- Email: [email protected]
Cloning of a Serratia marcescens DNA fragment that induces quinoprotein glucose dehydrogenase-mediated gluconic acid production in Escherichia coli in the presence of stationary phase Serratia marcescens
FEMS Microbiology Letters 205 (2001) 215^220
www.fems-microbiology.org
Cloning of a Serratia marcescens DNA fragment that induces quinoprotein glucose dehydrogenase-mediated gluconic acid production in Escherichia coli in the presence of stationary phase Serratia marcescens P.U. Krishnaraj 1 , A.H. Goldstein * Biomedical Materials Engineering Science Program Alfred University, Alfred, NY 14802, USA Received 17 July 2001; received in revised form 14 August 2001; accepted 8 October 2001 First published online 2 November 2001
Abstract Serratia marcescens ER2 was isolated from an endorhizosphere sample based on its high level of mineral phosphate solubilizing (MPS) activity. This phenotype was correlated with expression of the direct oxidation pathway. An ER2 plasmid library constructed in Escherichia coli strain DH5K was screened for MPS activity. A recombinant clone DH5K (pKG3791) was capable of gluconic acid (GA) production and tricalcium phosphate solubilization but only in the presence of stationary phase ER2 cells. GA production in DH5K (pKG3791) was apparently the result of the quinoprotein glucose dehydrogenase activity because AG121 (a Tn5 knockout of gcd) carrying pKG3791 did not produce GA under the same conditions. GA production by DH5K (pKG3791) was not observed when ER2 was replaced by another PQQ-producing strain bacterium. These data add to a growing body of evidence that E. coli contains some type of PQQ biosynthesis pathway distinct from those previously characterized in Gram-negative bacteria and that these genes may be induced under appropriate conditions. ß 2001 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved. Keywords : Mineral phosphate solubilization ; Direct oxidation ; Rhizobiology; PQQ; Quinoprotein; Serratia marcescens
1. Introduction For over a century agricultural microbiologists and microbial ecologists have been interested in the ability of some bacteria to dissolve poorly soluble mineral phosphates such as bone meal or hydroxyapatite [1,2]. We have termed this phenotype mineral phosphate solubilizing (MPS). In agriculture, phosphorus is second only to nitrogen as an essential mineral fertilizer for crop production, comprising V0.2% of plant dry weight [3,4]. Conversely, without fertilizer amendment, soluble phosphorus is often the limiting mineral nutrient in soil ecosystems [1,2]. At any given time, a substantial component of soil P is in the form of poorly soluble mineral phosphates. These mineral * Corresponding author. Tel. : +1 (607) 871-2645; Fax: +1 (607) 871-2359. E-mail address : [email protected] (A.H. Goldstein). 1
Present address: Department of Biotechnology, College of Agriculture, University of Agricultural Sciences, Dharwad 580 005, India.
phosphates are, in general, not bioavailable for nutritional transport and assimilation. Since most microorganisms growing in soils must obtain their P from the environment via membrane transport, insoluble P must be converted to 23 soluble ionic phosphate (Pi; H2 PO3 4 or HPO4 ). Evaluation of a large number of samples from a wide range of soil types has shown that, in general, highly e¤cacious MPS bacteria utilize the direct oxidation pathway to produce gluconic and 2-ketogluconic acids. These strong organic acids, in turn, provide the acidi¢cation of the external environment necessary to dissolve poorly soluble calcium phosphates such as tricalcium phosphate (TCP) or hydroxyapatite. The direct oxidation pathway occurs on the outer face of the cytoplasmic membrane. The ¢rst step is the oxidation of glucose to gluconic acid (GA) by the quinoprotein glucose dehydrogenase (PQQGDH). In Gram-negative bacteria studied to date, the holoenzyme is composed of a single polypeptide apoenzyme, the redox cofactor PQQ (2,7,9-tricarboxyl1H-pyrrolo[2,3-f]quinoline-4,5-dione) and a Ca2 or Mg2 [5,6].
0378-1097 / 01 / $20.00 ß 2001 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved. PII: S 0 3 7 8 - 1 0 9 7 ( 0 1 ) 0 0 4 7 2 - 4
FEMSLE 10218 10-12-01
216
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
Shotgun plasmid cloning into Escherichia coli has proven to be a useful strategy for isolating both the GDH apoenzyme and PQQ biosynthesis genes from other Gram-negative bacteria (Table 1). This is because E. coli constitutively produces the GDH apoenzyme (encoded by the gcd gene) but does not appear capable of independent PQQ biosynthesis [6]. There is some consensus that E. coli obtains PQQ from its external environment in vivo as a vitamin. Therefore, recombinant wild-type E. coli showing PQQGDH activity in vitro often contains PQQ biosynthesis genes in the foreign DNA insert of the plasmid (cf. [5]). Conversely, plasmid clones showing PQQGDH activity in a gcd knockout (e.g. AG121) growing in medium containing exogenous PQQ are candidates to carry the apoglucose dehydrogenase (GDH) gene. In a third scenario, our laboratory has now identi¢ed several clones from di¡erent Gram-negative bacteria that do not appear to encode (or have any homology with) known PQQ biosynthesis genes or apoglucose dehydrogenase genes (cf. [7]). Nevertheless, plasmids carrying these clones induce GA production in E. coli strain DH5K at levels equivalent to that seen in recombinants carrying cloned PQQ biosynthesis genes. Importantly, these clones do not function when placed in strain AG121, a Tn5 knockout of the gcd gene which has lead us to hypothesize that E. coli may have an alternative pathway for PQQ biosynthesis that is not expressed under `normal' metabolic conditions. In this paper we describe the cloning of a DNA fragment from Serratia marcescens ER2. The DNA sequence has no homology with any known PQQ or PQQGDH genes. However, the fragment induces PQQGDH in E. coli strain DH5K. Importantly, the clone DNA induces no GA when transformed into AG121. Additionally, GA is only produced by DH5K in the presence of stationary phase ER2, indicating a di¡usible agent. In addition to the potential ecophysiological signi¢cance of the observed signaling process for rhizosphere P cycling, this work adds another member to the family of genes capable of inducing PQQGDH activity in E. coli. Since, as discussed above, this cloned DNA has no homology to known PQQ biosynthesis genes, we again propose that E. coli must have
