A Rationally Designed Complex Replenishes the Transferrin Iron Pool Directly and With High Specificity

Abstract

Many forms of anemia are caused or complicated by pathologic restriction of iron (Fe). Chronic inflammation and certain genetic mutations decrease the activity of ferroportin, the only Fe-exporter protein, so that endogenously recycled or nutritionally absorbed Fe cannot be exported to the extracellular Fe carrier protein transferrin for delivery to the bone marrow. Diminished ferroportin activity renders anemia correction challenging, as Fe administered intravenously or through nutritional supplementation is trafficked through the ferroportin-transferrin axis. Utilizing judicious application of coordination chemistry principles, we designed an Fe complex (Fe-BBG) with solutions thermodynamics and Fe dissociation kinetics optimized to replenish the transferrin-Fe pool rapidly, directly, and with precision. Fe-BBG is unreactive under conditions designed to force redox cycling and production of reactive oxygen species. The BBG ligand has low affinity for divalent metal ions and does not compete for binding of other endogenously present ions including Cu and Zn. Treatment with Fe-BBG confers anemia correction in a mouse model of iron-refractory iron-deficiency anemia. Repeat exposure to Fe-BBG did not cause adverse clinical chemistry changes or trigger expression of genes related to oxidative stress or inflammation. Fe-BBG represents the first entry in a promising new class of transferrin-targeted Fe replacement drugs.

INTRODUCTION.

Many anemias, including anemia of inflammation and iron refractory iron deficiency anemia (IRIDA) are driven partially or entirely by pathologic Fe restriction.^1–3 In these conditions, chronic immune activation or genetic mutations upregulate expression of the Fe regulator hormone hepcidin, which inhibits activity of the only known Fe-exporter protein ferroportin.^4–6 Hepcidin excess thus inhibits Fe liberated by hemoglobin recycling in macrophages, dietary Fe absorbed through duodenal enterocytes, and other stored Fe from export to the extracellular Fe carrier protein transferrin.⁷ Hepcidin excess/ ferroportin deficiency also limits the efficacy of Fe replacement therapy, as nearly all intravenous (i.v.) formulations are comprised of Fe-carbohydrate or Fe-oxide nanoparticles, which are absorbed and metabolized in macrophages, and thus rely on ferroportin for Fe export.^7–11 Nutritional Fe supplementation suffers the same limitation and can also be complicated by gastrointestinal intolerance.¹²

There are no clinically available treatments that target the interaction between hepcidin and ferroportin.^13–15 Fe mobilization may be improved using erythropoiesis stimulation agents (ESAs), which indirectly lower hepcidin expression by enhancing erythropoiesis.¹⁶ However, additional Fe supplementation is needed by many patients receiving ESAs. Moreover, ESAs are associated with cardiovascular toxicity, increased risk for thromboembolic events, and malignancy.^17–21 For these reasons, ESAs carry a US Food and Drug Administration (FDA) boxed warning advising of increased risk of death and other serious adverse events.

Hypoxia inducible factor prolyl hydroxylase inhibitors (HIF-PHIs) are a new class of drug to stimulate endogenous erythropoietin production.²² HIF-PHIs may also increase intestinal Fe absorption and transport by promoting expression of Fe transport proteins.^{16, 23} However, it has not been established that HIF-PHIs improve Fe bioavailability or confer a safety advantage over ESAs in randomized controls trials, and one HIF-PHI failed to meet non-inferiority safety endpoints in global phase 3 clinical trials.²⁴ The first HIF-PHI agents submitted to the FDA for marketing authorization, Roxadustat and Vadadustat, were rejected based on safety concerns. The FDA just recently granted the first HIF-PHI approval for Daprodustat.

