JOURNAL OF BACrERIOLOGY, May 1993, p. 2727-27330021-9193/93/092727-07$02.00/0Copyright © 1993, American Society for Microbiology

Vol. 175, No. 9

ao-Keto Acids Are Novel Siderophores in the Genera Proteus,Providencia, and Morganella and Are Produced

by Amino Acid DeaminasesHARTMUT DRECHSEL,' ANDREA THIEKEN,2 ROLF REISSBRODT,3 GUNTHER JUNG,'

AND GUNTHER WINKELMANN2*Robert Koch Institut, Bundesgesundheitsamt, Wemigerode,3 and Institut fuir Organische Chemie

and Institut fur Biologie, Mikrobiologie und Biotechnologie,2 Universitat Tubingen,Auf der Morgenstelle 1, D-7400 Tubingen, Germany

Received 9 December 1992/Accepted 16 February 1993

Growth promotion and iron transport studies revealed that certain at-keto acids generated by amino aciddeaminases, by enterobacteria of the Proteus-Providencia-Morganella group (of the tribe Proteeae), showsignificant siderophore activity. Their iron-binding properties were confirmed by the chrome azurol S assayand UV spectra. These compounds form ligand-to-metal charge transfer bands in the range of 400 to 500 nm.Additional absorption bands of the enolized ligands at 500 to 700 nm are responsible for color formation.Siderophore activity was most pronounced with ce-keto acids possessing an aromatic or heteroaromatic sidechain, like phenylpyruvic acid and indolylpyruvic acid, resulting from deamination of phenylalanine andtryptophan, respectively. In addition, cK-keto acids possessing longer nonpolar side chains, like a-ketoisocap-roic acid or a-ketoisovaleric acid and even a-ketoadipic acid, also showed siderophore activity which wasabsent or negligible with smaller a-keto acids or those possessing polar functional groups, like pyruvic acid,ce-ketobutyric acid, or a-ketoglutaric acid. The fact that deaminase-negative enterobacteria, like Escherichiacoli and Salmonella spp., could not utilize a-keto acids supports the view that specific iron-carboxylatetransport systems have evolved in members of the tribe Proteeae and are designed to recognize ferric complexesof both ce-hydroxy acids and oe-keto acids, of which the latter can easily be generated by L-amino aciddeaminases in an amino acid-rich medium. Exogenous siderophores, like ferric hydroxamates (ferrichromes)and ferric polycarboxylates (rhizoferrin and citrate), were also utilized by members of the tribe Proteeae.

Siderophores are essential for iron nutrition in virtually allmicroorganisms. Catecholate- and hydroxamate-type sidero-phores are well known and have been described for variousbacterial genera (17). Members of the family Enterobacteri-aceae have been studied extensively for siderophore produc-tion and iron transport. It has been shown that certainspecies of enterobacteria, such as Escherichia coli, Salmo-nella spp., Shigella spp., and Kiebsiella spp., producepredominantly or exclusively the catecholate-type sidero-phore enterobactin (9-11). Recently, we have shown thatother members of this family, namely, Erwinia spp., Pan-toea spp., Enterobacter spp., and Hafnia spp., produceferrioxamine-type siderophores (1, 13). Enterobacter speciesand certain other enterobacteria have previously beenshown to produce a mixed functional carboxylate-hydroxa-mate-type siderophore, named aerobactin (5, 10). Moreover,enterobacteria are able to utilize a variety of exogenoussiderophores produced by fungi and other bacteria (12),suggesting that the spectrum of utilizable siderophoreswithin these bacteria is even greater.Although a-hydroxyisovaleric acid had been identified

earlier as a siderophore in Proteus mirabilis (4), its generalimportance for iron nutrition in these bacteria seems ques-tionable, especially in the presence of amino acid-rich me-dia. It is well known that the amino acid deaminase reactionis the most distinguishing feature of the tribe Proteeae (6)and is responsible for a wide variety of ox-keto acids pro-duced from amino acids prevailing in a complex medium

* Corresponding author.

(15). We therefore monitored the production of variousa-keto acids in the Proteus-Providencia-Morganella groupand studied siderophore activity by growth promotion andiron transport assays. This work presents evidence thatot-keto and a-hydroxycarboxylic acids proved most suitablefor iron nutrition in the Proteus-Providencia-Morganellagroup and may be regarded as typical siderophores ofmembers of the tribe Proteeae.

