Hemopexin decreases hemin accumulation and catabolism by neural cells

Hemopexin is a serum, CSF, and neuronal protein that is protective after experimental stroke. Its efficacy in the latter has been linked to increased expression and activity of heme oxygenase (HO)-1, suggesting that it facilitates heme degradation and subsequent release of cytoprotective biliverdin and carbon monoxide. In this study, the effect of hemopexin on the rate of hemin breakdown by CNS cells was investigated in established in vitro models. Equimolar hemopexin decreased hemin breakdown, as assessed by gas chromatography, by 60–75% in primary cultures of murine neurons and glia. Extracellular hemopexin reduced cell accumulation of ⁵⁵Fe-hemin by over 90%, while increasing hemin export or extraction from membranes by fourfold. This was associated with significant reduction in HO-1 expression and neuroprotection. In a cell-free system, hemin breakdown by recombinant HO-1 was reduced over 80% by hemopexin; in contrast, albumin and two other heme-binding proteins had no effect. Although hemopexin was detected on immunoblots of cortical lysates from adult mice, hemopexin knockout per se did not alter HO activity in cortical cells treated with hemin. These results demonstrate that hemopexin decreases the accumulation and catabolism of exogenous hemin by neural cells. Its beneficial effect in stroke models is unlikely to be mediated by increased production of cytoprotective heme breakdown products.     

Kinetics of the interaction of hemin liposomes with heme binding proteins

As a model for the transport of hemin across biological membranes, sonicated phosphatidylcholine liposomes with incorporated hemin were characterized. The interaction of the hemin liposomes with the heme binding proteins albumin, apomyoglobin, and hemopexin was examined as a function of liposome charge and cholesterol content. In all cases, there was an almost complete transfer of hemin from liposome to protein; a rapid phase and a slow phase were observed for the transfer. For negatively charged liposomes (with 11% dicetyl phosphate), the rapid and slow phases showed observed rates of transfer of ca. 2 and 0.01 s-1, respectively, for all three proteins. The presence of cholesterol in the liposomes decreased the observed rates by a factor of 2, and positively charged liposomes (with 11% stearylamine) showed about one-fifth the observed rates of negatively charged liposomes. The observed rates were independent of protein concentration, indicating that the rate-determining step is hemin efflux from the lipid bilayer. The hemin interaction with the phospholipid bilayer is suggested to be primarily hydrophobic with some electrostatic character. The two phases are suggested to arise from two different populations of hemin within the liposomes and are interpreted as arising from two different orientations of hemin within the bilayer.     

Long-term intercalation of residual hemin in erythrocyte membranes distorts the cell

The effect of long-term incubation of residual globin-free hemin on whole red blood cell and isolated cytoskeletal proteins was studied. Hemin at concentrations found in pathological red cells was inserted to fresh erythrocytes. Increased hemolysis developed in the hemin-containing cells after a few days at 37 degrees C and after about four weeks at 4 degrees C. Since lipid and hemoglobin peroxidation did not depend on the presence of hemin, time-dependent effects on the cytoskeleton proteins were studied. Observations were: (1) spectrin and protein 4.1 exhibited a time-dependent increasing tendency to undergo hemin-induced peroxidative crosslinking. (2) The ability of the serum proteins, albumin and hemopexin, to draw hemin from spectrin, actin and protein 4.1 decreased with time of incubation with hemin. These results were attributed to time-dependent hemin-induced denaturation of the cytoskeletal proteins. Albumin taken as a control for physiological hemin trap was unaffected by hemin. Small amounts of hemo-spectrin (2-5%) were analyzed in circulating normal cells, and this in vivo hemo-spectrin also failed to release hemin. It was concluded that slow accumulation of hemin, a phenomenon increased in pathological cells, is a toxic event causing erythrocyte destruction.     

