Advertisement

Might non-transferrin-bound iron in blood plasma and sera be a non-proteinaceous high-molecular-mass FeIII aggregate?

  • Shaik Waseem Vali
    Affiliations
    From the Department of Biochemistry and Biophysics, Texas A&M University, College Station Texas 77843
    Search for articles by this author
  • Paul A. Lindahl
    Correspondence
    To whom corresponding should be addressed: Paul A. Lindahl, Department of Chemistry, Texas A&M University, College Station TX 77843-3255. Phone, 979-845-0956 ; Fax, 979-845-4719, : .
    Affiliations
    From the Department of Biochemistry and Biophysics, Texas A&M University, College Station Texas 77843

    Department of Chemistry, Texas A&M University, College Station Texas 77843-3255
    Search for articles by this author
Open AccessPublished:November 02, 2022DOI:https://doi.org/10.1016/j.jbc.2022.102667

      ABSTRACT

      Blood of HFE(-/-) mice and of humans with hemochromatosis contains toxic non-transferrin-bound iron (NTBI) which accumulates in organs. However, the chemical composition of NTBI is uncertain. To investigate, HFE(-/-) mice were fed iron-deficient diets supplemented with increasing amounts of iron, with the expectation that NTBI levels would increase. Blood plasma was filtered to obtain retentate and flow-through-solution (FTS) fractions. Liquid chromatography detected by inductively coupled plasma mass spectrometry of FTSs exhibited low-molecular-mass iron peaks that did not increase intensity with increasing dietary iron. Retentates yielded peaks due to transferrin and ferritin, but much iron in these samples adsorbed onto the column. Retentates treated with deferoxamine (DFO) chelator yielded a peak that comigrated with the Fe-DFO complex and originated from iron that had adhered to the column. Additionally, plasma from younger and older 57Fe-enriched HFE mice were separately pooled and concentrated by ultrafiltration. After removing contributions from contaminating blood and transferrin, Mössbauer spectra were dominated by features due to magnetically-interacting FeIII aggregates, with greater intensity in the spectrum from the older mice. Similar features were generated by adding 57FeIII to “pseudo plasma”. Aggregation was unaffected by albumin or citrate at physiological concentrations, but DFO or high citrate concentrations converted aggregated FeIII into high-spin FeIII complexes. FeIII aggregates were retained by the cutoff membrane and adhered to the LC column, similar to NTBI. A model is proposed in which FeII entering blood is oxidized, and if apo-transferrin is unavailable, the resulting FeIII ions coalesce into FeIII aggregates, a.k.a. NTBI.

      Keywords

      Abbreviations:

      DFO (deferoxamine), EPR (electron paramagnetic resonance), FPN (ferroportin), FTN (ferritin), FTS (flow-through solution), HFE (common gene mutated in hereditary hemochromatosis), ICP-MS (inductively coupled plasma mass spectrometry), LC (liquid chromatography), LMM (low molecular mass), LN2 (liquid nitrogen), MB (Mössbauer spectroscopy), R24 and R36 (24 and 36 wk mice raised on a regular diet), TFN (transferrin)

