The mechanism and specificity of iron transport in Rhodotorula pilimanae probed by synthetic analogs of rhodotorulic acid.

The yeast Rhodotorula pilimanae produces the dihydroxamate siderophore rhodotorulic acid (RA) in prodigious amounts when starved for iron. Synthetic dihydroxamate analogs of RA have been prepared in which the diketopiperazine ring of RA is replaced by a simple chain of n methylene groups. It is found that R. pilimanae is able to accumulate iron using these achiral complexes, as well as from simple monohydroxamate analogs, at rates comparable to those of RA. While the Fe2RA3 complex does not enter the cell, there is a receptor system whose geometric requirements for siderophore recognition have been probed using analogs. In contrast to mono- or dihydroxamate ligands, the trihydroxamate siderophores such as ferrioxamine B are completely ineffective at delivering iron to R. pilimanae. This is ascribed to the greater stability of these complexes, which blocks release of the Fe(III) in a ligand exchange process that is required for uptake. To explore whether this ligand exchange involves redox catalysis, Ga(III) was substituted for Fe(III). The gallium was taken up at rates near those of iron and were also energy-dependent, as determined by metabolic inhibition with KCN.

The yeast Rhodotorula pilimanae produces the dihydroxamate siderophore rhodotorulic acid (RA) in prodigious amounts when starved for iron. Synthetic dihydroxamate analogs of RA have been prepared in which the diketopiperazine ring of RA is replaced by a simple chain of n methylene groups. It is found that R. pilimanae is able to accumulate iron using these achiral complexes, as well as from simple monohydroxamate analogs, at rates comparable to those of RA. While the Fe2RA, complex does not enter the cell, there is a receptor system whose geometric requirements for siderophore recognition have been probed using analogs. In contrast to mono-or dihydroxamate ligands, the trihydroxamate siderophores such as ferrioxamine B are completely ineffective at delivering iron to R.
pilimanae. This is ascribed to the greater stability of these complexes, which blocks release of the Fe(II1) in a ligand exchange process that is required for uptake. To explore whether this ligand exchange involves redox catalysis, Ga(1II) was substituted for Fe(II1). The gallium was taken up at rates near those of iron and were also energy-dependent, as determined by metabolic inhibition with KCN.
Since all life forms appear to have an absolute requirement for iron, the introduction of this element into a wide range of biochemical processes appears to have taken place very early in the evolution of life. For aerobic organisms the concentration of free, aqueous ferric ion is limited to lo-*' M at neutral pH due to the insolubility of Fe(OHh This is the driving force for the excretion by microbes of strong iron-chelating agents (siderophores) and the expression of high-affinity transport systems which provide a reliable cellular iron supply (1). Siderophore iron uptake has been investigated intensively in Escherichia coli (2), Ustilago sphaerogena (3,4), Neurospora crmsa (5, 6), Bacillus megaterium (7), and Streptomyces pilosus (8, 9). Synthetic siderophore analogs have been used as probes to explore the specificity of the siderophore transport systems. With synthetic enantioferrichrome and enantioenterobactin it was demonstrated in E. coli (10,11) that the uptake systems are able to discriminate between the mirror images of these ferric siderophore complexes. Through the * This research was funded by National Institutes of Health Grant AI 11744. This is Paper 31 in the series "Coordination Chemistry of Microbial Iron Transport Compounds." Ref. 9 is the previous paper in this series. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Recipient of a fellowship from the Deutsche Forshungsgemeinschaft.
4 Author to whom correspondence should be addressed. use of selectively varied synthetic enterobactin analogs in E. coli, results were obtained that suggested the primary importance of the catecholate iron portion in recognition and uptake at the membrane, but at most a minor role for the cyclic serine triester backbone (12).
Under iron-deficient growth conditions the yeast Rhodotorula pilimunae excretes vast amounts of rhodotorulic acid (RAI) (13). The structure of RA, a cyclic dipeptide (diketopiperazine) of 8-N-acetyl-L-(S)-rS-N-hydroxyornithine is shown in Fig. 1 For a receptor/uptake system several features of the ferric RA complexes may potentially be important for recognition: 1) the diketopiperazine rings, 2) the geometrical configuration about the metal centers and the direct surroundings (e.g. methyl groups), 3) one or both iron centers, 4) the distance between the iron centers, and 5 ) the entire molecular structure.
In order to characterize those molecular features of RA which are critical to the iron acquisition process of the yeast, we have synthesized analogs of RA with selectively modified sites. In these model ligands the hydroxamate groups are separated by varying numbers (n = 3 to 6 and 8) of methylene groups. These molecules are designed to probe the importance of the diketopiperazine rings and the iron-iron distance in the uptake process. These dihydroxamate ligands form complexes in aqueous solution with Fe3+ that are analogous to those of RA. The formation constants per ferric ion are -1031, identical within experimental error to that determined for Fe2RA3 (15).