MATERIALS AND METHODS

Strains and growth conditions. E. coli AB 2847 and Salmo-nella typhimurium TA 2700 are aroB and ent mutant strains,respectively, and are unable to synthesize enterobactin.These strains were used as a control for deaminase-negativeenterobacteria. Proteus vulgaris, P. mirabilis NM 12, P.mirabilis 8993, P. penneri, Providencia rettgeri 19, P. ret-tgeri 39, P. rettgeri 44, P. rettgeri 62, P. rettgeri 29395, P.rustigianii 2, P. alcalifaciens 4587, P. stuartii 167, P. stuartii712, P. stuartii 20137, Morganella morganii 13, M. morgandi22, and M. morganii SBK3 were all from the stock of theRobert Koch Institute in Wernigerode, Germany.Growth media. Strains were grown in Luria-Bertani me-

dium containing (per liter) 10 g of tryptone, 5 g of yeastextract, and 10 g of NaCl adjusted to pH 7.4. M9 minimalmedium, deferrated on a Chelex 100 column, plus 2%glucose (deferrated and autoclaved separately) and CaCl2(0.1 mM) and MgCl2 (1 mM), added separately, was used asa low-iron medium. M9 minimal medium was inoculatedwith 1% of a 24-h preculture in Luria-Bertani medium.

a-Keto acids (2-oxo acids), a-hydroxy acids, and other

2727

on February 13, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

2728 DRECHSEL ET AL.

siderophores. Pyruvic acid, a-ketobutyric acid, and a-keto-glutaric acid were from Merck GmbH, Darmstadt, Germany.a-Ketoadipic acid, a-ketoisovaleric acid (sodium salt), a-ke-toisocaproic acid (sodium salt), phenylpyruvic acid, andindolylpyruvic acid were obtained from Sigma, St. Louis,Mo. L-(-)-3-phenyllactic acid, D-(+)-3-phenyllactic acid,DL-3-indolyllactic acid, DL-3-(4-hydroxyphenyl)-lactic acid,L-2-hydroxyisocaproic acid, DL-2-hydroxybutyric acid so-dium salt, and (S)-(+)-2-hydroxy-3-methylbutyric acid werepurchased from Aldrich-Chemie, Steinheim, Germany. Fer-richrome, coprogen, ferrioxamine E, and rhizoferrin werefrom the stock of the Microbiology Department, Universityof Tubingen, Tubingen, Germany. The purity of sidero-phores was confirmed by high-performance liquid chroma-tography (HPLC) as described earlier (7).HPLC separation. HPLC separation was performed on a

reversed-phase column (Nucleosil C18, 5 pum, 4.6 by 250 mm)with a gradient of acetonitrile (10 to 50%) in water and 0.1%trifluoroacetic acid within 15 min on a Shimadzu HPLCsystem (two LC-9A pumps, an SCL-6B system controller,an SIL-6B autoinjector, a C-R4 AX-Chromatopac computer,and an SPD-6AV UV/vis spectrometric detector). The flowrate was 1 ml/min, and the detector wavelength was set at220 nm. Chromatograms were plotted after subtraction of asolvent gradient.Chrome azurol S (CAS) assay. The CAS assay was per-

formed as described by Schwyn and Neilands (14).UV and visible-light spectra. UV spectra of ferric a-keto

and a-hydroxy acids were recorded on an Ultrospec IIIspectrometer (Pharmacia LKB) connected to a CommodorePC40 computer. The acids were dissolved in water-methanol(3:1) and measured at concentrations of 1.5 (a-keto acids)and 1.45 to 1.7 (a-hydroxy acids) mM at an Fe/ligand ratio of1:3 at different pHs.Growth promotion tests. Growth promotion tests were

similar to those described earlier (12). A 5-p1 volume of a24-h culture grown in tryptone medium was used to inoculatetryptone agar containing 200 puM 2,2'-dipyridyl per plate.a-Keto or a-hydroxy acids (1 mg) were dissolved in 900 pl ofmethanol and mixed with a ferric chloride solution (2 mM,100 ,ul), resulting in a solution with a 0.2 mM iron complexconcentration (ratio of iron to ligand, 1:30). Filter discs (6mm in diameter) containing 10 pl (2 nmol) of the ferric a-ketoor a-hydroxy acid solution were used. Growth zones wereread after 24 h of incubation at 37°C.