      Introduction

      Non-transferrin-bound iron (NTBI) is a toxic form of iron in the blood of individuals with iron-overload diseases such as hereditary hemochromatosis (
      • Anderson G.J.
      • Bardou-Jacquet E.
      Revisiting hemochromatosis: genetic vs. phenotypic manifestations.
      ) and β-thalassemia (
      • Chauhan W.
      • Shoaib S.
      • Fatma R.
      • Zaka-ur-Rab Z.
      • Afzal M.
      Beta-thalassemia and the advent of new interventions beyond transfusion and iron chelation.
      ). It accumulates excessively in the liver and other organs, resulting in liver fibrosis, cirrhosis, cancer, endocrinopathies, and cardiomyopathies (
      • Knutson M.D.
      Non-transferrin-bound iron transporters.
      ). Treatments include low iron diets, frequent phlebotomies, and ingestion of iron-binding chelators. Effectiveness is limited, especially for β-thalassemia.
      Transferrin (TFN) is an iron-binding protein in the blood which serves as an iron buffer (
      • Bartnikas T.B.
      Known and potential roles of transferrin in iron biology.
      ). In healthy individuals, about one-third of TFN is in the holo- (two FeIII ions bound) form whereas most of the remainder is apo- (iron-free). Nutrient iron enters the blood via ferroportin (FPN), a membrane-bound FeII transporter that is highly expressed on the basolateral side of enterocytes in the duodenum (
      • Ganz T.
      Hepcidin and iron regulation, 10 years later.
      ). Holo-TFN is distributed throughout the body and enters cells via transferrin-receptor-mediated endocytosis. FPN is also highly expressed in macrophages, including Kuppfer cells in the liver and red-pulp macrophages in the spleen. In both cases, FPN functions to release stored iron into the blood.
      Iron import into the body is regulated at the systems’ level by hepcidin, a small peptide hormone produced in the liver in response to excessive bodily iron (
      • Ganz T.
      Hepcidin and iron regulation, 10 years later.
      ). Hepcidin binds FPN, causing its internalization and subsequent hydrolytic degradation. Individuals with hemochromatosis, most commonly harboring a mutation in the HFE gene, generate insufficient hepcidin, resulting in excessive levels of FPN and thus excessive import of nutrient iron into the blood. This saturates TFN such that the concentration of apo-TFN available to receive newly imported iron is insufficient. The excessive iron released into the blood becomes NTBI.
      Despite being recognized to exist for a half-century, the chemical identity of NTBI remains unestablished (
      • Anderson G.J.
      • Bardou-Jacquet E.
      Revisiting hemochromatosis: genetic vs. phenotypic manifestations.
      ,
      • Chauhan W.
      • Shoaib S.
      • Fatma R.
      • Zaka-ur-Rab Z.
      • Afzal M.
      Beta-thalassemia and the advent of new interventions beyond transfusion and iron chelation.
      ,
      • Knutson M.D.
      Non-transferrin-bound iron transporters.
      ,
      • Bartnikas T.B.
      Known and potential roles of transferrin in iron biology.
      ,
      • Ganz T.
      Hepcidin and iron regulation, 10 years later.
      ,
      • Faber M.
      • Jordal R.
      Presence of 2 iron-transport proteins in serum.
      ,
      • Sarkar B.
      State of iron(III) in normal human serum: low molecular weight and protein ligands besides transferrin.
      ,
      • Hershko C.
      • Graham G.
      • Bates G.W.
      • Rachmilewitz E.A.
      Nonspecific serum iron in thalassemia: An abnormal serum iron fraction of potential toxicity.
      ,
      • Graham G.
      • Bates G.W.
      • Rachmilwitz E.A.
      • Hershko C.
      Nonspecific serum iron in Thalassemia – quantitative and chemical reactivity.
      ,
      • Batey R.G.
      • Fracp P.
      • Fong L.C.
      • Shamir S.
      • Sherlock S.
      a Non-Transferrin-bound serum iron in idiopathic hemochromatosis.
      ,
      • Evans R.W.
      • Rafique R.
      • Zarea A.
      • Rapisarda C.
      • Cammack R.
      • Evans P.J.
      • Porter J.B.
      • Hider R.C.
      Nature of non-transferrinbound iron: studies on iron citrate complexes and thalassemic sera.
      ,
      • Silva A.M.
      • Kong X.
      • Parkin M.C.
      • Cammack R.
      • Hider R.C.
      Iron(III) citrate speciation in aqueous solution.
      ,
      • Brissot Pierre
      • Loreal Olivier
      Iron metabolism and related genetic diseases: A cleared land, keeping mysteries.
      ,
      • Silva A.M.N.
      • Rangel M.
      The (Bio)Chemistry of Non-Transferrin-Bound Iron.
      ). One reason for this is that its two most predominant characteristics – accumulating in organs and susceptibility to chelation – are indirect and difficult to quantify. Iron accumulates in organs primarily as ferritin, but NTBI is a different species that is likely altered via reduction and ligand exchange as it enters the cell and is converted into ferritin. NTBI is typically defined operationally as the non-transferrin-bound iron in plasma that reacts with a particular chelator under prescribed concentrations and durations. However, the size of the NTBI pool in plasma is affected by these details. A second problem is that the concentration of NTBI in diseased plasma is not exceptionally high (1 – 10 μM) and although NTBI concentration in healthy individuals is lower, it is still detectable and significant. A third problem arises from the gradual and ambiguous “spillover” conditions required to generate NTBI; between 40% - 70% TFN saturation is reportedly sufficient for NTBI levels to increase (
      • Batey R.G.
      • Fracp P.
      • Fong L.C.
      • Shamir S.
      • Sherlock S.
      a Non-Transferrin-bound serum iron in idiopathic hemochromatosis.
      ,
      • Aruoma O.I.
      • Bomford A.
      • Polson R.J.
      • Halliwell B.
      Nontransferrin-bound iron in plasma from hemochromatosis patients – effect of phlebotomy therapy.
      ,
      • Breuer W.
      • Ronson A.
      • Slotki I.N.
      • Hershko C.
      • Cabantchik Z.I.
      The assessment of serum nontransferrin-bound iron in chelation therapy and iron supplementation.
      ). A fourth problem is that the aqueous redox and coordination chemistry of iron is complicated, and NTBI may be heterogeneous (
      • Silva A.M.
      • Kong X.
      • Parkin M.C.
      • Cammack R.
      • Hider R.C.
      Iron(III) citrate speciation in aqueous solution.
      ,
      • Breuer W.
      • Hershko C.
      • Cabantchik Z.I.
      The importance of non-transferrin bound iron in disorders of iron metabolism.
      ).
      Two basic types of experiments have contributed to our understanding of NTBI, but perhaps they have also hindered it. One type of experiment has been to monitor the fate of added radioactive 59Fe to sera or plasma. Radioactive iron binds preferentially and tightly to available apo-TFN, and the excess is concluded to be (or become) NTBI. This assumes that the only iron-binding ligand in sera/plasma is the NTBI ligand (symbolized :LNTBI) and that this ligand is present in excess. Actually, there are many potential ligands in plasma (
      • May P.M.
      • Linder P.W.
      Computer simulation of metal-ion equilibria in biofluids: Models for the low-molecular-weight complex distribution of Calcium(II), Magnesium (II), Manganese(II), Iron(III), Copper(II), Zinc(II) and Lead(II) ions in human blood plasma.
      ) and the added 59Fe may bind any or all of them.
      In the other type of experiment, a chelator is added to sera/plasma and the resulting Fe-chelator complex is assumed to arise from the binding of NTBI; the assumed general reaction is {Fe:LNTBI + chelator ⇄ Fe-chelator + :LNTBI}. The problem is that NTBI is destroyed during this reaction making it unlikely that such experiments can ultimately be used to identify NTBI. Moreover, the added chelator might also sequester iron that is bound to other non-NTBI species, overestimating the size of the NTBI pool.
      To avoid these problems, we and others (
      • Neu H.M.
      • Alexishin S.A.
      • Brandis J.E.P.
      • Williams A.M.C.
      • Li W.J.
      • Sun D.J.
      • Zheng N.
      • Jiang W.L.
      • Zimrin A.
      • Fink J.C.
      • Polli J.E.
      • Kane M.A.
      • Michel S.L.J.
      Snapshots of Iron Speciation: Tracking the Fate of Iron Nanoparticle Drugs via a Liquid Chromatography-Inductively Coupled Plasma-Mass Spectrometric Approach.
      ) have used LC-ICP-MS chromatography of untreated blood plasma to detect and characterize endogenous iron-containing complexes - without adding iron or chelators. Neu et al. (
      • Neu H.M.
      • Alexishin S.A.
      • Brandis J.E.P.
      • Williams A.M.C.
      • Li W.J.
      • Sun D.J.
      • Zheng N.
      • Jiang W.L.
      • Zimrin A.
      • Fink J.C.
      • Polli J.E.
      • Kane M.A.
      • Michel S.L.J.
      Snapshots of Iron Speciation: Tracking the Fate of Iron Nanoparticle Drugs via a Liquid Chromatography-Inductively Coupled Plasma-Mass Spectrometric Approach.
      ) detected FeIII(citrate) in human blood plasma by ESI-MS, supporting the conclusion that NTBI = FeIII citrate. Dzubia et al. (
      • Dziuba N.
      • Hardy J.
      • Lindahl P.A.
      Low-molecular-mass iron in healthy blood plasma is not predominately ferric citrate.
      ) detected low concentrations of low-molecular-mass (LMM) iron species in plasma from healthy humans, horses, mice, and pigs, but the chromatographic properties of these species largely differed from those of FeIII(citrate). Unexpectedly, the 10 kDa plasma ultrafiltrate (or FTS) from human hemochromatosis patients did not exhibit any additional iron-detected LC peaks relative to healthy controls. However, the patients had been treated for the disease, suggesting that their NTBI levels might have been too low to detect.
      Dzubia et al. (
      • Dziuba N.
      • Hardy J.
      • Lindahl P.A.
      Low-molecular-mass iron complexes in blood plasma of iron-deficient pigs do not originate directly from nutrient iron.
      ) hypothesized that the low concentration of LMM iron species observed in the earlier study was due to the quantitative removal of NTBI by the liver, and so they surgically implanted catheters in the portal vein of pigs that had been starved for iron. Intestinal blood passes through this vein to the liver, and so they anticipated that blood removed from it would contain high concentrations of NTBI. Blood was also removed from the caudal/cranial vena cava as a control. A bolus of 57Fe was injected into the stomach via a feeding tube, and blood samples were removed from both catheters at increasing times. Since only 2% of natural-abundance iron is due to 57Fe, the fate of the injected enriched 57Fe could be followed. Surprisingly, the LMM iron complexes did not become enriched in 57Fe; rather the injected 57Fe bound apo-TFN. Dzubia et al. (
      • Dziuba N.
      • Hardy J.
      • Lindahl P.A.
      Low-molecular-mass iron complexes in blood plasma of iron-deficient pigs do not originate directly from nutrient iron.
      ) concluded that the detected LMM iron complexes arise from internal stores rather than directly from nutrient iron.
      Building off those results, we hypothesized that NTBI might only be detectable using iron-overloaded genetically-modified animals. Here, we investigated this by examining HFE(-/-) (heretofore called HFE) mice fed iron-deficient diets supplemented with 0, 50, 500 and 5000 mg of natural abundance FeIII citrate. We selected the HFE gene because a mutation in it is the most common origin of the disease (
      • Barton J.C.
      • Edwards C.Q.
      • Acton R.T.
      HFE gene: Structure, function, mutations, and associated iron abnormalities.
      ). Using our LC-ICP-MS system, we expected to observe increasing levels of LMM iron complexes in the blood plasma filtrates of these sick animals, especially as the concentration of iron in the diet increased. Once again, our results were unexpected, but they prompted a new and intriguing hypothesis as to the chemical nature of NTBI. Scheme 1 outlines the chronology of our study.

