Noncanonical interactions between serum transferrin and transferrin receptor evaluated with electrospray ionization mass spectrometry

Abstract

The primary route of iron acquisition in vertebrates is the transferrin receptor (TfR) mediated endocytotic pathway, which provides cellular entry to the metal transporter serum transferrin (Tf). Despite extensive research efforts, complete understanding of Tf-TfR interaction mechanism is still lacking owing to the complexity of this system. Electrospray ionization mass spectrometry (ESI MS) is used in this study to monitor the protein/receptor interaction and demonstrate the ability of metal-free Tf to associate with TfR at neutral pH. A set of Tf variants is used in a series of competition and displacement experiments to bracket TfR affinity of apo-Tf at neutral pH (0.2–0.6 μM). Consistent with current models of endosomal iron release from Tf, acidification of the protein solution results in a dramatic change of binding preferences, with apo-Tf becoming a preferred receptor binder. Contrary to the current models implying that the apo-Tf/TfR complex dissociates almost immediately upon exposure to the neutral environment at the cell surface, our data indicate that this complex remains intact. Iron-loaded Tf displaces apo-Tf from TfR, making it available for the next cycle of iron binding, transport and delivery to tissues. However, apo-Tf may still interfere with the cellular uptake of engineered Tf molecules whose TfR affinity is affected by various modifications (e.g., conjugation to cytotoxic molecules). This work also highlights the great potential of ESI MS as a tool capable of providing precise details of complex protein-receptor interactions under conditions that closely mimic the environment in which these encounters occur in physiological systems.

Iron is an essential nutrient for virtually all living systems, most notably due to its vital role in energy transfer processes (1). Since bioavailability of iron is severely limited under aerobic conditions due to the very poor solubility of its oxidized form (ferric ion, Fe³⁺), most living systems must rely on elaborate schemes to secure an adequate supply of this metal in soluble form, usually by synthesizing or recruiting efficient chelators. Sequestration of Fe³⁺ in various fluids and its transport are carried out in vertebrates by glycoproteins from the transferrin family, which includes serum transferrin, lactoferrin, and ovotransferrin. All proteins in this family have a distinct bilobal structure, with each lobe (usually termed N- and C-lobes) capable of binding one ferric ion synergistically with a carbonate or oxalate anion (2). Human serum transferrin (hTf) is an 80 kDa glycoprotein that transports iron through the blood stream to tissues which have the need for this metal and, therefore, express cell-surface receptors for hTf. Human transferrin receptor (TfR) is a 180 kDa homodimeric glycoprotein comprised of an ectodomain and a small cytosolic domain, connected by a single transmembrane segment (3). Upon hTf-TfR binding at the cell surface, the complex is internalized via endocytosis. Iron is removed from hTf in the mildly acidic environment of the endosome, and the metal-free hTf is recycled back to the cell surface and released to circulation for the next cycle of iron delivery (3).

Since hTf is one of relatively few proteins that can be internalized by rapidly growing cells, it is an attractive candidate for various therapeutic strategies aiming at specific delivery of imaging and cytotoxic agents to neoplastic cells (4). It is therefore not surprising that the TfR-mediated uptake pathway has been a focus of extensive research in recent years. Conjugation of common antineoplastic agents to hTf is known to affect its ability to bind iron (5), thereby placing a premium on our ability to characterize the influence of metal loading on Tf-TfR interaction both at the cell surface and inside the endosome. While it is commonly accepted that the diferric form of hTf has the highest affinity towards the receptor at neutral pH, the ability of TfR to recognize hTf forms lacking metal remains less clear. Several cell binding studies demonstrated the ability of apo-hTf to enter cells (6 and 7), and their TfR affinity was reported to be ∼20 times lower than that of Fe₂Tf (8 and 9).

However, other studies failed to detect apo-hTf binding to TfR at neutral pH (10), and similar conclusions came from work specifically targeting the hTf/TfR interaction using orthogonal biophysical techniques, such as surface plasmon resonance-based binding assays (11–13) and atomic force microscopy (14). As a result, the idea of apo-hTf/TfR association at neutral pH became increasingly apocryphal. This possibility is now either ignored (15), or attributed to contamination of commercial “metal-free” hTf with iron.