Transport measurements. Time-dependent uptake of "5Femediated by a-keto acids and other siderophores was deter-mined as described earlier (16). Cells were grown overnightat 37°C in M9 low-iron minimal medium, washed with M9medium, and suspended in fresh M9 medium (optical densityat 578 nm, 0.7 to 1.0). The transport assay mixture contained2 pM iron (55Fe; specific activity, 0.435 Bq/pmol) in thepresence of 40 pM ligand (Fe/ligand ratio, 1:20). "Fe-labelled ferric complexes were added at time zero to theculture (6 ml), and samples (0.6 ml) were taken at intervals,filtered through membrane filters, and washed with 5 ml ofsaline. Radioactivity was counted in a liquid scintillationcounter. Uptake was calculated as nanomoles of Fe permilligram of dry weight.

RESULTS

Iron-binding properties of ao-keto acids. Valine, leucine,tryptophan, and phenylalanine were used as representativesof amino acids that are amenable to oxidative deaminationand iron complex formation. The resulting ferric complexes

CH3

\11

Vdln NH3+

CCH3 C/\11

//S // 0

O -------Fe3/3

O3

CH3 I ° 1C-7

HO----F/

0

11

C C

I-

0

11R / 'ZcX/

11 I

11

0-O------ Fet3+

FIG. 1. Structural formulas of valine, leucine (R = isopropyl),phenylalanine (R = phenyl), and tryptophan (R = indolyl) and theirdeamination products (a-keto acids = 2-oxo acids) depicted as ferriccomplexes of the oxo form and the enolized form.

can be formulated as keto-carboxyl or enol-carboxyl biden-tates (Fig. 1). Although the formation constants of at-ketoand a-hydroxy acids with iron seem to be comparable,studies with CAS revealed that the a-keto acids give moreintense decoloration zones than the corresponding a-hy-droxy acids, indicating that iron binding by keto acids ismore pronounced (data not shown). Moreover, among thea-keto acids studied, a-ketoisovaleric and a-ketoisocaproicacids were not as effective as the two aromatic ligandsphenylpyruvate and indolylpyruvate, suggesting that iron-binding strength is also considerably influenced by thenature of the side chains.UV and visible-light spectra of the ferric complexes of

phenylpyruvate and ao-ketoisocaproic acid at pHs 4.8 and3.1, respectively, revealed two absorption bands at around300 to 500 and 500 to 700 nm (Fig. 2), suggesting the presenceof ferric complexes with characteristic ligand-to-metalcharge transfer bands. Similar spectra were recorded withferric complexes of indolylpyruvate and a-ketoisovalericacid (data not shown). In contrast, the ferric complexes ofthe corresponding a-hydroxy acids showed only one absorp-tion band, in the range of 300 to 400 nm (Fig. 2), and noadditional absorption bands in the visible region. The visibleabsorption band of the at-keto acids led to yellow-green, red,or grey-violet with the phenylpyruvate, indolylpyruvate, anda-ketoisocaproate complexes, respectively, and is due to theenolizing ability of the keto group. The concentrations usedwere 0.5 mM ferric chloride and 1.5 mM phenylpyruvic acidin a water-methanol (3:1) mixture which allowed formationof mononuclear 1:3 Fe-ligand complexes. UV measurementswere made in a slightly acidic solution to prevent competi-tion by hydroxide ions. Although acidic, these pHs are stillphysiological. Acid production is typical in enterobacteriawhich are methyl red positive (pH 4.4), suggesting that a

J. BACTERIOL.

on February 13, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

PROTEUS, PROVIDENCIA, AND MORGANELLA a-KETO ACIDS 2729

scant

0.80 Phenylpyruvic acid

a_~~~~~4seeea " 8

0.00

0.60 \ -Xetoiocaproic acid2W 3W 40 S" 000 700 I"

1,000 ,e .0 0. .