the ability to synthesize PQQ via an alternative metabolic pathway. 2. Materials and methods 2.1. Strains and plasmids The strains and plasmids used in the study were: E. coli DH5K [8], S. marcescens ER2 isolated from the endorhizosphere of 20-year-old bamboo (Dendrocalamus strictus) plants (Professor A.R. Alagawadi, Dharwad University, personal communication), E. coli DH5K (pMCG898) which carries Erwinia herbicola PQQ biosynthesis genes on the plasmid [5], and E. coli DH5K (pKG3791, described in this publication). All E. coli strains were grown at 37³C in LB broth [8]. Plasmid containing strains were grown in 50 Wg ml31 ampicillin (Ap). S. marcescens ER2 was maintained on LB, Luria agar and TCP medium (1% glucose, 0.05% yeast extract, 0.025% MgSO4 W7H2 O, 0.01% CaCl2 , 0.5% TCP from Fluka Biokemica). The pH of the TCP medium was adjusted to 7.0. ER2 was Ap resistant and this antibiotic was used at 50 Wg ml31 . The growth temperature for ER2 was 30³C. 2.2. Construction of plasmid library and the assay of MPS activity Total ER2 DNA was isolated using the Qiagen Genomic DNA kit (Qiagen Inc.) and was partially restricted with Sau3A (New England Biolabs, Inc.). The 10^15-kb fragments were isolated from low melting agarose gel (Seaplaque, Biowhittaker, Inc.) and puri¢ed via the Qiagen PCR puri¢cation kit (Qiagen, Inc.). The puri¢ed fragments were ligated into the BamHI site of Ready-To-Go pUC18 (Amersham Pharmacia) and the ligation product was transformed into competent E. coli DH5K [8]. Transformants were plated onto LB/Ap plates and individual recombinants were picked manually for spotting on TCP medium. In addition, 15 transformants were minipreped using the method of Kado and Liu [9], in order to deter-
Table 1 Strategies for `trapping' apoglucose dehydrogenase (GDH) or PQQ biosynthesis genes in E. coli by functional complementation E. coli strain
Genotype with respect to gcd
TCP solubilization
TCP solubilization with plasmid clone, (3) exogenous PQQ
TCP solubilization with plasmid clone, (+) exogenous PQQ
DH5K
+, constitutively produces apoenzyme
3
+, implies DNA fragment contains PQQ biosynthesis genes. Pick + clone for analysis
N/A for screening since exogenous PQQ will produce functional PQQGDH
AG121
3, Tn5 knockout
3
3, would require gcd and PQQ biosynthesis genes on same DNA fragment
+, implies DNA fragment contains gcd gene. Pick clone for further analysis
A shotgun library of DNA from a Gram-negative bacterium expressing the direct oxidation pathway is cloned into either an E. coli strain with a functional GDH gene (e.g. DH5K) or a GDH3 mutant (gcd). Recombinants are screened for strong acid production via solubilization of TCP or hydroxyapatite as described in Section 2.
FEMSLE 10218 10-12-01
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
mine the average size of the cloned ER2 DNA insert. This average insert size was V10 kb. Individual transformant colonies were inoculated directly into 100 Wl of LB/Ap liquid in 1.7 ml snap-cap microfuge tubes. These tubes were placed in a cryobox which was turned 90³ (to insure good aeration in the microtube) and secured to a gyrorotary shaker. These minicultures were grown overnight at 250 rpm, 37³C and 10 Wl of stationary phase culture of each clone was spotted on a TCP plate laid out on a 50-spot grid. In addition, 10 Wl of an equivalent overnight culture of ER2 was spotted on the four corners of each plate as a positive control. The plates were incubated at 37³C observed over a period of 8 days and then assayed visually. The solubilization of TCP appeared as a clear zone (transparent halo) around the colony resulting from the dissolution of the Pi in that region on the TCP agar. We have previously shown that E. coli cannot produce signi¢cant clearing zones on poorly soluble forms of calcium phosphate such as TCP or hydroxyapatite [2]. Any recombinant that showed clearing was picked, isostreaked, recultured, and reassayed in the manner described above. Since the genetic background was GDH (i.e. wild-type for the apoglucose dehydrogenase gene gcd) with no exogenous PQQ in the medium, zones of solubilization on TCP indicated candidate S. marcescens PQQ biosynthesis genes. One clone, pKG3791, was selected for further study for reasons discussed below. 2.3. Quantitation of GA production One hundred Wl of 24-h cultures of test strains was inoculated into 10 ml of LB medium [8] and incubated in a shaker at 250 rpm, at 30³C (ER2) or 37³C (E. coli strains) for 24 h. Ten Wl of the cells was spotted on the TCP plates, incubated at 30³C (ER2) or 37³C (E. coli strains) and assayed visually. When a clear zone of solubilization was formed, the cells were scraped o¡ and the agar containing the clearing zone removed, suspended in 500 Wl of nanopure water, and incubated overnight at 4³C. The supernatant obtained after centrifugation at 10 000Ug was withdrawn and the D-GA production was determined via an enzyme-linked spectrophotometric assay as previously described ([10], kit #428-191, Boehringer Mannheim). 2.4. Nucleotide sequence The DNA insert of pKG3971 was sequenced at Research Genetics Inc. by a double-stranded primer walking strategy. Contigs were assembled by Research Genetics using Sequencher software. The accuracy of the sequence was veri¢ed in our laboratory via restriction mapping. This sequence has been deposited in GenBank (accession AY028915) and may be obtained directly from NCBI.