An alternate approach to overcome hepcidin excess is to replenish the transferrin-Fe pool via mechanisms independent of the ferroportin-transferrin axis. Recent studies demonstrated that the natural product hinokitiol acts as a cell permeant Fe chelator and facilitates Fe redistribution across pathophysiologic gradients in rodents.^25–26 An aqueous formulation of iron pyrophosphate citrate (FPC), which directly replenishes the transferrin-Fe pool upon intravenous infusion, has an FDA indication for use in CKD patients receiving hemodialysis.^27–28 However, care must be taken to ensure that serum FPC concentrations do not exceed transferrin’s capacity to bind Fe, as exposure to toxic labile Fe occurs above this threshold.²⁷ To safely administer therapeutically meaningful quantities of Fe, FPC must be infused over the course of hours, often several times per week.

An ideal Fe replacement drug would replenish the transferrin-Fe pool directly, rapidly, and with high specificity. A transferrin-specific Fe delivery mechanism could enable the drug to be infused quickly and safely at doses that far exceed transferrin Fe binding capacity. Transferrin saturation could thus be achieved rapidly and maintained continuously until the drug is either consumed or eliminated entirely. We posit that such a treatment could be realized through judicious application of coordination chemistry principals. We envision a chelator that: (1) binds Fe³⁺ with lower affinity than transferrin, (2) but with sufficiently high stability to avoid ‘off-target’ Fe release, (3) rapidly delivers Fe³⁺ to transferrin, (4) is not reduced to Fe²⁺, (5) does not sequester or redistribute endogenously present metal ions; and since CKD patients represent one of the largest patient groups in need of such therapy,^{10–11, 29} (6) is eliminated through partial hepatobiliary excretion.

Here, we report the chelator BBG (N,N-bis(2-hydroxybenzyl)-L-glutamic acid, Chart 1A), and the corresponding Fe³⁺ complex Fe-BBG (Chart 2), which were conceived and synthesized to address the design elements outlined above. Drug design, characterization, in vitro screening, and preliminary in vivo evaluation of therapeutic efficacy and safety are discussed below. Taken together, our results establish proof of concept for Fe-BBG as the first entry in promising new class transferrin-targeted Fe replacement drug.

Chart 1.

The ligands BBG (A), NTA (B), EDTA (C), and EHPG (D).

Chart 2.

Fe-BBG speciation at pH 7.4 comprises a roughly 60:40 mixture of mono- and di-anionic Fe³⁺ complexes. We propose that this speciation comprises the structures [Fe(BBG)(H₂O)₂]^1- and [Fe(BBG)(H₂O)(OH)]^2-.

RESULTS AND DISCUSSION

Fe-BBG design and synthesis.

Prior literature evaluating Fe³⁺ trans-chelation reactions demonstrates rapid Fe transfer from nitrilotriacetic acid (NTA, Chart 1B) to apo-transferrin.^{28, 30} Fe-NTA is highly toxic, however, which has been attributed to the release of labile and redox active Fe.^31–35 The NTA ligand has also been shown to contribute toxicity through redistribution of endogenously present Zn.³¹ We reasoned that a structurally related tripodal tetradentate ligand functionalized instead with pendant arms to provide an increasingly hard anionic-O coordination sphere, such as 2-hydroxybenzyl, would maintain the rapid Fe³⁺ trans-chelation kinetics of NTA while simultaneously increasing Fe³⁺ binding affinity and stabilizing the Fe³⁺ oxidation state. The effect is illustrated by comparing the pH 7.4 stability constants (logK_{ML pH 7.4}) and Fe^3+/2+ redox potentials for the Fe³⁺ complexes of the structurally related ligands EDTA and EHPG (Chart 1C,D, Table 1).^36–38 We inferred that switching two acetic acid arms for 2-hydroxybenzyl would provide a logK_{ML pH 7.4} increase comparable to that of Fe³⁺-EHPG over Fe³⁺-EDTA - enough to disfavor “off target” trans-chelation with endogenous challenger ligands like citrate (accounts describing Fe³⁺-citrate thermodynamic parameters vary, with some reports indicating logK_{ML pH 7.4} as high as 16.5),³⁹ but also binding Fe³⁺ with orders of magnitude lower affinity than apo- and mono-ferric transferrin, which bind Fe³⁺ with logK_{ML pH 7.4} 22.3 and logK_{M2L pH 7.4} 21.5, respectively.⁴⁰ We also figured that the increasingly hard anionic donor set provided by BBG would decrease affinity for divalent metal ions relative to NTA at physiological pH, comparable to the differences observed between the Cu²⁺ and Zn²⁺ complexes of EDTA and EHPG (Table 1).³⁶

Table 1.