We investigate interactions between various forms of hTf (apo-hTf, diferric hTf, and the two monoferric forms) and TfR at near-physiological pH using electrospray ionization mass spectrometry (ESI MS). While ESI MS is rapidly gaining popularity in structural biology due to its ability to provide detailed information on the composition of large macromolecular assemblies (16), its application to characterize hTf/TfR interaction faces serious challenges. Unlike proteins derived from bacterial sources, for which high-quality ESI MS data can be obtained in the subMDa range and beyond, both hTf and TfR are glycosylated. The inherent heterogeneity of the carbohydrate component of these proteins makes it nearly impossible to confidently distinguish between metal-free and iron-bound forms of hTf in the context of the hTf/TfR complex. In this work we use various mutants of hTf, including authentic monoferric and apo-hTf in which iron-binding is prevented in one or both lobes. In addition, each mutant has a unique mass tag (e.g., due to the presence of a His-tag or removal of a carbohydrate chain), which allows unequivocal identification of the binding species even when several forms of hTf are present in solution.

Combining the capacity of ESI MS to preserve noncovalent protein complexes with our ability to selectively control iron binding in each lobe of hTf, we demonstrate that the apo-form of the protein does bind to the receptor at near-neutral pH. In addition, ESI MS can be used to rank various hTf forms according to their TfR affinities, allowing us to demonstrate that even though the apo-hTf/TfR interaction is weak compared to TfR association with the di- and monoferric forms of hTf, it is nonetheless specific and that apo-hTf competes successfully with Tf molecules from nonhuman sources.

A very important practical implication of this work is the conclusion that apo-hTf molecules are likely to remain bound to the receptor following the completion of the endocytotic metal delivery cycle, and their release back to circulation may require the presence of iron-bound hTf molecules with higher receptor affinity. Conjugation of cytotoxins to hTf may decrease its TfR affinity; obviously, if these molecules fail to displace apo-hTf bound to the receptor, their entrance to the cell may be compromised. Importantly, this unique methodology will also be useful in screening hTf-conjugated cytotoxic agents to determine how the conjugation affects TfR affinity and, therefore, the ability to enter neoplastic cells. It is also applicable to any soluble protein/receptor system, making it a particularly useful tool in situations when a variety of related (or indeed nearly identical) proteins interact with a single receptor.

Results

TfR Binds all hTf Forms at Neutral pH, Regardless of Their Iron Status.

ESI MS readily reveals binding of the di-ferric hTf (Fe₂Tf) to the receptor dimer (TfR) at pH 8.1 with a 2∶1 stoichiometry (Fig. 1A). No unbound TfR ions were detected in this case (the experiments were carried out with a slight excess of hTf molecules); in fact, all TfR molecules appear to be fully saturated with Fe₂Tf, as the ions representing an unsaturated complex Fe₂Tf·TfR are absent from the mass spectra as long as the Fe₂Tf/TfR molar ratio exceeds 2∶1.

Fig. 1.

ESI mass spectra of 6 μM TfR solutions in 100 mM NH₄HCO₃ containing 10 μM Fe₂Tf (A), aTf (B), Fe_NTf (C), and HSA (D) at pH 8.1. Gray traces represent mass spectra of TfR in the absence of binding partners. Asterisks represent hTf dimers.

Surprisingly, complex formation was also observed under these conditions between the iron-free hTf and TfR. Unfortunately, the structural heterogeneity of TfR due to extensive glycosylation makes it very difficult to verify the absence of iron in hTf molecules bound to TfR based on the mass measurements of the complexes. Although binding of a single Fe³⁺ ion to hTf results in a significant protein mass increase (ca. 0.07%), a difference that is easily discernable by ESI MS (17), the fractional mass increase of an hTf₂·TfR complex would be considerably smaller (< 0.02%). More importantly, the extent of structural heterogeneity exhibited by this complex is much more significant compared to that of hTf alone.

