THE JOURNAL OF BIOLOGICAL CHEMISTRY © 2004 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 279, No. 5, Issue of January 30, pp. 3885–3892, 2004 Printed in U.S.A.

RNA Polymerase Mutation Activates the Production of a Dormant Antibiotic 3,3 -Neotrehalosadiamine via an Autoinduction Mechanism in Bacillus subtilis*
Received for publication, September 8, 2003, and in revised form, November 11, 2003 Published, JBC Papers in Press, November 11, 2003, DOI 10.1074/jbc.M309925200

Takashi Inaoka‡, Kosaku Takahashi‡, Hiroshi Yada§, Mitsuru Yoshida§, and Kozo Ochi‡¶
From the ‡Microbial Function Laboratory and §Molecular Elucidation Laboratory, National Food Research Institute, Tsukuba, Ibaraki 305-8642, Japan

Bacillus and Streptomyces species possess the ability to produce a variety of commercially important metabolites and extracellular enzymes. We previously demonstrated that antibiotic production in Streptomyces coelicolor A3(2) and Streptomyces lividans can be enhanced by RNA polymerase (RNAP) mutations selected for the rifampicin-resistant (Rif r) phenotype. Here, we have shown that the introduction of a certain Rif r rpoB mutation into a B. subtilis strain resulted in cells that overproduce an aminosugar antibiotic 3,3 -neotrehalosadiamine (NTD), the production of which is dormant in the wild-type strain. Mutational and recombinant gene expression analyses have revealed a polycistronic gene ntdABC (formally yhjLKJ) and a monocistronic gene ntdR (formally yhjM) as the NTD biosynthesis operon and a positive regulator for ntdABC, respectively. Analysis of transcriptional fusions to a lacZ reporter revealed that NTD acts as an autoinducer for its own biosynthesis genes via NtdR protein. Our results also showed that the Rif r rpoB mutation causes an increase in the activity of A-dependent promoters including ntdABC promoter. Therefore, we propose that unlike the wild-type RNAP, the mutant RNAP efficiently recognized the A-dependent promoters, resulting in the dramatic activation of the NTD biosynthesis pathway by an autoinduction mechanism.

mutations are found within the rpoB gene that encodes the subunit of RNAP (3– 6). E. coli Rif r rpoB mutations that suppress the auxotrophy due to lack of stringent response were demonstrated to affect the transcription of stringently controlled genes by destabilizing the RNAP-stable RNA promoter complex (7). Recently, we successfully activated the secondary metabolism (antibiotic production) by introducing certain Rif r rpoB mutations in Streptomyces coelicolor A3(2) and Streptomyces lividans (8 –11). On the basis of those findings, we proposed that improvement of RNAP by introduction of a Rif r mutation can be useful to elicit bacterial ability by altering the gene expression in a variety of microorganisms. The members of the genus Bacillus produce several antibiotics to inhibit growth of the competing organisms in nature. Neotrehalosadiamine (3,3 -diamino-3,3 -dideoxy- , -trehalose; NTD), which is an aminosugar antibiotic produced by Bacillus pumilus (12) and Bacillus circulans (13), inhibits growth of Staphylococcus aureus and Klebsiella pneumoniae (but not Escherichia coli), although the inhibitory mechanism is not yet elucidated. Unlike B. pumilus and B. circulans, Bacillus subtilis normally does not produce this antibiotic. However, we found that one of the Rif r mutations isolated here activates the dormant ability to produce NTD in B. subtilis. In this paper, the mechanism by which production of this antibiotic is activated by the Rif r mutation is discussed.

Downloaded from at MAHIDOL UNIV, FAC SCIENCE on April 10, 2008

The ability of bacteria to survive in a wide range of adverse environmental conditions depends on diverse molecular mechanisms that adjust gene expression pattern in response to changing environment. DNA-dependent RNA polymerase (RNAP),1 which is composed of an essential catalytic core enzyme ( 2 ) and one of the sigma ( ) factors, is the central enzyme for expression of genomic information in all organisms. Rifampicin (Rif) inhibits transcription initiation by blocking the subunit of bacterial RNAP (1, 2). Many Rif-resistant (Rif r) mutants have been isolated and characterized in various bacteria, notably Escherichia coli. To our knowledge, most Rif r

* This work was supported by a grant from the Organized Research Combination System of the Science and Technology Agency of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ¶ To whom correspondence should be addressed: National Food Research Institute, 2-1-12 Kannondai, Tsukuba, Ibaraki 305-8642, Japan. Tel.: 81-29-838-8125; Fax: 81-29-838-7996; 1 The abbreviations used are: RNAP, RNA polymerase; Rif, rifampicin; NTD, 3,3 -neotrehalosadiamine; Erm, erythromycin; IPTG, isopropyl- -D-thiogalactopyranoside; ESI-MS, electrospray ionization mass spectrometry; HR-ESI-MS, high resolution ESI-MS.
This paper is available on line at

Bacterial Strains, Plasmids, and Culture Conditions—Strains of B. subtilis and plasmids used are listed in Table I. Oligonucleotides used in this study are listed in Table II. All of the B. subtilis strains are derived from strain 61884 (trpC2 aspB66) (14). A mutant TI91 is a bacilysin nonproducer strain (15). Strains UOT1285EN1 (amyE::rrnE-bgaB) and MF1 (rpoChis6 neo) were provided by F. Kawamura and M. Fujita, respectively. Plasmid pUB18 was provided by F. Kawamura. B. subtilis strains were grown aerobically in S7N medium or L medium as described previously (15). To monitor the promoter activities of ntdABC and ntdR, each upstream region (490 bp from the translational start codon) was amplified by PCR with the primer pairs PntdA-F and PntdA-R for ntdABC promoter and PntdR-F and PntdR-R for ntdR promoter. The amplified fragment was cloned into pCR2.1 and fully sequenced to confirm the correct sequence. Each EcoRI fragment containing the upstream region of ntdABC or ntdR was inserted into the EcoRI site of pDL2 (16), resulting in plasmids pDL2-PntdA and pDL2-PntdR, respectively. For overexpression of NTD biosynthesis operon in B. subtilis, the 430-bp fragment encoding N-terminal part of NtdA was synthesized by PCR with primers ntdA-F and ntdA-R and cloned at the EcoRV site of pHASH120 by a TA cloning method as described by Ohashi et al. (17), resulting in pHASH120-ntdA. Similarly, for overexpression of ntdBC (but not ntdA) in B. subtilis, the 240-bp fragment encoding the Nterminal part of NtdB was amplified by PCR with primers ntdB-F and ntdB-R and cloned into pHASH120, resulting in pHASH120-ntdB. To express the NTD biosynthesis operon in E. coli, the full length of the ntdABC coding region (3.2 kb) was amplified by PCR with the primers