217
3. Results and discussion Several recombinant DH5K clones from the plasmid library that showed clearing zones in the absence of exogenous PQQ, including pKG3791, were selected for further study. Initially, ER2 had been spotted at the four corners of each grid as a positive control both for the MPS trait (clearing zone formation) and for uniformity of TCP deposition (which could be determined by zone size). During re-screening of selected clones on TCP plates in the absence of ER2, DH5K (pKG3791) no longer showed the MPS phenotype. Other clones continued to show the MPS phenotype in the absence of ER2 and their characteristics will be the subject of another paper. E. coli DH5K (pKG3791) had the ability to TCP only in the presence of S. marcescens ER2. Upon closer examination, it was observed that the size of the zone of solubilization produced by the spotted DH5K (pKG3791) decreased with increased distance from ER2 (Fig. 1A). This gradation in MPS activity indicated a di¡usible substance. We further observed that induction of solubilization in the recombinant was related to the growth stage of the ER2. Initially, ER2 produced a large clearing zone but no zone was observed in DH5K (pKG3791) (Fig. 1B). Induction of solubilization was seen only after ER2 had reached its full growth potential on the solid medium, equivalent to stationary phase. At stationary phase, ER2 showed a large, mucoid colony that had fully overgrown the zone of solubilization generated by early growth (compare Fig. 1A,B). TCP solubilization is proportional to GA production not absolute numbers of cells. We have previously observed that, in liquid cultures of ER2 (1% glucose minimal medium), the level of GA in the medium drops dramatically as cells move from exponential to stationary phase (unpublished) so this phenotype on solid medium is not surprising. We also tried to assess whether the induction of activity in pKG3791 on the TCP agar was due to PQQ excretion by ER2. DH5K (pMCG898), containing the pqq genes isolated from E. herbicola [5], was spotted in the same orientation as ER2 but, as shown in Fig. 1A, no MPS activity was induced in DH5K (pKG3791). This is consistent with our normal experience, including the original isolation of the E. herbicola pqq genes [2] via plasmid library screening. E. coli spotted on grid plates near PQQ-producing clones does not produce zones of solubilization, indicating the PQQ is not di¡usible in these media. We have not determined if this is due to the intrinsic di¡usibility of PQQ or adsorption onto the calcium phosphate in the medium, since zones are produced throughout the grid when PQQ is incorporated uniformly throughout the medium. A GenBank homology search of the full sequence of the DNA fragment of pKG3791 was conducted using the
FEMSLE 10218 10-12-01
218
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
Fig. 1. A: Comparison of induction of the MPS trait by late phase S. marcescens ER2 vs. no induction by DH5K (pMCG898) which carries the PQQ genes from E. herbicola. Note the gradient of induction as shown by the smaller clearing zone displayed by the cells furthest from the ER2. Stationary phase in ER2 is shown by overgrowth of cells throughout the original zone of clearing (compare with B). B: Actively growing ER2 cells do not induce the MPS phenotype in DH5K (pKG3741). Note the MPS trait is displayed as a clearing zone around ER2 which is overgrown as the cells reach stationary phase (compare with A). The insoluble phosphate is TCP.
BLASTX algorithm. There are several theoretical open reading frames (ORFs) in the cloned fragment, using the default alignment parameters. The BLASTX search gave only three low `E value' alignments. The longest of these (V bases 2100^3300 of our insert) showed high homology to part of gi 5805199 (Expect = e3174, 73% identity in the 381 amino acids aligned). The identi¢ed sequence was an outer membrane protein (OmpP) of Salmonella enteritidis that may be involved in iron acquisition. This same region also showed relatively high homology to OmpF of E. coli. It is of interest to note that OmpF is involved in osmotic sensing and response regulation in E. coli [11], so that there is a homology between our putative ER2 gene and an E. coli gene that is part of an environmental sensor/ response regulator system. The other two high homology `hits' in GenBank were a small region of the E. coli lytic murein transglycosylase (Expect = e3174, 73% identity in the 141 amino acids aligned), and a hypothetical Pseudomonas aeruginosa ORF (Expect = 5e332, 45% identity in
Table 2 GA production induced in E. coli DH5K (pKG3791) by stationary phase ER2 on solid medium Strain
GA (g l31 )
E. coli DH5K (pKG3791) E. coli DH5K (pKG3791)+PQQ E. coli DH5K (pMCG898) E. coli DH5Ka E. coli AG121 (pKG3791)a Blank (medium sample, no bacteria)
0.352 0.508 0.187 0.0 0.0 0.0
GA production was normalized to a per liter basis as described in Section 2. All assays were carried out in the presence of S. marcescens ER2. Variability was high for GA-producing strains (as much as 20%) because the agar was removed and extracted manually (described in Section 2). However, there was no variability in the zero value strains or the blank. a Based on no observed zone of solubilization, previously shown to indicate no GA production.
