cm) of Sephacryl S300 equilibrated in 20 mM potassium phosphate (pH 7.0), to which 0.5 or 10 ml of plasma were applied and 0.5 or 4.5 ml fractions were collected respectively; or by HPLC on a Biosep 4000 column (Phenomenex, Torrance, CA, U.S.A.) equilibrated with 20 mM Tris/HCl (pH 7.4) and coupled directly to ICP-MS, as described previously . For the latter method, 100 μl of 5-fold diluted pre-filtered (0.45 μM) plasma was applied. The size standards used were thyroglobulin (670 kDa), ferritin (480 kDa), bovine γ-globulin (158 kDa), human ceruloplasmin (132 kDa), chicken ovalbumin (44 kDa), equine myoglobin (16.7 kDa) and vitamin B12 (1.34 kDa).
Comparative levels of copper, ceruloplasmin and ferroxidase activities in the blood plasma of wild-type mice, ceruloplasmin knockout mice, rats and humans
Since the copper components and activities of ceruloplasmin of mouse blood plasma have hardly been studied, we began a systematic examination to compare various aspects of mouse blood plasma with those of rats and humans. As a first step, we compared plasma from adult C57BL6 wild-type mice and their ceruloplasmin knockout siblings with that from adult Sprague–Dawley rats. This was essential in order to confirm results from the very limited number of previous reports on these parameters in other strains of mice and rats, and to compare the measurements for ceruloplasmin (Cp+/+) (wild-type) mice (C57BL6 background) with those of knockout [ceruloplasmin (Cp−/−)] mice under the same experimental conditions. Figure 1(A) shows that in these mice, the wild-type plasma had less than half as much copper as that of Sprague–Dawley rats. Knocking out ceruloplasmin further lowered the copper concentrations by approx. 70%. Similar differences were observed for ceruloplasmin oxidase activity measured using pPD (Figure 1B). Activity in wild-type mice was about half of that measured in rats. More importantly, knocking out ceruloplasmin virtually eliminated this oxidase activity (it was undetectable at the volumes available for measurements). This indicated that the pPD oxidase assay is highly specific for ceruloplasmin.
Next, the ferroxidase activity of mouse plasma was examined. For this, three different assays were established and applied. The first assay, adapted from that of Minotti and Ikeda-Saito , monitored the formation of Fe(III) at 310 nm. The results from this assay indicated that, as with total copper and pPD oxidase activity, the mouse had approx. 60% less ferroxidase activity than the rat (Figure 2A, dark grey bars). However, much to our surprise, plasma from the ceruloplasmin knockout mouse was no different from that of the wild-type mouse, suggesting that mouse ceruloplasmin might not have ferroxidase activity. To examine this further, we applied both the classical assay of Johnson et al. , which follows the appearance of Fe(III)–transferrin, and that of Erel , which measures the loss of Fe(II) colorimetrically with Ferene S. In all of these assays (including the first), measurements were taken during the linear phases of product accumulation and/or substrate decline, and under conditions where velocities were proportional to the volume of sample assayed. This is the first time that interspecies plasma ferroxidase activities have been measured side by side, and provide the first reports for mouse plasma we could find. Figures 2(B) and 2(C) (dark grey bars) show that, with these two assays, wild-type ferroxidase activities were not that different in mice and rats, although both were much lower than in human plasma [50% with the Erel assay (Figure 2B) and 75% with the transferrin assay of Johnson et al. (Figure 2C)] . This was in marked contrast with the results for the pPD oxidase assay for ceruloplasmin, where mouse activities were much lower than for rats, suggesting either that mouse ceruloplasmin has less reactivity with pPD and/or that mouse plasma has more non-ceruloplasmin ferroxidase activity. In contrast with what was determined using the Minotti and Ikeda-Saito assay, there was a reduction in ferroxidase activity in the ceruloplasmin knockout mouse detected using the Erel and Johnson et al. assays. However, the reduction was only 50%, indicating that in the normal adult mouse, at least half of the ferroxidase activity is not attributable to ceruloplasmin.