To completely eliminate the possibility of contaminating the apo-hTf sample with iron, we used an hTf mutant whose iron-binding capability was obliterated by replacing the two tyrosine ligands in each lobe (Y95 and Y188 in the N-lobe, and Y426 and Y517 in the C-lobe) with phenylalanine. Previous studies demonstrated that this mutant (aTf) completely lacks the ability to bind iron in either lobe (18). Furthermore, this construct has a mass that is significantly different from that of the wild-type hTf (76.8 vs. 79.7 kDa) due to the absence of the glycosylation and the presence of the His-tag, allowing distinction between (Fe₂Tf)₂·TfR and a putative (aTf)₂·TfR complex. Despite the total inability of aTf to bind ferric ion, it clearly is capable of making a complex with the receptor at pH 8.1 (Fig. 1B). The ability of aTf to form a stable complex with TfR was independently verified by SEC (see SI Text).

ESI MS provides clear evidence that aTf binds to TfR, revealing the presence of both partially unsaturated (aTf·TfR) and fully saturated (aTf₂·TfR) complexes alongside unbound aTf (Fig. 1B). A similar binding pattern is observed when TfR is mixed with excess monoferric transferrin (e.g., Fe_NTf, Fig. 1C), giving rise to protein/receptor complexes at both 1∶1 and 2∶1 stoichiometry in addition to free Fe_NTf. Although the apparent inability of aTf to completely saturate the receptor hints at less efficient binding (compared to Fe₂Tf), aTf-TfR binding is nonetheless specific, as suggested by control measurements carried out with another human serum protein, albumin (HSA). The mass spectrum of HSA/TfR mixture (Fig. 1D) acquired under the same conditions as Tf/TfR spectra (vide supra) shows unbound HSA and TfR as the principal species, while the contribution from the higher molecular weight species (putative nonspecific (HSA)_n.TfR complexes, m/z > 6,000) to the total ionic signal is negligible.

Ranking of Receptor Affinities for Various Forms of hTf.

While the direct ESI MS measurements clearly demonstrate the ability of aTf to form a stable complex with TfR at near-physiological pH, a quantitative assessment of the binding strength is not straightforward. In some favorable cases, relative abundance of complex ions and their constituents in ESI MS can be used to estimate the magnitude of the binding constant (19); however, nonlinearity of ionic signal dependence on solution concentration often raises concerns over the reliability of such calculations (20). We approached assessment of aTf receptor affinity by monitoring its interaction with TfR in the presence of other Tf species. It was accomplished by exploiting the unique ability of ESI MS to distinguish among various species in solution based on their mass differences (Fig. 2).

Fig. 2.

Competitive binding of aTf and Fe₂Tf to TfR at pH 8.1. The numbers on the right-hand side indicate Fe₂Tf/aTf/TfR molar ratios used in each measurement. The inset shows signals of unbound aTf and Fe₂Tf ions in the absence and presence of the receptor in solution.

Addition of a nearly equimolar mixture of aTf and Fe_NTf to the TfR solution results in formation of several complexes (aTf·Fe₂Tf·TfR, (aTf)₂·TfR, and (Fe₂Tf)₂·TfR) in addition to partially saturated complexes (aTf·TfR and Fe₂Tf·TfR), as long as the total Tf/TfR ratio remains below 2∶1. Further increase of total Tf/TfR ratio eliminates from the spectra not only the ions representing the partially saturated complexes, but also (aTf)₂·TfR complexes. Eventually, once the amount of Fe₂Tf becomes sufficient to saturate all receptor molecules in solution, only Fe₂Tf·TfR complexes are observed, despite the presence of large amounts of aTf in solution (Fig. 2, top trace). The low m/z region of the mass spectra (below m/z 4,000) contains ionic signals representing unbound transferrin molecules (inset in Fig. 2). Importantly, only aTf ions are present in the spectra, as long as the Fe₂Tf/TfR ratio remains below 2∶1. These measurements clearly demonstrate that despite being capable of binding to the receptor, aTf is outcompeted by Fe₂Tf.