0.56~ 2-Ktoioapoi acid

0.20

g PhWilectic acid

0.00

200 300 400 S" C00 700 600

Sam$

1.40

1.00 Pheylt-Hydraproicacid0.620

0.00200 300 400 SW 600 700 500

Usw.Loms(mlI

FIG. 2. UV and visible-light spectra of ferric complexes of twoa-keto acids (1.5 mM ligand and 0.5 MM ferric chloride), phe-nylpyruvic acid (pH 4.8) and 2-ketoisocaproic acid (pH 3.1), and twoa-hydroxy acids (1.4 mM ligand and 0.5 mM ferric chloride),L-phenyllactic acid (pH 3.5) and 2-hydroxyisocaproic acid (pH 4.0).Abs, absorbance.

slightly acidic environment does not interfere with ironacquisition by a-keto acids in members of the tribe Proteeae.

Isolation and characterization by HPLC. Figure 3 showstwo HPLC chromatograms of standard mixtures of com-monly found a-keto and a-hydroxy acids, with a gradient of

a -00

CMC4

0.U)c(A'San

IPl

0

KB KiC

a lS Xin KiV

10 20

Time (mini)

PL

b %

F.c0

e4C4

0

H

')W0

0 HE

iV

HiC

IL

0 5 10 15

Time (min)FIG. 3. HPLC separation of a standard mixture of a-keto acids

(a) and a-hydroxy acids (b). Column: Nucleosil C18, 5 pm, 4.6 by250 mm. Conditions: gradient of 10 to 50% acetonitrile in water(0.1% trifluoroacetic acid) within 15 min. Flow rate, 1 ml min-'.Detector wavelength, 220 nm. Designations: KB, ketobutyric acid;KiV, ketoisovaleric acid; KiC, ketoisocaproic acid, IP, indolylpyru-vic acid; PP, phenylpyruvic acid; HB, hydroxybutyric acid; HiV,hydroxyisovaleric acid; HiC, hydroxyisocaproic acid; PL, phenyl-lactic acid; IL, indolyllactic acid.

10 to 50% acetonitrile in water (plus 0.1% trifluoroaceticacid). With this HPLC system, we were able to identify thea-keto and a-hydroxy acids produced by selected strains inthe tribe Proteeae. To monitor a-keto acid production in

l1

VOL. 175, 1993

on February 13, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

2730 DRECHSEL ET AL.

PP

c)

PL

b)

a)

0 10 20 0

IL IP

C) K

b)

a)

10 20

Time (min)FIG. 4. HPLC separation of ethylactetate extracts of P. mirabilis

8993 grown in the presence of L-phenylalanine (0.1%) or L-trypto-phan (0.1%). Panels: a, extract; b, coinjection of phenylpyruvic acid(PP) or indolylpyruvic acid (IP); c, additional coinjection of phenyl-lactic acid (PL) or indolyllactic acid (IL). Conditions were as

described in the legend to Fig. 3.

cultures of P. mirabilis, phenylalanine and tryptophan were

added (1%) and the culture filtrate was extracted withethylacetate after 24 h of incubation at 370C. Phenylpyruvateand indolylpyruvate were then identified by HPLC separa-tion on reversed-phase columns by the coinjection proce-dure (Fig. 4). Peaks of phenylpyruvate (19.4 min) andindolylpyruvate (18.7 min) were clearly identified in ethylac-etate extracts by coinjection. Quantitative analysis by HPLCalso revealed that approximately 50% of the added aminoacid was transformed into the corresponding oa-keto acid,which would guarantee a sufficient iron supply via sidero-phores. However, the corresponding a-hydroxy acids 3-phe-nyllactic acid and 3-indolyllactic acid were absent in theseextracts, suggesting that simultaneous reduction to the hy-droxy acids did not take place.While a-keto acids prevailed in amino acid-rich media,

a-hydroxyisocaproic acid was the predominant acid in M9minimal medium (plus 1% inoculum, Luria-Bertani medium)containing ammonium sulfate instead of amino acids. Asproven by coinjection (Fig. 5), a-hydroxyisocaproic acidwas the principal hydroxy acid in culture filtrates of P.mirabilis 8993. This was also true for some other strains(data not shown). However, in M9 minimal medium noo-keto acids were detected by HPLC. Thus, the predomi-nant a-hydroxy acid seems to originate from intracellularpools of leucine after subsequent reduction, while a-ketoacids seem to appear only when amino acids are exoge-

nously present in the incubation medium.Siderophore activity. Growth promotion tests are very

sensitive bioassays for determination of nutritionally avail-able iron sources for bacteria. As shown in Fig. 6, cc-ketoacids behaved differently, depending on chain length and