FEMSLE 10218 10-12-01
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
the 111 amino acids aligned). Further subcloning will be required to identify the Serratia gene regulating induction of hologlucose dehydrogenase in the recombinant E. coli. There are currently three questions of major interest with respect to this system: (1) the sequence of pKG3791 has no homology with any identi¢ed PQQ biosynthesis genes. However, DH5K (pKG3791) produced GA, whereas under identical conditions no GA was produced by either DH5K or AG121 (pKG3791), as shown in Table 2. (2) As previously discussed, AG121 is a speci¢c knockout in the gcd gene which provides very strong evidence that pKG3791 is acting via induction of the essential PQQ cofactor. Since there are multiple ORFs in the ER2 fragment, how many are necessary to the production of this apparent PQQGDH activity? (3) What is the diffusible substance produced by ER2 that promotes the MPS activity in DH5K (pKG3791) and why does its production appear to be a function of the ER2 growth stage ? It is generally accepted that the rhizosphere has a unique microbial population and that this population may play a role in important root functions from disease resistance to growth and development (cf. [12^15]). The MPS trait may have important implications for P cycling within the rhizosphere. Workers at the University of Agricultural Sciences, Dharwad, India, have identi¢ed S. marcescens as a superior MPS organism in rhizosphere samples from diverse ecosystems. We have previously reported on the regulation or unique expression of the MPS trait in other rhizobacterial isolates. With speci¢c reference to regulation of the direct oxidation pathway by a di¡usible signal, Goldstein et al. [10] recently showed that a material washed from the roots of Helianthus sp. growing in the Mojave Desert was necessary to induce PQQGDH activity in Enterobacter cloacae isolated from the roots of this plant. This alkaline soil (pH of 10) was high in insoluble calcium phosphate but soluble Pi was undetectable. In another experiment, Goldstein [12] showed that MPS bacteria isolated from an alkaline desert soil expressed the direct oxidation pathway in vitro whereas non-MPS bacteria isolated from the same rhizosphere did not. Given the known complexity of phosphate regulons and stimulons in bacteria, it is possible that the availability of Pi may play a role in some type of cell-to-cell signaling process that can regulate expression of the direct oxidation pathway for the MPS function. Clearly there are other possibilities such as some form of quorum sensing (although no homology to lux-type genes was found during the BLAST search), or the simple leakage of metabolites from dead cells in older ER2 colonies. It is also clear that we have added another member to the growing number of DNA fragments capable of inducing holoenzyme GDH (PQQGDH) activity in E. coli. Since the cloned ER2 fragment contained a gene or genes that regulate GA production in E. coli, it is likely that a similar function is performed in the wild-type S. marces-
219
cens. As with the E. cloacae system [10], it appears that the direct oxidation pathway may be regulated under appropriate conditions by signal molecules that originate with the host plant and/or by the growth stage or cell density of the bacteria itself. As has been pointed out previously [16], the rhizosphere is possibly the only region of the soil where glucose is routinely available to drive acid-mediated insoluble mineral phosphate solubilization. However, neither the plant nor the bacterium would want to utilize glucose if su¤cient Pi was available in the soil solution. Therefore, these types of signaling systems are reasonable and certainly merit further investigation.
References [1] Goldstein, A.H. (1986) Bacterial mineral phosphate solubilization : Historical perspective and future prospects. Am. J. Altern. Agric. 1, 57^65. [2] Goldstein, A.H. and Liu, S.T. (1987) Molecular cloning and regulation of a mineral phosphate solubilizing gene from Erwinia herbicola. Bio/Tech 5, 72^74. [3] Armstrong, D.L. (1999) Phosphorus nutrition improves plant disease resistance. In: Better Crops with Plant Food. Potash and Phosphate Institute, GA. [4] Schachtman, D.P., Reid, R.J. and Ayling, S.M. (1998) Phosphorus uptake by plants: From soil to cell. Plant Physiol. 116, 447^453. [5] Liu, S.T., Lee, L.-Y., Tai, C.-Y., Horng, C.-H., Chang, Y.-S., Wolfram, J.H., Rogers, R. and Goldstein, A.H. (1992) Cloning of an Erwinia herbicola gene necessary for gluconic acid production and enhanced mineral phosphate solubilization in Escherichia coli HB101. J. Bacteriol. 174, 5814^5819. [6] Duine, J.A. (1991) Quinoproteins: enzymes containing the quinoid cofactor pyrroloquinoline quinone, topaquinone or tryptophan-tryptophan quinone. Eur. J. Biochem. 200, 271^284. [7] Babu-Khan, S., Yeo, T.C., Martin, W.L., Duron, M.D., Rogers, R.D. and Goldstein, A.H. (1995) Cloning of a mineral phosphate solubilizing gene from Pseudomonas cepacia. Appl. Environ. Microbiol. 61, 972^978. [8] Sambrook, J., Fritsch, E.F. and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, New York. [9] Kado, C.I. and Liu, S.-T. (1981) A rapid procedure for determination and isolation of large and small plasmids. J. Bacteriol. 145, 1365^ 1373. [10] Goldstein, A.H., Braverman, K.E. and Osorio, N. (1999) Evidence for mutualism between a plant growing in a phosphate-limited desert environment and a mineral phosphate solubilizing (MPS) rhizobacteria. FEMS Microbiol. Ecol. 30, 295^300. [11] Pratt, L.A. and Silhavy, T.J. (1994) Porin regulon of Escherichia coli. In: Two-Component Signal Transduction (Hoch, J.A. and Silhavy, T.J., Eds.), pp. 105^128. American Society for Microbiology, Washington, DC. [12] Goldstein, A.H. (1994) Solubilization of exogenous phosphates by Gram negative bacteria. In: Cellular and Molecular Biology of Phosphate and Phosphorylated Compounds in Microorganisms (Silver, S. et al., Eds.), pp. 197^203. American Society for Microbiology, Washington, DC. [13] Bolton, H. Jr., Fredrickson, J.K. and Elliott, L.F. (1993) Microbial ecology of the Rhizosphere. In: Soil Microbial Ecology (Metting, F.B. Jr., Ed.), pp. 27^64. Marcel Dekker, New York. [14] Bayliss, C., Lasby, B. and Wood, J.M. (1993) Mutant derivatives of
FEMSLE 10218 10-12-01
220
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
Pseudomonas putida GR 12-2R3 defective in nutrient utilization or cell surface structures show reduced ability to promote canola root elongation. Can. J. Microbiol. 39, 1111^1119. [15] Howie, W.J. and Suslow, T.V. (1991) Role of antibiotic biosynthesis in the inhibition of Pythium ultimatum in the cotton spermosphere
and rhizosphere by Pseudomonas £uorescens. Mol. Plant Microbe Interact. 4, 393^399. [16] Goldstein, A.H. (1995) Recent progress in understanding the molecular genetics and biochemistry of calcium phosphate solubilization by Gram-negative bacteria. Biol. Agric. Hortic. 12, 185^193.
