Background
Iron is an indispensable micronutrient for almost all species in the six kingdoms of life [
1]. The unicellular protozoan parasites that cause malaria are no exception and are wholly dependent on their host as a source of the vital metal [
2]. The redox properties of iron that enable it to readily cycle between the predominant ferric Fe
3+ and ferrous Fe
2+ oxidation states underpin biochemical reactions that function in a variety of cellular processes including those involved in energy production, respiration, and DNA replication [
1,
3]. However, those same redox properties that have been exploited by organisms for beneficial purposes, and that make iron an essential constituent of many biological macromolecules, also render it potentially cytotoxic: oxidation of excess ferrous iron to the ferric state in the cell cytoplasm by Fenton/Haber–Weiss chemistry results in production of reactive oxygen species that are injurious to nucleic acids, lipids and proteins [
4]. To prevent such cytotoxicity but at the same time ensure an adequate supply of essential iron, cells have evolved integrated mechanisms for maintenance of iron homeostasis through a carefully choreographed regulation of the systems that control iron acquisition and storage [
1]. Cellular iron acquisition is commonly mediated by the activities of membrane transporters or receptor-mediated endocytosis [
5‐
7]. Most bacteria, archaea, plant and animal cells accomplish intracellular iron storage by binding excess cytosolic Fe
2+ to ferritin, a protein that oxidises the ferrous iron and stores it in an essentially unreactive state [
1,
8]. However, some organisms do not express cytosolic ferritin and instead detoxify the cell cytoplasm by storing excess iron in vacuoles [
1]. In yeast, for example, the movement of ferrous ions into the vacuole is facilitated by activity of the vacuolar membrane-bound Ca
2+-sensitive cross complementer protein, CCC1 [
9,
10]. Plants synthesize CCC1 homologues—integral membrane polypeptides of between 250 and 400 amino acids that belong to the vacuolar iron transporter (VIT) family of proteins—that localize Fe
2+ to the vacuole [
11‐
19]. Homologues of plant VITs have been identified in other eukaryotes (but, significantly, not in animals), bacteria and archaea [
20]; and an orthologue of plant VIT is encoded by the genomes of several apicomplexan parasites, including those involved in pathogenesis of human disease states such as toxoplasmosis, cryptosporidiosis, and malaria [
20,
21].
The malaria parasite has evolved a complex life cycle that constantly alternates between vertebrate hosts and anopheline mosquito vectors. Within human hosts,
Plasmodium species are obligate intracellular parasites that transition through three different life cycle stages, including liver and erythrocytic stages during which the parasite must manage potentially toxic concentrations of iron [
22]. The asexual intra-erythrocytic stage of the parasite life cycle offers a particular challenge to iron homeostatic processes due to release of labile Fe
2+ into the cell cytosol by proteolytic digestion of host haemoglobin [
23] and the cytosolic concentration of labile Fe
2+ increases as the parasite matures from ring to schizont form [
24]. A VIT orthologue plays a major role in cytosolic iron detoxification of
Plasmodium species [
21]. Although this transporter is not critical for viability of the parasite, and it is expressed during all life cycle stages, experiments on mice infected with VIT-deficient
Plasmodium showed a significant reduction in parasite load in both liver and blood stages of infection compared to mice infected with wild type parasites [
21,
25]. These data suggest PfVIT could provide an efficient target for chemoprophylactic treatment of malaria.
The most virulent malaria parasite of humans [
26],
Plasmodium falciparum, expresses a 273 amino acid, ~ 31 kDa plant VIT orthologue, named PfVIT, that functions in cytoplasmic iron detoxification by transport of ferrous ions, via Fe
2+/H
+ exchange, into the endoplasmic reticulum [
21,
27]. In contrast to plant VITs and yeast CCC1, which have broader divalent metal cation specificity, PfVIT appears to function specifically as a Fe
2+ transporter [
21,
27]. To better understand the structural basis of substrate recognition and binding by PfVIT, the individual amino acid residues that constitute the Fe
2+ binding site of the protein were identified. This was aided by a recently determined x-ray crystal structure of the vacuolar iron transporter
EgVIT1 from the plant
Eucalyptus grandis with metal ion substrate bound [
28]. Using the structure of
EgVIT1 as a template, a high-quality homology model of PfVIT was constructed to identify the amino acid composition of the transporter’s substrate binding site and to act as a guide for subsequent mutagenesis studies. To test the effect of mutation of the putative substrate binding-site residues on PfVIT function a yeast complementation assay was used to assess the ability of overexpressed, recombinant wild type and mutant PfVIT to rescue an iron-sensitive deletion strain (
ccc1∆) of
Saccharomyces cerevisiae from the toxic effects of high concentrations of Fe
2+ [
9]. This combined in silico and mutagenesis approach enabled identification of individual amino acid residues essential for coordination of Fe
2+ to the PfVIT metal binding domain and speculation about the structural basis for the Fe
2+-specificity of the transporter.