Activation of B. subtilis Neotrehalosadiamine Biosynthesis
TABLE I Bacterial strains and plasmids used in this study

Strains and plasmids

Genotype or description

Construction or source

B. subtilis 61884 84R2 84R5 84R6 84R32 TI91R5 Y8–19 T5–31 E2–10 TI122 TI122R5 TI123 TI123R5 TI125R5 TI127 TI127R5 TI127R2 TI127R6 TI127R32 TI130R5 TI131R5 TI132R5 TI139 TI148R5 TI149 TI149R5 TI149R2 TI149R6 TI149R32 TI150 TI150R5 TI151R5 Plasmids pDL2-PntdA pDL2-PntdR pMutinT3- ntdR pHASH120-ntdA pHASH120-ntdB pUC18-ntdR::neo pUC18-ntdABC pUB18-ntdR pQE80L-ntdR

trpC2 aspB66 trpC2 aspB66 rpoB2 trpC2 aspB66 rpoB5 trpC2 aspB66 rpoB6 trpC2 aspB66 rpoB32 trpC2 aspB66 rpoB5 ywfB::pMutinT3 (Ermr) trpC2 aspB66 rpoB5 ntdA::Tn10 (Spcr) trpC2 aspB66 rpoB5 ntdB::Tn10 (Spcr) trpC2 aspB66 rpoB5 ntdC::Tn10 (Spcr) trpC2 aspB66 amyE::(PntdA-lacZ, Cmr) trpC2 aspB66 rpoB5 amyE::(PntdA-lacZ, Cmr) trpC2 aspB66 amyE::(PntdR-lacZ, Cmr) trpC2 aspB66 rpoB5 amyE::(PntdR-lacZ, Cmr) trpC2 aspB66 rpoB5 ntdR::pMutinT3(Pspac-ntdR) trpC2 aspB66 ntdR::neo amyE::(PntdA-lacZ, Cmr) trpC2 aspB66 rpoB5 ntdR::neo amyE::(PntdA-lacZ, Cmr) trpC2 aspB66 rpoB2 ntdR::neo amyE::(PntdA-lacZ, Cmr) trpC2 aspB66 rpoB6 ntdR::neo amyE::(PntdA-lacZ, Cmr) trpC2 aspB66 rpoB32 ntdR::neo amyE::(PntdA-lacZ, Cmr) trpC2 aspB66 rpoB5 ntdA::Tn10 amyE::(PntdA-lacZ, Cmr) trpC2 aspB66 rpoB5 ntdB::Tn10 amyE:: (PntdA-lacZ, Cmr) trpC2 aspB66 rpoB5 ntdC::Tn10 amyE:: (PntdA-lacZ, Cmr) trpC2 aspB66 ntdA::pHASH120(PS10-ntdABC) trpC2 aspB66 rpoB5 ntdA::Tn10 ntdB::pHASH120(PS10-ntdBC) trpC2 aspB66 amyE::(PrrnE-bgaB Cmr) trpC2 aspB66 rpoB5 amyE::(PrrnE-bgaB Cmr) trpC2 aspB66 rpoB2 amyE::(PrrnE-bgaB Cmr) trpC2 aspB66 rpoB6 amyE::(PrrnE-bgaB Cmr) trpC2 aspB66 rpoB32 amyE::(PrrnE-bgaB Cmr) trpC2 aspB66 rpoChis6 Nmr trpC2 aspB66 rpoB5 rpoChis6 Nmr trpC2 aspB66 rpoB5 ntdR::neo amyE::(PntdR-lacZ, Cmr) pDL2 containing the 490 bp of ntdABC upstream region pDL2 containing the 490 bp of ntdR upstream region pMutinT3 containing 5’-region of ntdR pHASH120 containing 5’-region of ntdA pHASH120 containing 5’-region of ntdB pUC18 containing ntdR::neo pUC18 containing complete ntdABC pUB18 containing complete ntdR pQE80L containing complete ntdR

Ochi and Ohsawa (14) This study This study This study This study 84R5 3 TI91a This study This study This study pDL2-PntdA 3 61884 84R5 3 TI122 pDL2-PntdR 3 61884 84R5 3 TI123 pMutinT3-ntdR 3 61884 pUC18-ntdR::neo 3 TI122 84R5 3 TI127 84R2 3 TI127 84R6 3 TI127 84R32 3 TI127 Y8–19 3 TI122R5 T5–31 3 TI122R5 E2–10 3 TI122R5 pHASH120-ntdA 3 61884 pHASH120-ntdB 3 Y8–19 UOT1285EN1 3 61884 84R5 3 TI149 84R2 3 TI149 84R6 3 TI149 84R32 3 TI149 MF1 3 61884 84R5 3 TI150 TI127 3 TI123R5 This This This This This This This This This study study study study study study study study study

Downloaded from at MAHIDOL UNIV, FAC SCIENCE on April 10, 2008

The latter was transformed with chromosomal DNA or plasmid of the former. TABLE II Oligonucleotides used in this study
Purpose Primer Oligonucleotide sequence (5 3 3 )

Transcriptional fusion analysis

Forced expression of ntdABC Forced expression of ntdBC Expression of ntdABC in E. coli IPTG-dependent induction of ntdR Insertional mutation of ntdR Purification of His6-tagged NtdR

PntdA-F PntdA-R PntdR-F PntdR-R ntdA-F ntdA-R ntdB-F ntdB-R ntdA-F2 ntdC-R ntdR-F ntdR-R ntdR-F ntdR-R ntdR-F2 ntdR-R2


These primers have the EcoRI, BamHI, HindIII, or SalI site as indicated by underlines. For disruption of the ntdR gene, the partial internal coding region of ntdR gene (510 bp) was amplified by PCR with the primers ntdR-F and ntdR-R and cloned into plasmid pCR2.1 (Invitrogen). The EcoRI fragment containing the partial ntdR gene was subcloned into pUC18 (Takara), resulting in pUC18-ntdR. A 1.3-kb SmaI fragment containing the neo gene derived from pBEST501 (19) was inserted at the EcoRV site within the ntdR gene in pUC18-ntdR, creating pUC18-ntdR::neo. To determine the transcriptional start point of ntdR, the PCR frag-

ntdA-F2 and ntdC-R. The EcoRI-BamHI fragment was cloned into the corresponding region of pUC18, generating pUC18-ntdABC. A part of the fragment containing PCR errors was replaced with the other corresponding fragment. To generate pMutinT3-ntdR, the DNA fragment containing a ShineDalgarno sequence, and the N-terminal coding region was amplified by PCR with the primers ntdR-F and ntdR-R. A HindIII-BamHI fragment was cloned into pMutinT3 (18), resulting in a plasmid pMutinT3-ntdR.