FEMSLE 10218 10-12-01
www.fems-microbiology.org
Cloning of a Serratia marcescens DNA fragment that induces quinoprotein glucose dehydrogenase-mediated gluconic acid production in Escherichia coli in the presence of stationary phase Serratia marcescens P.U. Krishnaraj 1 , A.H. Goldstein * Biomedical Materials Engineering Science Program Alfred University, Alfred, NY 14802, USA Received 17 July 2001; received in revised form 14 August 2001; accepted 8 October 2001 First published online 2 November 2001
Abstract Serratia marcescens ER2 was isolated from an endorhizosphere sample based on its high level of mineral phosphate solubilizing (MPS) activity. This phenotype was correlated with expression of the direct oxidation pathway. An ER2 plasmid library constructed in Escherichia coli strain DH5K was screened for MPS activity. A recombinant clone DH5K (pKG3791) was capable of gluconic acid (GA) production and tricalcium phosphate solubilization but only in the presence of stationary phase ER2 cells. GA production in DH5K (pKG3791) was apparently the result of the quinoprotein glucose dehydrogenase activity because AG121 (a Tn5 knockout of gcd) carrying pKG3791 did not produce GA under the same conditions. GA production by DH5K (pKG3791) was not observed when ER2 was replaced by another PQQ-producing strain bacterium. These data add to a growing body of evidence that E. coli contains some type of PQQ biosynthesis pathway distinct from those previously characterized in Gram-negative bacteria and that these genes may be induced under appropriate conditions. ß 2001 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved. Keywords : Mineral phosphate solubilization ; Direct oxidation ; Rhizobiology; PQQ; Quinoprotein; Serratia marcescens
1. Introduction For over a century agricultural microbiologists and microbial ecologists have been interested in the ability of some bacteria to dissolve poorly soluble mineral phosphates such as bone meal or hydroxyapatite [1,2]. We have termed this phenotype mineral phosphate solubilizing (MPS). In agriculture, phosphorus is second only to nitrogen as an essential mineral fertilizer for crop production, comprising V0.2% of plant dry weight [3,4]. Conversely, without fertilizer amendment, soluble phosphorus is often the limiting mineral nutrient in soil ecosystems [1,2]. At any given time, a substantial component of soil P is in the form of poorly soluble mineral phosphates. These mineral * Corresponding author. Tel. : +1 (607) 871-2645; Fax: +1 (607) 871-2359. E-mail address : [email protected] (A.H. Goldstein). 1
Present address: Department of Biotechnology, College of Agriculture, University of Agricultural Sciences, Dharwad 580 005, India.
phosphates are, in general, not bioavailable for nutritional transport and assimilation. Since most microorganisms growing in soils must obtain their P from the environment via membrane transport, insoluble P must be converted to 23 soluble ionic phosphate (Pi; H2 PO3 4 or HPO4 ). Evaluation of a large number of samples from a wide range of soil types has shown that, in general, highly e¤cacious MPS bacteria utilize the direct oxidation pathway to produce gluconic and 2-ketogluconic acids. These strong organic acids, in turn, provide the acidi¢cation of the external environment necessary to dissolve poorly soluble calcium phosphates such as tricalcium phosphate (TCP) or hydroxyapatite. The direct oxidation pathway occurs on the outer face of the cytoplasmic membrane. The ¢rst step is the oxidation of glucose to gluconic acid (GA) by the quinoprotein glucose dehydrogenase (PQQGDH). In Gram-negative bacteria studied to date, the holoenzyme is composed of a single polypeptide apoenzyme, the redox cofactor PQQ (2,7,9-tricarboxyl1H-pyrrolo[2,3-f]quinoline-4,5-dione) and a Ca2 or Mg2 [5,6].
0378-1097 / 01 / $20.00 ß 2001 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved. PII: S 0 3 7 8 - 1 0 9 7 ( 0 1 ) 0 0 4 7 2 - 4
FEMSLE 10218 10-12-01
216
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
Shotgun plasmid cloning into Escherichia coli has proven to be a useful strategy for isolating both the GDH apoenzyme and PQQ biosynthesis genes from other Gram-negative bacteria (Table 1). This is because E. coli constitutively produces the GDH apoenzyme (encoded by the gcd gene) but does not appear capable of independent PQQ biosynthesis [6]. There is some consensus that E. coli obtains PQQ from its external environment in vivo as a vitamin. Therefore, recombinant wild-type E. coli showing PQQGDH activity in vitro often contains PQQ biosynthesis genes in the foreign DNA insert of the plasmid (cf. [5]). Conversely, plasmid clones showing PQQGDH activity in a gcd knockout (e.g. AG121) growing in medium containing exogenous PQQ are candidates to carry the apoglucose dehydrogenase (GDH) gene. In a third scenario, our laboratory has now identi¢ed several clones from di¡erent Gram-negative bacteria that do not appear to encode (or have any homology with) known PQQ biosynthesis genes or apoglucose dehydrogenase genes (cf. [7]). Nevertheless, plasmids carrying these clones induce GA production in E. coli strain DH5K at levels equivalent to that seen in recombinants carrying cloned PQQ biosynthesis genes. Importantly, these clones do not function when placed in strain AG121, a Tn5 knockout of the gcd gene which has lead us to hypothesize that E. coli may have an alternative pathway for PQQ biosynthesis that is not expressed under `normal' metabolic conditions. In this paper we describe the cloning of a DNA fragment from Serratia marcescens ER2. The DNA sequence has no homology with any known PQQ or PQQGDH genes. However, the fragment induces PQQGDH in E. coli strain DH5K. Importantly, the clone DNA induces no GA when transformed into AG121. Additionally, GA is only produced by DH5K in the presence of stationary phase ER2, indicating a di¡usible agent. In addition to the potential ecophysiological signi¢cance of the observed signaling process for rhizosphere P cycling, this work adds another member to the family of genes capable of inducing PQQGDH activity in E. coli. Since, as discussed above, this cloned DNA has no homology to known PQQ biosynthesis genes, we again propose that E. coli must have