Kinetics of hemoprotein reduction and interprotein heme transfer

The transfer of hemin from one protein to another is an event biologically important for the conservation of heme iron. Hemin entering the circulation (or added to serum) is mainly bound by albumin and then transferred to hemopexin [Morgan, W.T., Liem, H.H., Sutor, R.P., & Muller-Eberhard, U. (1976) Biochim. Biophys. Acta 444, 435-445], and we are now investigating which mechanisms may be operative in enhancing this process. The presence of imidazole has been demonstrated to accelerate hemin transfer from albumin to hemopexin [Pasternack, R.F., Gibbs, E.J., Hoeflin, E., Kosar, W.P., Kubera, G., Skowronek, C. A., Wong, N.M., & Muller-Eberhard, U. (1983) Biochemistry 22, 1753-1758]. The present work is an examination of the effect of the reduction of albumin-bound hemin on the rate of its transfer to hemopexin. Hemin (HmIII., ferriprotoporphyrin IX) was reduced to HmII (ferroprotoporphyrin IX) by the addition of sodium dithionite under argon. The reduction kinetics of HmIII to HmII were studied separately in the two complexes: with human serum albumin (HSA), which binds up to 20 mol of heme/mol (the first mole with K congruent to 10(8)), and with hemopexin (HHx), which binds heme equimolarly (K congruent to 10(13)). The rate of reduction of HmIII to HmII on HSA was first order over several half-lives and linearly dependent on [S2O4(2-)]1/2. At [HSA]0/[HmIII] = 3, the kobsd was (5 X 10(-3) + 0.75[S2O4(2-)]1/2, and with [HSA]/[HmIII] approximately 25, the kobsd was (2 X 10(-3)) + 0.25[S2O4(2-)]1/2. The reduction of HmIII to HmII on human hemopexin (HHx) is much more rapid with kobsd = (2.5 X 10(3))[S2O4(2-)]1/2.(ABSTRACT TRUNCATED AT 250 WORDS)     

Kinetics of hemin distribution in plasma reveals its role in lipoprotein oxidation.

Hemin is a powerful in vitro inducer of low-density lipoprotein (LDL) oxidation, implicated in development of atherosclerosis. To support the proposed role of hemin in atherogenesis, the question of whether hemin has any chance of getting together with LDL in vivo, must be addressed. A stopped-flow technique was employed in order to investigate the fast kinetics of hemin binding to LDL and to other plasma hemin-binding proteins: high-density lipoprotein (HDL), albumin and hemopexin. Based on the measured rate constants of hemin association with and dissociation from each of these proteins, time-dependent hemin distribution in plasma was analyzed. The analysis shows that as much as 80% of total hemin binds initially to LDL and HDL, the plasma components which are most susceptible to oxidation. Only then hemin partially transfers to the antioxidants albumin and hemopexin. The half time of the hemin-LDL complex in plasma, initially comprising 27% of total hemin, was more than 20 s. Not only transient, but also oxidatively active steady-state hemin-lipoprotein complexes in plasma were both predicted from the kinetic analysis and found in experiment. Our data suggest that the hemin-LDL complex may exist in vivo and that its oxidative potential should be considered pro-atherogenic.     

Accumulation and drainage of hemin in the red cell membrane

The subject of hemin intercalation in red cell membranes and the correlation of the accumulated hemin level with the membrane pathology was studied. Methods which made use of dioxan and octan-2-ol mixtures to quantitate small amounts of hemin in membranes were developed. Applying these methods, hemin levels were measured in the cytoskeleton and the remaining lipid core of various red cell membranes. The amount of hemin, in both membrane fractions, was higher in pathological cells of sickle cell anemia and beta-thalassemia as compared to normal circulating cells. Correlation exists between the amount of the membrane-accumulated hemin and the severity of the disease. The level of hemin in the membrane was found to be age dependent, old cells in circulation accumulating more hemin than young cells. The level of hemin in all cells tested was much lower than the amount found previously to cause immediate hemolysis when applied externally (Kirschner-Zilber, I., Rabizadeh, E. and Shaklai, N. (1982) Biochim. Biophys. Acta 690, 20-30). This was explained by the differences between the process leading to immediate lysis and membrane changes recognized as pathological by the in-vivo sequestration mechanism. In search of a physiological mechanism which may drain the cell membrane from the hazardeous hemin, albumin, the main serum protein, was found capable of serving as an efficient agent for extracting hemin trapped in red cell membranes. It is suggested that under normal conditions albumin extracts enough hemin to leave the erythrocyte with unharmful hemin amounts, however, under pathological conditions greater amounts accumulate leading to a shorter cell life span.     

The effects of heme-binding proteins on the peroxidative and catalatic activities of hemin.