In parallel tests, we examined the effects of azide on ferroxidase activity, since in rats and humans, ceruloplasmin pPD and o-dianisidine oxidase activities are inhibited by azide [13,26], and this has been reported for ferroxidase in humans previously . Using the Erel (Figure 2B) and classical (transferrin-based) (Figure 2C) assays (light grey bars), azide had by far the most inhibitory effect in the case of human plasma, virtually eliminating it in the Erel assay and reducing it 75% in the transferrin assay. With both assays, the lowest level of inhibition was observed in the ceruloplasmin knockout mouse (30–50%), and an intermediate degree of inhibition was evident for the wild-type mouse and rat. The Minotti and Ikeda-Saito assay showed no inhibition by azide, suggesting that it was not measuring ferroxidase activity attributable to ceruloplasmin. Taken together, all of these enzyme assay results suggest that rat and mouse ceruloplasmin might have small but significant structural differences compared with that of human ceruloplasmin. A comparison of previous reports indicates that rat plasma has just as much pPD oxidase activity as that of humans [2,26], yet it had 50–75% less ferroxidase activity as determined by the Erel and Johnson et al. assays. Mouse plasma had half as much pPD oxidase activity as human plasma, but was similar to rat plasma in having 50–75% less ferroxidase activity. Human ferroxidase activity was inhibited almost completely by azide, whereas that in the plasma of rats and mice was inhibited by 70–80%.
Does mouse ceruloplasmin have less ferroxidase activity and sensitivity to azide?
To determine whether the differences in enzyme activities atrributed to ceruloplasmin might be the result of structural differences in the pPD-, iron- and azide-binding sites, mouse ceruloplasmin was modelled in silico based on homology with the known X-ray crystal structure of human ceruloplasmin [38,44]. Owing to the high level of sequence and predicted structural homology, a homology model was expected to be accurate and of predictive value . We looked for potential structural differences in the regions known to be involved in iron and azide binding, as well as in the ferro-oxidation and binding of pPD [38,44,46]. Our modelling showed first that the mononuclear copper sites in domains 2, 4 and 6, and residues associated with the proposed electron transfer pathway in mouse ceruloplasmin, are highly conserved (Figure 3A) and align well structurally with human ceruloplasmin, as would be expected . The iron (cation)-binding site in the mouse was also highly conserved (Figure 3B). We next turned our attention to the region surrounding the azide-binding site. Although the environment of this site is generally conserved, several potentially significant sequence differences between mouse and human ceruloplasmin emerged when comparing our mouse model with the unrefined structure of azide-bound human ceruloplasmin (V.N. Zaitzev, personal communication). Figure 3(C) shows the altered microenvironment of the azide, highlighting substitutions E1032A, V1027I, Q729S and I1016T in mouse ceruloplasmin, which may modulate azide binding and thus affect its ability to inhibit ferroxidation. With regard to the pPD-binding site (Figure 3D), we observed that the striking π–stacking interaction between His667 and Trp669 found in human ceruloplasmin, which is implicated in electron transfer , was absent from the mouse ceruloplasmin model, where these residues are leucine and asparagine respectively: compare the positions of brown/lavender (human) and green/mauve (mouse) residues in Figure 3. These subtle but significant differences in the azide- and pPD-binding sites thus help to explain the reduced pPD oxidase activity and reduced azide inhibition of ferroxidation in mouse (versus human) plasma.
The question of whether mouse ceruloplasmin has ferroxidase activity was also investigated experimentally. First, fresh samples (1.0 ml) of wild-type and ceruloplasmin knockout mouse plasma (as well as human and rat plasma) were subjected to partial purification by anion-exchange chromatography (using DEAE–Sepharose), and the resulting ceruloplasmin-containing extracts were assayed for in-gel ferroxidase activity after separation by native PAGE. Samples from humans and rats showed a single (clear) ferroxidase band migrating at approx. one-quarter to one-third of the way from the top of the gel (Figure 4A). Wild-type mouse plasma purified simultaneously consistently showed a clear but less intense band in the same position as human ferroxidase on the gel. The position of the ferroxidase band in mouse plasma corresponded to that of ceruloplasmin, as determined by Western blotting after native PAGE (Figure 4B), providing strong evidence that mouse ceruloplasmin does possess ferroxidase activity. This was further confirmed by assaying a small amount of pure mouse ceruloplasmin (Figure 4C) obtained using the one-step procedure of Calabrese et al. , that had ferroxidase activity as determined using the Erel assay.