      Results

      Our initial objective was to detect and characterize NTBI in the blood of HFE mice (and controls) raised on 0, 50, 500, and 5000 mg of iron per kg of iron-deficient chow. Some HFE mice were raised on 50 mg 57Fe/kg chow. Mice were sacrificed at various ages, blood was collected, and the plasma portion was filtered through a 10 kDa cut-off membrane, resulting in retentate and FTS fractions. A flow-chart showing an overview of the study is given in Figure S1.
      We initially focused on detecting NTBI in FTSs since we expected it to be a LMM species. FTSs of the plasma from both HFE and control mice were subjected to LC-ICP-MS chromatography using a size-exclusion column that could resolve species with masses ranging from 100 – 10,000 Da. In a typical experiment performed over the course of a day, 8 HFE or control mice were sacrificed, two from each diet group. Adult mice yielded only 0.5 – 1 mL of blood, and so blood of 2 mice from each group (3 for younger mice) were combined. After centrifugation, plasma fractions, representing ∼ 50% of total volume, were collected and filtered, yielding ∼ 70 μL of retentate and ∼ 400 μL of FTS.
      The iron concentration in HFE FTS (from six 12-16 wk mice) was modest; 1.9 ± 0.2 μM Fe. Plasma contains hundreds of mM of salt which fouled the ICP-MS instrument and reduced the detector response. Significant amounts of iron adsorbed onto the column, and subsequent cleaning had limited effectiveness. Small differences in daily tuning of the ICP-MS and changes in columns caused shifts in peak intensities and/or elution volumes. As a result, traces obtained on different days were not easily compared, so analyses were limited to assessing overall patterns within a group of traces obtained on the same day.
      Chromatograms of FTS from HFE and control mice exhibited 1 - 4 low-intensity iron peaks (Figure 1). Results of 3 separate experiments are shown, including from 3-wk HFE mice (Panel A), 12 wk HFE mice (Panel B), and 16 wk control mice (Panel C). Iron-detected traces from 4 additional experiments are shown in Figure S2, including FTS from 3 wk HFE (Panel B), 24 wk HFE (Panel C) 32 wk HFE (Panel A), and 24 wk control (Panel D) mice. Specific peaks within a group were nearly identical in terms of elution volumes and intensities. Peaks in SI experiments were more intense; however, some of that intensity was due to contaminating iron that had desorbed from the column with each sample injection. In the experiment of Figure S2, Panel D, we verified this by running a “ghost” column (i.e. peek tubing in place of the column) and measuring the area under the resulting iron peaks. Although this indicated substantial contaminating iron in LC traces, the nearly identical intensity of the ghost column peaks confirmed that the concentration of iron in the FTS did NOT increase proportionately with dietary iron. All such experiments indicated the same.
      Figure thumbnail gr1
      Figure 1Iron-detected Chromatograms of Plasma FTSs. Panel A: 3 wk HFE mice. Panel B: 12wk HFE mice. Panel C: 16 wk control mice. Food group of mice are shown in the figure.
      In the experiment of Figure S2, Panel C, an FeIII citrate standard run on the same day did not comigrate with the Fe peaks from plasma, supporting our previous conclusion (
      • Dziuba N.
      • Hardy J.
      • Lindahl P.A.
      Low-molecular-mass iron in healthy blood plasma is not predominately ferric citrate.
      ,
      • Dziuba N.
      • Hardy J.
      • Lindahl P.A.
      Low-molecular-mass iron complexes in blood plasma of iron-deficient pigs do not originate directly from nutrient iron.
      ) that FeIII citrate was not the dominant LMM Fe species in plasma FTS. The LMM iron species in the FTS of HFE (or control) mice were present at low concentrations and were similar regardless of whether the sample originated from HFE or control mice, and regardless of the iron level in the diet. Overall, these results forced us to conclude that plasma FTS from HFE mice do not accumulate a LMM form of iron even on a high-iron diet.
      With some reluctance, we shifted the search for NTBI to retentate fractions. Such fractions from 24 wk control and HFE mice were run on a different column which resolved species with masses between 10 kDa – 100 kDa (Figure 2, A and B). Matching ghost column traces (×0.1) are in red. Two major peaks were observed, which were assigned to ferritin (FTN) and TFN by running protein standards of horse spleen ferritin (Sigma) and human transferrin (Athens Research). The pattern in Figure 2A suggested that the TFN peak intensity in HFE retentates increased from 0 – 50 mg Fe and then remained relatively constant. The peak assigned to FTN increased gradually with dietary iron. For control retentates, the FTN peak also increased with dietary iron (Figure 2B). No such trend was evident for TFN. Other batches of retentate analyzed by LC-ICP-MS showed increasing TFN and FTN saturation with increasing Fe in the diet. Ghost column areas in Figure S3, Panel A indicated little contamination; however, 35-70% of sample iron was typically adsorbed on the column. Retentates from control and HFE mice were devoid of low-molecular-mass species (Figure S4). The intensity of the void volume peaks increased with increasing nutrient iron, especially for the HFE samples.
      Figure thumbnail gr2
      Figure 2Chromatograms of Plasma Retentates on high-mass column. Panel A: 24 wk old HFE mice. Flow rate was 0.6 ml/min. Matching ghost column traces (×0.1) are in red. Also shown are traces of ferritin and TFN (×25). Ferritin was from horse spleen which may have resulted in a slight shift in elution volume. Panel B: 24 wk old control mice. Flow rate was 0.5 ml/min.
      NTBI is commonly quantified by its reaction with DFO, and so we treated retentate samples with this chelator to assess the presence of NTBI in those fractions. Some HFE mice were raised on a “regular” diet containing ∼250 mg iron/kg chow. Retentate samples from 24 and 36 wks “regular’ mice (called R24 and R36) were diluted 80-fold and divided in halves. Half of each sample was treated with DFO and the other half was untreated. All four samples were run on the high-mass and ghost columns (Figure 3). A trace of an Fe-DFO standard was also collected. The dominant Fe-detectable peak for both R24 and R36 mice originated from TFN, and its intensity was similar for all four traces. Traces from the two samples that had been treated with DFO exhibited a second major peak which comigrated with the Fe-DFO standard. The intensity of these peaks did not increase at the expense of the TFN peak; rather the peaks simply appeared. We concluded that HFE retentates contain a “sticky” form of iron that adsorbed onto the column but was also chelatable by DFO. We will ultimately conclude that this material is NTBI. The corresponding ghost column peaks indicated that the R36 retentate contained more than twice as much iron as the R24 retentate, suggesting that the DFO-chelatable iron increased with age, as expected for NTBI.
      Figure thumbnail gr3
      Figure 3Chromatograms of retentates with/without DFO on high-mass column. Top, R36 (“regulars”, 36 wk) retentate was diluted 80× and treated with 10 μM DFO for 1 h; Second, R24 retentate treated similarly; Third, untreated R36 retentate diluted 80×; Fourth, untreated R24 retentate; Bottom, Fe-DFO standard (2 μM FeSO4 + 10 μM DFO). Corresponding ghost column traces (0.5×) are in red. A diagram outlining this experiment is given in .
      Citrate is the most popular candidate for the NTBI ligand, and so we performed experiments to explore this possibility. We attempted to remove LMM species from plasma, including FeIII citrate, by filtering plasma from 36-wk HFE “regular” mice using a 10 kDa cutoff membrane, and washing the retentate twice with water. The retentate was then divided in two, half was treated with citrate, and half was untreated. Both halves were passed through the high-mass column. In both traces, the dominant peak was TFN (Figure 4 Panel A). Quantification of the ghost-column trace indicated that 40% of the iron in the washed untreated retentate sample was not bound to TFN and was not detected in the trace. This undetected form of iron must have adsorbed to the column (ultimately, we will assign it to NTBI). The citrate-treated sample exhibited new peaks in the LMM region (at ∼ 22 mL) with approximately the intensity expected if citrate coordinated the high-mass “sticky” (NTBI) iron that was undetected in the other trace. We write this as {FeIII:LNTBI + citrate → FeIIIcitrate + :LNTBI}.
      Figure thumbnail gr4
      Figure 4Chromatograms of Plasma Retentates and FTSs with added citrate on high-mass (A) and low-mass (B) columns. Panel A: washed and diluted plasma retentate from 36 wk old “regular” HFE mice. Top chromatogram, after treatment with 50 μM Na citrate; bottom before treatment. Panel B: FTS from combined 24-36 wk 0, 50, 500 and 5000 mg HFE mice. Top chromatogram, after adding 2 μM FeIII citrate. Middle chromatogram, before treatment. Bottom trace, FeIII citrate standard (2 μM FeSO4 + 20 μM Na citrate).
      Two similar experiments using plasma FTSs were analyzed using the low-mass column. In one experiment, FTSs from plasma of 6 mice were combined, and then divided in two. Half was treated with an FeIII citrate solution (which contained an excess of citrate) and half was untreated. An FeIII citrate standard was also run. The trace of the untreated half exhibited ∼ 3 resolvable peaks at 16 – 21 mL elution volume plus a broad peak at longer elution volumes (Figure 4B, middle trace). The trace of the treated sample (upper trace) was qualitatively similar but more intense. It included a peak that nearly comigrated with the FeIII citrate standard (perhaps the slight shift was due to the salt present only in the FTS sample), but the intensity of the “FeIII citrate” peak was modest relative to the increased intensities of the other peaks in the trace. (Note: the elution volume of FeIII citrate that migrated through the low-mass column differed from that through the high-mass column.) We concluded that most (∼ 90%) of the iron from the added FeIII citrate underwent ligand-exchange with other species in the FTS. This suggested that FeIII citrate is not the most stable iron complex formed when citrate is added to plasma; much of it converts to other forms resulting in a distribution of species.
      If NTBI were present in retentate fractions rather than in FTSs, we realized that it might be possible to detect it using Mössbauer spectroscopy. With a limited amount of 57Fe-enriched plasma available from a recent study (

      Vali, S.W. and Lindahl, P.A. (2022) Mössbauer spectroscopic characterization of organs from HFE(-/-) mice: relevance to hereditary hemochromatosis. In preparation.