0 to 20

Time (min)FIG. 5. HPLC separation of an ethylacetate extract from a

culture of P. mirabilis 8993 grown in M9 minimal medium. Theinoculum contained 1 ml of an overnight culture grown in Luria-Bertani medium. Panels: a, extract of P. mirabilis 8993; b, coinjec-tion of extract plus a-hydroxyisocaproic acid (HiC) (mixture, 1:1[vol/vol]).

lipophilicity. While good growth response was observedwith the ferric complexes of indolylpyruvate, phenylpyru-vate, a-ketoisovaleric acid, a-ketoisocaproic acid, and a-ke-toadipic acid, virtually no activity was found with the morepolar pyruvic acid and a-keto-glutaric acid. A small butsignificant amount of activity was also found with o-ketobu-tyric acid after prolonged incubation. Comparable resultswere obtained for a variety of other strains among membersof the tribe Proteeae (Table 1). These results suggest that thereceptors for ferric ao-keto acids require not only a ferriccarboxylate complex but also a lipophilic or aromatic sidechain. Contrary to the results obtained with a-keto acids, thecorresponding a-hydroxy acids were only weakly effectiveat identical concentrations and under the same conditions.Thus, because of their absence in amino acid-rich media, thecorresponding a-hydroxy acids seem to be replaced byae-keto acids in complex media. On the other hand, growth inM9 minimal medium resulted in the absence of a-keto acidsand the presence of a-hydroxy acids.

Figure 6 also shows the growth response with four exog-enous siderophores on an agar plate seeded with M. morga-nii 13. While the siderophore activity of ferrioxamine E andcoprogen was quite low, there was a remarkable growthresponse with ferrichrome and Fe-rhizoferrin, two fungalsiderophores. Ferrichrome is a hydroxamate-type sidero-phore isolated from ascomycetous and basidiomycetousfungi, while rhizoferrin is a polycarboxylate siderophorecharacteristically produced by members of the class Zygo-mycetes (18). The pattern of siderophore utilization could be

b HiC

e0(N

0

0.C6

0-

a

Tow,

0

CN4

U,0c,

a X6"W0C.)4)a)

J. BACT1ERIOL.

on February 13, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

PROTEUS, PROVIDENCL4, AND MORGANELLA a-KETO ACIDS 2731

FIG. 6. Growth promotion tests with M. morganii 13. Ferriccomplex designations are as follows: PA, pyruvic acid; KB, a-ke-tobutyric acid; KG, a-ketoglutaric acid; KA, a-ketoadipic acid;KiV, a-ketoisovaleric acid; KiC, a-ketoisocaproic acid; PP, phe-nylpyruvic acid; IP, indolylpyruvic acid; FE, ferrioxamine E; CP,coprogen; FC, ferrichrome; RF, Fe-rhizoferrin (ligand-to-iron ratio,5:1).

confirmed in most strains of Proteus, Providencia, andMorganella spp., although some strains of Morganella andProvidencia spp. deviated from this general scheme (Table2).

Transport measurements with 55Fe-labelled a-keto acids(a-ketoisovaleric and phenylpyruvic acids) were performedin M9 minimal medium with cells of P. mirabilis 8993, P.rettgeri 39, and M. morganii 13 grown for 24 h in the samemedium. Only the transport data on M. morganii 13 areshown in Fig. 7. For comparison, iron uptake mediated bythe polycarboxylate siderophore rhizoferrin was also mea-sured. Time-dependent iron uptake mediated by a-ketoiso-valeric acid and that mediated by phenylpyruvic acid were

very similar, yielding uptake rates of 0.025 nmol of Fe permin per mg during the linear transport phase. Transport datawere somewhat lower when Fe-rhizoferrin was used, but theactual rates of iron uptake (0.025 nmol/min/mg) were equiv-alent.