These experiments were extended to include both monoferric species of hTf (Fe_CTf and Fe_NTf) with the goal of ranking all four forms of hTf according to their affinity for the receptor (Fig. 3). Monoferric hTf binds to the receptor in the presence of Fe₂Tf only if the latter is present at subsaturating levels (Fig. 3A). Increase of the Fe₂Tf/TfR ratio above 2∶1 completely abolishes Fe_CTf/TfR binding. At the same time, monoferric hTf out-competes aTf for receptor binding, since the latter only binds to TfR in the presence of subsaturating amounts of Fe_CTf (Fig. 3C). The substantial mass difference between the two forms of monoferric hTf used in this work (Fe_CTf and Fe_NTf, see SI Text) allows a distinction to be made among the (Fe_NTf)₂·TfR, Fe_NTf·Fe_CTf·TfR, and (Fe_CTf)₂·TfR complexes. The observed abundance distribution of these three complexes (1∶2∶1) under conditions when no unbound or partially saturated TfR molecules remain in solution (Fig. 3B) suggests that each monomeric species of hTf binds to the TfR with similar affinity.

Fig. 3.

TfR (6 μM) binding to Fe_CTf (10 μM) in the presence of 10 μM Fe₂Tf (A), Fe_NTf (B), and aTf (C).

Solution Acidification Reverses Binding Preferences of TfR.

While binding of aTf to TfR in the presence of subsaturating amounts of Fe₂Tf at neutral pH is restricted to filling the remaining vacant binding sites on TfR (Fig. 4A), acidification of the protein solution to endosomal pH 5.6 results in complete reversal of the receptor’s binding preferences. The most abundant species present in solution under these conditions is the (aTf)₂·TfR complex (Fig. 4B). The ability of holo-hTf to bind to the receptor is not completely obliterated by pH reduction, but is dramatically weakened. Indeed, Fe₂Tf loses its ability to compete with aTf, and is present in the spectrum shown in Fig. 4B only as a constituent of the mixed complex aTf·Fe₂Tf·TfR, apparently filling the vacant binding sites.

Fig. 4.

Acid-induced reversal of TfR binding preferences: ESI MS of aTf, Fe₂Tf (10 μM each) and TfR (6 μM) at pH 8.1 (A) and the same mixture acidified to pH 5.6 (B). The blue and red traces in each spectrum represent ionic signals of (aTf)₂·TfR and (Fe₂Tf)₂·TfR recorded under the same conditions in the absence of a competing transferrin.

Low Receptor Affinity hTf Is Displaced from TfR by a Higher Affinity Form of the Protein.

The competition binding assays described in the previous sections establish the receptor affinity scale for various forms of hTf. The ability of Fe₂Tf to displace aTf from the (aTf)₂·TfR complex at near-physiological pH was evaluated by titrating the preformed aTf/TfR complexes with Fe₂Tf, i.e., by adding incremental amounts of the latter to the aTf/TfR mixture (Fig. 5). Prior to any additions of Fe₂Tf, the mass spectrum of the aTf/TfR mixture (24 and 13 μM, respectively) contains contributions from both aTf/TfR complexes and unbound aTf species (gray trace in Fig. 5B). Addition of Fe₂Tf (raising its concentration from 0 to 2 μM) gives rise to ionic signal representing the mixed complex aTf·Fe₂Tf·TfR. Further increase of Fe₂Tf concentration leads not only to a significant increase in the abundance of this ionic species, but also to the appearance of (Fe₂Tf)₂TfR, while the relative abundance of the (aTf)₂·TfR species monotonically decreases. Once the Fe₂Tf∶TfR molar ratio in solution exceeds 1∶1, the abundance of the mixed complexes begins to decrease as well, and once this ratio exceeds 2∶1, (Fe₂Tf)₂TfR becomes the only high-mass species present in solution.

Fig. 5.

Nano-ESI MS monitoring of aTf displacement from the receptor by Fe₂Tf at pH 8.1. Concentration of aTf (24 μM) and TfR (13 μM) remained constant, while concentration of Fe₂Tf was increased incrementally as indicated by the numbers shown on the right-hand side of A. B shows full-range mass spectra prior to Fe₂Tf addition (gray) and following addition of Fe₂Tf raising its concentration in solution to 34 μM (black). Relative abundances of ionic species representing all three transferrin/receptor complexes, (aTf)₂·TfR, aTf·Fe₂Tf·TfR and (Fe₂Tf)₂TfR, as a function of the amount of Fe₂Tf added to the solution are shown in C. Relative abundance for each species was calculated by fitting the raw data (A) with Gaussian curves and summing up their heights for all observed charge states.