      ), we decided to combine plasma from the 4 youngest HFE mice that were available (4, 6, 10, and 14 wks), concentrate the sample using a 10 kDa cutoff membrane, and load it into a MB cup for analysis. We did the same for plasma from 4 older HFE mice (18, 24, 32, and 52 wks). Since no 57Fe-enriched plasma from control mice were available to serve as a control, we anticipated that NTBI might be present in greater amounts in the old mice sample.
      The iron concentrations in the young and old retentates (after removing residual red-blood cell contributions) were 96 ± 6 μM and 136 ± 5 μM, respectively. Since the retentates were concentrated ∼ 5-fold for this experiment, we estimate that that the plasma iron concentration would be ∼ 20 μM in young and ∼ 27 μM in old plasma, similar to our previous values (
      • Dziuba N.
      • Hardy J.
      • Lindahl P.A.
      Low-molecular-mass iron in healthy blood plasma is not predominately ferric citrate.
      ,
      • Dziuba N.
      • Hardy J.
      • Lindahl P.A.
      Low-molecular-mass iron complexes in blood plasma of iron-deficient pigs do not originate directly from nutrient iron.
      ).
      The dominating features in the raw 6 K 0.05 T MB spectra (Figure S5) were quadrupole doublets originating from contaminating deoxy and oxy FeII hemoglobin; these were simulated and removed. The resulting difference spectra were dominated by a magnetic feature which was simulated using the spin Hamiltonian parameters of di-ferric TFN (
      • Kretchmar S.A.
      • Teixeira M.
      • Huynh B.H.
      • Raymond K.N.
      Mössbauer studies of electrophoretically purified monoferric and diferric human transferrin.
      ). The only remaining spectral feature (Figure 5, A and B respectively) was narrow absorption in the central region. Spectral noise was significant even after > 200 hr of data collection. Spectral noise was too severe to firmly assign the absorption in the central region, but it could be approximately simulated by a quadrupole doublet with δ = 0.5 ± 0.1 mm/s and ΔEQ = 0.5 ± 0.1 mm/s. The parameters approximated those of FeIII aggregates such as the oxyhydroxide phosphate associated nanoparticles found in diseased mitochondria (δ = 0.52 mm/s and ΔEQ = 0.63 mm/s; Ref 25). The intensity of this material was ∼ 2× higher in the spectrum obtained from old vs. young HFE mice. In animals with iron-overload diseases, NTBI concentration increases with age (
      • Lee S.M.
      • Loguinov A.
      • Fleming R.E.
      • Vulpe C.D.
      Effects of strain and age on hepatic gene expression profiles in murine models of HFE-associated hereditary hemochromatosis.
      ).
      Figure thumbnail gr5
      Figure 55.5 K 0.05 T Mössbauer Spectra of Plasma Retentates from Young vs. Old 57Fe-enriched HFE Mice. A, young; B, old; C, 150 μM 57FeCl3 added to blood plasma from 24 wk mice fed on regular diet not enriched in 57Fe. The red lines in A and B are simulations for the feature due to transferrin (S = 5/2; D = -0.25 cm-1; E/D = 0.30; ΔEQ = 0.40 ± 0.02 mm/s; δ = 0.56 ± 0.02 mm/s; η = -1.5; Aiso = -30.18 MHz; Γ = 0.70 mm/s. The red line in C is a simulation assuming ΔEQ = 0.60 mm/s, δ = 0.55 mm/s, and Γ = 0.45 mm/s.
      Based on these observations, we hypothesized that NTBI was an FeIII aggregate in plasma. We considered that such aggregates formed due to the oxidizing, salty, pH-neutral conditions of plasma. To consider this further, we added acidic 57FeIII ions dropwise to blood plasma (to 250 μM) from 24 wk HFE mice that had been fed a regular natural-abundance-iron diet. The resulting spectrum (Figure 5C) exhibited a broad quadrupole doublet again typical of 57FeIII aggregates (simulated in red). The minor magnetic species evident from the baseline was probably from added 57Fe that bound apo-TFN. We next prepared a “pseudo plasma FTS” solution and added acidic 57FeIII in similar fashion. The corresponding MB spectrum (Figure 6D) was once again a quadrupole doublet with similar parameters (δ = 0.55 ± 0.02 mm/s and ΔEQ = 0.60 ± 0.03 mm/s). In contrast, control spectra of acidic FeIII, FeIII citrate, and FeIII-DFO were typical of magnetically isolated non-aggregated mononuclear high-spin FeIII (Figure 6, A-C).
      Figure thumbnail gr6
      Figure 65.5 K 0.05 T Mössbauer spectra of iron standards and pseudo-plasma. A, 250 μM acidic 57FeIII; B, same as A but mixed with 150 μM sodium citrate; C, same as A but mixed with 500 μM DFO; D, same as A but mixed with a pseudo-plasma salts; E, same as D plus 0.2 g/L bovine serum albumin; F, same as D plus 150 μM citrate; G, same as D plus 400 μM DFO (affording 200 μM 57Fe). The red line indicates baseline and the expanded region highlights the magnetic features.
      Albumin and citrate are popular candidates for NTBI ligands and both are present in plasma; including either candidate in pseudo-plasma FTS had no effect on the resulting Mössbauer spectra (Figure 6, E and F). In another experiment, we added DFO (Millipore) to the pseudo-plasma FTS that contained the FeIII aggregate. In this case, the resulting Mössbauer spectra (Figure 6G) revealed some coordination of FeIII to DFO indicating that the aggregate can be partially broken up by the DFO chelator. Perhaps additional coordination would have been observed if incubation conditions had been optimized.
      To determine whether aggregated iron in the pseudo plasma behaved like the retentate of mice plasma, we filtered the sample of Figure 6E using the 10 kDa cutoff membrane. The resulting retentate was diluted, passed through the SEC column, and the resulting LC traces were analyzed like those of mice plasma retentates. All the aggregated FeIII evident in the ghost trace shown in red adsorbed onto the column (Figure 7, Panel A). Thus, the aggregated FeIII material prepared in pseudo-plasma was “sticky”, a property shared with NTBI.
      Figure thumbnail gr7
      Figure 7LC-ICP-MS analysis of 57FeIII aggregate in pseudo-plasma. Panel A, retentate from sample used in E after being diluted 80× and analyzed using the high-mass column. Red chromatogram, same sample run through a ghost column. Panel B, trace: i, 150 μM sodium citrate run on a cleaned column; ii, 3 μM 57FeIII citrate (150 μM citrate); iii, 3 μM of 57FeIII plus pseudo-plasma salts plus 150 μM sodium citrate; iv, 3 μM 57FeIII plus pseudo-plasma salts; v, same as i but after the other samples were run. Samples were incubated aerobically for 1 h prior to analysis.
      This property was modestly affected by including 150 μM sodium citrate in pseudo plasma salts containing 3 μM Fe. After incubating the samples 1 hr aerobically, samples with and without citrate were analyzed by LC-ICP-MS (Figure 7, Panel B). In the absence of citrate, all of the Fe adsorbed on the column while in its presence, a peak appeared representing ∼27% of the absorption area of an Fe-citrate standard containing the same concentration of Fe. These results suggested that citrate at physiological concentration might coordinate as much as 27% the iron that would otherwise form the FeIII aggregate. However, sodium citrate by itself chelated some adsorbed iron from the column, so 27% is an upper limit. In the same experiment sodium citrate was passed through the column before (Citrate 1) and after (Citrate 2) the other samples. The higher intensity after the other samples were run reflected the adsorption of iron from those other samples and the ability of citrate to remove some of the adsorbed iron. Thus, some of the 27% FeIII citrate peak intensity was due to citrate’s ability to remove iron from the column.
      Higher concentrations of citrate were able prevent FeIII aggregation to a greater degree. When an extremely high concentration citrate (2500 μM, ca. 25× physiological) was included in pseudo plasma, followed by 250 μM 57FeIII, Mössbauer spectra showed that the added iron did not aggregate (Figure S6). When 500 μM citrate (5× physiological) was used, some 57FeIII aggregated and some formed an uncharacterized magnetic species (perhaps an intermediate). We are uncertain whether the products formed in these experiments were kinetically or thermodynamically controlled, but they indicate the variety of outcomes depending on the concentration of citrate present in pseudo (and real) plasma. Under physiological conditions (ca. 100 μM citrate), the FeIII aggregate will dominate or exclusively constitute the non-transferrin non-ferritin (nonproteinaceous) fraction of plasma iron.
      Finally, we thawed the MB samples used to generate Figure 6, B, E, and G, and transferred them to EPR tubes. The FeIII citrate sample exhibited a g = 4.3 EPR signal (Figure 8, Panel A, top spectrum), similar to those previously reported (
      • Evans R.W.
      • Rafique R.
      • Zarea A.
      • Rapisarda C.
      • Cammack R.
      • Evans P.J.
      • Porter J.B.
      • Hider R.C.
      Nature of non-transferrinbound iron: studies on iron citrate complexes and thalassemic sera.
      ,
      • Simpson R.J.
      • Cooper C.E.
      • Raja K.B.
      • Halliwell B.
      • Evans P.J.
      • Aruoma O.I.
      • Singh S.
      • Konijn A.M.
      Non-transferrin-bound iron species in the serum of hypotransferrinaemic mice.
      ). Such signals arise from S = 5/2 states with rhombic symmetry. In contrast, the FeIII aggregate was EPR-silent (Figure 8, Panel A, bottom spectrum), as expected for magnetically-interacting FeIII nanoparticles. The spectrum of the DFO-treated sample (Figure 8, Panel A, middle spectrum) exhibited a g = 4.3 signal that was about half the intensity of the FeIII citrate signal, consistent with the corresponding MB spectrum (Figure 6G). All three samples contained 250 μM iron.
      Figure thumbnail gr8
      Figure 8X-band EPR spectra of plasma-related iron species. Panel A: top, FeIII citrate standard from B; middle, FeIII in pseudo-plasma treated with DFO, from G; bottom, FeIII in pseudo-plasma, from E. Parameters: microwave frequency, 9.354 GHz; temperature, 11K; microwave power, 0.02 mW, modulation amplitude, 10 G; modulation frequency, 100 kHz; sweep time, 120s; 4 scans averaged. Panel B; EPR spectra of sera isolated from normal and hypotransferrinemic mice. Taken from Simpson et al. (
      • Simpson R.J.
      • Cooper C.E.
      • Raja K.B.
      • Halliwell B.
      • Evans P.J.
      • Aruoma O.I.
      • Singh S.
      • Konijn A.M.
      Non-transferrin-bound iron species in the serum of hypotransferrinaemic mice.
      ) with permission from the publisher.