Activation of B. subtilis Neotrehalosadiamine Biosynthesis
ment synthesized with PntdR-F and ntdR-R2 was cloned into pCR2.1 and fully sequenced. An EcoRI-SalI fragment was subcloned into plasmid pUB18, which is plasmid pUB110 (20) containing multicloning sites derived from pUC18, generating pUB18-ntdR. For the expression of His-tagged NtdR, the PCR fragment synthesized with ntdR-F2 and ntdR-R2 was cloned into pCR2.1 and fully sequenced. A BamHI-SalI fragment containing the full length of the ntdR coding region was inserted into the expression vector pQE80L (Qiagen), generating pQE80L-NtdR. E. coli harboring the resultant plasmid can express His6-tagged NtdR protein, which is extended with Met-Arg-Gly-Ser-His6-Gly-Ser at the N terminus. For the selection of B. subtilis transformants, Rif (2 g/ml), erythromycin (Erm; 0.5 g/ml), spectinomycin (100 g/ml), neomycin (3 g/ml), and chloramphenicol (5 g/ml) were used. For the selection of E. coli transformants, ampicillin (100 g/ml) was used. NTD Production—B. subtilis cells were precultured in S7N medium at 37 °C for 12 h with vigorous shaking. The cultures were then diluted 50-fold in a fresh S7N medium and cultivated for the indicated time under the same conditions. The growth of E. coli was monitored in M9 medium supplemented with 0.1% yeast extract. Isopropyl- -D-thiogalactopyranoside (IPTG; final concentration 10 mM) was added to the growth medium when the culture reached A650 0.5. Antibiotic activity was determined by the paper disk-agar diffusion assay using S. aureus 209P as the test organism. The culture supernatants (50 l) obtained after centrifugation were applied onto a paper disk (diameter 8.0 mm; Advantec), and the paper disk was placed on a half-strength Mueller Hinton agar (Difco) plate inoculated with S. aureus 209P. Isolation and Purification of NTD—The culture supernatant (400 ml) of strain TI91R5 prepared after 24-h cultivation was adsorbed on a column of Dowex 50W 8 (NH4 form). The column was eluted with 250 ml of 50 mM ammonium acetate and 100 ml of 1 M NH4OH solutions, successively. The second eluate was concentrated in vacuo, subjected to a SiO2 gel column chromatography, and then eluted with 300 ml of 65% CH3CN solution. The active fractions were collected and evaporated to yield a brown material. This material was chromatographed using a column packed with Sephadex LH-20 resin in 50% CH3OH to obtain a brown solid (140 mg). The brown solid (20 mg) was further purified by high pressure liquid chromatography (Capcell pack NH2, 4.6 150 mm; Shiseido) using 90% CH3CN as a mobile phase at a flow rate of 1 ml/min and monitored by refractive index. The resulting compound (purity 95%) was analyzed by mass spectrometry and 13C NMR spectrometry. High resolution electrospray ionization mass spectrometry (HR-ESIMS) was performed on a Brucker Daltonics ApexII 70e mass spectrometer. 13C NMR spectra were recorded on a Brucker DRX600 spectrometer operating at 150.09 MHz. Transposon Mutagenesis—For transposon mutagenesis, the temperature-sensitive mini-Tn10-containing vector pIC333 was used as described by Steinmetz and Richter (21). We selected colonies that were resistant to spectinomycin but sensitive to Erm and then tested for their ability to produce NTD by cultivating the isolates for 24 h in S7N medium. The mutants resistant to both spectinomycin and Erm were not subjected to the NTD production test. NTD-nonproducing mutants were selected for further analysis. Assay for -Galactosidase Activity—Strains were grown aerobically in S7N medium, and samples were withdrawn every 1 h during 10-h cultivation. -Galactosidase activity was measured as described previously (22). Thermostable -galactosidase (BgaB) activity was also measured by the same method, except that the samples were incubated at 62 °C. Primer Extension—Strain 84R5 for ntdABC or 61884 harboring pUB18-ntdR for ntdR was grown in S7N medium and harvested at t0 (the time points of transition from exponential to stationary phase) or t4 (4 h after t0), with corresponding A650 of 1.5 and 8.0, respectively. The total cellular RNAs were prepared using the Isogen reagent (Nippon Gene). To remove the contaminated DNA, the prepared RNAs were treated with 10 units of DNase I (amplification grade; Invitrogen) at 25 °C for 15 min. DNase I was inactivated by the addition of EDTA and heat treatment at 65 °C for 10 min. After ethanol precipitation, a total of 50 g of RNA was incubated with 1 pmol of an infrared dye-labeled primer, IRD800-ntdA (5 -IRD800-TTGTTTCCACCGCTCGTTTAC-3 ) (Aloka) for ntdABC or IRD800-ntdR (5 -IRD800-TCCGAGCTAAATAATTGGGAGTG-3 ) for ntdR, at 80 °C for 15 min. After annealing, 5 l of 0.1 M dithiothreitol, 10 l of 5-fold First-Strand buffer (250 mM Tris-HCl (pH 8.3), 375 mM KCl, and 15 mM MgCl2) and 5 l of dNTP mix (10 mM each of dATP, dGTP, dCTP, and dTTP) were added. To 50 l of this reaction mixture, 200 units of SuperScript II (Invitrogen) was added followed by incubation at 42 °C for 1 h. Subsequently, 1 l of RNase mixture (Ambion) was added to the reaction mixture and further