Methods
All chemicals and reagents were Sigma-Aldrich brand purchased from Merck (UK) unless stated otherwise.
Homology modelling
Initially, a number of homology models of PfVIT protein were built using the automated comparative modelling servers ModWeb [
28], SWISS-MODEL [
29] and I-TASSER [
30]. The 273 amino acid primary sequence of PfVIT from the 3D7 isolate of
P. falciparum (PlasmoDB ID: PF3D7_1223700) was used as the target input and three-dimensional crystal structures of
EgVIT1, a 234-residue protein from the plant
Eucalyptus grandis (PDB IDs: 6IU3 and 6IU4) [
31], were used as structural templates. The best-scoring models from each server were selected for further filtering and subjected to additional validation analyses using PROCHECK [
32] and WHAT IF [
33], and visualized by the PyMOL Molecular Graphics System, Version 1.8.4.0 (Schrödinger, LLC). On the basis of these quality tests, the best-performing model was judged as one generated by SWISS-MODEL that used the 3.5 Å crystal structure of
EgVIT1 (PDB code 6IU4) as the template.
A separate model of isolated cytoplasmic metal binding domain (MBD) of PfVIT, which consisted of 79 residues (Ala101-Leu179), was generated with MODELLER 9.24 [
34] and the EasyModeller 4.0 GUI [
35] using the 3.0 Å structure of the MBD fragment of
EgVIT1 bound to Fe
2+ (PDB: 6IU9) as the template. The initial best model was further optimised, refined and energy minimised in MODELLER. Statistical quality assessment of the model was performed using PROCHECK [
32]. The PfVIT MBD homology model was then superposed onto the isolated
EgVIT1 MBD crystal structure using the ‘align’ command of PyMOL. The atomic coordinates of the bound Fe
2+ ion from the superposed
EgVIT1 MBD structure were subsequently extracted and transferred to the PfVIT MBD homology model to yield a final model of PfVIT cytoplasmic MBD bound to Fe
2+ ion. Binding of Fe
2+ to the final model was analysed using the CheckMyMetal (CMM): Metal Binding Site Validation Server [
36] to identify individual amino acid residues that could potentially coordinate Fe
2+ and form the binding site.
Gene synthesis and site-directed mutagenesis
Previous work demonstrated that full-length
Plasmodium VIT was expressed at low levels and was functionally impaired in a yeast heterologous expression system. In contrast, N-terminal truncation versions of the transporter demonstrated increased expression levels and activity and were easily detectable in vacuolar membrane fractions of the same system [
21]. Therefore, an N-terminal truncation mutant of PfVIT, designated sPfVIT, was designed for this study. The 711 bp sequence encoding the 237 amino acid, D
2-36 N-terminal truncation mutant of the VIT from
P. falciparum 3D7 (PlasmoDB gene ID: PF3D7_1223700) was codon-optimized for expression in
S. cerevisiae and synthesized using a commercially available service (GenScript, USA). To facilitate detection of expressed protein and future protein purification, an in-frame thrombin cleavage site,
myc epitope and His
10 affinity tag were engineered into the C-terminal end of the protein to give a final construct of 822 bp (see Additional File
1: Figure S1).