To test the influence of steric hindrance around the hydroxamate metal centers, dimerum acid ( Fig. 1) was also used as a probe. Furthermore, in order to test whether one or two metal centers are necessary for iron transport, the uptake characteristics of synthetic monomeric trishydroxamate complexes such as acetohydropamate and l-hydroxy-2-(1H)-pyridinones ( Fig. 1) were also investigated.

MATERIALS AND METHODS'
The abbreviation used is: R A , rhodotorulic acid.

Iron ~p~~ in R. p~~i~~e ~e~~e d by ~i~~ Iron
C o~p~~~-~g . 2 shows the c o m p~s o n of =Fe uptake mediated by dimeric hydroxarnates and t~h y~o x~a t e siderophores. All complexes were added to cell cultures at 10 ELM concentrations. Iron uptake only occurred in the presence of dimeric complexes; no uptake occurred with trihydroxamate siderophores such as ferrioxamine €3, coprogen, and ferricrocin. The species own siderophore, RA, e x h i b i~ the highest uptake rate; the ferric complex of dimerum acid gave the lowest rate (28% of the ferric RA uptake, measured &r 10 min). Iron was supplied to the celis by the synthetic analogs n z: 6 iso, n = 6 me, and n = 8 me at rates between these extrema2 even though these analogs lack the diketopiperazine ring.  complexes on the =Fe uptake in R. ~i l~~~, the series of Nisopropyl analogs in which the n u m~r of bridging methylene groups varies from 3 to 6 was invest~ated Although the n = 6 compound is still shorter than the chain length of RA, the higher chain length analogs are too insoluble for these experimenta. Time-dependent uptake measurements with the ferric complexes of the synthetic isopropyl analogs n = 3 to 6 (Fig. 1) showed that n = 6 and n = 5 supply iron at 80 and 74% the rate of 55Fe2R&, respectively. The n = 4 is0 and n = 3 is0 compounds mediated the iron transport at rates similar to monomeric ferric tris(acetohydrox~ate), which was between 50 and 60% of ferric RA. All experiments were p e~o~~ at 10 ELM substrate concentrat~ons, where ~€~s i o n into the cell membrane contribuless than 1.5% to the overall iron accumulation after 10 min, as measured a h r addition of 1 m, KCN, a respiratory poison (data not shown). Fig. 3 illustrates that the concentration dependence of uptake of the RA analogs is quite different from that for ferric RA. A maximum in uptake rate for 55Fez& was reached at IO to 20 ELM. However, the analogs did not show sa~ration kinetics in supplying iron to R. ~~~~~~. At substrate concentrations less than 10 to 20 FM the uptake rates were lower Fe uptake rates about 10% lower than the n = 6 is0 compound. The n = 3 is0 compound (also not shown) gave uptake about 10% greater than the n = 4 is0 compound.
than for 55FezRA3; however, above that concentration the iron transport rates mediated by the analogs exceeded the values of ferric RA accumulation. The active portion of this was probed using the respiratory pbison KCN. The same concentration-dependent kinetic experiments with the analog and RA complexes revealed that, in contrast to FezR&, iron uptake from the analogs is not completely inhibited by KCN (Fig. 3). In order to show the portion of uptake that is due to active transport, the rate of transport remaining after KCN inhibition must be subtracted from the uninhibited rate curves. When this is done, the difference curve for the n = 6 is0 analog is the same as that for the uninhibited RA-mediated uptake curve of Fig. 3. The difference curve for the n = 4 is0 analog is similar, but reaches a saturation rate that is 70% that of RA. The active transport portions of the ferric analogs n = 4 is0 and n = 6 is0 are, like RA, transported by a saturable process; the n = 6 is0 derivative is also essentially the same as FezRA3. The concentration-dependent uptake kinetics were also measured for the n = 6 me and n = 8 me analogs. They showed s u b s~t i a~y the same kinetics as the isopropyl derivatives (data not shown). 55Fe Uptuke from Monomeric Iron Complexes-When cultivated under iron-deficient conditions at low pH, R. pilimanae produces rhodbtorulic acid and a second cyclic monohydroxamic acid which has been characterized as l-hydroxy-3(S)-amino-2-piperidone (22). The latter has been shown to arise from cyclization of 6-N-hydroxyornithine. In this study we used analogous cyclic monohydroxamic acids, N,N-dimethyl-1-hydroxy-2-(1H)-pyridinone-6-carboxamide and 1hydroxy-2-(1H)-pyridinone-6-carboxylic acid (Fig. 1) to probe whether these compounds might serve as iron transport agents in R. ~i l i m a~-.