FIG. 1. Antibiotic production by parental (61884) and Rif r mutant (84R5) strains of B. subtilis. Antibiotic activity was determined by the paper disk-agar diffusion assay. Strains 61884 and 84R5 (rpoB5) were grown in S7N medium supplemented with amino acids as required (50 g/ml of tryptophan and 20 mM aspartate) at 37 °C for the indicated times. incubated at 37 °C for 30 min. After ethanol precipitation, primer extension and sequencing reactions with the corresponding primer were run on polyacrylamide gel. Analysis was done using DNA sequencing system LIC-4200L(S)-2 (LI-COR). Purification of His6-tagged Protein—E. coli BL21(DE3) harboring pQE80L-NtdR was grown aerobically to A650 0.6 at 37 °C in L medium supplemented with 1% glucose. IPTG was then added to a final concentration of 2 mM, and the culture was further incubated for 4 h. E. coli cells were harvested and disrupted by sonication. The cell lysate was centrifuged (8,000 g for 10 min) to remove insoluble material. The soluble His6-tagged protein was purified using the TALON purification kit (Clontech) as described in the manufacturer’s manual. The purified protein was dialyzed against 20 mM potassium phosphate buffer (pH 7.5). The RNAP holoenzyme containing His6-tagged subunit was purified as described by Fujita and Sadaie (23). Electrophoretic Mobility Shift Assay—DIG Gel Shift Kit (Roche Applied Science) was used for the gel retardation assay. The internal region (100 bp) between ntdA and ntdR was amplified with primers PntdR-R and PntdA-R (see Table II) and used for the labeling reaction. The 100-fold diluted probe was incubated at 25 °C for 15 min with or without His6-tagged NtdR, varying amounts of NTD, trehalose, and neotrehalose in 15 l of binding buffer (20 mM Hepes (pH 7.6), 30 mM KCl, 1 mM EDTA, 10 mM (NH4)2SO4, 1 mM dithiothreitol, 0.2% Tween 20, and poly(dA-dT)). DNA and DNA-protein complexes were separated on 7% nondenaturing PAGE, transferred onto a membrane (HybondN ; Amersham Biosciences), and detected as described in the manufacturer’s manual.

Downloaded from at MAHIDOL UNIV, FAC SCIENCE on April 10, 2008

A Rif RNAP Mutation Activates Neotrehalosadiamine Production—To investigate the effect of rpoB mutation in B. subtilis, we isolated a number of spontaneous Rif r mutants on L agar plate containing 100 g/ml Rif. Of these isolates, 40 strains were tested for their ability to produce antibiotic. As a result, a total of 16 mutants were found to produce a significantly increased amount of antibiotic. Antibiotic production by the mutant 84R5 is shown in Fig. 1 as an example. Nucleotide sequence analysis revealed that all the mutants with enhanced antibiotic production had an identical mutation at Ser-487 (corresponding to Ser-531 in E. coli) to Leu (rpoB5) in RNAP -subunit. The other mutants were found to have a substitution mutation of His-482 (corresponding to His-526 in E. coli) to Arg (rpoB2), Tyr (rpoB6), or Pro (rpoB32) (Table III). Since all of the rpoB5-back-cross transformants tested produced antibiotic as much as 84R5, rpoB5 mutation was responsible for the observed activation of antibiotic production. When B. subtilis strain 61884 is grown in S7N medium, cells produce the dipeptide antibiotic bacilysin after 8 h of cultivation (15). A mutant TI91, which does not produce bacilysin, cannot accumulate antibiotic activity against S. aureus (15). Surprisingly, the introduction of the rpoB5 mutation into the mutant TI91, designated TI91R5, activated the antibiotic production (data not shown). Moreover, the rpoB5 mutation did not influence the expression of the bacilysin biosynthesis operon (data not shown). Consequently, the rpoB5 mutation was thought to activate a novel gene(s) for antibiotic production in B. subtilis. Therefore, we isolated the antimicrobial sub-



Activation of B. subtilis Neotrehalosadiamine Biosynthesis
TABLE III Mutations found in rpoB gene in Rif r mutants


Relevant genotype

Nucleotide substitution

Amino acid substitution

Homologous rpoB positions E. coli S. coelicolor

Rif resistancea g/ml

Antibiotic productionb

61884 84R5 84R2 84R6 84R32
a b

wt(rpoB ) rpoB5 rpoB2 rpoB6 rpoB32

1460C3 T 1445A3G 1444C3 T 1445A3C

487S3L 482H3R 482H3Y 482H3P

531S 526H 526H 526H

442N 437H 437H 437H

0.1 100 100 100 100 , no inhibition zone; ,

Determined after 12 h of incubation on L agar plate. Antibiotic production was determined after 24 h of cultivation in S7N medium as described in the legend to Fig. 1. inhibition zone of 25 mm. TABLE IV Chemical properties of the isolated compound and NTD
Isolated compound


Appearance Elemental analysis Electron impact mass spectrometry (m/z) ESI-MS (m/z) HR-ESI-MS (m/z) Calculated for C12H24N2O9Na Ninhydrin 13C NMR in ppm (in D2O)
a b

White amorphous powder NDb ND 363 M Na 363.1368 363.1374 Positive 54.1, 58.1, 61.1, 61.3, 70.1, 70.1, 71.8, 72.2, 74.8, 77.4, 92.3, 97.2

White amorphous powder C42.54, H7.43, N8.60 340 M

Downloaded from at MAHIDOL UNIV, FAC SCIENCE on April 10, 2008

Positive 54.7, 58.6, 61.7, 61.8, 70.6, 70.7, 72.2, 72.9, 75.6, 77.8, 93.0, 98.3

The data on the physico-chemical property are from Tsuno et al. (12). ND, not determined.