BamH1 and
Xho1 restriction sites were introduced into the 5′ and 3′-ends of the synthetic DNA, respectively, and the codon optimized sequence was ligated into the multiple cloning site of pESC-Leu expression vector (Agilent, UK) that contained the
LEU2 selectable marker. This gave rise to pESC-Leu-sPfVIT that encoded a 273 amino acid construct of molecular mass 30.9 kDa, expression of which was placed under control of the
GAL1 promoter. Site- directed mutants of sPfVIT used in this study were engineered using a QuikChange II XL Site-Directed Mutagenesis Kit (Agilent, UK) according to the manufacturer’s protocol with the DNA primers listed in Additional File
2: Table S1, and with pESC-Leu-sPfVIT as template DNA. The fidelity of all constructs was verified by DNA sequence analysis (Macrogen Europe, Amsterdam). Plasmids were propagated in
Escherichia coli XL10-Gold Ultracompetent cells (Agilent, UK) using carbenicillin for selection, then purified, quantitated and conserved at – 20 °C until required.
The budding yeast
S. cerevisiae BY4741 derivative strain (
MATa
his3Δ
1 leu2Δ
0 met15Δ
0 ura3Δ
0,
ccc1Δ::
KanMX) that lacked CCC1 vacuolar iron transporter was used as the model organism for this study. The recipient yeast strain was transformed using a previously published method [
37] except incubation temperatures were adjusted to 30 °C. Electrocompetent cells (45 µl aliquots) were transferred into pre-cooled sterile 1.5 ml Eppendorf tubes and 1–3 µg of DNA (either ‘empty’ pESC-Leu vector, wild type pESC-Leu-sPfVIT or mutant pESC-Leu-sPfVIT plasmid) added. The cells were then incubated on ice for 10 min prior to transfer to electroporation cuvettes. A single 1.5 kV, 200 Ω, 25 μF pulse was applied to the mixture using a MicroPulser Electroporator (Bio-Rad, UK) then 950 μl of YPD media (1% w/v Bacto™-yeast extract, 2% w/v Bacto™ -peptone, 2% w/v D-glucose) pre-warmed to 30 °C was added immediately into each cuvette. Cells were regenerated by incubation for 1 h at 30 °C then harvested by centrifugation (2200×
g, 5 min). Harvested cells were resuspended in 100 μl of sterile water, plated on solid synthetic complete media lacking leucine (SC-Leu; 6.68 g/l yeast nitrogen base, 1.4 g/l yeast synthetic drop-out medium supplements without leucine, 20 g/l select agar, 2% w/v
d-glucose) for selection and incubated for 2 days at 30 °C. Transformants were subsequently used for production of yeast cultures that overexpressed
P. falciparum VIT.
Overexpression of sPfVIT
Single colonies of each transformant were cultured in liquid SC-Leu medium containing 2% (w/v) D-glucose and incubated overnight at 30 °C with 180 rpm shaking. The overnight culture was diluted with SC-Leu to an OD600 of 0.2 in a sterile conical flask and further incubated until OD600 of 1.0. Protein expression was induced by addition of d-galactose to a final concentration of 3.5% (w/v) and the cultures grown for a further 20 h at 30 °C with shaking at 180 rpm. At the end of the induction period, samples of culture were taken for western blot analysis of protein expression levels and for use in qualitative and quantitative complementation assays designed to assess the ability of overexpressed protein to rescue the iron-sensitive ccc1∆ strain of S. cerevisiae from the cytotoxic effects of high concentrations of Fe2+.
Isolation of vacuoles from Saccharomyces cerevisiae
Isolation of vacuoles from
ccc1Δ S. cerevisiae cells that harboured ‘empty’ pESC-Leu vector (negative control cells) or that overexpressed wild type sPfVIT transporter from pESC-Leu-sPfVIT was performed based on a method described previously [
38]. 500 ml of galactose-induced yeast culture was centrifuged at 3000×
g for 5 min at room temperature. Harvested cells were resuspended in 50 ml of wash buffer (0.1 M Tris–HCl pH 9.4, 10 mM dithiothreitol) by gentle vortexing. The suspension was then incubated at 32 °C for 10 min then centrifugated at 2240×
g for 6 min. The resulting pellet was resuspended in spheroplasting buffer (0.75 mM KH
2PO
4, 0.2% SC-Leu medium, 0.09 M sorbitol) containing 1 tablet of cOmplete™ EDTA-free Protease Inhibitor Cocktail (Roche) to a volume that contained 2 × 10
9 cells/ml. Next, 50 U of Zymolyase 20T (MP Biomedicals, UK)/g wet weight of yeast cells was added and the mixture was incubated at 32 °C for 30 min with occasional swirling prior to centrifugation at 1000×
g for 2 min at 4 °C. Spheroplasts were resuspended by gentle vortexing in 2.5 ml of 15% Ficoll solution (10 mM Pipes-KOH pH 6.8, 0.2 M sorbitol, 150 g/l Ficoll). An appropriate volume of Dextran solution (the µl volume of which was calculated by taking the final OD
600 of a culture after 20 h of galactose induction × culture volume × 0.15) was added and the suspension held on ice for 2 min with occasional swirling. The suspension was then incubated at 32 °C for 3 min and placed back on ice. The spheroplasts were carefully layered onto a density step gradient consisting of 0%, 4%, 8% and 15% Ficoll solutions. The gradient was centrifuged at 175,000×
g for 90 min at 4 °C and the vacuoles carefully harvested from the 0 to 4% Ficoll interface. The total protein content of the recovered vacuoles was quantified subsequent to analysis by western blot.