Relative abundance of ionic species representing all three Tf/TfR complexes, (aTf)₂·TfR, aTf·Fe₂Tf·TfR, and (Fe₂Tf)₂TfR, as a function of the amount of Fe₂Tf added to the solution clearly indicate consecutive replacement of each of the two aTf molecules from the receptor by Fe₂Tf (Fig. 5C). No unbound Fe₂Tf is observed in the low m/z region of mass spectra until the amount of added Fe₂Tf exceeds the amount of TfR present in solution by more than 2∶1, indicating that each incremental addition of Fe₂Tf results in displacement of a stoichiometric amount of aTf from TfR.

Similar experiments were carried out in order to determine if aTf itself has any ability to displace other weak binders from TfR in the absence of iron-bound forms of hTf (Fig. 6). Human TfR was mixed and incubated with iron-saturated Tf molecules from nonhuman sources (bovine serum transferrin, bTf, and chicken ovo-transferrin, oTf). Consistent with earlier cell binding studies (21), ESI MS analysis of this solution reveals weak bTf/TfR interaction and virtually no interaction of oTf with the human receptor (Fig. 6A). Addition of aTf to this solution results in facile displacement of bTf from TfR, with the most abundant ionic species in the high m/z region of the spectrum representing the (aTf)₂·TfR complex (Fig. 6B).

Fig. 6.

A: nano-ESI MS analysis of complexes generated by incubating bTf and oTf in 13.5 µM TfR at pH 8.1 (total transferrin concentration 22 μM, oTf∶bTf ratio ca. 2.5∶1). B: same mixture following addition of 22 μM aTf. Peaks corresponding to 1∶1 complexes are labeled with single triangles, and 2∶1 complexes are labeled with double triangles (color indicates the bound transferrin(s): bTf, black; oTf, gray; aTf, white).

Discussion

It is well established that the iron status of hTf is a major determinant of its interaction with TfR both at the cell surface and in the endosome. The results of earlier cell binding assays suggested that iron removal from hTf results in a dramatic reduction of its receptor affinity (6–9), while the results of more recent biophysical studies were consistent with a complete loss of the receptor-binding competence for the apo-hTf under conditions mimicking the extra cellular environment (11–14). ESI MS provides conclusive evidence that aTf has the ability to form relatively stable complexes with TfR at neutral and mildly basic pH. The observed aTf/TfR complexes have the same stoichiometry as the complexes formed by Fe₂Tf (Fig. 1). Specificity of the aTf/TfR association at neutral pH is confirmed not only by demonstrating the inability of other ubiquitous serum proteins such as HSA to bind TfR under similar conditions (Fig. 1D), but also by the fact that associations of aTf and Fe₂Tf to each monomeric unit of TfR are mutually exclusive. Indeed, when Fe₂Tf is present in solution at levels insufficient for complete TfR saturation, aTf binds to the receptor only by filling the vacancies left by Fe₂Tf (Fig. 2). Once the Fe₂Tf/TfR molar ratio exceeds 2∶1 (complete receptor saturation), no association between aTf and TfR is observed, consistent with the notion of either identical or significantly overlapping binding interfaces in Fe₂Tf/TfR and aTf/TfR complexes.

The specificity of the aTf/TfR interaction provides a unique opportunity to evaluate it in a semiquantitative fashion by ranking it among other TfR binders. Competition experiments (Fig. 3) provide clear evidence that iron status of hTf does determine the rank of its receptor affinity: Fe₂Tf≫Fe_NTf ≈ Fe_CTf > aTf, effectively placing an upper limit on TfR affinity of aTf at neutral pH as 0.2 μM (9). At the same time, the ability of aTf to displace bTf from TfR (Fig. 6) indicates that the 0.6 μM affinity of bTf to human receptor (9) can be used as a lower limit of the aTf/TfR binding strength. Importantly, this bracketing approach may be used to evaluate receptor affinity for any Tf-related protein, including covalent Tf-cytotoxin conjugates and chimeric proteins.