DISCUSSION

In microbiological identification tests, the activity of thephenylalanine or tryptophan deaminase in the Proteus-Prov-

idencia-Morganella group is well known and generally de-tected by color formation after addition of ferric ions (6). Thedeaminase reaction test is based on formation of a ferriccomplex after simple addition of ferric chloride to an incu-bation mixture in which tryptophan or phenylalanine hasbeen deaminated to indolylpyruvic acid and phenylpyruvicacid. The resulting iron complex is green with phenylpyru-vate and red-brown with indolylpyruvate (15). However, thedeaminase reaction is not restricted to phenylalanine andtryptophan and the predominant a-keto acids produced inthe incubation medium depend on the presence of theprevailing amino acids.A large collection of members of the tribe Proteeae was

studied in the present investigation, and there was noexception with regard to iron utilization mediated by ax-ketoacids. Short-chain and polar a-keto acids, like pyruvate,a-ketobutyrate, or a-ketoglutarate, were generally not wellaccepted by the iron carboxylate transport systems, al-though a-ketobutyrate seemed to be effective after pro-longed incubation. It should be emphasized that a-keto acidsdo not form very stable complexes with ferric iron comparedwith the hydroxamate siderophores. However, complexstability seems to be high enough to be detected by theirabsorption spectra and with CAS. Therefore, in the absenceof catecholate and hydroxamate siderophores, neither ofwhich is produced by members of the tribe Proteeae, thea-keto acids are stable enough to function as iron transport,agents. Interestingly, the ferric complexes are more stable ina slightly acidic environment, as revealed by their UVspectra. This correlates well with the positive methyl redreaction of members of the tribe Proteeae, which is charac-teristic for acid production down to pH 4.4.

In the present investigation, a-hydroxyisocaproic acidwas the predominant hydroxy acid in low-iron cultures of P.mirabilis 8993 and other members of the tribe Proteeae,suggesting that in minimal media like M9, a-hydroxy acidscan take over the role of siderophores. A smaller analog,a-hydroxyisovaleric acid, has been isolated earlier as asiderophore of P. mirabilis in low-iron medium (4). How-ever, we found that the chain length is different, yieldinga-hydroxyisocaproic acid as the predominant a-hydroxyacid; this may be due to differences in low-iron media andthe corresponding changes of metabolic pathways for aminoacid biosynthesis. Both valine and leucine are biosyntheti-cally closely related, as they originate from a commonprecursor, acetolactate. The preference for leucine as aprecursor may also rely on the specificity of the intracellulardeaminase, as was shown for the eukaryotic L-amino acidoxidase (8). The preferred production of a-hydroxyisocap-roic acid in all genera of the tribe Proteeae argues in favor ofa main pathway for its biosynthesis in low-iron media.Although the iron-binding properties of most a-hydroxy

and a-keto acids seem to be comparable, phenylpyruvic acidand indolylpyruvic acid were significantly better iron-bind-ing ligands than the corresponding phenyllactic and indolyl-lactic acids, as revealed by the CAS assay. While growth-promoting activity was observed with all members of thetribe Proteeae, no siderophore activity was found in E. coliand Salmonella strains. E. coli AB2748 is unable to produceenterobactin because of an aroB mutation. Enterohactin isknown to be an extraordinarily strong iron chelator whichmight interfere with iron complexation by a-keto acids. S.typhimurium TA 2700 is also an aroB mutant and in additioncarries sidA4, which, like the tonB mutation, prevents thetransport of any hydroxamate-type siderophore. Thus thenegative results obtained with E. coli and S. typhimurium

VOL. 175, 1993

on February 13, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

2732 DRECHSEL ET AL.

TABLE 1. Growth-promoting activities of ferric complexes of a-keto acids

Growth4 in the presence of:Strain

PA KB KG KiV KiC KA PP IP

P. vulgaris - - - + + + + +P. vulgaris Ox 19 - - - + + + + +P. mirabilis NM 12 - - - + + ++ + +P. mirabilis 8993 - - - + + + + +P. pennen - - - + ++ + + ++