Consistent with our findings that knocking out ceruloplasmin only eliminated approximately half of the total ferroxidase activity, plasma from the ceruloplasmin knockout mice consistently showed an additional ferroxidase band in the in-gel assays (Figure 4A, KO lane). The migration of this band was always slower than that of ceruloplasmin.
Other ferroxidases in plasma
Since knocking out ceruloplasmin did not eliminate ferroxidase activity in mouse plasma, it was clear that one or more additional ferroxidases must be present. In a first attempt to separate and identify other ferroxidases, fresh wild-type mouse plasma was subjected to large pore size-exclusion chromatography on Sephacryl S300, and fractions were assayed for activity with the Erel assay. Several individual and pooled samples were run on small (25 ml) and large (500 ml) columns. Four peaks of ferroxidase activity were usually detected, which varied in relative amounts among the samples. The results for two such column runs for mouse plasma are shown in Figure 5(A). In these 500 ml columns, the first ferroxidase peak (thick line) eluted at approx. 200 ml, and had an apparent molecular mass of 700 kDa. The second and third ferroxidase peaks eluted just before and with ceruloplasmin (elution volumes of ∼250 and 270 ml respectively). These ferroxidase peaks (and particularly the latter one) coincided with the elution of oxidase activities attributable to ceruloplasmin: o-dianisidine oxidase activity (Figure 5A, upper panel, dashed/dotted line) and pPD oxidase activity (Figure 5A, lower panel, dashed/dotted line), and the main copper peak (results not shown). Usually there was also a ferroxidase peak near the end of the column (elution volume of ∼400 ml; Figure 5A, upper panel). Similar results were obtained for fresh human plasma using 25 ml columns of Sephacryl S300 (Figure 5B).
Plasma from ceruloplasmin knockout mice consistently contained two major ferroxidase peaks (Figure 5C), corresponding to the second and fourth ones seen in wild-type plasma. Evidence for a copper-binding component in knockout plasma corresponding to the first of these peaks (the second peak in wild-type plasma) was obtained by large pore size-exclusion HPLC coupled directly to ICP-MS. Figures 6(A) and 6(B) show the copper elution profiles obtained for wild-type and ceruloplasmin knockout mice (thin and thick line traces respectively) averaged (Figure 6A) and for individual samples (Figure 6B). For the wild-type mice, the copper elution profiles were similar to those obtained in the larger open columns, showing a central large peak and additional ones on either side, corresponding in molecular mass to transcuprein (180 kDa macroglobulin α1-inhibitor 3 of rodents) and albumin (69 kDa). Approx. 55% of the total copper eluted in the main (‘ceruloplasmin’) peak (Figures 6A and 6B), and approx. 30% eluted in the position of albumin (Figure 6A). The elution position of albumin was verified by comparing the elution of haemoglobin (which has almost the same molecular mass) and by examining the effect of prior Cibacron Blue treatment, which selectively removes albumin .
Elution profiles for the plasma from ceruloplasmin knockout mice (Figure 6, thick line traces) indicated that loss of this protein lowered the copper content of the main (central) peak by approximately two-thirds, confirming that ceruloplasmin eluted at this peak. However, to our surprise, a major copper peak was still present (180 kDa), eluting with the early part of the wild-type ‘ceruloplasmin’ peak. Its elution thus corresponds to that of the second ferroxidase peak detected, which also elutes just ahead of ceruloplasmin.
In the copper profile for the ceruloplasmin knockout mice, it may also be noted that not only the ‘ceruloplasmin’ peak, but also the ‘albumin’ peak was reduced in terms of copper content, suggesting that the latter contained not just albumin but also ceruloplasmin, and that their elution profiles overlapped. This was confirmed by analysing fractions from an open column separation of wild-type mouse plasma for ceruloplasmin and albumin by immunoblotting .
We have shown that, with respect to copper-binding proteins and their activities, the blood plasma of the mouse has similarities to and also major differences from that of humans and rats. More importantly, we show, for the first time, that a significant portion of the copper in the plasma peak traditionally ascribed to ‘ceruloplasmin’ is contributed by another (unknown) plasma protein that may also have ferroxidase activity, and that several ferroxidases additional to ceruloplasmin are present in the blood plasma.