Specific binding of aTf to TfR at mildly basic (near-physiological) pH with modest, but measurable (sub-μM) affinity is rather surprising, since Tf conformation was suggested as the major factor enabling its recognition by the receptor (12, 13, 22, 23). Iron dissociation from Tf triggers large-scale conformational changes in each of its lobes, which may interfere with the ability of this protein to make contacts with key TfR residues in the binding interface. Recent analysis of the low-resolution structure of the Tf/TfR complex suggests that only conformational changes triggered by iron removal from the N-lobe may be restricted by hTf binding to TfR (22). We note that despite the existence of two distinct conformations (open in the iron-free protein and closed in the holo-form), Tf molecules are highly dynamic in solution, and may sample both open and closed states in the absence of iron (24).

The existence of a fine balance between open and closed conformations of Tf in the absence of iron, with very little energy difference between them, would allow the protein to sample both states frequently, as long as they are separated by a small activation energy barrier (24). Therefore, aTf may transiently populate the conformation required for TfR recognition at neutral pH. Binding to TfR may further increase the Boltzmann weight of the “noncanonical” closed conformation of aTf, although it apparently fails to lock aTf in this TfR-favored conformation, as evidenced by the ease of displacing aTf from TfR with Fe₂Tf (Fig. 5). Interestingly, recent attempts to obtain a structure of apo-hTf bound to the receptor using electron microscopy produced electron density maps, whose resolution could not be improved beyond 16 Å (23), attributed to the significant structural variability of iron-free hTf.

The canonical (open) conformation of apo-hTf is stabilized by the TfR in the mildly acidic endosomal environment, a phenomenon which is often viewed as an indirect facilitator of endosomal iron release from Tf (13). The dramatic reversal of the TfR affinities of aTf and Fe₂Tf is indeed evident upon mild acidification of the aTf/Fe₂Tf/TfR mixture (Fig. 4). The molecules of aTf present in solution at mildly basic pH displace Fe₂Tf from TfR, although the Fe₂Tf/TfR binding is not completely eliminated under these conditions. Indeed, in a similar experiment carried out in the absence of aTf, Fe₂Tf molecules remained bound to TfR for at least half an hour and their iron status was not changed based on the inductively coupled plasma - optical emission spectrometry (ICP-OES) measurements. The low-resolution electron density maps of the apo-hTf/TfR complex crystallized at endosomal pH suggested the preference of TfR-bound apo-hTf to adopt an open conformation (23). Recently, evidence was presented that the preferred recognition of the open conformation of Tf by TfR at endosomal pH is a result of acid-triggered conformational transition in the remote regions of the receptor, which are then transmitted to the binding site (13). The ability of Fe₂Tf to interact with the TfR under these conditions (e.g., by filling the vacancies left by aTf) should not be surprising, and its explanation is also likely to invoke the notion of noncanonical conformations.

The ability of the holo-protein to sample a noncanonical open conformation at endosomal pH was postulated a long time ago based on the results of an ESI MS study of iron dissociation from the N-lobe of hTf under mildly acidic conditions (25), and the existence of this state was recently confirmed by hydrogen/deuterium (H/D) exchange measurements (26). It appears that the receptor alone cannot trigger iron dissociation from Fe₂Tf under endosomal conditions, and additional (still unidentified) players must be considered in order to achieve full understanding of the iron release process. Additional stabilization of the noncanonical open conformation of Fe₂Tf by TfR may nonetheless be a critical component of this process, as it should enhance the access of organic chelators and/or other metalloproteins, such as ferrireductase, to the iron-binding sites of Tf (22).