The plasma proteins hemopexin (Hx) and albumin (Alb) are known to bind heme with high and medium affinity, respectively. To study how this binding modifies heme catalytic reactivity, the effects of Hx, human serum Alb (HSA), and bovine serum Alb (BSA) on the peroxidase- and catalaselike activities of hemin were investigated. These hemin activities were found to be inhibited by 50 to 60% with either HSA or BSA, and by 80 to 90% with Hx. The heme complexes with Hx or Alb (1:1 = protein:heme) therefore had a much lower reactivity toward H2O2 and Cum-OOH than the nonprotein heme. A kinetic analysis suggested that binding to Hx or Alb inhibited the primary activation of heme by H2O2, the step common for both peroxidase- and catalaselike activities of hemin. It is thought that by complexing heme, the Hx and Alb can prevent the toxic effects of extracellular heme in blood plasma.     

Hemin-induced lysis of rat erythrocytes. Protective action of ceruloplasmin and different serum albumins

Hemin-induced lysis of rat erythrocytes is markedly reduced by ceruloplasmin (human) and serum albumins from different species, the order of effectiveness beings: bovine albumin approximately equal to ceruloplasmin greater than human albumin approximately equal to dog albumin greater than apotransferrin (human). Although the proteins studied had hemin binding capacity, the best protective agents, ceruloplasmin and bovine albumin, did bind hemin less strongly than human and dog albumin. The results suggest the existence of another protective mechanism, possibly involving an interaction between erythrocyte membranes and serum proteins.     

Association of hemin with protein 4.1 as compared to spectrin and actin

The interaction of hemin with protein 4.1 isolated from red cell membrane cytoskeleton has been studied. Spectrophotometric titration has shown one strong binding site and additional lower affinity sites for hemin. From fluorescence quenching data an association binding constant of 1.3 . 10(7) M-1 has been calculated for the primary site. The conformation of cytoskeletal proteins after hemin binding was followed by the use of far UV circular dichroism and compared to that of the serum hemin trap, albumin. The secondary structure of albumin was unchanged in the presence of high hemin concentrations. Both spectrin and actin lost their conformation upon hemin binding in a ligand-concentration and time-dependent manner. Unlike spectrin and actin, the secondary structure of protein 4.1 appeared. The findings of this study suggest that protein 4.1 may serve as the cytoskeletal temporary sink for small amounts of membrane-intercalated hemin similarly to the function of albumin in the serum. However, an increased release of hemin under pathological conditions may cause hemin association with the cytoskeletal proteins and as a result the cell membrane is expected to be distorted.     

Ascorbate protects against tert-butyl hydroperoxide inhibition of erythrocyte membrane Ca2+ + Mg2(+)-ATPase

The incubation of erythrocyte suspensions or isolated membranes containing a residual amount of hemoglobin (0.04% of original cellular hemoglobin) with tert-butyl hydroperoxide (tBHP, 0.5 mM) caused significant inhibition of basal and calmodulin-stimulated Ca2+ + Mg2(+)-ATPase activities and the formation of thiobarbituric acid reactive products measured as malondialdehyde. In contrast, the treatment of white ghosts (membranes not containing hemoglobin) with tBHP (0.5 mM) did not lead to appreciable enzyme inhibition within the first 20 min and did not result in malondialdehyde (MDA) formation. However, the addition of either 10 microM hemin or 100 microM ferrous chloride + 1 mM ADP to white ghosts produced hydroperoxide effects similar to those in pink ghosts (membranes with 0.04% hemoglobin). The concentrations of hemin and ferrous chloride which caused half-maximal inhibition of Ca2+ + Mg2(+)-ATPase activity at 10 min were 0.5 and 30 microM, respectively. The effects of several antioxidants (mannitol, thiourea, hydroxyurea, butylated hydroxytoluene, and ascorbate) were investigated for their protective effects against oxidative changes resulting from tBHP treatment. Over a 30-min incubation period only ascorbate significantly reduced the enzyme inhibition, MDA formation, and protein polymerization. Thiourea and hydroxyurea decreased MDA formation and protein polymerization but failed to protect against the enzyme inhibition. Butylated hydroxytoluene was similar to thiourea and hydroxyurea but with better protection at 10 min. Mannitol, under these conditions, was an ineffective antioxidant for all parameters tested.     