the ability to synthesize PQQ via an alternative metabolic pathway. 2. Materials and methods 2.1. Strains and plasmids The strains and plasmids used in the study were: E. coli DH5K [8], S. marcescens ER2 isolated from the endorhizosphere of 20-year-old bamboo (Dendrocalamus strictus) plants (Professor A.R. Alagawadi, Dharwad University, personal communication), E. coli DH5K (pMCG898) which carries Erwinia herbicola PQQ biosynthesis genes on the plasmid [5], and E. coli DH5K (pKG3791, described in this publication). All E. coli strains were grown at 37³C in LB broth [8]. Plasmid containing strains were grown in 50 Wg ml31 ampicillin (Ap). S. marcescens ER2 was maintained on LB, Luria agar and TCP medium (1% glucose, 0.05% yeast extract, 0.025% MgSO4 W7H2 O, 0.01% CaCl2 , 0.5% TCP from Fluka Biokemica). The pH of the TCP medium was adjusted to 7.0. ER2 was Ap resistant and this antibiotic was used at 50 Wg ml31 . The growth temperature for ER2 was 30³C. 2.2. Construction of plasmid library and the assay of MPS activity Total ER2 DNA was isolated using the Qiagen Genomic DNA kit (Qiagen Inc.) and was partially restricted with Sau3A (New England Biolabs, Inc.). The 10^15-kb fragments were isolated from low melting agarose gel (Seaplaque, Biowhittaker, Inc.) and puri¢ed via the Qiagen PCR puri¢cation kit (Qiagen, Inc.). The puri¢ed fragments were ligated into the BamHI site of Ready-To-Go pUC18 (Amersham Pharmacia) and the ligation product was transformed into competent E. coli DH5K [8]. Transformants were plated onto LB/Ap plates and individual recombinants were picked manually for spotting on TCP medium. In addition, 15 transformants were minipreped using the method of Kado and Liu [9], in order to deter-
Table 1 Strategies for `trapping' apoglucose dehydrogenase (GDH) or PQQ biosynthesis genes in E. coli by functional complementation E. coli strain
Genotype with respect to gcd
TCP solubilization
TCP solubilization with plasmid clone, (3) exogenous PQQ
TCP solubilization with plasmid clone, (+) exogenous PQQ
DH5K
+, constitutively produces apoenzyme
3
+, implies DNA fragment contains PQQ biosynthesis genes. Pick + clone for analysis
N/A for screening since exogenous PQQ will produce functional PQQGDH
AG121
3, Tn5 knockout
3
3, would require gcd and PQQ biosynthesis genes on same DNA fragment
+, implies DNA fragment contains gcd gene. Pick clone for further analysis
A shotgun library of DNA from a Gram-negative bacterium expressing the direct oxidation pathway is cloned into either an E. coli strain with a functional GDH gene (e.g. DH5K) or a GDH3 mutant (gcd). Recombinants are screened for strong acid production via solubilization of TCP or hydroxyapatite as described in Section 2.
FEMSLE 10218 10-12-01
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
mine the average size of the cloned ER2 DNA insert. This average insert size was V10 kb. Individual transformant colonies were inoculated directly into 100 Wl of LB/Ap liquid in 1.7 ml snap-cap microfuge tubes. These tubes were placed in a cryobox which was turned 90³ (to insure good aeration in the microtube) and secured to a gyrorotary shaker. These minicultures were grown overnight at 250 rpm, 37³C and 10 Wl of stationary phase culture of each clone was spotted on a TCP plate laid out on a 50-spot grid. In addition, 10 Wl of an equivalent overnight culture of ER2 was spotted on the four corners of each plate as a positive control. The plates were incubated at 37³C observed over a period of 8 days and then assayed visually. The solubilization of TCP appeared as a clear zone (transparent halo) around the colony resulting from the dissolution of the Pi in that region on the TCP agar. We have previously shown that E. coli cannot produce signi¢cant clearing zones on poorly soluble forms of calcium phosphate such as TCP or hydroxyapatite [2]. Any recombinant that showed clearing was picked, isostreaked, recultured, and reassayed in the manner described above. Since the genetic background was GDH (i.e. wild-type for the apoglucose dehydrogenase gene gcd) with no exogenous PQQ in the medium, zones of solubilization on TCP indicated candidate S. marcescens PQQ biosynthesis genes. One clone, pKG3791, was selected for further study for reasons discussed below. 2.3. Quantitation of GA production One hundred Wl of 24-h cultures of test strains was inoculated into 10 ml of LB medium [8] and incubated in a shaker at 250 rpm, at 30³C (ER2) or 37³C (E. coli strains) for 24 h. Ten Wl of the cells was spotted on the TCP plates, incubated at 30³C (ER2) or 37³C (E. coli strains) and assayed visually. When a clear zone of solubilization was formed, the cells were scraped o¡ and the agar containing the clearing zone removed, suspended in 500 Wl of nanopure water, and incubated overnight at 4³C. The supernatant obtained after centrifugation at 10 000Ug was withdrawn and the D-GA production was determined via an enzyme-linked spectrophotometric assay as previously described ([10], kit #428-191, Boehringer Mannheim). 2.4. Nucleotide sequence The DNA insert of pKG3971 was sequenced at Research Genetics Inc. by a double-stranded primer walking strategy. Contigs were assembled by Research Genetics using Sequencher software. The accuracy of the sequence was veri¢ed in our laboratory via restriction mapping. This sequence has been deposited in GenBank (accession AY028915) and may be obtained directly from NCBI.
