FEMS Microbiology Letters 100 (1992) 227-232 © 1992 Federation of European Microbiological Societies 0378-1097/92/$05.00 Published by Elsevier

227

FEMSLE 80037

The effect of iron limitation on expression of the aerobic and anaerobic electron transport pathway genes in Escherichia coli Peggy A. Cotter, Silvia Darie and R o b e r t P. Gunsalus Department of Microbiology and Molecular Genetics, and the Molecular Biology Institute, Unit;ersity of California, Los Angeles, California, USA Received 12 June 1992 Accepted 15 June 1992

Key words: Iron limitation; Bacterial electron transport; Anaerobic gene expression; fnr Gene; fur Gene; Escherichia coli

1. SUMMARY To determine whether the aerobic and anaerobic respiratory pathways of Escherichia coli are regulated in response to iron availability, strains containing lacZ reporter fusions to the cydAB, cyoABCDE, narGHJl, dmsABC and frdABCD operons were grown in medium limited for iron by use of the chelator, 2,2'-dipyridyl. Under anaerobic conditions, expression of the anaerobic respiratory pathway operons, narG-lacZ, dmsAlacZ and frdA-lacZ, was reduced 14-16 fold when iron was limited. In contrast, expression of the aerobic pathway operons, cyoA-lacZ and cydA-lacZ, was elevated modestly. Iron-dependent transcriptional control of these operons was independent of the fur gene which encodes an iron-and-DNA-binding regulatory protein. The

Correspondence to: R.P. Gunsalus, Department of Microbiology and Molecular Genetics, 5304 LS, University of California, Los Angeles, CA 90024, USA.

expression of fnr-lacZ was relatively unaffected by iron limitation suggesting that Fnr levels in the cell do not change in response to iron. The above findings suggest that in addition to Fur, some other cellular protein may bind iron for reporting and regulating iron-dependent cell functions.

2. INTRODUCTION The aerobic and anaerobic terminal respiratory enzymes of Escherichia coli contain one or more molecules of heme a n d / o r non-heme iron as co-factors (Table 1). The two aerobic enzymes, cytochrome o oxidase encoded by cyoABCDE and cytochrome d oxidase encoded by cydAB, contain 2 and 3 hemes, respectively. Nitrate reductase, encoded by narGHJI, contains 4 hemes plus 16 non-heme irons while D M S O / T M A O reductase (dmsABC) and fumarate reductase (frdABCD) contain 12 and 9 irons, respectively, but lack heme. Synthesis of these enzymes has been shown to be controlled at the transcriptional level in response to respiratory substrate

228 Table 1 Heme and Fe content of terminal respiratory enzymes in E.

coli Enzyme

Molecules per complex a Heine

Cytochrome

o oxidase

Cytochrome d oxidase Nitrate reductase DMSO/TMAO reductase Fumarate reductase

Non-heineiron

2

-

3 4 -

16 12 9

b

" From ref. 10. b _: no iron or heme detected.

availability. The two cytochrome oxidases are expressed optimally during aerobic or microaerophilic growth [1,3,6,7] while the anaerobic respiratory enzymes for nitrate, fumarate or D M S O / T M A O use are produced only in the absence of oxygen [2,8,16]. The presence of nitrate further induces nitrate reductase (narGHJI) gene expression and represses expression of the D M S O / T M A O reductase (dmsABC) and fumarate reductase (frdABCD) genes [3,8,16]. Since the formation of active respiratory enzyme complexes requires insertion of either heine or non-heine iron prosthetic groups, we wished to determine if iron availability was also important, in addition to the respiratory substrate, for controlling expression of the respiratory genes. In this study we examined the effect of iron limitation on gene expression as monitored by use of lacZ reporter fusions. Each terminal respiratory operon was shown to be transcriptionally responsive to iron limitation in varying degrees. This control was independent of the fur gene product, which is known to regulate iron uptake by the cell, and suggests that a second cellular mechanism exists for iron detection to insure iron-dependent regulation of the respiratory genes during anaerobiosis.

3. M A T E R I A L S A N D M E T H O D S

3.1. Bacterial strains, bacteriophages, and plasmids The isogenic E. coli strains MC4100 (wild type)

and PC2 (Afnr), plus the bacteriophage used in this study have been previously described [3]. High-titer lysates of the fusion phage were used to lysogenize MC4100 and PC2, and single lysogens were identified by/3-galactosidase assay. The fur strains were constructed by P1 transduction of the fur kan R region from W3110 fur kan g [4] into the MC4100 lysogens [3].