Hemin toxicity: a preventable source of brain damage following hemorrhagic stroke

Abstract

Hemorrhagic stroke is a common cause of permanent brain damage, with a significant amount of the damage occurring in the weeks following a stroke. This secondary damage is partly due to the toxic effects of hemin, a breakdown product of hemoglobin. The serum proteins hemopexin and albumin can bind hemin, but these natural defenses are insufficient to cope with the extremely high amounts of hemin (10 mM) that can potentially be liberated from hemoglobin in a hematoma. The present review discusses how hemin gets into brain cells, and examines the multiple routes through which hemin can be toxic. These include the release of redox-active iron, the depletion of cellular stores of NADPH and glutathione, the production of superoxide and hydroxyl radicals, and the peroxidation of membrane lipids. Important gaps are revealed in contemporary knowledge about the metabolism of hemin by brain cells, particularly regarding how hemin interacts with hydrogen peroxide. Strategies currently being developed for the reduction of hemin toxicity after hemorrhagic stroke include chelation therapy, antioxidant therapy and the modulation of heme oxygenase activity. Future strategies may be directed at preventing the uptake of hemin into brain cells to limit the opportunity for toxic interactions.     

Detection and comparison of specific hemin binding by Porphyromonas gingivalis and Prevotella intermedia.

A radioligand assay was designed to detect and compare specific hemin binding by the periodontal anaerobic black-pigmenting bacteria (BPB) Porphyromonas gingivalis and Prevotella intermedia. The assay included physiological concentrations of the hemin-binding protein rabbit serum albumin (RSA) to prevent self-aggregation and nonspecific interaction of hemin with cellular components. Under these conditions, heme-starved P. intermedia cells (two strains) expressed a single binding site species (4,100 to 4,600 sites/cell) with a dissociation constant (Kd) of 1.0 x 10(-9) M. Heme-starved P. gingivalis cells (two strains) expressed two binding site species; the higher-affinity site (1,000 to 1,500 sites/cell) displayed a Kd of between 3.6 x 10(-11) and 9.6 x 10(-11) M, whereas the estimated Kd of the lower-affinity site (1.9 x 10(5) to 6.3 x 10(5) sites/cell) ranged between 2.6 x 10(-7) and 6.5 x 10(-8) M. Specific binding was greatly diminished in heme-replete cells of either BPB species and was not displayed by iron-replete Escherichia coli cells, which bound as much hemin in the absence of RSA as did P. intermedia. Hemin binding by BPB was reduced following treatment with protein-modifying agents (heat, pronase, and N-bromosuccinimide) and was blocked by protoporphyrin IX and hemoglobin but not by Congo red. Hemopexin also inhibited bacterial hemin binding. These findings indicate that both P. gingivalis and P. intermedia express heme-repressible proteinaceous hemin-binding sites with affinities intermediate between those of serum albumin and hemopexin. P. gingivalis exhibited a 10-fold-greater specific binding affinity and greater heme storage capacity than did P. intermedia, suggesting that the former would be ecologically advantaged with respect to heme acquisition.     

Degradation of Host Heme Proteins by Lysine- and Arginine-Specific Cysteine Proteinases (Gingipains) of Porphyromonas gingivalis

Porphyromonas gingivalis can use hemoglobin bound to haptoglobin and heme complexed to hemopexin as heme sources; however, the mechanism by which hemin is released from these proteins has not been defined. In the present study, using a variety of analytical methods, we demonstrate that lysine-specific cysteine proteinase of P. gingivalis (gingipain K, Kgp) can efficiently cleave hemoglobin, hemopexin, haptoglobin, and transferrin. Degradation of hemopexin and transferrin in human serum by Kgp was also detected; however, we did not observe extensive degradation of hemoglobin in serum by Kgp. Likewise the beta-chain of haptoglobin was partially protected from degradation by Kgp in a haptoglobin-hemoglobin complex. Arginine-specific gingipains (gingipains R) were also found to degrade hemopexin and transferrin in serum; however, this was observed only at relatively high concentrations of these enzymes. Growth of P. gingivalis strain A7436 in a minimal media with normal human serum as a source of heme correlated not only with the ability of the organism to degrade hemoglobin, haptoglobin, hemopexin, and transferrin but also with an increase in gingipain K and gingipain R activity. The ability of gingipain K to cleave hemoglobin, haptoglobin, and hemopexin may provide P. gingivalis with a usable source of heme for growth and may contribute to the proliferation of P. gingivalis within periodontal pockets in which erythrocytes are abundant.     