217
3. Results and discussion Several recombinant DH5K clones from the plasmid library that showed clearing zones in the absence of exogenous PQQ, including pKG3791, were selected for further study. Initially, ER2 had been spotted at the four corners of each grid as a positive control both for the MPS trait (clearing zone formation) and for uniformity of TCP deposition (which could be determined by zone size). During re-screening of selected clones on TCP plates in the absence of ER2, DH5K (pKG3791) no longer showed the MPS phenotype. Other clones continued to show the MPS phenotype in the absence of ER2 and their characteristics will be the subject of another paper. E. coli DH5K (pKG3791) had the ability to TCP only in the presence of S. marcescens ER2. Upon closer examination, it was observed that the size of the zone of solubilization produced by the spotted DH5K (pKG3791) decreased with increased distance from ER2 (Fig. 1A). This gradation in MPS activity indicated a di¡usible substance. We further observed that induction of solubilization in the recombinant was related to the growth stage of the ER2. Initially, ER2 produced a large clearing zone but no zone was observed in DH5K (pKG3791) (Fig. 1B). Induction of solubilization was seen only after ER2 had reached its full growth potential on the solid medium, equivalent to stationary phase. At stationary phase, ER2 showed a large, mucoid colony that had fully overgrown the zone of solubilization generated by early growth (compare Fig. 1A,B). TCP solubilization is proportional to GA production not absolute numbers of cells. We have previously observed that, in liquid cultures of ER2 (1% glucose minimal medium), the level of GA in the medium drops dramatically as cells move from exponential to stationary phase (unpublished) so this phenotype on solid medium is not surprising. We also tried to assess whether the induction of activity in pKG3791 on the TCP agar was due to PQQ excretion by ER2. DH5K (pMCG898), containing the pqq genes isolated from E. herbicola [5], was spotted in the same orientation as ER2 but, as shown in Fig. 1A, no MPS activity was induced in DH5K (pKG3791). This is consistent with our normal experience, including the original isolation of the E. herbicola pqq genes [2] via plasmid library screening. E. coli spotted on grid plates near PQQ-producing clones does not produce zones of solubilization, indicating the PQQ is not di¡usible in these media. We have not determined if this is due to the intrinsic di¡usibility of PQQ or adsorption onto the calcium phosphate in the medium, since zones are produced throughout the grid when PQQ is incorporated uniformly throughout the medium. A GenBank homology search of the full sequence of the DNA fragment of pKG3791 was conducted using the
FEMSLE 10218 10-12-01
218
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
Fig. 1. A: Comparison of induction of the MPS trait by late phase S. marcescens ER2 vs. no induction by DH5K (pMCG898) which carries the PQQ genes from E. herbicola. Note the gradient of induction as shown by the smaller clearing zone displayed by the cells furthest from the ER2. Stationary phase in ER2 is shown by overgrowth of cells throughout the original zone of clearing (compare with B). B: Actively growing ER2 cells do not induce the MPS phenotype in DH5K (pKG3741). Note the MPS trait is displayed as a clearing zone around ER2 which is overgrown as the cells reach stationary phase (compare with A). The insoluble phosphate is TCP.
BLASTX algorithm. There are several theoretical open reading frames (ORFs) in the cloned fragment, using the default alignment parameters. The BLASTX search gave only three low `E value' alignments. The longest of these (V bases 2100^3300 of our insert) showed high homology to part of gi 5805199 (Expect = e3174, 73% identity in the 381 amino acids aligned). The identi¢ed sequence was an outer membrane protein (OmpP) of Salmonella enteritidis that may be involved in iron acquisition. This same region also showed relatively high homology to OmpF of E. coli. It is of interest to note that OmpF is involved in osmotic sensing and response regulation in E. coli [11], so that there is a homology between our putative ER2 gene and an E. coli gene that is part of an environmental sensor/ response regulator system. The other two high homology `hits' in GenBank were a small region of the E. coli lytic murein transglycosylase (Expect = e3174, 73% identity in the 141 amino acids aligned), and a hypothetical Pseudomonas aeruginosa ORF (Expect = 5e332, 45% identity in
Table 2 GA production induced in E. coli DH5K (pKG3791) by stationary phase ER2 on solid medium Strain
GA (g l31 )
E. coli DH5K (pKG3791) E. coli DH5K (pKG3791)+PQQ E. coli DH5K (pMCG898) E. coli DH5Ka E. coli AG121 (pKG3791)a Blank (medium sample, no bacteria)
0.352 0.508 0.187 0.0 0.0 0.0
GA production was normalized to a per liter basis as described in Section 2. All assays were carried out in the presence of S. marcescens ER2. Variability was high for GA-producing strains (as much as 20%) because the agar was removed and extracted manually (described in Section 2). However, there was no variability in the zero value strains or the blank. a Based on no observed zone of solubilization, previously shown to indicate no GA production.