Thermodynamic Parameters^a for Metal Complexes of BBG and Ligands Discussed in Reference to BBG Design.

	BBGb	NTA42–43 c	EDTA38, 44 c	EHPG36–37 c
logKFeL pH 7.4	19.5	13.7	22.1	26.2
logKCuL pH 7.4	10.5	11.1	16.0	15.9
logKZnL pH 7.4	4.5	8.5	13.9	9.3
E1/2 for Fe3+/2+ (V vs NHE)	< −0.6 V	0.38 V	0.11 V	< −0.6 V

^alogK_{ML pH 7.4} values calculated from ligand protonation and metal complex formation constants.

^b37 °C, I = 0.1M NaCl.

^c25 °C, I = 0.1M KCl.

BBG is constructed from an L-glutamic acid core. We reasoned that the anionic amino acid side chain would ensure that Fe-BBG remains negatively charged at pH 7.4, thus promoting biodistribution within the extracellular spaces where transferrin is found. We further reasoned that the modestly amphiphilic nature of the complex would promote partial hepatobiliary elimination.⁴¹

The ligand BBG was synthesized by through a double reductive amination reaction comprising L-glutamic acid and 2 mol. equiv. salicylaldehyde, and purified by preparative scale RP-HPLC (Figures S1,S2A). Fe-BBG was synthesized by mixing BBG with 1 molar equiv. FeCl₃ in water, then adjusting to pH 6–8. The complex is isolated as a red solid (l_max = 462 nm, e = 2,200 M⁻¹cm⁻¹ at pH 7.4, Figures S2B,S3) following RP-HPLC purification (Details in SI). HPLC analysis of solutions containing Fe-BBG indicates that the complex exists as a single isomer. However, we note that a mixture of diastereomeric complexes can be potentially envisioned upon complexation of the Fe³⁺ by the chiral BBG ligand and the limited HPLC analysis performed for this study cannot definitively rule out the possibility of an equilibrium mixture of diastereomers.

Fe-BBG solutions thermodynamics.

The aqueous solutions thermodynamics of Fe-BBG were interrogated by pH-potentiometric titration of BBG in the absence and presence of 1 molar equiv. Fe³⁺(Figures 1A, S4). Spectrophotometric titrations in which relative concentrations of Fe³⁺ and BBG are varied indicated that 1:1 metal:ligand (ML) binding is favored under equimolar conditions (Figure S3B), and the pH-titration data were modelled accordingly. The entirety of the thermodynamic parameters are tabulated in Table S1, and from these values a logK_{ML pH 7.4} value 19.5 was calculated for Fe-BBG (Table 1). The relative increase in logK_{ML pH 7.4} of Fe-BBG compared to Fe-NTA are consistent with the effect predicted based on the Fe³⁺ complexes of EHPG and EDTA. Under the titration conditions, >99% of the Fe³⁺ is chelated by pH ≥ 3.0, and the final H⁺ attributable to the BBG ligand is consumed with pKa 4.42, resulting in a monoanionic complex. Continued titration consumes an additional three molar equivalents H⁺ with pKa’s of 7.66, 9.65, and 11.05. The pKa 7.66 and 9.65 events are tentatively assigned to deprotonation of putative cis-divacant water co-ligands, the pKa 11.05 event likely corresponds to coordination of third hydroxide ligand, mirroring speciation reported for Fe³⁺-NTA.⁴²

Figure 1.