Life-span and size of the trans-membrane channel formed by large doses of complement

To evaluate the life-span and size of trans-membrane channels in complement (C) treated membranes, resealed erythrocyte ghosts containing trapped native protein markers, as well as residual hemoglobin, were treated with anti-Forssman antibody and large doses of guinea pig C. Ovalbumin and hemoglobin were released slowly through the channels so produced, whereas human serum albumin was not. Release of hemoglobin was not blocked by extracellular bovine serum albumin. Release of hemoglobin continued for at least 72 hr at 4 degrees C. Semi-logarithmic plots of ovalbumin or hemoglobin release showed gradual diminution of the rate constant, which indicates slow loss of channels during the experimental period. These experiments demonstrate that the channels produced in erythrocyte ghost membranes by large C doses have a long, although finite, life-span. Their effective diameter is al least 55 A on the basis of ovalbumin and hemoglobin release, and not more than 150 A, since serum albumin was not released. However, an upper limit of 100 A would be more reasonable in light of electronmicroscopic observations by others. These results are compatible with the doughnut model.     

Interaction of hemopexin with Sn-protoporphyrin IX, an inhibitor of heme oxygenase. Role for hemopexin in hepatic uptake of Sn-protoporphyrin IX and induction of mRNA for heme oxygenase

Sn-protoporphyrin IX (SnPP), an inhibitor of heme oxygenase and a potential therapeutic agent for neonatal hyperbilirubinemia, is bound tightly by hemopexin. The apparent dissociation constant (Kd) at pH 7.4 is 0.25 +/- 0.15 microM, but estimation of the Kd for the SnPP-hemopexin complex is hampered by the fact that at physiological pH SnPP exists as monomers and dimers, both of which are bound by hemopexin. SnPP is readily displaced from hemopexin by heme (Kd less than 1 pM). The hemopexin-SnPP interaction, like that of heme-hemopexin, is dependent on the histidine residues of hemopexin. However, as expected from the differences in the coordination chemistries of tin and iron, the stability of the histidyl-metalloporphyrin complex is lower for SnPP-hemopexin than for mesoheme-hemopexin. Nevertheless, when SnPP binds to hemopexin, certain of the ligand-induced changes in the conformation of hemopexin which increase the affinity of the protein for its receptor are produced. Binding of SnPP produces the conformational change in hemopexin which protects the hinge region of hemopexin from proteolysis, but SnPP does not produce the characteristic increase in the ellipticity of hemopexin at 231 nm that heme does. Competition experiments confirmed that human serum albumin (apparent Kd = 4 +/- 2 microM) has a significantly lower affinity for SnPP than does hemopexin. Appreciable amounts of SnPP (up to 35% in adults and 20% in neonates) would be bound by hemopexin in the circulation, and the remainder of SnPP would be associated with albumin due to the latter's high concentration in serum. Essentially no non-protein-bound SnPP is present. Importantly, SnPP-hemopexin binds to the hemopexin receptor on mouse hepatoma cells with an affinity comparable to that of heme-hemopexin and treatment of the hepatoma cells with SnPP-hemopexin causes a rapid increase in the steady state level of heme oxygenase messenger RNA. These results show that hemopexin participates in the transport of SnPP to heme oxygenase and in its regulation by SnPP.     

Immunodepletion of albumin for two-dimensional gel detection of new mouse acute-phase protein and other plasma proteins

Immunodepletion of albumin to improve the 2-D gel resolution of human plasma proteins has recently been described. With the importance of mouse models in many studies in which serum or plasma is often analyzed, we have adopted this approach to immunoprecipitate mouse albumin and evaluated its effectiveness for 2-D separation of mouse plasma proteins. Purified polyclonal antibodies against mouse albumin were effective depleting intact albumin as well as its numerous fragments from mouse plasma samples. Removal of albumin resulted in better resolution of mouse plasma proteins. Three proteins, alpha2-macroglobulin, coagulation factor XII, and hemopexin, that were previously either undetectable or poorly resolved, were identified from albumin-depleted 2-D gels by peptide mass fingerprinting. Albumin depletion also led to partial loss of several other proteins such as clusterin and gelsolin. This loss can be attributed to the interaction with albumin itself because the specificity of the antibody was demonstrated by Western blot. When applying this method to the 2-D separation of plasma from inflamed mouse induced by cutaneous burn injury with superimposed Pseudomonas aeruginosa infection, the upregulation of inter alpha-trypsin inhibitor heavy chain 4 (ITIH4) and hemopexin was unambiguously detected along with other mouse acute-phase proteins (APP), including haptoglobin and serum amyloid A. Based on the significant increase of ITIH4, we propose that this protein is a new member of mouse APP that are upregulated during the inflammatory response.     