FEMSLE 10218 10-12-01
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
the 111 amino acids aligned). Further subcloning will be required to identify the Serratia gene regulating induction of hologlucose dehydrogenase in the recombinant E. coli. There are currently three questions of major interest with respect to this system: (1) the sequence of pKG3791 has no homology with any identi¢ed PQQ biosynthesis genes. However, DH5K (pKG3791) produced GA, whereas under identical conditions no GA was produced by either DH5K or AG121 (pKG3791), as shown in Table 2. (2) As previously discussed, AG121 is a speci¢c knockout in the gcd gene which provides very strong evidence that pKG3791 is acting via induction of the essential PQQ cofactor. Since there are multiple ORFs in the ER2 fragment, how many are necessary to the production of this apparent PQQGDH activity? (3) What is the diffusible substance produced by ER2 that promotes the MPS activity in DH5K (pKG3791) and why does its production appear to be a function of the ER2 growth stage ? It is generally accepted that the rhizosphere has a unique microbial population and that this population may play a role in important root functions from disease resistance to growth and development (cf. [12^15]). The MPS trait may have important implications for P cycling within the rhizosphere. Workers at the University of Agricultural Sciences, Dharwad, India, have identi¢ed S. marcescens as a superior MPS organism in rhizosphere samples from diverse ecosystems. We have previously reported on the regulation or unique expression of the MPS trait in other rhizobacterial isolates. With speci¢c reference to regulation of the direct oxidation pathway by a di¡usible signal, Goldstein et al. [10] recently showed that a material washed from the roots of Helianthus sp. growing in the Mojave Desert was necessary to induce PQQGDH activity in Enterobacter cloacae isolated from the roots of this plant. This alkaline soil (pH of 10) was high in insoluble calcium phosphate but soluble Pi was undetectable. In another experiment, Goldstein [12] showed that MPS bacteria isolated from an alkaline desert soil expressed the direct oxidation pathway in vitro whereas non-MPS bacteria isolated from the same rhizosphere did not. Given the known complexity of phosphate regulons and stimulons in bacteria, it is possible that the availability of Pi may play a role in some type of cell-to-cell signaling process that can regulate expression of the direct oxidation pathway for the MPS function. Clearly there are other possibilities such as some form of quorum sensing (although no homology to lux-type genes was found during the BLAST search), or the simple leakage of metabolites from dead cells in older ER2 colonies. It is also clear that we have added another member to the growing number of DNA fragments capable of inducing holoenzyme GDH (PQQGDH) activity in E. coli. Since the cloned ER2 fragment contained a gene or genes that regulate GA production in E. coli, it is likely that a similar function is performed in the wild-type S. marces-
219
cens. As with the E. cloacae system [10], it appears that the direct oxidation pathway may be regulated under appropriate conditions by signal molecules that originate with the host plant and/or by the growth stage or cell density of the bacteria itself. As has been pointed out previously [16], the rhizosphere is possibly the only region of the soil where glucose is routinely available to drive acid-mediated insoluble mineral phosphate solubilization. However, neither the plant nor the bacterium would want to utilize glucose if su¤cient Pi was available in the soil solution. Therefore, these types of signaling systems are reasonable and certainly merit further investigation.
References [1] Goldstein, A.H. (1986) Bacterial mineral phosphate solubilization : Historical perspective and future prospects. Am. J. Altern. Agric. 1, 57^65. [2] Goldstein, A.H. and Liu, S.T. (1987) Molecular cloning and regulation of a mineral phosphate solubilizing gene from Erwinia herbicola. Bio/Tech 5, 72^74. [3] Armstrong, D.L. (1999) Phosphorus nutrition improves plant disease resistance. In: Better Crops with Plant Food. Potash and Phosphate Institute, GA. [4] Schachtman, D.P., Reid, R.J. and Ayling, S.M. (1998) Phosphorus uptake by plants: From soil to cell. Plant Physiol. 116, 447^453. [5] Liu, S.T., Lee, L.-Y., Tai, C.-Y., Horng, C.-H., Chang, Y.-S., Wolfram, J.H., Rogers, R. and Goldstein, A.H. (1992) Cloning of an Erwinia herbicola gene necessary for gluconic acid production and enhanced mineral phosphate solubilization in Escherichia coli HB101. J. Bacteriol. 174, 5814^5819. [6] Duine, J.A. (1991) Quinoproteins: enzymes containing the quinoid cofactor pyrroloquinoline quinone, topaquinone or tryptophan-tryptophan quinone. Eur. J. Biochem. 200, 271^284. [7] Babu-Khan, S., Yeo, T.C., Martin, W.L., Duron, M.D., Rogers, R.D. and Goldstein, A.H. (1995) Cloning of a mineral phosphate solubilizing gene from Pseudomonas cepacia. Appl. Environ. Microbiol. 61, 972^978. [8] Sambrook, J., Fritsch, E.F. and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, New York. [9] Kado, C.I. and Liu, S.-T. (1981) A rapid procedure for determination and isolation of large and small plasmids. J. Bacteriol. 145, 1365^ 1373. [10] Goldstein, A.H., Braverman, K.E. and Osorio, N. (1999) Evidence for mutualism between a plant growing in a phosphate-limited desert environment and a mineral phosphate solubilizing (MPS) rhizobacteria. FEMS Microbiol. Ecol. 30, 295^300. [11] Pratt, L.A. and Silhavy, T.J. (1994) Porin regulon of Escherichia coli. In: Two-Component Signal Transduction (Hoch, J.A. and Silhavy, T.J., Eds.), pp. 105^128. American Society for Microbiology, Washington, DC. [12] Goldstein, A.H. (1994) Solubilization of exogenous phosphates by Gram negative bacteria. In: Cellular and Molecular Biology of Phosphate and Phosphorylated Compounds in Microorganisms (Silver, S. et al., Eds.), pp. 197^203. American Society for Microbiology, Washington, DC. [13] Bolton, H. Jr., Fredrickson, J.K. and Elliott, L.F. (1993) Microbial ecology of the Rhizosphere. In: Soil Microbial Ecology (Metting, F.B. Jr., Ed.), pp. 27^64. Marcel Dekker, New York. [14] Bayliss, C., Lasby, B. and Wood, J.M. (1993) Mutant derivatives of
FEMSLE 10218 10-12-01
220
P.U. Krishnaraj, A.H. Goldstein / FEMS Microbiology Letters 205 (2001) 215^220
Pseudomonas putida GR 12-2R3 defective in nutrient utilization or cell surface structures show reduced ability to promote canola root elongation. Can. J. Microbiol. 39, 1111^1119. [15] Howie, W.J. and Suslow, T.V. (1991) Role of antibiotic biosynthesis in the inhibition of Pythium ultimatum in the cotton spermosphere
and rhizosphere by Pseudomonas £uorescens. Mol. Plant Microbe Interact. 4, 393^399. [16] Goldstein, A.H. (1995) Recent progress in understanding the molecular genetics and biochemistry of calcium phosphate solubilization by Gram-negative bacteria. Biol. Agric. Hortic. 12, 185^193.
FEMSLE 10218 10-12-01