Yeast complementation assays
Cultures from single colonies of the ccc1∆ strain of S. cerevisiae transformed with either ‘empty’ pESC-Leu vector, or pESC-Leu that encoded wild type or mutant sPfVIT, were grown and protein expression induced as described above. At the end of the induction period the cultures were diluted to OD600 of 0.2 with SC-Leu medium. The cultures were then twofold serially diluted and 10 µl of each spotted onto solid medium induction plates (SC-Leu agar containing 3.5% w/v galactose and 2 mM ascorbic acid) with/without 7.5 mM ammonium iron (II) sulphate. The ascorbic acid and ammonium iron (II) sulphate solutions were freshly prepared then argon flushed to minimize oxidation of the ferrous iron during handling. Plates were incubated at 30 °C for 3 days prior to imaging with an Azure Biosystems C200 gel documentation imager (Cambridge Bioscience, UK).
A quantitative analysis of the ability of wild type and mutant sPfVIT to rescue the ccc1∆ strain of S. cerevisiae was performed using a colony count assay. Cells were grown and protein expression induced as described above. At the end of the induction period samples of each culture were taken and diluted with SC-Leu medium to OD600 of 0.002. Aliquots (50 µl) of diluted cultures were spread onto Petri dishes that contained solid media (SC-Leu agar, 3.5% w/v galactose, 2 mM ascorbate) with/without 7.5 mM ammonium iron (II) sulphate. The plates were incubated at 30 °C for 3 days then individual colonies counted. All assays were performed in triplicate. The number of colonies formed per ml of OD600 1.0 was calculated for each and the data analysed using one-way analysis of variance (ANOVA) and Dunnett’s multiple comparison tests.
Protein quantitation and western blot analysis
Total protein in yeast whole cell samples and in vacuoles was determined by Pierce BCA Protein Assay Kit (ThermoFisher Scientific, UK) used according to the manufacturer’s instructions. Samples of yeast cell cultures that had sPfVIT expression induced for 20 h were prepared for SDS-PAGE by adjustment of OD600 to 1.0 in a volume of 1 ml with ultrapure H2O. Cells were pelleted by centrifugation, resuspended in 200 μl of 0.2 M NaOH and incubated at RT for 10 min prior to addition of 75 μl of 5× concentrated SDS-PAGE sample loading buffer (300 mM Tris–HCl pH 6.8, 50% v/v glycerol, 5% w/v SDS, 0.05% v/v bromophenol blue, 250 mM dithiothreitol). Samples containing 30 µg of total protein (for vacuole samples 7 µg of total protein was added) were electrophoresed on a 10-well Novex WedgeWell 4 to 20% Tris–Glycine gel (Invitrogen, UK) in Tris–Glycine SDS running buffer at constant 200 V for 45 min. Proteins were then transferred from the gel onto a 0.22 µm nitrocellulose membrane (MDI Membrane Technologies, USA) at constant 30 V for 16 h at 4 °C using a Mini Trans-Blot apparatus (Bio-Rad, UK). After transfer the membrane was blocked for 1 h at RT in TBS-T buffer (50 mM Tris pH 7.5, 0.01% v/v Tween 20, 150 mM NaCl) containing 1.0% w/v BSA then washed twice in TBS-T buffer. For detection of the His10 tag of the recombinant transporter the membrane was probed at RT for 1 h with HisProbe-HRP conjugate (ThermoFisher Scientific, UK) diluted 1:5000 in TBS-T buffer pre-treated with 0.1% w/v BSA. The membrane was then washed four times in TBS-T buffer prior to detection of the probe with SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoFisher Scientific, UK). The membrane was subsequently imaged with a G:BOX Chemi XRQ gel doc system (Syngene, USA).