(A) pH-titration profile of BBG in the absence (L) and presence (FeL) of 1 molar equiv. Fe³⁺ ([Fe] = [BBG] = 6.1 mM, 0.1 M NaCl, 37 °C) The molar equiv. NaOH are plotted relative to BBG in the charge neutral (H₄L) state. (B) pH-dependence of Fe³⁺ speciation in the presence of 1 molar equiv. BBG and a concentration of 6.1 mM. The purple trace corresponds to unchelated Fe³⁺, the green trace corresponds to the HML species, the black trace to ML (tentatively assigned as the [Fe(BBG)(H₂O)₂]^1-), the red trace to ML(OH) ([Fe(BBG)(H₂O)(OH)]^2-), the teal trace to ML(OH)₂, and the brown trace to ML(OH)₃. The dotted vertical line is at pH 7.4.

The simulated pH dependence on Fe³⁺ speciation in the presence of 1 mol equiv. BBG between pH 2–10 is shown in Figure 1B. Bulk magnetic susceptibility measurements made using Evan’s NMR method yield μ_eff = 5.9 B.M. under conditions of 10 mM Fe-BBG, pH 7.4 and 37 °C (Table S2), consistent with high-spin Fe³⁺ and suggesting that μ-oxo bridged, anti-ferromagnetically coupled complexes of higher nuclearity do not contribute appreciably to Fe³⁺ speciation.^45–46 Based on the titration data and high spin electron configuration we tentatively propose that within the range of concentrations studied, Fe-BBG speciation at pH 7.4 is comprised of a roughly 60:40 mixture of the complexes [Fe(BBG)(H₂O)₂]^1- and [Fe(BBG)(H₂O)(OH)]^2-, shown in Chart 2.

Fe-BBG stoichiometrically and rapidly delivers Fe³⁺ to apo-transferrin.

At pH 7.4, BBG binds Fe³⁺ with roughly 10³-fold and 10²-fold lower than affinity than apo- and mono-ferric transferrin.⁴⁰ In this regard, Fe³⁺ trans-chelation from BBG to transferrin was observed through quenching of apo-transferrin intrinsic fluorescence^47–48 and by urea gel electrophoresis. Figures 2A and 2B respectively show λ_ex = 280 nm fluorescence spectra and urea gel electrophoresis ladders recorded for solutions comprising apo-transferrin incubated with varying molar equivalents Fe-BBG. The data indicate stoichiometric transfer of Fe³⁺ from Fe-BBG to transferrin up to the point of transferrin saturation. No Fe³⁺ trans-chelation was observed when holo-transferrin incubated with the BBG ligand (Figure S5). The time course of the Fe³⁺ trans-chelation reaction (instrument temporal resolution = 6s) recorded via fluorescence emission intensity at 330 nm indicates that that Fe³⁺ transfer reaction is complete within 6s of mixing (Figure S6).

Figure 2.

(A and B) Fluorescence spectra (A) and urea gel electrophoresis (B) recorded on solutions containing 20 μM apo-transferrin incubated with varied concentrations of Fe-BBG ranging between 0 and 40 μM (0 to 2 molar equiv), pH 7.4 50 mM HEPES buffer with 25 mM carbonate added, 25 °C, incubation time: 2h. The concentrations of transferrin and carbonate approximate serum concentrations in vivo. Fe_C-Tf and Fe_N-TF correspond to mono-ferric transferrin with Fe³⁺ bound by the C- and N- lobes, respectively. (C) Concentrations of thiobarbituric acid reactive substances (TBARS, generated by oxidation of 5-deoxyribose) measured in solutions that contain either no Fe or 5 mM of different Fe³⁺ complexes recorded 1h after incubation with 2.8 mM 5-deoxyribose under conditions designed to force redox cycling and production of ROS (2.8 mM H₂O₂, 0.1 mM ascorbate. pH 7.4 phosphate 100 mM, 37 °C). (D-H) Solutions of human blood plasma were incubated with varied concentration of BBG ranging between 0 to 5 mM for 1h at 37 °C and then subject to ultrafiltration through a 10 kDa molecular weight cutoff filter. The concentrations of Cu (D), Zn (E), Mn (F), Mg (G), and Ca (H) in the plasma concentrate (open circles) and ultrafiltrate (filled black circles) were not affected even by large excess of BBG.

Fe-BBG is robust against “off-target” trans-chelation.