      Discussion

      In the 1960’s, all nonheme iron in serum was generally thought to be bound to TFN, though some experiments were suggesting otherwise (
      • Faber M.
      • Jordal R.
      Presence of 2 iron-transport proteins in serum.
      ). In 1970, Sarkar (
      • Sarkar B.
      State of iron(III) in normal human serum: low molecular weight and protein ligands besides transferrin.
      ) titrated sera with 59FeCl3, and removed the TFN-bound fraction by ultracentrifugation. Early in the titration, essentially all added 59Fe bound to apo-TFN, but once this protein became saturated, a significant portion of 59Fe remained in the protein-free supernatant. This suggested that sera contained a LMM ligand (:LNTBI) capable of binding the excess added 59Fe. After attempting to remove this ligand by dialysis, the titration behavior could only be replicated by adding citrate to dialyzed sera, highlighting the possibility that citrate was the sought-after :LNTBI ligand. Viewed from our current perspective, the results of Sarkar imply that endogenous NTBI is a high-mass species that can be mobilized by coordination with citrate to form a low-mass complex. Sarkar suggested that the high-mass species was/were iron-binding proteins other than TFN. Although not considered at the time, an FeIII aggregate would also be consistent with his results.
      In 1978, Hershko et al. (
      • Hershko C.
      • Graham G.
      • Bates G.W.
      • Rachmilewitz E.A.
      Nonspecific serum iron in thalassemia: An abnormal serum iron fraction of potential toxicity.
      ) connected NTBI to iron-overload diseases. They detected a chelatable form of iron (2-7 μM), not bound to TFN, in sera of patients with β-thalassemia that was absent in controls. They found that the NTBI concentration was related to the degree of TFN saturation past a threshold, and that NTBI could be pried from its ligand with strong chelators or apo-TFN. A year later, Graham et al. (
      • Graham G.
      • Bates G.W.
      • Rachmilwitz E.A.
      • Hershko C.
      Nonspecific serum iron in Thalassemia – quantitative and chemical reactivity.
      ) quantified NTBI by adding EDTA to sera from thalassemia patients and passing the solution through a 25 kDa cutoff membrane. In the absence of EDTA, the FTS lacked NTBI, again suggesting that NTBI was a high-mass iron species that could not pass through the membrane unless mobilized by a chelator. Based on these considerations, Graham et al. concluded, like Sarkar, that NTBI was “bound to serum proteins”. However, their results are also consistent with NTBI being a high-molecular-mass nonproteinaceous FeIII aggregate that could be mobilized by chelators like EDTA, citrate, DFO etc.
      In 1980, Batey et al. (
      • Batey R.G.
      • Fracp P.
      • Fong L.C.
      • Shamir S.
      • Sherlock S.
      a Non-Transferrin-bound serum iron in idiopathic hemochromatosis.
      ) reported that NTBI adheres strongly to DEAE-Sephadex chromatography columns, and used this “stickiness” to determine the percentage of serum iron that was NTBI. They added 59FeIII citrate which preferentially bound apo-TFN, but once TFN was saturated, the added 59FeIII citrate adhered to the column (perhaps some of the added complex converted to 59FeIII aggregates).
      Many subsequent NTBI studies focused on standardizing an NTBI assay and quantifying the concentration of the NTBI pool, as obtained by incubating sera or plasma with various chelators under different reaction conditions (
      • de Swart L.
      • Hendriks J.C.M.
      • van der Vorm L.N.
      • Cabantchik Z.I.
      • Evans P.J.
      • Hod E.A.
      • Brittenham G.M.
      • Furman Y.
      • Wojczyk B.
      • Janssen M.C.H.
      • Porter J.B.
      • Mattijssen V.E.J.
      • Biemond B.J.
      • MacKenzie M.A.
      • Origa R.
      • Galanello R.
      • Hider R.C.
      • Swinkels D.W.
      Second international round robin for the quantification of serum non-transferrin bound iron and labile plasma iron in patients with iron-overload disorders.
      ). Other studies investigated the NTBI = FeIII citrate hypothesis. In 1987, Craven et al. (
      • Cravens C.M.
      • Alexander J.
      • Eldridge M.
      • Kushner J.P.
      • Bernstein S.
      • Kaplan J.
      Tissue distribution and clearance kinetics of non-transferrin-bound iron in the hypotransferrenemic mouse: A rodent model for hemochromatosis.
      ) injected 59Fe citrate into genetically-modified hypotransferrinemic mice (which have low TFN levels) and controls. They found that in these mutant mice, nearly all of the injected 59Fe that was absorbed by the body localized in the liver and pancreas. This reinforced the idea that NTBI was FeIII citrate. Perhaps some of the added FeIII citrate may have been converted to an FeIII aggregate once in the blood.
      Our currents results reinforce our earlier conclusion that the FTS of blood from both healthy and hemochromatosis individuals contain little iron that could be NTBI (
      • Dziuba N.
      • Hardy J.
      • Lindahl P.A.
      Low-molecular-mass iron in healthy blood plasma is not predominately ferric citrate.
      ,
      • Dziuba N.
      • Hardy J.
      • Lindahl P.A.
      Low-molecular-mass iron complexes in blood plasma of iron-deficient pigs do not originate directly from nutrient iron.
      ). Although the source of iron giving rise to these minor species remains unknown, our current results indicate that the concentrations of these species are independent of the concentration of iron in the diet. The idea that NTBI might not be a LMM species was prompted by our realization that only the column used to run retentate fractions was becoming fouled with excess iron. Unfortunately, this realization occurred towards the end of the study, limiting the follow-up experiments that could be performed.
      Probing Fe aggregation in pseudo plasma using LC-ICP-MS showed that citrate at high concentrations can inhibit formation of the FeIII aggregate. However, the required citrate concentration is higher than the typical range of citrate concentrations in blood plasma. This suggests an equilibrium {FeIII aggregate + citrate ⇄ FeIIIcitrate + nH2O + other nonproteinaceous ligands}, with the FeIII aggregate dominating under physiological conditions. Whether the distribution (FeIII aggregate vs FeIII citrate) is kinetically or thermodynamically controlled in blood plasma remains uncertain. A similar equilibrium involving transferrin {FeIII aggregate + apo-TFN → TFN + nH2O etc} is probably also likely, but with rapid kinetics and thermodynamics favoring products.
      According to our model (Figure 9) bare FeII ions enter the blood through FPN and are quickly hydrated and oxidized by hephaestin or ceruloplasmin. A similar result would obtain if exogenous FeII or FeIII salts were injected into the blood via syringe. If apo-TFN were available, the newly formed FeIII ions preferentially coordinate to this protein, but when none is available (saturated or nearly so) most will aggregate to form a high-molecular-mass species that is retained by 10 kDa cutoff membranes and adheres readily to chromatography columns. Entry of NTBI into bodily cells requires pre-reduction to the FeII state followed by entry via Zip14, DMT1, or other FeII importers (
      • Knutson M.D.
      Non-transferrin-bound iron transporters.
      ).
      Figure thumbnail gr9
      Figure 9Model for the formation and metabolism of NTBI. See text for details. HFN, hephaestin; CPN, ceruloplasmin; FPN, ferroportin; TFN, transferrin.
      Grootveld et al. (
      • Grootveld M.
      • Bell J.D.
      • Halliwell B.
      • Aruoma 0. I.
      • Bomford A.
      • Sadler P.J.
      Non-transferrin-bound iron in plasma or serum from patients with idiopathic hemochromatosis.
      ) is typically cited as strong evidence that NTBI = FeIII citrate. However, we find this conclusion only partially supported by their data. These authors examined plasma and plasma ultrafiltrates (equivalent to our FTSs) from hemochromatosis patients and healthy controls, using NMR spectroscopy and reverse-phase HPLC. Hemochromatosis plasma contained an average of 370 ± 60 μM citrate, ∼ 3× higher than in healthy controls. They detected corresponding NMR spectra of hemochromatosis plasma with signals due to unbound citrate (high-spin FeIIIcitrate is undetectable due to paramagnetic line broadening). They reported that the NTBI concentration in such plasma was ∼ 5 μM Fe, and that adding DES resulted in “small increases” in citrate NMR intensities. They assumed that this increase arose from the chelation reaction {FeIIIcitrate + DES → FeIIIDES + citrate}. However, according to concentrations that they reported, complete coordination of citrate in hemochromatosis plasma would have only increased the NMR citrate peak intensities by < 2%, well within reported uncertainties. Also, DES treatment affected the intensity of other NMR signals (e.g. from acetate) to an even greater extent. This implies that the observed increase in NMR intensity was due to causes other than the suggested reaction, and they do not lend support to the NTBI = FeIIIcitrate hypothesis. The NMR spectra actually imply that most citrate in plasma is not coordinated by FeIII.
      In another experiment, Grootveld et al. added 100 – 500 μM FeCl3 to plasma in a titration monitoring the free citrate NMR peaks and observed a gradual broadening of these peaks. They reasonably concluded that broadening arose from the binding of the added FeIII to free citrate, but the concentration of Fe used was 20 - 100× higher than physiological, and so does not seem relevant to actual conditions in blood.
      They also performed LC studies of plasma and ultrafiltrates (equivalent to our FTS). In one LC experiment (Fig 7, ref 30), a peak due to ferrioxamine (FeIII-DES) was obtained when hemochromatosis plasma ultrafiltrate was treated with DES, from which they reasonably concluded that DES mobilizes (and coordinates) NTBI. However, their results do not impact whether NTBI = FeIII citrate.
      In other LC experiments, plasma FTS from healthy controls exhibited a minor shoulder (at 8.25 min; Fig. 4A, ref 30) that comigrated with FeIII citrate, and its intensity increased slightly in traces of hemochromatosis FTS or in control FTS when FeCl3 was added (Fig. 4C, ref 30). These results support the NTBI = FeIII citrate hypothesis, but concerns remain. They used a column in which polar hydrophilic species like FeIII citrate are eluted unresolved at the void volume possibly along with other 340 nm-absorbing species. Thus, the intensity of the peak at 8.25 min may not be solely due to FeIII citrate. Second, their elution profiles of plasma FTS exhibited numerous features with intense “negative absorption” which is physically impossible; the problem was not mentioned.
      May et al. (
      • May P.M.
      • Linder P.W.
      Computer simulation of metal-ion equilibria in biofluids: Models for the low-molecular-weight complex distribution of Calcium(II), Magnesium (II), Manganese(II), Iron(III), Copper(II), Zinc(II) and Lead(II) ions in human blood plasma.
      ) is another oft-cited study supporting the NTBI = FeIII citrate hypothesis. In their recent review, Silva and Rangel (
      • Silva A.M.N.
      • Rangel M.
      The (Bio)Chemistry of Non-Transferrin-Bound Iron.
      ) correctly reported that May et al. predicted that 99% of iron in plasma was associated with citrate (we assume they meant 99% of non-transferrin-bound plasma iron). However, left unmentioned was that May et al also predicted that the total low-molecular-weight iron concentration in plasma should be 6.68×10-13 M (Table 5 of Ref. 18), ca. 7 orders of magnitude lower than the estimated size of the NTBI pool.
      The tendency of ferric ions to form high mass aggregates in aqueous solutions at neutral pH is well known. Such nanoparticles can have masses of ∼ 150 kDa (Ref 11 and references therein). Using the Ksp of ferric hydroxide and Keq’s for competing iron binding reactions, Evans et al. found that FeIII hydroxide dominates under equilibrium conditions. However, they found that the precipitation reaction was slow and concluded that the aggregate would not dominate in plasma where product distributions would be kinetically control. They considered that the high mass NTBI species might be FeIII bound to serum albumin, but ultimately concluded that binding was weak and easily disrupted. A later study by Silva and Hider (
      • Silva A.M.N.
      • Hider R.C.
      (2009) Influence of non-enzymatic post-translation modifications on the ability of human serum albumin to bind iron Implications for non-transferrin-bound iron speciation.
      ) suggested the NTBI might be a heterogenous mixture of FeIII:albumin and FeIII citrate. Silva and Rangel (
      • Silva A.M.N.
      • Rangel M.
      The (Bio)Chemistry of Non-Transferrin-Bound Iron.
      ) suggested that FeIII in plasma might bind non-specifically to the 98 carboxylate groups on the surface of albumin. Their model of NTBI assumes binding to citrate and serum albumin in a fine-tuned equilibrium depending on FeIII concentration, the iron:citrate molar ratio, and albumin post-translational modifications.