Whatever the specific mechanism for iron removal from hTf in the endosomal environment is, apo-hTf is likely to be recycled back to the cell surface in the TfR-bound form. On average, each TfR molecule spends less than 3 min on the cell surface before being internalized again (27). Although the commonly accepted model of iron delivery to tissues implies that the apo-hTf/TfR complex dissociation occurs within seconds upon its exposure to the neutral environment at the cell surface (28), our data indicate that the protein may remain bound to TfR for extended periods of time (vide supra). However, this does not interfere with the iron delivery process, since the iron-loaded hTf molecules displace iron-free Tf from the receptor (Fig. 5). Such displaced apo-Tf molecules would contribute to the circulating pool of metal-free transporters available for the next cycle of intestinal iron binding, transport, and delivery to tissues needing this metal.

While the ability of apo-hTf to remain bound to TfR at near-physiological pH does not interfere with the normal iron delivery cycle, its modest receptor affinity may nonetheless be sufficient to pose a challenge for cellular entry for other forms of Tf molecules whose binding to TfR is less efficient than that of the wild-type hTf. Indeed, our studies provide strong evidence that aTf is perfectly capable of displacing Tf molecules of nonhuman origin from TfR regardless of their iron status (Fig. 6). Recently, Tf has emerged as an attractive delivery vehicle for cytotoxic agents that target cells with elevated demand for iron, e.g., those in neoplastic tissues (4). The underlying assumption in designing these therapeutic strategies utilizing Tf as a delivery vector is that the conjugated form of Tf should retain its ability to enter the cell as long as the receptor recognition is not abolished by the protein modification. The results of our study suggest that this assumption may be too optimistic, as the efficient cellular entry may not only require the ability to bind to a free receptor at the cell surface, but also to displace iron-free hTf molecules from the receptor. This challenge is multiplied due to the vast amounts of hTf in the blood (ca. 35 μM), against which the modified forms of transferrin would have to compete. Therefore, minimizing the influence of Tf conjugation on its receptor affinity must become a critical part in the design of efficient antitumor therapies.

Materials and Methods

Materials.

HSA was purchased from Sigma-Aldrich Chemical Co. and the diferric form of hTf (Fe₂Tf) from Intergen, Inc. Iron-free hTf (residual iron content less than 5% of the fully saturated level as shown by ICP-OES) was prepared by acid-denaturing the protein in the presence of EDTA followed by centrifugal ultrafiltration (Centricon microconcentrators) to 100 mM ammonium bicarbonate. Proper folding was verified by the analysis of protein ion charge state distributions in ESI MS. Recombinant mutant forms of hTf (Fe_NTf, Fe_CTf, and aTf) and TfR were produced as described previously (18 and 29). Amino acid sequences of all proteins used in this work are presented in the SI Text section.

Mass Spectrometry.

ESI MS analyses of Tf/TfR binding were carried out with a JMS-700 MStation (JEOL) two-sector mass spectrometer. Typically, a protein solution in 100 mM NH₄HCO₃ was injected into the ESI source at 3 μL/min flow rate. Displacement of low affinity Tf from TfR was monitored with a QStar-XL (ABI-Sciex) hybrid Q-TOF MS equipped with a nanospray source. Medium borosilicate emitters (Proxeon Biosystems A/S) were used to spray protein solutions.

TfR Binding Studies.

Typically, TfR was added to 10 μM Tf solutions (in 100 mM NH₄HCO₃, pH 8.1) to achieve a final TfR concentration of 6 μM. Displacement of aTf from TfR by Fe₂Tf was performed by incremental addition of Fe₂Tf to the solution containing 13.5 μM TfR and 24 μM aTf (each addition raising Fe₂Tf concentration by 2 μM). All protein mixtures were incubated for 30 min prior to ESI MS analyses. Displacement of bTf from TfR by aTf was accomplished by addition of 22 μM aTf to the solution containing 13.5 μM TfR₂, 22 μM of oTf and bTf, and incubating the mixture for 24 h at 4 °C. The sample was equilibrated at room temperature prior to the ESI MS analysis for ca. 10 min. prior to the ESI MS analysis. Acidification of the aTf/Fe₂Tf/TfR mixture entailed a 1∶16 dilution of the initial protein solution (in 100 mM NH₄HCO₃) with 100 mM NH₄CH₃CO₂ (pH 5.0) to a final pH ∼5.6, followed by 30 min incubation at room temperature and preconcentration prior to ESI MS analysis.