stance produced by TI91R5 (see “Experimental Procedures”). The purified antimicrobial substance was positive for ninhydrin reaction and adsorbed in a cation exchange column chromatography. No UV maximum was observed in the aqueous solution. HR-ESI-MS and 13C NMR spectral data demonstrated that this antimicrobial substance is identical to NTD which was previously isolated from B. pumilus (12) and B. circulans (13). The chemical properties are summarized in Table IV. NTD production of strain 84R5 reached as high as 500 –1000 g/ml after 24 h of cultivation, whereas the parental rpoB strain produced less than the detectable level ( 1 g/ml) (data not shown). Identification of the Genes Involved in NTD Production—To identify the genes involved in NTD synthesis, we performed transposon mutagenesis of strain 84R5 using the mini-Tn10 delivery vector pIC333 (21). Of the 1,800 mutant isolates with mini-Tn10 insertion tested for NTD production, we found three mutants were unable to produce NTD. These three mutants were all found to have a Tn10-insertional mutation within the polycistronic operon yhjLKJ (we renamed ntdABC in this study) (Fig. 2). Even when the ntdA downstream genes (ntdB and ntdC) was expressed from a powerful S10 promoter originating from the large S10 ribosomal gene cluster in strain Y8 –19 (rpoB5 ntdA::Tn10), NTD production was no longer detected, indicating that blocking the NTD synthesis by Tn10 insertion was not simply due to polar effect on ntdB and ntdC (Fig. 2). The gene products of ntdA, ntdB, and ntdC were putatively identified to be pyridoxal phosphate-dependent aminotransferase, hydrolase, and NADH-dependent dehydrogenase, respectively. To confirm whether NTD synthesis is dependent upon ntdABC transcription, we replaced its promoter in rpoB strain 61884 with a S10 promoter. The resulting recombinant TI139 carrying S10 promoter-dependent ntdABC produced NTD just as strain 84R5 (Fig. 2), suggesting that the activation of NTD production by rpoB5 mutation resulted from the activation of ntdABC promoter. The ntdR gene (formerly yhjM), which locates inversely in the upstream region of ntdABC operon, encodes a putative LacI family transcriptional regulator. To elucidate the possible func-

tion of NtdR protein on NTD production, we constructed the ntdR interruption mutant (TI125R5) carrying IPTG-inducible spac promoter-dependent ntdR (Fig. 2). Strain TI125R5 did not accumulate NTD without the ntdR induction (in the absence of IPTG). On the other hand, the NTD production in TI125R5 was induced when IPTG was added into the culture broth. These results suggest that NtdR protein takes part in the regulatory mechanism for NTD production. The ntdABC Operon Encodes a Complete Set of Genes for NTD Biosynthesis—We also investigated whether or not the ntdABC operon was the structural gene for NTD biosynthesis. The ntdABC operon was cloned into a high copy number plasmid, pUC18, and expressed in E. coli BL21(DE3) under the control of lac promoter (see “Experimental Procedures”). An E. coli strain harboring pUC18-ntdABC displayed a prolonged growth lag phase compared with the vector control (Fig. 3A), although no growth inhibition was observed when a high concentration (1,000 g/ml) of NTD was added exogenously. TLC analysis revealed that a ninhydrin-positive spot with a corresponding retention factor of NTD was detected in the culture broth of cells expressing ntdABC (Fig. 3B). Moreover, the peak at m/z 363 that represents the mass of NTD [M Na] was also detected in the culture broth, as determined by ESI-MS spectrometry analysis (Fig. 3C). In contrast, no peak was detected at m/z 363 when a sample from the cells harboring the vector control was analyzed (data not shown). The amount of NTD produced by the ntdABC-harboring E. coli was 5 g/ml. These results strongly indicate that ntdABC operon contains a complete set of genes required for NTD biosynthesis. This is another instance of “exogenous” antibiotic production in E. coli, following the previous report on erythromycin production in E. coli (24). Transcriptional Analysis of NTD Biosynthesis Gene and Its Regulator—The pDL2-derivative plasmids allow the creation of the promoter-lacZ fusion at the amyE locus via a double crossover event (16). To monitor the transcription of ntdABC and ntdR, we constructed a strain carrying ntdABC-lacZ or ntdRlacZ transcriptional fusion. The transcription level of ntdABC in rpoB5 mutant was 20 –50-fold higher than that in the rpoB

Activation of B. subtilis Neotrehalosadiamine Biosynthesis


FIG. 2. The B. subtilis mutant constructions and the corresponding phenotypes for NTD production. Genes are shown as thick arrows. Pspac, PS10, and stem-loop structure indicate IPTG-dependent spac promoter, S10 promoter, and the transcriptional terminator, respectively.

Downloaded from at MAHIDOL UNIV, FAC SCIENCE on April 10, 2008

FIG. 3. NTD production in E. coli transformed with the plasmid carrying the ntdABC gene. A, E. coli strain BL21(DE3) harboring pUC18 (open circles) or pUC18-ntdABC (closed circles) were grown in M9 medium supplemented with 0.1% yeast extract. IPTG was added into the culture when cells reached A650 0.5, as indicated by the arrows. After a 4-h incubation, cells were removed by centrifugation, and the supernatants were subjected to silica gel TLC analysis (B). The area corresponding to the retention factor of NTD was scraped and then eluted with water. The eluted sample was analyzed by ESI-MS (C).

strain throughout the growth (Fig. 4A). No change in transcription level was observed for the other rpoB mutations, namely rpoB2, rpoB6, and rpoB32 (data not shown). The disruption of the ntdR gene almost completely shut off the ntdABC transcription in the rpoB5 mutant (Fig. 4A). In contrast, it resulted in an increase in its own promoter activity (Fig. 4B). These results indicate that NtdR protein apparently acts positively on the ntdABC transcription and negatively on its own transcription. However, no significant difference in the ntdR expression was observed between rpoB strain and rpoB5 mutant (Fig.

FIG. 4. Transcriptional analysis of ntdABC operon and ntdR gene in B. subtilis wild-type and mutant (rpoB5) strains. The transcription of ntdABC and ntdR were analyzed using transcriptional lacZ fusion constructs with ntdABC (A) and ntdR (B). A, strains TI122 (squares), TI122R5 (circles), and TI127R5 (triangles) were grown in S7N medium supplemented with the required amino acids (50 g/ml Trp and 20 mM of Asp). Culture samples were withdrawn at the indicated times, and the -galactosidase activity was measured. B, strains TI123 (squares), TI123R5 (circles), and TI151R5 (triangles) were grown as in A. C, nucleotide sequence of the intergenic region between ntdR and ntdABC. The transcriptional start points for the ntdABC and ntdR are indicated as bending arrows. The 10 and 35 hexamers and ribosome binding site (RBS) are shown by capital boldface letters with underline. The inverted repeat structure is represented by arrows facing one another.