The solution chemistry of the iron complexes of these powerful chelating agents has been well characterized (20). At neutral pH and excess ligand (L) a 31 complex is formed (FeL3) with ferric ion. AI1 complexes were assayed at 10 pM concentrations. The uptake of labeled iron from these compounds, RA, and acetohydroxamate (a simple monohydroxamate) is linear for the first 10 min. Regardless of differences in the structure of these compounds, the uptake rates were about 50% of the iron accumulation from dimeric RA complexes. After 10 min about 2 nmol of Fe/mg of cells were absorbed, compared to 4 for Fe2R&. Incubation of the cells with 10 mM NaN3, an inhibitor of oxidative phosphorylation, resulted in total inhibition of iron uptake (less than 1% after 10 min, data not shown). This indicates that iron from the monomeric compounds,t as well as free RA, is transported by a process requiring energy. This is the f i t time it has been shown that hydrox~~idinones can act as siderophores.
Inhaition of Iron Uptake-The inhibition of iron hydroxamate uptake by increasing concentrations of chromic RA complexes should test whether RA analogs deliver iron to the cells by the same uptake system as RA. The chromic complexes, which are kinetically inert to Iigand exchange, were separated into geometrical and optical isomers as described elsewhere (17). Fig. 4 (Miniprint Section) shows the inhibition of 55Fe uptake mediated by RA, dimerum acid, and the n = 6 me and n = 6 is0 analogs (2.5 pi%) in the presence of increasing concentrations of the A-truns-CrzRA3 pr~paration. At that concent~tion in~bition of uptake occurred for all complexes except for n = 5 iso; it was only inhibited if added at 0.5 BM to the assay, in which case iron accumulation decreased to 80% at 30 PM CrzRA3 concentration.
Incorporation of iron from complexes of tris(acetohydr0xa-

mate) (Ac) and tris l-hydroxy-Z-(lH)-pyridinone (HOPO)
was also impeded in the presence of A-cis-CrzRAa; 40 CrzRA3 diminished Fe(HOP0)a uptake to 75% and F~( A C )~ uptake to 68% (data not shown). In summary, CrzRA3 inhibits uptake of iron from synthetic monomeric and dimeric hydroxamate in the same way as it does from RA. This indicates that the same receptor system is used in the iron acquisition mediated by these ligands.
U p~k e of 67Ga-Since Ga(II1) has no stable +2 oxidation state and thus cannot be reduced in biological systems, Ga(Il1) complexes of RA are excellent probes to elucidate dependence of the transport mechanism upon reduction. As shown in Fig.   5 (Miniprint Section), 67Ga is accumulated from RA, n .= 6 is0 and n = 6 me at exactly the same rates as the 55Fe label of ferric complexes of these ligands. Transport is dependent on metabolic energy as shown by addition of KCN (1 mM). At a substrate concentration of 10 p M diffusion into the membrane is low for n = 6 me and n = 6 is0 and after 20 min has only reached 2.5% of the total of 55Fe or 67Ga added to the cell suspension. From these results one can conclude that the synthetic ligands as well as RA supply iron to the cells by a nonreductive transport mechanism.

DISCUSSION
In this study it has been shown that R. pilimunue is able to accumulate iron at good rates from analogs of the siderophore rhodotorulic acid which include achiral synthetic dimeric complexes as well as from synthetic monomeric iron complexes. As with RA, iron uptake mediated by RA analogs involves an active transport system. Diffusion or passive accumulation only becomes significant at very high substrate concentrations and accounts for less than 1.5% of the overall transport at 10 p~ complex concentration in the assay. In transport of both RA and the analogs, reduction is not re-quired, as shown with the 67Ga complexes. Of particular importance is the inhibitory effect of chromic RA on the accumulation of analog-bound ferric ion. This indicates that the synthetic analogs and dimerum acid provide iron by the same transport system as RA. In addition, all synthetic ligands lack the diketopiperazine ring. Therefore, we conclude that the diketopiperazine ring does not play a major role in the recognition and uptake process.
The influence of the residues of the hydroxamate carbonyl group is demons~ated by dimenun acid (Fig. 2). The dimenun acid is identical to RA except that the terminal methyl groups are substituted by isopentenol residues. Compared to RA, the uptake of ferric dimerum acid is reduced by 70%, indicating a steric hindrance of the isopentenol groups for recognition and uptake. A further strong argument for the hypothesis that it is the terminal metal center which is of primary importance in recognition and uptake is the comparison of dimerum acid and the synthetic analogs n = 6 me and n = 6 iso. Iron from complexes of the analogs is taken up at higher rates than from dimerum acid. As shown in another investigation, there exists a strong discrimination in iron transport between RA and its enantiomer (17). Since the diketopiperazine ring is not of primary importance for recognition/uptake, the mirror image hydroxamate "propellors" and the opposite twist of the me$hylene chains must account for the reduced uptake of enantio-RA. These results again illustrate the importance of the terminal region in the accumulation process.