      The NTBI = FeIII aggregate hypothesis:

      Our results and evaluation of the NTBI literature support the idea that NTBI is an FeIII aggregate composed of magnetically interacting FeIII ions. Further studies are required to determine the exact chemical composition of this material. Although the size of the aggregated particles is also uncertain, they appear large enough to be retained by a 10 kDa cutoff membrane filter. These aggregates also adhere to our LC column and can be mobilized by FeIII chelators like DFO or citrate (but only at high concentrations). We suspect that the nanoparticles dissolve upon reduction, allowing FeII chelators to become effective.
      Our last line of evidence for this hypothesis comes from EPR spectra of sera from WT and hypotransferrinemic mice (
      • Simpson R.J.
      • Cooper C.E.
      • Raja K.B.
      • Halliwell B.
      • Evans P.J.
      • Aruoma O.I.
      • Singh S.
      • Konijn A.M.
      Non-transferrin-bound iron species in the serum of hypotransferrinaemic mice.
      ) in conjunction with spectra of FeIII citrate and other standards (
      • Evans R.W.
      • Rafique R.
      • Zarea A.
      • Rapisarda C.
      • Cammack R.
      • Evans P.J.
      • Porter J.B.
      • Hider R.C.
      Nature of non-transferrinbound iron: studies on iron citrate complexes and thalassemic sera.
      ). Spectra of sera from WT mice are dominated by a g = 4.3 signal which arises from high-spin FeIII ions bound to TFN that are not magnetically interacting. In Figure 8, panel B, top spectrum, we have reproduced the published spectrum of Simpson et al. (
      • Simpson R.J.
      • Cooper C.E.
      • Raja K.B.
      • Halliwell B.
      • Evans P.J.
      • Aruoma O.I.
      • Singh S.
      • Konijn A.M.
      Non-transferrin-bound iron species in the serum of hypotransferrinaemic mice.
      ). Similar g = 4.3 signals are observed for FeIII citrate, FeIII albumin, and FeIII DFO. In contrast, EPR spectra of sera from hypotransferrinaemic mice, which lack TFN, are devoid of a g = 4.3 signal (
      • Simpson R.J.
      • Cooper C.E.
      • Raja K.B.
      • Halliwell B.
      • Evans P.J.
      • Aruoma O.I.
      • Singh S.
      • Konijn A.M.
      Non-transferrin-bound iron species in the serum of hypotransferrinaemic mice.
      ), identical to what we obtained using an FeIII aggregate sample (Figure 8). Their spectrum is reproduced in Figure 8 Panel B, bottom. Simpson et al (
      • Simpson R.J.
      • Cooper C.E.
      • Raja K.B.
      • Halliwell B.
      • Evans P.J.
      • Aruoma O.I.
      • Singh S.
      • Konijn A.M.
      Non-transferrin-bound iron species in the serum of hypotransferrinaemic mice.
      ) suggested “that the majority of the iron (in sera of these TFN-deficient mice) is polynuclear” as would be the case for our NTBI = FeIII aggregate hypothesis. The lack of g = 4.3 signal also excludes the presence of (high concentrations of) FeIII citrate or FeIII albumin in sera; such species would also exhibit g = 4.3 signals. Evans et al. (
      • Evans R.W.
      • Rafique R.
      • Zarea A.
      • Rapisarda C.
      • Cammack R.
      • Evans P.J.
      • Porter J.B.
      • Hider R.C.
      Nature of non-transferrinbound iron: studies on iron citrate complexes and thalassemic sera.
      ) found that FeIII citrate solutions were devoid of g = 4.3 signals when the Fe:citrate ratio was 1:10 and 1:100, but that the signal developed when ratios were increased to 1:1000 and then 1:10,000. They concluded that “oligomeric and polymeric species containing oxygen-bridged iron(III) ions” were present at low iron:citrate ratios, and that this was converted to magnetically isolated complexes at higher ratios. This suggests that FeIII aggregates shift to mononuclear FeIII citrate complexes as the concentration of citrate increases. All of these past results support and confirm the results presented here as well as our hypothesis regarding the nature of NTBI. Additional studies are being planned to test our hypothesis further. We hope that these insights and the eventual chemical identification of NTBI will help in developing new treatments for iron-overload diseases.