4B), although the rpoB5 mutation caused a transcriptional burst of ntdABC throughout whole growth phase of the rpoB5 mutant (Fig. 4A). To confirm the results from the lacZ fusion analysis, we conducted Northern blot experiments using probes for ntdA and ntdR. The ntdABC transcript ( 3.2 kb) was detected in rpoB5 mutant 84R5 but not in rpoB strain 61884


Activation of B. subtilis Neotrehalosadiamine Biosynthesis

FIG. 5. Effect of the NTD addition on ntdABC transcription. A, strains TI122 (rpoB ), TI122R5 (rpoB5), TI127R5 (rpoB5 ntdR::neo), TI130R5 (rpoB5 ntdA:: Tn10), TI131R5 (rpoB5 ntdB::Tn10), and TI132R5 (rpoB5 ntdC::Tn10) were grown for 10 h in S7N medium containing varying concentrations of NTD (purity 95%), and -galactosidase activity was measured. The average values for five separate experiments are shown with their S.D. values. B, gel mobility shift analysis of the ntdR-ntdABC intergenic region with His-tagged NtdR and NTD. DIG-labeled DNA fragment of the ntdR-ntdABC intergenic region was incubated with or without His-tagged NtdR, NTD, trehalose (T), or neotrehalose (NT).

Downloaded from at MAHIDOL UNIV, FAC SCIENCE on April 10, 2008

(data not shown). Consistent with the results from lacZ fusion analysis, there was no significant difference in the amount of ntdR transcript ( 1.0 kb) between the 84R5 and 61884 strains (data not shown). Thus, the ntdABC transcription was dramatically activated by the rpoB5 mutation without increasing the expression of the positive transcriptional regulator ntdR. We therefore deduced that a specific co-factor is required (or plays a role) for the transcriptional activation of the ntdABC by NtdR protein. To investigate the regulation of ntdABC transcription by NtdR protein, we first determined the transcriptional start sites for ntdABC operon and ntdR gene. For ntdABC primer extension, total RNA of the 84R5 cells was isolated at t0 (the time point of transition from exponential to stationary phase) and t4 (representing the stationary phase) and used for primer extension experiments. Transcription of ntdABC was initiated at the G residue 42 bases upstream of the translational start codon (Fig. 4C). The most probable 10 and 35 promoter sequences, as recognized by housekeeping factor ( A), were found in the upstream region of the transcriptional start site. Interestingly, the A-T-rich inverted repeat structure was found in this region, overlapped with the 10 hexamer and the transcriptional start point, suggesting that this inverted repeat may be a regulatory element for ntdABC transcription. For ntdR primer extension, we used strain 61884, harboring a

multicopy number plasmid pUB18-ntdR, because the ntdR expression was too low to detect the transcriptional start site. An intensive signal was detected at the A residue 81 bases upstream of its translational start codon, although no obvious consensus sequence was found (Fig. 4C). Recently, Bandow et al. (25) reported that the addition of Rif to the exponentially growing B. subtilis cells induces B and D regulons, eliciting the cell’s tolerance to moderate concentrations of Rif. To test whether these regulon genes were involved in the NTD production, we grew the rpoB strain TI122 in S7N medium containing varying concentrations (10 –100 g/ml) of Rif. However, we found that Rif had no effect on the activation of ntdABC transcription. Moreover, the disruption of these alternative factors, B and D, did not affect NTD production in the rpoB5 mutant (data not shown). NTD Functions as an Autoinducer for Its Own Biosynthesis—It was observed that the insertional mutation of ntdA, ntdB, or ntdC abolishes their own gene expression in rpoB5 mutant (Fig. 5A), suggesting that the transcription of ntdABC can be controlled by an autoregulation mechanism. Furthermore, the addition of purified NTD into the culture of these mutants actually induced the ntdABC transcription in a dosedependent manner (Fig. 5A). Even in the rpoB strain, ntdABC transcription was activated by the addition of NTD, although the effect in the rpoB cells was lower than that in the rpoB5

Activation of B. subtilis Neotrehalosadiamine Biosynthesis
cells. No such effect was observed in the ntdR-disrupted mutant. In contrast, the addition of trehalose or neotrehalose (also known as , -trehalose) (500 g/ml) instead of NTD failed to activate the promoter for the NTD biosynthesis gene (data not shown). Accordingly, we examined the ability of NtdR protein to interact with the internal region between the ntdABC operon and ntdR gene in the presence or absence of NTD. The 100 bp of DNA fragment containing the deduced regulatory element as mentioned above was used to investigate possible interaction with His6-tagged NtdR. A gel mobility shift assay revealed that NtdR protein apparently bound to this region even in the absence of NTD (Fig. 5B, lane 2). A complex (NtdR-NTD-DNA) mobilizing more slowly than the NtdR-DNA complex was detected in dose-dependent manner when NTD was added (lanes 4 –7). The addition of trehalose or neotrehalose instead of NTD did not affect the mobility of NtdR-DNA complex (lanes 8 and 9). Thus, it is concluded that NTD (but not trehalose and neotrehalose) probably acts as an autoinducer for its own biosynthesis operon by directly interacting with NtdR protein. Mutant RNAP Enhances the Activity of Promoters Recognized by A—Our results as described above suggest that rpoB5 mutation may render RNAP more potent for recognition of the ntdABC promoter. To assess this hypothesis, we first compared the amount of A binding to core RNAP in rpoB and rpoB5 strains, since ntdABC promoter was recognized by A (see Fig. 4C). For isolation of RNAP holoenzyme, a strain carrying a His6-tagged subunit of RNAP was used. Western blot analysis showed that the amount of A co-eluted with core RNAP in rpoB5 strain was virtually the same as that in the rpoB strain (Fig. 6A). A similar result was obtained when whole-cell lysates were fractionated by gel filtration (data not shown). Next, we measured the basal promoter activity (without autoinduction) of the NTD biosynthesis operon in rpoB and rpoB5 strains. To avoid the effect of autoinduction by endogenous NTD, we used the ntdR-disrupted mutants. The basal promoter strength of ntdABC in rpoB5 mutant was 2.5-fold higher than that in the wild-type strain (Fig. 6B), despite no observed increase in the A level associated with core RNAP (see above). The basal promoter strength was not affected when other rpoB mutants (rpoB2, rpoB6, and rpoB32) were used, indicating that the observed increase in promoter strength is specific to the rpoB5 mutation. Likewise, the activities of stable RNA promoter recognized by A (examined for rrnA, rrnB, rrnD, rrnE, rrnI, rrnJ, and rrnO) in the rpoB5 strain were also found to be about 2-fold higher than that in the wild-type strain. The result for rrnE is presented in Fig. 6C as an example. These results suggest that rpoB5 mutation results in an increase in the A-dependent promoter activity without facilitating the A-binding to core RNAP. Thus, it is thought that the rpoB5 mutation renders RNAP highly potent to enhance the transcription from the A -dependent promoters, including the ntdABC promoter, eventually leading to the dramatic induction of NTD biosynthesis operon via the autoinduction system.