Discussion
Despite the gradual annual decline in malaria cases and deaths over the past decade, emerging and growing resistance to frontline drugs used to treat the disease has highlighted a need for identification and characterisation of novel anti-malarial drug targets [
25,
40]. A vacuolar iron transporter family homologue that functions in iron detoxification during liver and asexual blood stages of
Plasmodium species could represent just such a target for novel prophylactic drugs [
21,
41]. In the absence of an experimentally determined three-dimensional structure of any
Plasmodium VIT, a high-quality homology model of PfVIT from the human malaria parasite
P. falciparum was built and used as a guide for mutagenesis studies aimed at identification of individual amino acid residues that constitute the substrate binding site of the protein; information critical not only for a better understanding of PfVIT function but also for structure-based design of novel inhibitors of the transporter.
The structural similarities between the PfVIT homology model and plant VIT1 crystal structure suggest a common gross architecture for eukaryotic VITs. This similarity is particularly evident in the cytoplasmic MBD of the transporters; several glutamate and methionine residues that form the metal binding site(s) of plant VIT1 are not only fully conserved in the primary sequence of PfVIT, but the binding sites also exhibit a striking degree of structural conservation. Despite this, the malaria and plant VITs possess different metal cation substrate recognition profiles. PfVIT is specific for Fe
2+ whereas plant VITs are more promiscuous and can bind and transport divalent Zn, Mn, Co or Ni cations in addition to Fe
2+ [
12,
21,
27,
31]. Yeast CCC1 is also capable of transporting Mn
2+ and Ca
2+ ions [
10]. Differences between
Plasmodium and plant or yeast VITs may also extend to metal ion occupancy of the cytoplasmic MBD. Although our PfVIT model suggested only one Fe
2+ ion is bound to the MBD, the crystal structure of plant VIT1 revealed that two Zn
2+ ions and a water molecule played a role in coordination of Fe
2+ to that transporter [
31]. While it is intriguing to speculate if PfVIT can bind more than one metal ion in the MBD, or if Zn
2+ is a cofactor necessary for PfVIT function, previous work demonstrated that purified recombinant PfVIT in detergent solution did not bind exogenously added zinc [
27].
The results of yeast complementation assays that assessed the ability of mutant PfVIT to rescue the
ccc1Δ iron sensitive phenotype of
S. cerevisiae suggested a plasticity of the metal binding site with respect to substrate binding and highlighted additional differences between the
P. falciparum and plant VITs. Taken together, the assay results clearly demonstrated that a glutamate residue at positions 113, 116, 124, 127 or 165 in the PfVIT metal binding domain is not essential for transporter function. Individual removal of the negative charge at each of these positions by alanine scanning mutagenesis revealed only E165 to be sensitive to replacement by alanine. However, introduction of a neutral glutamine at the same position completely recovered the iron-tolerant yeast phenotype. Furthermore, substitution of glutamate by glutamine at positions 113, 116 and 127 greatly diminished but did not completely abolish the ability of
ccc1Δ yeast cells that expressed the mutant transporters to grow on medium that contained added iron. Introduction of a glutamine at position 124 had no significant effect on PfVIT function. In contrast, loss of the negative charge by individual replacement of the equivalent MBD glutamate residues with glutamine in plant VIT1 resulted in loss of function of all the mutant transporters [
31]. The resilience exhibited by PfVIT toward substitution of the conserved MBD glutamate residues suggests the malaria protein can remodel its binding site to utilise different coordination geometries for binding the Fe
2+ substrate in response to mutation. This contention is supported by a study of transition metal binding selectivity in proteins which showed that although Fe
2+ had a preference for octahedral coordination geometry, other Fe
2+-binding geometries with coordination numbers of 4 or 5 were also well represented in the set of test proteins [
42]. It is also possible that water molecules in OH
− or O
2− form could replace individual amino acid side chains as coordinating ligands to bind metal substrate in the mutant PfVIT proteins, as observed in the plant VIT1 structure [
31]. Although the homology model of PfVIT suggested a single Fe
2+ was bound to each MBD of the protein, the possibility of the presence of more than one distinct Fe
2+ binding site within the domain cannot be excluded; clarification of this will likely have to await the availability of an experimentally-determined, high resolution structure of PfVIT with substrate bound.