      Experimental Procedures

      All procedures involving mice were approved by the Animal Use Committee at Texas A&M University (AUP 2018-0204). HFE mice (stock number 017784, B6.129S6-Hfe<tm2Nca>/J) and control mice (C57BL/6J) were purchased from The Jackson Laboratory (www.jax.org). Animals were housed in the LAAR facility in the School of Veterinary Medicine at Texas A&M University. Mice were raised in disposable all-plastic cages (Innovive model MVX1) containing synthetic bedding (Alpha-Dri Irradiated; Lab Supply, Houston) and all-plastic water bottles. Room temperature was 74 ± 2 °F. Lighting was on a 12/12 hr cycle. Mice were initially bred on an iron-deficient mouse diet (TD.80396.PWD; www.envigo.com) spiked with 0, 50, 500, and 5000 mg of FeIII(citrate) per kg chow. Each diet was prepared by weighing out the appropriate mass of FeSO4 dissolved in solutions of 10× excess sodium ascorbate and mixed with the Fe deficient chow powder. The resulting moistened material was pelleted manually by pressing into a plastic pipe with a tight-fitting glass rod. Pellets were baked in a glass pan at 75 oC for 4-7hrs and then refrigerated until used. Mice were offered food and distilled water ad libitum. Only breeding pairs given the 50 mg/Kg diet bred successfully, so upon weaning, mice from that food group were switched to a designated diet to balance the study.
      Animals ranging from 3 – 52 wks were transported to the Chemistry department at TAMU where they were sacrificed. Mice were anesthetized by injecting ketamine (5 mg/20 gm mouse) and xylazine (1 mg/20 gm mouse) subcutaneously. Exsanguination was by cardiac puncture once tests for pain (foot-pad squeeze) showed no response. Between 0.5 – 1.2 mL of blood was removed from each animal. Blood samples from 2 animals grown on the same diets were routinely combined. Samples were spun by centrifugation (2500×g for 15 min using a Sorvall RC centrifuge) and imported into a refrigerated N2-atmosphere glove box (Mbraun Labmaster 130) containing 1 – 20 ppm O2. Plasma was collected by syringe, transferred to epi-tubs, removed from the box, and stored at -80 °C.
      LC-ICP-MS: FTSs and retentate fractions were obtained from plasma using Amicon Ultra filters (2 ml) 10 kDa filters (Millipore). 500 μL plasma sample was loaded onto the activated filter and centrifuged at 2500×g until ∼ 400 μL solution was collected as the FTS (25 – 60 min). The retained fraction (retentate) was diluted using 20 mM ammonium acetate buffer to the original volume of the plasma. FTSs and retentates were passed through a 0.2 μm filter (to remove red blood cells and any debris), and then on LC-ICP-MS system which consisted of an Agilent Bioinert LC housed in an anaerobic glovebox (Mbraun Labmaster) at 5-10 oC interfaced with an online ICP-MS (Agilent 7700x). Two size exclusion chromatography columns were used, including Superdex 200 C and Superdex Peptide 10/300 GL (Cytiva). These will be referred to as the “high-mass” and “low-mass” columns, respectively. Columns were equilibrated with filtered and degassed 20 mM ammonium acetate pH 6.5 at a flow rate of 0.6 ml/min. In later experiments, retentate fractions were diluted 80× using 20 mM ammonium acetate and then analyzed by LC-ICP-MS. 56Fe and/or 57Fe were detected in He collision mode with a dwell time of 0.1 s. A ‘ghost column’ composed of PEEK tubing replaced actual columns to assess the total iron present in the samples and indirectly the portion adsorbed on actual columns.
      Mössbauer Spectroscopy: Retentate fractions from HFE mice raised on diets containing 50 mg of 57Fe-enriched per kg chow were combined based on age. The “young” sample combined retentates from 4, 6, 10, and 14 wks old mice; the “old” sample combined retentates from 18, 24, 32, and 48 wks. Each sample (∼ 2 mL) was concentrated using 10 kDa cutoff Amicon Ultra filters to a final volume of 400-500 μL. These were transferred to MB cups and frozen in LN2 until analyzed. Unlike the LC-ICP-MS experiments, MB samples were not passed through a 0.2 μm filter and so they had minor red-blood cell contamination.
      Pseudo plasma was prepared as a 2× stock of plasma salts, including 5 mM potassium chloride (all final 1× concentrations), 28 mM sodium bicarbonate, 1.45 mM potassium phosphate, 0.1g/L calcium chloride, 1 mM magnesium chloride, and 112 mM sodium chloride, adjusted to pH 7.4 with HCl. For some preparations, bovine serum albumin (Pierce) and sodium citrate were added at final concentrations of 0.2 g/L and 150 μM respectively. Acidic 57FeIII was prepared by dissolving 1.00 g of 57Fe2O3 (Cambridge Isotope Laboratories; 95.5% enrichment) in 3-5 mL of concentrate HCl followed by dilution to 40 mM final Fe concentration. Solutions were stored at 4 oC until used to prepare samples. The 57Fe stock was mixed with pseudo-plasma salts, yielding 1× plasma salts and 250 μM 57Fe final concentrations. Samples were frozen in Mӧssbauer cups after a 1 hr incubation on the bench top at room temperature. Mӧssbauer spectra were collected on an MS4 WRC spectrometer (SEE Co., Edina, MN) at 5-6 K with a field of 0.05 T applied parallel to the gamma radiation. Instruments were calibrated at room temperature with α-Fe foil. Spectra were simulated using WMOSS (http://www.wmoss.org/) software.
      Metal analysis using ICP-MS: FTS from 6 mice were used to prepare 3 replicates of samples by combining 2 FTS at a time. The samples were digested with 2× volume of trace-metal-grade HNO3 (Fisher) at 70 oC for 16 h in sealed plastic falcon tubes. Following digestion, samples were cooled to room temperature and diluted to obtain a 5% final HNO3 concentration. Metal content was analyzed with ICP-MS. A series of 5 ICP-MS iron calibration standards were prepared with a custom-made TEXASAM-15REV3 stock (Inorganic ventures). The concentrated stock contained 1mg/L of natural abundance Fe. The remaining standards were obtained by diluting the previous standard 10×.
      Figure 4 experiment: Plasma from 36 wk old regular HFE mice (∼ 500 μL) was concentrated using a 10 kDa cutoff filter, and FTS was collected. To wash the sample, 450 μL of water was added to the ∼ 50 μL of the retentate. The sample was again concentrated and FTS collected. The washed retentate was diluted 80× with 20 mM AA pH 6.5 and divided in half. Half was treated with 50 μM sodium citrate and half was untreated. A similar experiment was performed using plasma from 24 wk old regular HFE mice. In another experiment, FTSs from six 24-36wk HFE mice (0, 50, 500 and 5000 mg) were combined and then divided in half. Half was treated with 50 μM Na citrate and half was untreated.

      Supporting Information

      This article contains supporting information: Figure S1, flow-chart of experiments; Figure S2, LC-ICP-MS chromatograms of FTS from different batches run on the low-mass SEC column. Figure S3, LC-ICP-MS chromatograms of RTN from HFE and control mice. Figure S4, LC-ICP-MS of control and HFE RTN run on the SP Low Mass Column. Figure S5, Raw Mössbauer spectra of HFE retentate plasma before subtracting oxy- and deoxyhemoglobin contributions. Figure S6, Mössbauer spectra of pseudo-plasma salts with added Fe-citrate.

      Data Availability Statement

      All data are contained within the manuscript and SI.

      Uncited reference

      • Miao R.
      • Martinho M.
      • Morales J.G.
      • Kim H.
      • Ellis E.A.
      • Lill R.
      • Hendrich M.P.
      • Münck E.
      • Lindahl P.A.
      EPR and Mössbauer spectroscopy of intact mitochondria isolated from Yah1p depleted Saccharomyces cerevisiae.
      .