Fe-BBG is also very robust against Fe³⁺ trans-chelation with endogenously present ligands commonly implicated in formation of labile and redox active Fe complexes. The UV-vis spectra of 0.1 mM Fe-BBG solutions incubated for 2h with up 10 mM (100 molar equiv.) challenger ligands such as acetate, lactate, phosphate, and carbonate reveal no evidence of trans-chelation, Figure S7A–D. Analysis of equilibrium mixtures of Fe-BBG challenged with between 0 and 10 molar equiv. citric acid yield K_comp = 0.006±0.002, indicating that BBG binds Fe³⁺ with 230±40-fold higher affinity than citrate, Figure S7E.

Fe-BBG does not redox cycle or generate reactive oxygen species.

Cyclic voltammetry measurements reveal no evidence of Fe-BBG reduction in scans down to −0.6 V vs. NHE (Figure S8). To screen for spontaneous redox cycling and generation of reactive oxygen species (ROS), Fe-BBG was incubated with 5-deoxyribose for 1h under redox forcing conditions, and the solution assayed for 5-deoxyribose oxidation products using a previously reported assay for thiobarbituric acid reactive substances (TBARS, Figure 2C).^49–50 Solutions containing Fe³⁺-NTA, Fe³⁺-EDTA, and FPC were assayed as positive controls. Solutions containing no added Fe complex were assayed as negative control. Unlike the other Fe complexes, no ROS were generated in samples containing Fe-BBG.

BBG does not compete for metals endogenously present in serum.

The thermodynamic stability of the Zn²⁺ and Cu²⁺ complexes of BBG were determined using pH-potentiometry, Figures S9–11, Tables 1 and S1. Based on the logK_{ML pH 7.4} values for Zn²⁺-BBG and Cu²⁺-BBG (4.45 and 10.49, respectively), we do not expect the BBG ligand to sequester or redistribute metal ions present in blood serum and the extracellular spaces. For example, serum Zn²⁺ speciation is comprised predominantly of Zn-bound to albumin with logK_{ML pH 7.4} = 7.0,⁵¹ whereas serum Cu speciation comprises Cu²⁺ bound to proteins including albumin, a-macroglobulin, and low molecular weight ligands, each with logK_{ML pH 7.4} ~13.⁵²

Given the Cu-BBG and Zn-BBG stability constants and high abundance of serum albumin (~660 μM) we posit that even millimolar concentrations of BBG should not disrupt homeostasis of these metal ions. To illustrate this point, human blood plasma samples containing between 0–5 mM BBG were incubated for 1h at 37 °C before separation of the low-molecular weight solution components by ultrafiltration through a 10 kDa molecular weight cutoff filter. The plasma concentrate and ultrafiltrate were assayed for concentrations of Cu and Zn, as well as Mn, Mg, and Ca by ICP-MS, Figure 2D–H. Incubation with BBG had little effect on the metal ion content remaining in the plasma concentrate, nor did BBG substantially increase concentrations of any metal ion in the ultrafiltrate relative to control samples not treated with any ligand.

Fe-BBG Pharmacokinetics.

Complexes of high-spin Fe³⁺ are potent generators of paramagnetic relaxation enhancement.⁵³ We leveraged this physical property to monitor Fe-BBG blood clearance and excretion following intravenous injection of 7.8 μg Fe per g body weight as Fe-BBG using dynamic magnetic resonance imaging. We note that post-injection blood signal increase may receive contributions from both an increase in transferrin-bound Fe and from intact Fe-BBG. In this regard, we administered a large enough dose of Fe-BBG so that MR signal early post-injection time points would be dominated by contributions from intact Fe-BBG (see Supporting Information for calculations).

Figures 3A,B show coronal abdominal T₁-weighted images at the level of the liver (3A) and kidneys (3B) prior to, 1 min, and 10 min after injection of 7.8 μg Fe per g body weight as Fe-BBG. The image series in Figure 3B shows strong enhancement of kidneys with accumulation of paramagnetic Fe in the renal pelvis 1 min after injection. This enhancement diminishes substantially by 10 min, consistent with rapid elimination through the kidneys. The image series in Figure 3A shows liver enhancement 1 min after injection which persists out to the 10 min time point. The red arrow in Figure 3A denotes strong enhancement of the gall bladder 10 min after injection, consistent hepatobiliary excretion of the injected Fe. Intact Fe-BBG was recovered in urine and bile after injection (Figures S12,13). Mono-exponential fits to plots of vena cava signal intensity vs. time indicate that post-injection blood MR signal decreases with a half-life of 5.5±2.8 min (N=4), Figures 3C (corresponding image series in Figure S14).