3.2. Cell growth For strain manipulations and maintenance, ceils were grown in Luria broth or on solid media. When required, tetracycline or kanamycin was added to the medium at a concentration of 20 or 50 m g / l , respectively. For assay of cells grown with iron supplementation, FeCI 2 was added at a final concentration of 50 ~M. To limit iron 2,2'-dipyridyl was added to the culture medium at a final concentration of 200 ~ M [13]. For enzyme assay, cells were grown in glucose (40 mM) minimal medium (pH 7.0) [8] or buffered L broth as indicated [3]. Aerobic and anaerobic growth was performed as previously described [1]. Flasks or tubes of the indicated medium were inoculated from overnight cultures grown under identical conditions, and the cells were allowed to double four to five times in mid-exponential phase prior to harvest (OD600 of 0.40-0.45; Kontron Uvikon 810 Spectrophotometer).

3.3. /3-Galactosidase assay /3-Galactosidase assays were performed as previously described [1]. Protein concentration was estimated by assuming that a cell density at OD600 of 1.4 is equal to 150 /xg of protein per ml [9]. Units of /3-galactosidase are expressed as nmol O N P G hydrolysed per min per mg protein. /3Galactosidase values represent the average of at least three experiments with a variation of no more than _+ 10%.

3. 4. Materials o-Nitrophenyl-/3-D-galactopyranoside (ONPG), tetracycline, kanamycin, and 2,2'-dipyridyl were purchased from Sigma (St. Louis, MO). Analytical reagent grade FeC12 was purchased from Mallinckrodt (Paris, KY). All other chemicals were of reagent grade.

229 4. R E S U L T S

4.1. Effect of iron limitation on anaerobic respiratory gene expression To determine if iron limitation affects expression of the narGHJl, frdABCD and dmsABC operons, we measured /3-galactosidase activity in strains carrying narG-lacZ, f r d A - l a c Z , or d m s A - l a c Z fusions following anaerobic growth in media containing the iron chelator 2,2'-dipyridyl (Table 2), and either with or without 50 ~ M FeC12. Expression of each lacZ reporter fusion was reduced 14-16 fold under iron-limiting conditions compared to cells supplemented with iron. Thus, during anaerobic growth, E. coli cells respond to iron limitation by restricting synthesis of the terminal electron transport pathway enzymes required for nitrate-, D M S O / T M A O - and fumarate-dependent respiration. The level of /3galactosidase was unchanged in the various reporter strains grown with or without dipyridyl as long as FeC12 was present in the medium (data not shown) indicating that the metal chelator did not cause additional alterations in gene expression. To test whether the effect of iron limitation was independent or dependent of the fnr gene product, similar experiments were performed with dmsA-lacZ, f r d A - l a c Z and narG-lacZ lysogens that contained a Afnr chromosomal mutation (Table 2). As anticipated, the activation of narG-lacZ and d m s A - l a c Z expression seen in wild-type cells was nearly abolished in the afnr strains while f r d A - l a c Z expression was reduced

2.5-fold. Expression of the narG-lacZ fusion was not further affected upon iron limitation. However, d m s A - l a c Z and f r d A - l a c Z expression was reduced during iron limitation an additional 2.7and 7.5-fold, respectively, which was Fnr-independent. Interestingly, a 2- to 3-fold increase in f r d A - l a c Z expression seen in the afnr strain following a shift from aerobic to anaerobic growth conditions was abolished by the addition of dipyridyl to the culture medium (data not shown). These observations suggest the existence of a second form of anaerobic control for frdABCD expression that is also responsive during ironlimited cell growth.

4.2. Effect of iron limitation on aerobic respiratory gene expression The pattern of cydA-lacZ and cyoA-lacZ expression in response to iron limitation was different than that observed for the anaerobic respiratory genes (Table 2). In MC4100, cyoA-lacZ expression was elevated 3-fold during iron limitation. However, in the Afnr strain, cyoA-lacZ expression was reduced 6-fold during iron-limited vs. iron-supplemented conditions. A similar pattern was observed for cydA-lacZ expression. When iron was limited, the activities about equal in either genetic background.