Relation between Ca2+-ATPase and endogenous calmodulin of human erythrocyte membranes

Short incubation of erythrocyte membranes with oleic acid releases Ca2+-independently bound endogenous calmodulin together with a minor fraction of membrane-associated proteins without destruction of the membranes. The released endogenous calmodulin is similar if not identical to cytosolic calmodulin reversibly bound to ghosts in a Ca2+-dependent manner. The release of endogenous calmodulin proceeds without affecting the activity of Ca2+-ATPase when ghosts are incubated with oleic acid in the presence of Ca2+ plus ATP and thereafter freed from oleic acid by washings with serum albumin. Kinetic parameters of Ca2+-ATPase of ghosts with and without endogenous calmodulin are identical as are amounts of exogenous calmodulin bound to these ghosts. Thus, endogenous calmodulin does not function as an essential part of Ca2+-ATPase.     

Interaction of rabbit hemopexin with bilirubin

The interaction of hemopexin with bilirubin was characterized by spectrophotometric, fluorimetric and circular dichroic techniques. Hemopexin rapidly forms an equimolar complex with libirubin that has an apparent dissociation constant Kd, of 7.5.10(-7) M. The association alters the absorption band of bilirubin near 150 nm, quenches the fluorescence of tryptophan residues of hemopexin, enhances the fluorescence of bilirubin, and induces strong ellipticity extrema in bilirubin of --60 . 10(3) deg . cm2 . dmol-1 at 465 nm and +70 . 10(3) deg . cm2 . dmol-1 at 415 nm. However, the conformation-sensitive ellipticity aband at 231 nm of hemopexin is not altered. In displacement experiments using circular dichroism, heme readily replaced bound bilirubin, indicating that bilirubin and heme are bound at the same site on hemopexin. Even at molar ratios of hemopexin to albumin of 3 to 1, human serum albumin removes bilirubin from hemopexin. Hemopexin is thus unlikely to have a role in the transport of bilirubin in serum.     

Abnormal membrane protein methylation and merocyanine 540 fluorescence in sickle erythrocyte membranes

Sickle cell erythrocytes exhibit reduced carboxyl methylation of membrane proteins compared to normal erythrocytes. This altered methylation in sickle membrane proteins is also observable when extracted membranes, both intact and alkali treated, were used as substrates for the homologous protein methylase II (S-adenosylmethionine:protein-carboxyl O-methyltransferase, EC. 2.1.1.24). However, when glycophorin A, one of the major methyl acceptors in both membranes, was extracted by lithium diiodosalicylate and used as the methyl acceptor, the proteins from both membranes were methylated equally, suggesting an involvement of membrane structure in membrane-bound protein methylation. Merocyanine 540 (MC-540), a fluorescent probe, was used to determine if the membranes differed in organization. Incubation of both normal and sickle erythrocytes membranes with MC-540 produced a marked increase in extrinsic fluorescence, reflecting a relatively nonpolar environment for the dye bound to the membranes. The fluorescence from sickle cell ghosts was only 87% as intense as that from normal ghosts, while the actual amount of MC-540 associated with sickle cell membranes was only 62% of normal. These data suggest that differences exist in the distribution of surface charges on these plasma membranes. These results are consistent with the hypothesis that abnormal levels of membrane protein methylation observed in sickle erythrocytes may be a result of abnormal membrane organization characteristic to sickle cell anemia.     

The utilization of heme iron by Yersinia pestis in human blood and blood sera

Y. pestis, the causative agents of plague, have been found to be incapable of using heme iron bound to haptoglobin and hemopexin complexes in human blood and blood serum, and protein components of the serum are not the factors inhibiting this process. At the same time iron of free hemoglobin can be successfully utilized by Y. pestis in the systems used in this study. On the contrary, hemin not only produces any stimulating effect on the growth of Y. pestis in blood serum, but leads to the death of these bacteria [correction of lasteria].