P. rettgeri 19 - - - + + + + +P. rettgeri 39 - - - + + + + ++P. rettgeri 44 - - - + + + + ++P. rettgeri 62 - - - (+) (+) + ++ +P. stuartii 167 - - - + + + ++ ++P. stuartii 712 (+) - - + + ++ + +++P. stuartii 20137 (+) - - ++ + ++ + +++P. rustigianii 2 - - - + +++ + + +P. alcalifaciens 5487 (+) - - + ++ + + +

M. morganii SBK3 - - - + + + + +M. morganii 13 - - - + ++ ++ ++ ++M. morganii22 - - - + +++ ++ +++ ++

E. coli AB2847 (aroB) - - -

S. typhimupium TA2700 - - -

a -, no growth; (+), weak growth; +, growth zone of 10 to 15 mm; ++, growth zone of 16 to 20 mm; +++, growth zone of 21 to 25 mm. Abbreviations: PA,Fe-pyruvate; KB, Fe-a-ketobutyrate; KG, Fe-a-ketoglutarate; KiV, Fe-a-ketoisovalerate; KiC, Fe-a-ketoisocaproate; KA,' Fe-c-ketoadipate; PP, Fe-phenylpyruvate; IP, Fe-indolylpyruvate.

confirm that under these conditions at strong iron limitationdeaminase-negative enterobacteria do not use a-keto acidsas siderophores and that this route is a very particular routeof the tribe Proteeae.As in other enterobacterial genera, uptake of some exog-

enous hydroxamate siderophores was also observed. Forexample, ferrichrome was one of the most efficient sidero-phores in all members of the tribe Proteeae, whereas copro-gen and ferrioxamines were not as effective and hydroxam-ate siderophore activity was even lacking in M. morganiiSBK3, as reported earlier (17). Surprisingly, the newlydescribed polycarboxylate siderophore rhizoferrin, isolatedfrom fungi of the class Zygomycetes (2, 3, 16), as well as

citrate (data not shown), revealed strong siderophore activ-ity in all genera of the tribe Proteeae. As an ancient bacterialgroup showing only little relationship to other enterobacte-rial genera, members of the tribe Proteeae are unable tosynthesize hydroxamates and catecholates but instead havedeveloped an efficient iron uptake system based on keto orhydroxycarboxylate iron complexes. Although the a-keto-ot-hydroxycarboxylate bidentates represent comparablyweak iron binders, they are effective iron transport agents inthe tribe Proteeae and their concentration can be signifi-cantly increased by the action of amino acid deaminases inan amino acid-rich medium. Whether or not a-hydroxy acidproduction, ot-keto acid production, and synthesis of the

TABLE 2. Growth-promoting activities of ferric complexes of hydroxamates and rhizoferrin

Growth' in the presence of:Strain

Ferrioxamine E Coprogen Ferrichrome Rhizoferrin

P. vulgaris (+) + +++ +++P. mirabilis 8993 - + +++ +++

P. rettgen 19 - +++ +++P. rettgeri 39 -+P. rettgeri 44 - +++ +++P. stuartii 167 - + ++P. stuartii 712 (+) + + ++P. rustigianii 2 (+)

M. morganii SBK3 - -+++M. morganii 13 -+++ +++M. morganii 22 (+) + + +++

E. coli AB2847 (aroB) (+) ++ +++S. typhimurium TA2700 (ent-1 SidAl)

- , no growth; (+,weak growth; +, growth zone of 10 to 15 mm; ++ growth zone of 16 to 20 mm; +++ growth zone of 21 to 25 mm.

J. BACTERIOL.

on February 13, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

PROTEUS, PROVIDENCLA, AND MORGANELLA a-KETO ACIDS 2733

0.7

0~.2-*

0

D0.1-0.3-

0.1

5 10 15

Time (min)FIG. 7. Transport of 55Fe in M. morgandi 13 cells mediated by

phenylpyruvic acid, a-ketoisovaleric acid, and rhizoferrin in M9minimal medium. Symbols: *, Fe-phenylpyruvate; A, Fe-a-keto-isovaleric acid; M, Fe-rhizoferrin. Conditions are described inMaterials and Methods.

cognate receptors are iron-repressible phenomena remainsan open question which will be dealt with in a furtherinvestigation.As P. mirabilis and P. vulgaris may cause primary and

secondary infections in humans, their iron metabolism andability to enhance iron acquisition by the deaminase reactionhave to be taken into consideration. Urinary tract infectionsare often associated with underlying metabolic disorders, suchas diabetes, or mechanical injury caused by catheterization,which can lead to increased local concentration of hydroxy andketo acids or even amino acids amenable to deamination.