Downloaded from at MAHIDOL UNIV, FAC SCIENCE on April 10, 2008

FIG. 6. Effect of rpoB mutations on the A-dependent promoter activity. A, Western blot analysis of the RNAP component. TI150 (wild type; WT) and TI150R5 (rpoB5) were grown in S7N medium until A650 0.5 (midexponential phase). Each His6-tagged RNAP holoenzyme was isolated and quantified by antibodies for core RNAP or A . B, effect of rpoB mutations on the promoter activity of ntdABC. Strains TI127 (rpoB ntdR::neo), TI127R5 (rpoB5 ntdR::neo), TI127R2 (rpoB2 ntdR::neo), TI127R6 (rpoB6 ntdR::neo), and TI127R32 (rpoB32 ntdR::neo) were grown for 10 h (early stationary phase) in S7N medium, and -galactosidase activity was measured for each of the strains. C, effect of rpoB mutations on the promoter activity of rrnE. Strains TI149 (rpoB ), TI149R5 (rpoB5), TI149R2 (rpoB2), TI149R6 (rpoB6), and TI149R32 (rpoB32) were grown in S7N until A650 0.5, and -galactosidase activity was determined. The average values for five separate experiments are shown with their S.D. values.

We have previously reported that antibiotic production by Streptomyces spp. is dramatically activated by introducing certain mutations into rpoB that confer resistance to Rif (8 –11). In the present study, we showed that the introduction of Rif r rpoB mutation into a B. subtilis strain activates the dormant ability for NTD synthesis. We also successfully identified the genes required for NTD synthesis and for transcriptional activation. Of these genes, it is highly likely that the ntdA gene product encoding a putative aminotransferase took part in amination of sugar moieties at positions 3 and 3 . Although it is difficult at present to speculate upon the role of ntdB and ntdC gene

products, either or both proteins might be involved in the formation of , -linkage of sugar moieties. Since the addition of trehalose ( , -linkage) or neotrehalose ( , -linkage) at a concentration of 1,000 g/ml did not increase the productivity of NTD regardless of the co-existence or absence of exogenously added NTD (data not shown), amination of sugar moieties may proceed prior to the formation of , -linkage. Thus, the seemingly simple biosynthetic process of NTD may offer a good system to elucidate the regulation of bacterial secondary metabolism. Another point of interest is that NTD acts as an autoinducer for its own biosynthetic process. The mechanism of autoinduction has been extensively studied in several bacteria in relation to quorum-sensing systems that are important for various physiological processes, such as acquisition of competence, sporulation, motility, biofilm formation, bioluminescence, and virulence (26 –28). In Gram-positive bacteria, most of these signaling molecules are peptides or modified peptides including subtilin (29) and nisin (30). In the model for the autoinduction of subtilin and nisin production, these antibiotics induce the transcription of their own biosynthesis genes via a two-component signal transduction system. In contrast, it is noteworthy that NTD bound directly to the positive regulator NtdR in vitro (Fig. 5B) and induced the transcription of NTD biosynthesis operon in vivo (Fig. 5A). As shown in Fig. 5B, NtdR apparently bound to the ntdABC promoter region that contains an A-Trich inverted repeat sequence overlapped with the 10 hexamer of its promoter and its transcriptional start point (Fig. 4C). Although we have no experimental evidence at this stage, it is likely that this inverted repeat structure is a regulatory