Figure 3.

(A and B) Dynamic series of coronal abdominal T₁-weighted images recorded prior to and after injection of 7.8 μg Fe per gram body weight as Fe-BBG show that the injected Fe is rapidly excreted through the liver (A) and kidneys (B). The red arrow shows paramagnetic Fe accumulating in gall bladder. (C) Time course of MR signal intensity normalized to pre-injection value (nSI) recorded in the vena cava. The blood signal decreases with a half-life of 5.5±2.8 min.

Serum nonheme Fe parameters after Fe-BBG injection.

To gain further insight into the in vivo fate of Fe-BBG, we analyzed blood serum harvested from mice following intravenous administration of the compound. The dose levels and timing of blood harvest were designed so that serum would be analyzed both under conditions where Fe-BBG concentrations are well in excess of transferrin and where intact Fe-BBG has been largely eliminated. Mice received a single injection comprising either 0.0, 0.56 μg Fe or 5.6 μg Fe per gram body weight Fe-BBG (placebo, low, and high doses). The mice were euthanized either 5-, 60-, or 100-min post-injection (N=3/dose/time point), blood serum was harvested, and serum nonheme Fe parameters (serum total Fe concentration, unbound iron binding capacity (UIBC), and serum total iron binding capacity (TIBC)) quantified using the Ferrozine spectrophotometric assay, Figure 4 and described in Supporting Information. In placebo treated mice, TIBC is comprised entirely of transferrin.⁵⁴ However, because Fe-BBG also contributes to TIBC readings, we measured serum transferrin concentrations directly using an ELISA capture assay and estimated transferrin iron binding capacity (Tf-IBC) from these values (SI). We also analyzed representative serum samples from each group for labile and redox active Fe by spectrophotometrically assaying the kinetics dihydrorodamine 123 oxidation in serum incubated with added ascorbate.⁵⁵

Figure 4.

Nonheme Fe parameters recorded on serum harvested from mice 5 min after i.v. injection of saline (placebo, grey filled circles) and 5 and 60 min after injection of 0.56 μg (low dose, open circles) or 5.6 μg Fe (high dose, black filled circles) per gram body weight as Fe-BBG: (A) Serum total nonheme Fe, (B) unbound iron binding capacity (UIBC), (C) total iron binding capacity (TIBC), (D) transferrin iron binding capacity (Tf-IBC), and (E) percentage transferrin saturation (%TSAT). Serum Fe, UIBC, and TIBC were measured using the ferrozine spectrophotometric assay, Tf-IBC was estimated based on serum transferrin concentration determined using ELISA capture assay (Details in SI).

Figure 4A shows that serum Fe levels are elevated substantially after Fe-BBG injection. Serum Fe levels in samples collected from mice receiving the high dose exceed spectroscopically determined TIBC recorded in placebo treated mice, or Tf-IBC estimated from the ELISA capture assay.

Figure 4B shows that serum UIBC was not overwhelmed in any sample analyzed. Even for the blood serum samples where Fe levels far exceed Tf-IBC, serum maintained its capacity to absorb Fe. We contend that UIBC preservation even under conditions where serum Fe >> Tf-IBC is attributable to the presence of the BBG ligand remaining after Fe release to transferrin. Furthermore, labile Fe concentrations remained beneath the level of our lowest concentration calibration standard (<1.1 μg/dL, or <0.20 μM), including for samples where serum Fe concentrations were elevated to >1500 μg/dL (>270 μM), Table S2.