4.3. Effect of dipyridyl and iron on fnr gene expression We also examined whether iron affected fnr gene expression. A mild derepression in fnr-lacZ expression was seen in the wild-type strain under

Table 2 Effect of FeC12 and the iron chelator, dipyridyl, on anaerobic expression of the aerobic and anaerobic respiratory genes Reporter fusion

narG- lacZ dmsA - lacZ frdA - lacZ cyoA - lacZ cydA - lacZ fnr - lacZ

a

/3-Galactosidase activity b (nmol ONPG hydrolysedmin I mg-I protein)

Afnr

MC4100 - Fe

+ Fe

(fold) b

-- Fe

+ Fe

(fold) b

130 37 54 3 757 1330

1820 600 735 1 481 1 150

(14) (16) (14) (0.3) (0.6) (0.8)

21 9 40 4 760 1470

26 24 300 24 1490 1850

(1) (2.7) (7.5) (6) (2) (1.2)

a The wild-type and fnr strains were grown anaerobically on glucose minimal medium in the presence of 200/~M dipyridyl. b Values obtained by dividing the plus iron by the minus iron value.

230

4.6. Effect of iron limitation on NarXL-dependent nitrate control

iron-limited vs. iron-supplemented growth conditions (Table 2). Like the pattern for cyd-lacZ and cyo-lacZ expression in the Afnr strains, fnr-lacZ expression was slightly elevated in a afnr strain during iron-supplemented vs. ironlimited growth.

We tested to see whether iron limitation affected NarX- and NarL-mediated nitrate control of narG-lacZ, dmsA-lacZ and frdA-lacZ expression (data not shown). We found that while iron limitation had no direct effect on the ability of NarX and NarL to activate narGHJI expression or to repress frdABCD or dmsABC gene expression in response to nitrate, heine did (Cotter, unpublished).

4.4. Effect of fur on electron transport pathway gene expression To determine whether iron limitation of respiratory gene expression was affected by the fur gene product, which is an iron-and-DNA-binding regulatory protein [4], narG-lacZ, frdA-lacZ, dmsA-lacZ, cyoA-lacZ, and cydA-lacZ expression was examined in strains deleted for fur. Regardless of the growth condition tested (aerobic vs. anaerobic), expression of each reporter fusion remained relatively unchanged compared to the isogenic wild-type cells (Table 3). Only for dmsA-lacZ expression did activity vary significantly in a fur background (approx. 2.5-fold decrease during anaerobic growth).

5. DISCUSSION We examined whether cellular iron limitation is an important variable in regulating the expression of electron transport pathway genes in E. coli. Addition of 2,2'-dipyridyl, an iron chelator commonly used to limit iron in iron-uptake studies [4,13], to the growth medium caused a 14- to 16-fold reduced expression of the anaerobic respiratory genes, narGHJl, dmsABC, and frdABCD. The aerobic respiratory pathway genes (cyoABCDE, and cydAB) were also affected but in the opposite direction (2- to 3-fold increase). However, when examined in a Afnr strain, iron limitation caused a 2- to 6-fold reduction in cyo and cyd expression. Expression of fnr-lacZ, however, did not vary significantly in response to iron limitation (Table 2) which demonstrates that the iron-regulatory phenomenon is not due to the

4.5. Effect of the fur gene product on fnr-lacZ expression We also examined whether fnr-lacZ expression and thus the level of Fnr protein in the cell was altered by a defect in the fur gene (Table 3). It was not. As anticipated, fnr-lacZ expression was mildly autoregulated in response to anaerobiosis as has been previously reported [8,12,14]. Autoregulation was also seen in the fur mutant.

Table 3 Effect of the fur gene product on aerobic and anaerobic expression of respiratory genes Reporter fusion a

/3-Galactosidase activity (nmol ONPG hydrolysed m i n - i mg i protein) MC4100

n a r G - lacZ dmsA - lacZ frdA lacZ cyoA - lacZ cydA lacZ f n r - lacZ

A fur

+ Oe

- 0 2

(fold) b

+0 2

-- 0 2

(fold) b

19 12 102 250 95 1250

1080 1 180 1990 27 441 798

(57) (98) (20) (0.1) (4.6) (0.6)

25 11 97 355 122 1070

1290 426 2140 25 568 775

(52) (39) (22) (0.07) (4.6) (0.7)

~ Wild-type and fur strains were grown aerobically and anaerobically on buffered L broth medium. ~' Values obtained by dividing the anaerobic value by the aerobic value.