Activation of B. subtilis Neotrehalosadiamine Biosynthesis
is of current interest not only in basic microbiology but also in industrial microbiology. This is because it provides a powerful tool for the screening of novel compounds and for strain improvement to overproduce useful compounds. Recently, rpoB mutations (including S487L mutation corresponding to rpoB5) have been demonstrated to be effective for overproduction of extracellular enzymes such as amylase and protease by several Bacillus species.3 The present study, together with our previous works (8, 9), may be helpful in constructing the strategies applicable to industrial microbiology.
Acknowledgments—We thank Shin-ichi Eto, Jun-ichi Yasuda, and Chie Kobayashi for performing several experiments, Fujio Kawamura (Rikkyo University, Tokyo, Japan) for providing the antibodies and strain 1285EN1, and Masaya Fujita (Harvard University) for providing strain MF1.
1. Wehrli, W., Knusel, F., Schmid, K., and Staehelin, M. (1968) Proc. Natl. Acad. Sci. U. S. A. 61, 667– 673 2. Campbell, E. A., Korzheva, N., Mustaev, A., Murakami, K., Nair, S., Goldfarb, A., and Darst, S. A. (2001) Cell 104, 901–912 3. Ezekiel, D. H., and Hutchins, J. E. (1968) Nature 220, 276 –277 4. Wehrli, W., Knusel, F., and Staehelin, M. (1968) Biochem. Biophys. Res. Commun. 32, 284 –288 5. Ingham, C. J., and Furneaux, P. A. (2000) Microbiology 146, 3041–3049 6. Nicholson, W. L., and Maughan, H. (2002) J. Bacteriol. 184, 4936 – 4940 7. Zhou, Y. N., and Jin, D. J. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 2908 –2913 8. Hu, H., and Ochi, K. (2001) Appl. Environ. Microbiol. 67, 1885–1892 9. Hu, H., Zhang, Q., and Ochi, K. (2002) J. Bacteriol. 184, 3984 –3991 10. Lai, C., Xu, J., Tozawa, Y., Okamoto-Hosoya, Y., Yao, X., and Ochi, K. (2002) Microbiology 148, 3365–3373 11. Xu, J., Tozawa, Y., Lai, C., Hayashi, H., and Ochi, K. (2002) Mol. Genet. Genomics 268, 179 –189 12. Tsuno, T., Ikeda, C., Numata, K., Tomita, K., Konishi, M., and Kawaguchi, H. (1986) J. Antibiot. (Tokyo) 39, 1001–1003 13. Numata, K., Satoh, F., Hatori, M., Miyaki, T., and Kawaguchi, H. (1986) J. Antibiot. (Tokyo) 39, 1346 –1348 14. Ochi, K., and Ohsawa, S. (1984) J. Gen. Microbiol. 130, 2473–2482 15. Inaoka, T., Takahashi, K., Ohnishi-Kameyama, M., Yoshida, M., and Ochi, K. (2003) J. Biol. Chem. 278, 2169 –2176 16. Fukuchi, K., Kasahara, Y., Asai, K., Kobayashi, K., Moriya, S., and Ogasawara, N. (2000) Microbiology 146, 1573–1583 17. Ohashi, Y., Ohshima, H., Tsuge, K., and Itaya, M. (2003) FEMS Microbiol. Lett. 221, 125–130 18. Moriya, S., Tsujikawa, E., Hassan, A. K., Asai, K., Kodama, T., and Ogasawara, N. (1998) Mol. Microbiol. 29, 179 –187 19. Itaya, M., Kondo, K., and Tanaka, T. (1989) Nucleic Acids Res. 17, 4410 20. Gruss, A., and Ehrlich, S. D. (1989) Microbiol. Rev. 53, 231–241 21. Steinmetz, M., and Richter, R. (1994) J. Bacteriol. 176, 1761–1763 22. Inaoka, T., and Ochi, K. (2002) J. Bacteriol. 184, 3923–3930 23. Fujita, M., and Sadaie, Y. (1998) Gene (Amst.) 221, 185–190 24. Pfeifer, B. A., Admiraal, S. J., Gramajo, H., Cane, D. E., and Khosla, C. (2001) Science 291, 1790 –1792 25. Bandow, J. E., Brotz, H., and Hecker, M. (2002) J. Bacteriol. 184, 459 – 467 26. Kaiser, D., and Losick, R. (1993) Cell 73, 873– 885 27. Kleerebezem, M., and Quadri, L. E. (2001) Peptides 22, 1579 –1596 28. Fuqua, C., Parsek, M. R., and Greenberg, E. P. (2001) Annu. Rev. Genet. 35, 439 – 468 29. Stein, T., Borchert, S., Kiesau, P., Heinzmann, S., Kloss, S., Klein, C., Helfrich, M., and Entian, K. D. (2002) Mol. Microbiol. 44, 403– 416 30. Kuipers, O. P., Beerthuyzen, M. M., de Ruyter, P. G., Luesink, E. J., and de Vos, W. M. (1995) J. Biol. Chem. 270, 27299 –27304 31. Cashel, M., Gentry, D. R., Hernandez, V. J., and Vinella, D. (1996) in Escherichia coli and Salmonella: Cellular and Molecular Biology (Neidhardt, F. C., Curtiss, R., III, Ingraham, J. L., Lin, E. C. C., Low, K. B., Magasanik, B., Reznikoff, W. S., Riley, M., Schaechter, M., and Umbarger, E., eds) Vol. 1, 2nd Ed., pp. 1458 –1496, American Society for Microbiology Press, Washington, D. C. 32. Chang, D. E., Smalley, D. J., and Conway, T. (2002) Mol. Microbiol. 45, 289 –306 33. Eymann, C., Homuth, G., Scharf, C., and Hecker, M. (2002) J. Bacteriol. 184, 2500 –2520

element for ntdABC transcription. In our model, the NtdR protein dramatically stimulates the ntdABC transcription when its promoter region is occupied by NTD-bound NtdR protein. In line with this notion, NTD-unbound NtdR protein results in a failure to express ntdABC. For further understanding the function of this intriguing protein, elucidation of the ternary structure by x-ray analysis may be helpful. One of the most important regulations for gene expression (including that for antibiotic production) in bacteria is stringent control that enables cells to adapt to nutrient-limiting conditions (31). The effector molecule (called bacterial alarmone) of the stringent control is a hyperphosphorylated guanosine nucleotide, ppGpp, which binds to the and subunits of core RNAP. Alarmone takes part in the global control of gene expression in bacteria (32, 33). In E. coli, it was shown that certain Rif r mutants behave like “stringent” RNAPs even in the absence of the stringent response (7). Similarly, in S. coelicolor A3(2) and S. lividans, alteration at His437 (corresponding to His-482 in B. subtilis) to Arg or Tyr can suppress the deficiency of antibiotic production resulting from the lack of ppGpp due to relA or relC mutation (9 –11). These results suggest that the mutant RNAPs functionally mimic “stringent” RNAP, leading to the activation of a stringent response-dependent antibiotic biosynthesis pathway (10, 11). Although rpoB5 mutation also effectively activated NTD production, the activation mechanism for NTD production in B. subtilis appears quite different from that in Streptomyces spp. in the following ways. (i) Unlike the case for NTD production, rpoB5 mutation did not enhance the bacilysin biosynthesis pathway (see “Results”), which is controlled in the stringent response-dependent manner (15); (ii) the rpoB mutations altering His-437 to Arg or Tyr (corresponding to rpoB2 or rpoB6 in B. subtilis) effectively activated antibiotic production in S. lividans (9), whereas these mutations were ineffective in B. subtilis NTD production (Table I); (iii) unlike bacilysin production, NTD production was not enhanced by provoking the stringent response upon amino acid shift-down (data not shown). It is also notable that, unlike the case of Streptomyces spp. (9, 11), any Rif r rpoB mutations (including rpoB5) we tested so far were ineffective to suppress the ppGpp-deficient phenotype on bacilysin production.2 Thus, we propose that activation of NTD production by rpoB5 mutation emerges from events independent of the stringent response, by which bacterial secondary metabolism is often controlled. This suggestion is supported by the fact that rpoB5 mutation caused an increase in the promoter activity of stable RNA genes as exemplified for rrnE (Fig. 6B), the expression of which is normally reduced by stringent response. It is conceivable that the RNAP of B. subtilis rpoB5 mutant acquired a superior ability (apparently by alteration of its ternary structure) to efficiently transcribe A-dependent promoters. For a precise understanding of the mechanism of gene activation by rpoB5 mutation, it may be useful to perform comparative studies between B. subtilis and B. pumilus (or B. circulans) that are capable of producing NTD (12, 13). How to activate dormant antibiotic biosynthetic gene cluster

Downloaded from at MAHIDOL UNIV, FAC SCIENCE on April 10, 2008

J. Yasuda, T. Inaoka, and K. Ochi, unpublished results.

S. T. Jørgensen, personal communication.