ACKNOWLEDGMENTSWe thank the Deutsche Forschungsgemeinschaft for financial

support, Alexander Cansier for performing HPLC analyses, andBrigitte Doez for expert technical assistance.

REFERENCES

1. Berner, I., S. Konetschny-Rapp, G. Jung, and G. Winkelmann.1988. Characterization of ferrioxamine E as the principal sidero-phore of Erwinia herbicola (Enterobacter agglomerans). Bio-Metals 1:51-56.

2. Drechsel, H., G. Jung, and G. Winkelmann. 1992. Stereochem-ical characterization of rhizoferrin and identification of itsdehydration products. BioMetals 5:141-148.

3. Drechsel, H., J. Metzger, S. Freund, G. Jung, J. R. Boelaert, andG. Winkelmann. 1991. Rhizoferrin-a novel siderophore fromthe fungus Rhizopus microsporus var. rhizopodifonnis. BioMet-als 4:238-243.

4. Evanylo, L. P., S. Kadis, and J. R. Maudsley. 1984. Siderophoreproduction by Proteus mirabilis. Can. J. Microbiol. 30:1046-1051.

5. Gibson, F., and D. I. Magrath. 1969. The isolation and charac-terization of a hydroxamic acid (aerobactin) formed by Aer-obacter aerogenes 62-1. Biochim. Biophys. Acta 192:175-184.

6. Henriksen, S. D. 1950. A comparison of the phenylpyruvic acidreaction and the urease test in the differentiation of Proteusfrom other enteric organisms. J. Bacteriol. 60:225-231.

7. Konetschny-Rapp, S., H. Huschka, G. Winkelmann, and G.Jung. 1988. High-performance liquid chromatography of sidero-phores from fungi. BioMetals 1:18-25.

8. Lichtenberg, L. A., and D. Wellner. 1968. A sensitive fluorimet-ric assay for amino acid oxidases. Anal. Biochem. 26:313-319.

9. O'Brien, I. G., G. B. Cox, and F. Gibson. 1970. Biologicallyactive compound containing 2,3-dihydroxybenzoic acid andserine formed by Escherichia coli. Biochim. Biophys. Acta201:453-460.

10. Payne, S. M. 1988. Iron and virulence in the family Enterobac-teriaceae. Crit. Rev. Microbiol. 16:81-111.

11. Pollack, J. R., and J. B. Neilands. 1970. Enterobactin, an irontransport compound from Salmonella typhimuium. Biochem.Biophys. Res. Commun. 38:989-992.

12. Rabsch, W., and G. Winkelmann. 1991. The specificity ofbacterial siderophore receptors probed by bioassays. BioMetals4:244-250.

13. Reissbrodt, R., W. Rabsch, A. Chapeaurouge, G. Jung, and G.Winkelmann. 1990. Isolation and identification of ferrioxamineG and E in Hafnia alvei. BioMetals 3:54-60.

14. Schwyn, B., and J. B. Neilands. 1987. Universal chemical assayfor the detection and determination of siderophores. Anal.Biochem. 160:47-56.

15. Singer, J., and B. E. Volcani. 1955. An improved ferric chloridetest for differentiating Proteus-Providence group from otherEnterobacteriaceae. J. Bacteriol. 69:303-306.

16. Thieken, A., and G. Winkelmann. 1992. Rhizoferrin: a complex-one type siderophore of the Mucorales and Entomophthorales(Zygomycetes). FEMS Microbiol. Lett. 94:37-42.

17. Winkelmann, G. 1991. Specificity of iron transport in bacteriaand fungi, p. 65-105. In G. Winkelmann (ed.), Handbook ofmicrobial iron chelates. CRC Press, Inc., Boca Raton, Fla.

18. Winkelmann, G. 1992. Structures and functions of fungal sidero-phores containing hydroxamate and complexone type iron bind-ing ligands. Mycol. Res. 96:529-534.

VOL. 175, 1993

on February 13, 2020 by guest

http://jb.asm.org/

Dow

nloaded from