With regard to the first points, we corroborated the scant previous evidence on total serum or plasma copper being approx. 400 ng/ml, our previous values for Swiss and Balb/c mouse serum  being 392 ng/ml (for three mice), values for C57BLx129S6/SvEv (‘young’) mice being 291–584 ng/ml, and those for 6–44-week-old C57BL6 mice being 401–728 ng/ml , thus being approximately half as much as in humans and rats. This suggests either that there is much less ceruloplasmin present and/or that ceruloplasmin has less copper (there is more of the apo form). Our measurements of enzyme activity, considered to be somewhat specific for this copper protein (pPD oxidase), were consistent with much less holoceruloplasmin and less copper, with mouse plasma demonstrating half as much activity as that of rats. Mice lacking ceruloplasmin expression had virtually no activity, confirming the specificity of this ceruloplasmin assay.
Although interest in ceruloplasmin has focused on its ferroxidase activity, we could find no published reports of its measurement in the plasma of mice. In the present study, total plasma ferroxidase activity (measured by three different assays) did not follow the same pattern of differences as those for copper and pPD oxidase between species and with knocking out ceruloplasmin. With two of the three ferroxidase assays employed [the classical transferrin assay, and an assay which follows the loss of Fe(II)], there was little or no difference between rats and mice, but the activity was much less in the rodents than in humans. Thus even though the rat has approximately the same amount of total plasma copper and pPD oxidase activity as humans , we found it had much less ferroxidase activity. Also, although rats have much more plasma copper than mice, and much more copper associated with the ‘ceruloplasmin’ peak eluting in size-exclusion chromatography (Figure 6) [2,5], the mice and rats had approximately the same amount of total plasma ferroxidase activity. The results from the third ferroxidase assay (Minotti and Ikeda-Saito), however, were consistent with the differences in copper and pPD oxidase between mice and rats. The various discrepancies suggest there could be species differences between the active sites for the enzymatic substrates [for example Fe(II) and pPD], which were investigated in the present study by modelling mouse ceruloplasmin on the structure of the human protein. The iron-binding site appeared to be unaltered, and thus cannot explain the much lower ferroxidase activity in mice. However, the pPD-binding site showed potentially significant differences between the human and mouse proteins, which could contribute to the lower pPD oxidase activity in mice.
Another difference we observed was in the binding site for azide, azide inhibition of oxidase and ferroxidase activities having been considered to be a characteristic of ceruloplasmin [14,26]. The proportion of ferroxidase activity inhibited by azide was much greater in humans than in mice (and rats), which would be consistent with differences in azide-binding affinity. In the case of the human protein, inhibition was 90–100%, and 60–70% in the mouse protein (except with the Minotti and Ikeda-Saito assay, where no inhibition was detected). Comparison of the plasma of wild-type and ceruloplasmin knockout mice confirmed that ceruloplasmin was responsible for a portion of the total ferroxidase activity in mouse plasma: an absence of ceruloplasmin expression reduced ferroxidase activity to 50–60% as measured by the classical and Erel assays, but did not change it in the case of the Minotti and Ikeda-Saito assay. The presence of ferroxidase activity in mouse ceruloplasmin was confirmed by purifying to homogeneity a small quantity of the mouse protein (which had activity) and demonstrating an in-gel ferroxidase-active band coinciding with ceruloplasmin by immunoblotting after native PAGE.