231

amount of Fnr synthesized and thus presumably present in the cell. The pattern of respiratory pathway gene expression in response to iron limitation (decrease in narGHJI, dmsABC, and frdABCD but increase in cyoABCDE and cydAB expression) is opposite to the pattern of transcriptional control mediated by the fnr regulatory system in response to anaerobiosis [1,3]. It is interesting to speculate that if Fnr is unable to bind iron under iron-limiting conditions, it would be unable to provide the cell with a means to sense anaerobiosis and to either activate or repress transcription. It was noteworthy that in a Afnr background, an additional effect of iron limitation was still observed for five of the operons examined (dmsABC, frdABCD, cyoABCDE, cydAB, and fnr). This iron effect was also shown to be independent of the fur gene product (Table 3) and may be due to another iron regulator in the cell (see below). Is this iron-dependent transcriptional control of the respiratory genes physiologically relevant? Yes, if one accepts the experimental rationale commonly applied for fur-dependent iron-uptake studies in E. coli (i.e. the use of iron chelators in the medium to deprive the cell of iron) [4,13]. Similar frdA-lacZ expression experiments using different metal chelators were reported by Unden [11] and by Guest [15]. Their studies addressed whether metals were involved in Fnr-dependent gene regulation during anaerobic cell growth. They concluded that iron was somehow required to monitor the anaerobic state and this proposal is supported by the recent demonstration of equimolar iron in purified Fnr [5,11]. Unden and co-workers suggested that the iron dependency may not be physiologically important because fur-dependent genes were derepressed under conditions that frdA-lacZ expression remained unchanged [11]. However, from the respiratory operon expression studies described above (Table 2) it is clear that iron limitation is being detected in some manner by the cell and this results in decreased transcription of the anaerobic respiratory genes. A physiological role may be to provide the cell with a second level of iron sensing to aid in survival. If the fur-dependent iron-uptake systems fail to provide sufficient iron to the cell for

normal cellular functioning, the second iron sensor might then respond to the lower iron level in the cell and reduce transcription of the anaerobic and aerobic respiratory genes. The cell could then reduce synthesis of the apoenzymes which require from 9 to 16 iron molecules for assembly of the iron-sulfur centers contained in the mature respiratory enzymes. A similar argument can be made for heme-containing enzymes. Whether this control involves Fnr or some other cellular ironbinding protein is presently unclear. These possibilities are currently being examined. Thus, like yeast and higher organisms, iron availability in E. coli appears to be critical for the proper expression of respiratory genes. It is likely that a variety of other procaryotic microorganisms regulate respiratory functions in response to iron similar to E. coli.

ACKNOWLEDGEMENTS We thank K. Postle for advice and comments regarding the iron-limitation studies. This work was supported in part by grant AI21678 from the National Institutes of Health. P.C. was supported in part by an E. Warsaw Graduate Fellowship and by NIH Training Grant GM07185.

REFERENCES [1] Cotter, P.A., Chepuri, V., Gennis, R.B. and Gunsalus, R.P. (1990) J. Bacteriol. 172, 6333-6338. [2] Cotter, P.A. and Gunsalus, R.P. (1989) J. Bacteriol. 171, 3817-3823. [3] Cotter, P.A. and Gunsalus, R.P. (1992) FEMS Microbiol. Lett. 91, 31-36. [4] De Lorenzo, V., Herrero, M., Giovannini, F. and Neilands, J.B. (1988) Eur. J. Biochem. 173, 537-546. [5] Engel, P., Trageser, M. and Unden, G. (1991) Arch. Microbiol. 156, 463-470. [6] Fu, H.A., Iuchi, S. and Lin, E.C.C. (1991) Mol. Gen. Genet. 226, 209-213. 17] Iuchi, S., Chepuri, V., Gennis, R.B. and Lin, E.C.C. (1990) J. Bacteriol. 172, 6020-6025. [8] Jones, H.M. and Gunsalus, R.P. (1987) J. Bacteriol. 169, 3340-3348. [9] Miller, J.H. (1972) Experiments in Molecular Genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.

232 [10] Neidhardt, F.C. (1987) Escherichia coli and Salmonella typhimurium; Cellular and Molecular Biology. pp. 170220. Am. Soc. Microbiol., Washington, DC. [11] Niehaus, F., Hantke, K. and Unden, G. (1991) FEMS Microbiol. Lett. 84, 319-324. [12] Pascal, M.C., Bonnefoy, V., Fons, M. and Chippaux, M. (1986) FEMS Microbiol. Lett. 36, 35-39.

[13] Postle, K. (1990) J. Bacteriol. 172, 2287-2293. [14] Spiro, S. and Guest, J.R. (1987) J. Gen. Microbiol. 133, 3279-3288. [15] Spiro, S., Roberts, R.E. and Guest, J.R. (1989) Mol. Microbiol. 3, 601-608. [16] Stewart, V. (1982) J. Bacteriol. 151, 1320-1325.