Unexpectedly, the n = 6 iso analog shows even higher uptake than la = 6 me (Fig. 2). The latter possesses methyl residues, like RA, whereas the isopropyl groups of n = 6 is0 are bulkier. Competition experiments with these analogs and chromic RA at low and high substrate concentration indicate that n = 6 is0 has a different interaction with the RA transport system than n = 6 me. Since n = 8 me is taken up at almost the same rate as n = 6 me, the deviant behavior of the n = 6 is0 iron complexes cannot be due to changes in the length of the molecule. It must be ascribed to the increase of hydrophobic portions around the metal centers. Emery and Emery (23) have observed a similar effect with semisynthetic analogs of ferrichrome, in which substitution of the hydroxamate-linked methyl groups by butyl functionalities resulted in increased iron supply to U. sphaerogena (23) compared to the native ferrichrome. Thus, lipophilic groups seem to be one important part of the recognition/transport process.
In order to measure the possible influence of the iron-iron distances on uptake rates, homologous analogs (n = 3 is0 to n = 6 iso) have been compared. With decreasing chain length a decrease in uptake rates is observed. Accumulation of iron from analogs n = 3 is0 and n = 4 is0 are in the range of acetohydroxamate, whereas analogs n = 5 is0 and n = 6 is0 display 74 and 80% of the ferric RA uptake. Since the geometry of the metal center is the same for all of these dimeric compounds, the decrease of uptake rates is due to the reduced molecular length.
All hydroxamic analogs show lower uptake rates compared to RA. Since the diketopiperazine ring plays no major role in accumulation, this must be due to additional structural features. In contrast to R A , the analogs have the hydroxamate carbonyl and N-hydroxyl groups interchanged, and they are achiral. For the racemic mixtures of the analogs the concentration of receptor active species might be as low as 50% compared to RA complexes which are predominantly A. In summary, we suggest three features to account for the observed differences in uptake rates: 1) the terminal residues, 2) the iron-iron distances, and 3) the chirality at the metal center and its effect on the position of the terminal groups.
It has been shown in an earIier paper, using 'H-labeled ligand and chromic RA that the ligand is not taken up in R. pilimnue (16). However, it remained unclear whether the iron removal involves reduction of the metal or merely ligand exchange by a superior substrate. The 67Ga experiments in this study eliminate the reduction mechanism for metal incorporation.
In R~o~o r u~ gracilis and another fungus, N. crasscb, it wm shown that the potential of the cycloplasmic membrane is caused by a proton gradient. Low pH regions at the membrane surface, produced by the proton motive force, could catalyze a shift in species distribution of ferric RA and analogs and promote the ligand exchange process which must accompany uptake. In fact, the results with hydmxamate siderophores, such as coprogen, ferricrocin, and ferrichrome B (all of which are not taken up), strongly suggest that the higher complex formation constants and/or the stability of these complexes over a wide pH range (2 to 8) are crucial for the lack of any ferric ion removal. The absence of metal accumulation from these complexes cannot be due to different stereochemistry. Ferricrocin, for example, displays the same structural features at the metal center part of the molecule as enantio-RA (21). The latter compound exhibits uptake rates which are half as high as those for ferric RA, whereas ferricrocin-bound iron is not taken up at all. Similarly, d i m e m acid shows low accumulation rates and coprogen does not release iron to the membrane-associated fe&ic ion acceptor. From uptake measurements with monomeric ferric complexes of h~d m x~a t e and l-hydroxy-2-(~)-p~idinones it has been demonstrated that two iron center substrate molecules are not absolutely required for transport by the RA uptake system. However, it cannot be completely excluded that these molecules might be small enough to occupy simultaneously two iron binding centers at the same recognition site. There is only poor resemblance of ferric 1-hydroxy-Z-(~) -p~d i n o n e to ferric RA. The two systems have in common a hydrophobic portion adjacent to the hydroxamate group of the ring and the solution chemistry is comparable. Lowering the pH leads to shifts in the species distribution favoring FeL2+ and FeLz* Species. Finally, it should be mentioned that hydroxypyridones are similar to the piperidone isolated by Akers and Neilands (22) from R. pilimnae lowiron cultures. This compound may serve as a second siderophore or as part of a membrane-bound system which functions as a ferric ion acceptor.