Spectrophotometrically determined TIBC values recorded after the low dose of Fe-BBG mirror TIBC recorded in placebo treated mice, Figure 4C, and also reflect relative differences in Tf-IBC determined from direct measurements of transferrin content, Figure 4D. Together, these data are consistent with transferrin as the predominant species contributing to TIBC recorded 5–100 min after treatment with the lower dose of Fe-BBG. On the other hand, TIBC values after the high Fe-BBG dose are roughly 10-fold than from placebo 5 min post-injection and 2-fold higher between 60–100 min after injection, indicating that Fe-BBG and/ or BBG contribute substantially to overall TIBC. In this regard, percentage transferrin saturation were estimated only for samples harvested after placebo or low-dose Fe-BBG, Figure 4E.

The serum nonheme Fe data, taken together with the Fe-BBG thermodynamics parameters, transferrin delivery kinetics, and dynamic MRI data, indicate that Fe-BBG rapidly replenishes the transferrin-Fe pool in vivo. When the dose of Fe-BBG is high enough that serum Fe levels exceed Tf-IBC, injected Fe that was not delivered to transferrin remains chelated and is rapidly eliminated.

Fe-BBG corrects iron refractory iron deficiency anemia in mice.

We evaluated the therapeutic efficacy of Fe-BBG in Tmprss6 knockout mice, which exhibit aberrantly high levels of hepcidin and suffer Fe restricted erythropoiesis, phenocopying human patients with IRIDA due to TMPRSS6 mutations, and recapitulating salient features of anemia of inflammation. Tmprss6 knockout mice received 29 intraperitoneal injections each containing 25 μg Fe as Fe-BBG (1.3±0.18 μg Fe per gram body weight, N=5 mice) or containing saline (placebo, N=5 mice) over 15 days. The cumulative dose of Fe was 725 μg Fe (37±0.53 μg Fe per gram body weight). The mice were euthanized 120 min after the final injection. Blood was harvested for measurement of complete blood counts, clinical chemistry, and nonheme Fe parameters. Bone marrow and liver tissue were analyzed by qPCR for expression of genes related to erythropoietic activity, Fe exposure, oxidative stress, and inflammation (Table S5). The entirety of the ex vivo analysis is tabulated in Tables S5–8.

Figures 5A–H show representative hematologic and serum nonheme Fe parameters recorded in mice treated with Fe-BBG vs. placebo, with the entirety of the data are tabulated in Tables S5–6. Treatment with Fe-BBG resulted in a significant elevation of hemoglobin compared to placebo, Figure 5A. Hematocrit levels were also increased significantly (5B). Fe-BBG treatment increased mean corpuscle volume and red cell width, 5C-D, which we attribute to newly increased production of larger hemoglobin replete erythrocytes. Taken together, these changes are consistent with restoration of normal hematology. No noteworthy differences were detected between any other hematology parameters, Table S5.

Figure 5.

Representative hematology parameters (A-D), serum nonheme Fe parameters (E-H), and expression of genes related to erythropoietic activity (I,J), and Fe excess/ oxidative stress (K-M) recorded after treatment of Tmprss6 knockout male (triangles) and female (circles) mice with 725 μg Fe as Fe-BBG (37±0.53 μg Fe per g body weight) administered as 29 separate intraperitoneal doses containing 25 μg Fe each (1.3±0.18 μg Fe per g body weight) over the course of 15 days, and comparison to values recorded after treatment with placebo on an identical dosing schedule. Hemoglobin (A), hematocrit (B), mean corpuscle volume (C), red cell distribution width (D), serum total nonheme Fe (E), UIBC (F), TIBC (G), TSAT (H), relative expression of bone marrow erythroferrone (Erfe) (I) and transferrin receptor 1 (Tfrc) (J), relative liver expression of heme oxygenase 1 (Hmox1) (K), glutamate-cysteine ligase catalytic subunit (Gclc) (L), and NAD(P)H quinone dehydrogenase 1 (Nqo1) (M). * P < 0.05, ** P < 0.01, ***P < 0.001 ****P < 0.0001.