One of our major findings of the present study, however, was that knocking out ceruloplasmin did not abolish ferroxidase activity in mice, with 50% or more activity remaining, as demonstrated by all three of the ferroxidase assays. It proves that ceruloplasmin is not the only ferroxidase in the blood plasma. Further proof was obtained by fractionating mouse and human plasma by size-exclusion chromatography and assaying the fractions for activity. As mentioned above, evidence for an additional ferroxidase (ferroxidase II) in human and rabbit plasma had been presented previously [14,15], but this has largely been ignored. Consistent with these earlier reports, we detected a large ferroxidase eluting with an apparent molecular mass of approx. 700 kDa in both mouse plasma and that of humans. We did not find this activity in every sample of fresh plasma, although it was present in more than half. We are currently in the process of purifying the unidentified protein possessing this activity for identification. It had been reported previously that ferroxidase had two large subunits (200–300 kDa) and that one of them contained a single copper atom [14,15]. Topham et al.  also reported that, like ceruloplasmin, this protein (which is expressed at higher levels in rabbits) was capable of facilitating the release of iron from storage organs. However, the present study of plasma fractionated by size-exclusion chromatography indicated that there were additional ferroxidases. The most consistent was that eluting just ahead of ceruloplasmin, with a molecular mass of approx. 180 kDa. Some overlap with pPD and o-dianisidine oxidase activities suggests the possibility that it has some homology with ceruloplasmin, which would also be consistent with the native PAGE results. This showed a ferroxidase band in the plasma of ceruloplasmin knockout mice migrating slightly slower than ceruloplasmin. In addition, this ferroxidase co-purified with ceruloplasmin during ion-exchange chromatography. Moreover, it was particularly active in plasma from knockout mice (first peak, Figure 5C), as was a fourth ferroxidase which eluted at the end of the column. The largest ferroxidase (II), however, was not present prominently in the two column runs of knockout plasma analysed. Thus there is a possibility that the second (and possibly fourth) ferroxidases are up-regulated in the absence of ceruloplasmin. This would explain the observed absence (or lower activity) of the band seen in the knockout plasma in samples compared with wild-type mice separated on native gels.
Of particular interest is the evidence of an additional major copper-binding protein in blood plasma that may normally be hidden under the ‘ceruloplasmin’ peak. This is the first time such a component has been reported (other than by us at a recent scientific meeting). Moreover, the copper component elutes in the position of the 180 kDa ferroxidase in the plasma of ceruloplasmin knockout mice and could thus be another copper-containing ferroxidase. If so, as mentioned above, it could be homologous with ceruloplasmin: it may have copper as well as ferroxidase activity and may co-purify with it in ion-exchange chromatography. Attempts to purify and identify this protein are also underway.
The work in the present study with mouse plasma has thus been instructive, not just in demonstrating apparent species-specific differences in the expression of copper-binding proteins and their activities, but also in identifying new copper components and new ferroxidases in the blood plasma that may or may not be dependent on copper. Our recent study comparing the copper ‘profiles’ for blood plasma from mice, rats and humans (separated by the same type of size-exclusion HPLC coupled directly to ICP-MS as shown here for wild-type and ceruloplasmin knockout mice)  indicates that the major difference between mice and the other species is in the amount of copper in the ‘ceruloplasmin’ peak, which we now know also contains another copper protein. Whereas numerous results from our laboratory and from others have reported that ceruloplasmin accounts for approx. 70% of the total plasma copper in human and rat plasma [1,2,4,5], in mice, this value is more like 55%, or approx. 25% less. Total copper in mouse plasma is at least 50% lower. Like the reduced levels of pPD oxidase activity, these results imply that the levels of ceruloplasmin in mice are considerably lower than in humans and rats. In addition, there could be a lower proportion of copper-containing (holo) ceruloplasmin . The contribution of the unknown component to levels of copper in wild-type mouse plasma and that of other species remains to be discovered.
This work was supported by the National Institutes of Health [grant number RO1-HD46949] (to M. C. L.); by a grant from the Minority Access to Research Careers [grant number NIH 5 T34 GM008612] (to California State University Fullerton, supporting L. A. G.); and by the National Science Foundation [grant numbers REU CHE 0354159 (to K. P., S. A. and A. N.), OCE-9978806, DBL-9978806 (both to A. Z. M.)]. J. B. C. was supported in part by a Medical Scientist Training Grant from the National Institutes of Health [grant number GM08014].
We thank Dr Z. Leah Harris (Department of Anesthesiology and Critical Care Medicine, Johns Hopkins University, Baltimore, MD, U.S.A.) for providing the heterozygous ceruloplasmin knockout mice used for breeding our colony, and Dr V. N. Zaitsev (Centre of Biomolecular Sciences, University of St. Andrews, St. Andrews, Fife, U.K.) for sharing information about the unrefined azide–human ceruloplasmin structure.
Abbreviations: ICP-MS, inductively coupled plasma-MS; pPD, p-phenylene diamine
- © The Authors Journal compilation © 2009 Biochemical Society