      Conflict of interest

      The authors declare that they have no conflicts of interest with the contents of this article.

      Acknowledgments

      We thank Jean Kovar (TAMU) for help maintaining the mice used in this study and Nathaniel Dziuba for help with early experiments.

      References

        • Anderson G.J.
        • Bardou-Jacquet E.
        Revisiting hemochromatosis: genetic vs. phenotypic manifestations.
        Annals Translational Med. 2021; 9 (Article 731)https://doi.org/10.21037/atm-20-5512
        • Chauhan W.
        • Shoaib S.
        • Fatma R.
        • Zaka-ur-Rab Z.
        • Afzal M.
        Beta-thalassemia and the advent of new interventions beyond transfusion and iron chelation.
        Brit. J. Clin. Pharmacology. 2022; https://doi.org/10.1111/bcp.15343
        • Knutson M.D.
        Non-transferrin-bound iron transporters.
        Free Rad. Biol. Med. 2019; 133: 101-111
        • Bartnikas T.B.
        Known and potential roles of transferrin in iron biology.
        Biometals. 2012; 25: 677-686https://doi.org/10.1007/s10534-012-9520-3
        • Ganz T.
        Hepcidin and iron regulation, 10 years later.
        Blood. 2011; 117: 4425-4433
        • Faber M.
        • Jordal R.
        Presence of 2 iron-transport proteins in serum.
        Nature. 1961; 192: 181
        • Sarkar B.
        State of iron(III) in normal human serum: low molecular weight and protein ligands besides transferrin.
        Canadian J. biochem. 1970; 48: 1339-1350
        • Hershko C.
        • Graham G.
        • Bates G.W.
        • Rachmilewitz E.A.
        Nonspecific serum iron in thalassemia: An abnormal serum iron fraction of potential toxicity.
        Br J Haematology. 1978; 50: 255-263
        • Graham G.
        • Bates G.W.
        • Rachmilwitz E.A.
        • Hershko C.
        Nonspecific serum iron in Thalassemia – quantitative and chemical reactivity.
        Am. J. Hematol. 1979; 6: 207-217
        • Batey R.G.
        • Fracp P.
        • Fong L.C.
        • Shamir S.
        • Sherlock S.
        a Non-Transferrin-bound serum iron in idiopathic hemochromatosis.
        Digestive Diseases and Sciences. 1980; 25: 340-346
        • Evans R.W.
        • Rafique R.
        • Zarea A.
        • Rapisarda C.
        • Cammack R.
        • Evans P.J.
        • Porter J.B.
        • Hider R.C.
        Nature of non-transferrinbound iron: studies on iron citrate complexes and thalassemic sera.
        J. Biol. Inorg. Chem. 2008; 13: 57-74
        • Silva A.M.
        • Kong X.
        • Parkin M.C.
        • Cammack R.
        • Hider R.C.
        Iron(III) citrate speciation in aqueous solution.
        Dalton Trans. 2009; 40: 8616-8625
        • Brissot Pierre
        • Loreal Olivier
        Iron metabolism and related genetic diseases: A cleared land, keeping mysteries.
        J. Hepatology. 2016; 64: 505-515
        • Silva A.M.N.
        • Rangel M.
        The (Bio)Chemistry of Non-Transferrin-Bound Iron.
        Molecules. 2022; 27: 1784
        • Aruoma O.I.
        • Bomford A.
        • Polson R.J.
        • Halliwell B.
        Nontransferrin-bound iron in plasma from hemochromatosis patients – effect of phlebotomy therapy.
        Blood. 1988; 72: 1416-1419
        • Breuer W.
        • Ronson A.
        • Slotki I.N.
        • Hershko C.
        • Cabantchik Z.I.
        The assessment of serum nontransferrin-bound iron in chelation therapy and iron supplementation.
        Blood. 2000; 95: 2975-2982
        • Breuer W.
        • Hershko C.
        • Cabantchik Z.I.
        The importance of non-transferrin bound iron in disorders of iron metabolism.
        Transfusion Sci. 2000; 23: 185-192
        • May P.M.
        • Linder P.W.
        Computer simulation of metal-ion equilibria in biofluids: Models for the low-molecular-weight complex distribution of Calcium(II), Magnesium (II), Manganese(II), Iron(III), Copper(II), Zinc(II) and Lead(II) ions in human blood plasma.
        J.C.S Dalton. 1977; : 588-595
        • Neu H.M.
        • Alexishin S.A.
        • Brandis J.E.P.
        • Williams A.M.C.
        • Li W.J.
        • Sun D.J.
        • Zheng N.
        • Jiang W.L.
        • Zimrin A.
        • Fink J.C.
        • Polli J.E.
        • Kane M.A.
        • Michel S.L.J.
        Snapshots of Iron Speciation: Tracking the Fate of Iron Nanoparticle Drugs via a Liquid Chromatography-Inductively Coupled Plasma-Mass Spectrometric Approach.
        Mol. Pharmaceutics. 2019; 16: 1272-1281
        • Dziuba N.
        • Hardy J.
        • Lindahl P.A.
        Low-molecular-mass iron in healthy blood plasma is not predominately ferric citrate.
        Metallomics. 2018; 10: 802-817
        • Dziuba N.
        • Hardy J.
        • Lindahl P.A.
        Low-molecular-mass iron complexes in blood plasma of iron-deficient pigs do not originate directly from nutrient iron.
        Metallomics. 2019; 11: 1900-1911
        • Barton J.C.
        • Edwards C.Q.
        • Acton R.T.
        HFE gene: Structure, function, mutations, and associated iron abnormalities.
        Gene. 2015; 574: 179-192
      1. Vali, S.W. and Lindahl, P.A. (2022) Mössbauer spectroscopic characterization of organs from HFE(-/-) mice: relevance to hereditary hemochromatosis. In preparation.

        • Kretchmar S.A.
        • Teixeira M.
        • Huynh B.H.
        • Raymond K.N.
        Mössbauer studies of electrophoretically purified monoferric and diferric human transferrin.
        Biol. Metals. 1988; 1: 26-32
        • Miao R.
        • Martinho M.
        • Morales J.G.
        • Kim H.
        • Ellis E.A.
        • Lill R.
        • Hendrich M.P.
        • Münck E.
        • Lindahl P.A.
        EPR and Mössbauer spectroscopy of intact mitochondria isolated from Yah1p depleted Saccharomyces cerevisiae.
        Biochemistry. 2008; 47: 9888-9899
        • Lee S.M.
        • Loguinov A.
        • Fleming R.E.
        • Vulpe C.D.
        Effects of strain and age on hepatic gene expression profiles in murine models of HFE-associated hereditary hemochromatosis.
        Genes and Nutrition. 2015; 10 (Article Number 443)
        • Simpson R.J.
        • Cooper C.E.
        • Raja K.B.
        • Halliwell B.
        • Evans P.J.
        • Aruoma O.I.
        • Singh S.
        • Konijn A.M.
        Non-transferrin-bound iron species in the serum of hypotransferrinaemic mice.
        Biochim. Biophys. Acta. 1992; 1156: 19-26
        • de Swart L.
        • Hendriks J.C.M.
        • van der Vorm L.N.
        • Cabantchik Z.I.
        • Evans P.J.
        • Hod E.A.
        • Brittenham G.M.
        • Furman Y.
        • Wojczyk B.
        • Janssen M.C.H.
        • Porter J.B.
        • Mattijssen V.E.J.
        • Biemond B.J.
        • MacKenzie M.A.
        • Origa R.
        • Galanello R.
        • Hider R.C.
        • Swinkels D.W.
        Second international round robin for the quantification of serum non-transferrin bound iron and labile plasma iron in patients with iron-overload disorders.
        Haematologica. 2016; 10: 38-45
        • Cravens C.M.
        • Alexander J.
        • Eldridge M.
        • Kushner J.P.
        • Bernstein S.
        • Kaplan J.
        Tissue distribution and clearance kinetics of non-transferrin-bound iron in the hypotransferrenemic mouse: A rodent model for hemochromatosis.
        Proc. Natl. Acad. Sci. 1987; 84: 3457-3461
        • Grootveld M.
        • Bell J.D.
        • Halliwell B.
        • Aruoma 0. I.
        • Bomford A.
        • Sadler P.J.
        Non-transferrin-bound iron in plasma or serum from patients with idiopathic hemochromatosis.
        J. Biol. Chem. 1989; 264: 4417-4422
        • Silva A.M.N.
        • Hider R.C.
        (2009) Influence of non-enzymatic post-translation modifications on the ability of human serum albumin to bind iron Implications for non-transferrin-bound iron speciation.
        Biochim. Biophys. Acta – proteins and proteomics. 1794; : 1449-1458