Regardless of the limitations of the PfVIT homology model, it can still provide a useful framework for interpretation of the mutagenesis experiments. The statistically significant gain of function observed for the PfVIT E127A mutation can be rationalised if consideration is given to the location of this residue on the cytoplasmic MBD of the protein. Inspection of the PfVIT MBD structure (Fig.
3b) revealed the acidic E127 to be located about midway on H1 of the MBD and facing toward the cytosol. Such positioning could allow E127 to function as a ‘gatekeeper residue’ to control entry/exit of Fe
2+ to/from the metal binding site. It is plausible, therefore, that the gain of function observed for the E127A mutant was due to removal of side chain bulk. Furthermore, it suggests that the negative charge on E127 may play an important role in controlling the binding of Fe
2+ to the MBD. Consistent with the notion of E127 as a gatekeeper residue was the loss of function observed when the acidic, negatively charged glutamate at this position was substituted with uncharged glutamine.
The information provided by the PfVIT homology model also allowed the effects of mutation of the E165 residue to be placed in a structural context. The loss of function observed when E165 was substituted with alanine, combined with the ability of the PfVIT E165Q mutant to rescue the iron-sensitive of
ccc1Δ yeast phenotype, suggested an important role for side chain bulk at position 165 for transporter function. The location of E165 at the C-terminal end of helix 3 of the MBD (see Fig.
3b), and with its side chain facing towards the substrate translocation pore of PfVIT, places this glutamate residue at an ideal position to facilitate diffusion of Fe
2+ along the protein surface from the binding site to the central pore of the transporter. PfVIT functions as an Fe
2+/H
+ exchanger [
27] and competition between protons and substrate is central to the transport mechanism of these proteins [
43]. The apparent necessity of an amino acid side chain that is capable of undergoing (de)protonation at position 165 of PfVIT raises the possibility that E165 could be a component of a proton relay that functions in translocation of H
+ countersubstrate across the internal compartment membrane. Further biochemical studies aimed at measurement of substrate binding to purified wild type and mutant transporter in detergent solution combined with transport measurements of protein reconstituted into liposomes will probably be required to confirm if this is indeed the case.
In contrast to the conserved glutamate residues of the PfVIT cytoplasmic MBD, a methionine at position 161 is indispensable to transporter function. Sulphur-containing residues play an important role in coordination of metal ions in the binding sites of many metalloproteins [
44,
45], and a conserved methionine in the metal binding site of plant VIT1 [
31] and the Nramp-family of transporters [
46] acts to confer metal ion selectivity for transition metal ions while discriminating against the alkaline earth metals. It is probable, therefore, that M161 performs the same selectivity function in PfVIT. Insight into the role of this methionine in metal ion selectivity can be provided by viewing substrate binding through the lens of the Lewis concept of acids and bases [
47]. Transition metal ions such as Fe
2+ are regarded by Lewis theory as borderline acids with properties that are intermediate to those of the electrophilic hard Lewis acids and the nucleophilic soft Lewis acids [
48]. The intermediate acid nature of Fe
2+ means it can bind to both hard and soft base ligands such as the carboxylate oxygens of glutamate side chains, and the thioether sulphur atom of methionine side chains, respectively. The methionine side chain, however, is rather hydrophobic and an unsuitable ligand for the hard acid alkaline earth metals. Although this characteristic imparts a discriminatory function to methionine with respect to metal ion recognition, it does not fully explain why plant VIT1 can recognize and transport Co
2+, Ni
2+ and Zn
2+ as well as Fe
2+ [
31] whereas PfVIT is selective for Fe
2+ [
21,
27]. Therefore, other subtle specificity determinants must be at play within the PfVIT binding site to enable discrimination between Fe
2+ and the other divalent transition metal cations. It may be that specific physicochemical attributes of Fe
2+, such as size and electron configuration, are exploited by the transporter to sculpt the composition and geometry of the metal binding site and the immediate surroundings to afford selectivity.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.