Serum Fe levels were elevated in Fe-BBG treated mice (5E), which was mirrored by concomitant decrease in UIBC (5F). TIBC levels in Fe-BBG treated mice were unchanged relative to placebo (5G), indicating that Fe-BBG has cleared entirely 120 min after the final injection and that serum Fe increase arises from replenishment of the transferrin Fe pool. In this regard, %TSAT is increased significantly in Fe-BBG treated mice, with mean %TSAT increased by nearly 7-fold (5H).

Figures 5I,J compare relative expression of genes related to erythropoietic activity and Fe demand in mice treated with placebo vs Fe-BBG. The data demonstrate that bone marrow erythroferrone (Erfe, 5K) is significantly downregulated in Fe-BBG treated mice, consistent with re-hemoglobinization. Transferrin receptor type 1 (Tfrc, 5J) and glycophorin A (GypA, Table S7), proteins abundantly expressed by erythroid precursor cells, were also downregulated in Fe-BBG treated mice, consistent with restoration of normative red blood cell maturation processes.

Our preliminary pharmacokinetics data indicate that Fe-BBG elimination is rapid compared to the rate of transferrin-Fe turnover in mice (~45 min reported for normal mice),⁵⁶ and thus much of the injected compound is excreted intact. As the dose is increased, the relative fraction of injected Fe absorbed and utilized by the erythron will decrease. For this reason, we opted to treat IRIDA in Tmprss6 knockout through a series of multiple repeated small doses of Fe, rather than fewer large doses. The serum half-life of transferrin-Fe in iron deficiency humans is reported to be ~50 min,⁵⁷ whereas based on dose-conversion principals we estimate Fe-BBG elimination half-life in humans will longer by over an order of magnitude.⁵⁸ In other words, we expect a substantially larger fraction of the injected Fe to be absorbed by humans receiving an equivalent dose of Fe-BBG.

Preliminary assessment of Fe-BBG safety.

Repeat dosing of Fe-BBG to Tmprss6 knockout mice did not generate significant differences in any clinical chemistry parameters, including serum markers routinely used in drug toxicity screens, Table S8. Fe-BBG did not trigger upregulation of liver heme oxygenase 1 (Hmox1), glutamate-cysteine ligase catalytic subunit (Gclc), or NAD(P)H quinone dehydrogenase 1 (Nqo1), which taken together indicate that Fe-BBG did not result in oxidative stress, Figures 5K–M. Expression of Serum amyloid A1 (Saa1), a sensitive marker of inflammation, was also unchanged after repeat dosing with Fe-BBG, Table S7.

CONCLUSION.

We designed Fe-BBG as a potential drug to replenish transferrin Fe pool directly, rapidly, and with high specificity. Fe-BBG binds Fe³⁺ with low thermodynamic stability relative to transferrin, but with high enough stability to avoid “off target” trans-chelation reactions. The BBG ligand stabilizes the Fe³⁺ oxidation state exclusively and remains unreactive even under conditions designed to force redox cycling and to generate ROS. The BBG ligand also binds to divalent metal ions with low affinity, and in vitro assays show that even a large excess of BBG does not strip Cu²⁺ or Zn²⁺ from ligands endogenously present in blood. Studies to interrogate Fe pharmacokinetics and serum Fe speciation indicate that Fe-BBG efficiently delivers Fe³⁺ to transferrin in vivo, and that injected Fe remains chelated and is rapidly eliminated under conditions of transferrin saturation. Fe-BBG is eliminated through mixed renal and hepatobiliary excretion as the intact complex, indicating that the compound can be efficiently cleared even from patients with advanced kidney disease. Treatment with Fe-BBG was shown to provide anemia correction in Tmprss6 knockout mice, which suffer anemia driven by hepcidin excess, recapitulating human IRIDA and salient features of anemia of inflammation. Repeat dosing with Fe-BBG did not adversely affect any CBC parameter, significantly alter any clinical chemistry parameters, or trigger genes related to oxidative stress and inflammation. Fe-BBG represents the first entry for a new class of transferrin-targeting Fe replacement drug. This new class of therapeutic could profoundly impact the quality of care for patients suffering anemia driven entirely or in part by pathologic restriction of Fe.