The Structure and Biosynthesis of New Tetrahydropyrimidine Derivatives in Actinomycin D Producer Streptomyces parvulus

Two novel compounds, 2-methyl, 4-carboxy, 5-hydroxy-3,4,5,6-tetrahydropyrimidine (THP(A)) and 2-methyl, 4-carboxy-3,4,5,6-tetrahydropyrimidine (THP(B)) have been identified in the pool of Streptomycesparvulus by in vivo and in vitro studies. 13C and 16N were introduced into the compounds by feeding S. parvulus with "Nand "C-labeled L-glutamate. High resolution 13C and "N NMR have been applied to elucidate their structure and biosynthesis in S. parvulus. The splitting patterns and coupling constants of adjacent nitrogen-carbon molecular fragments enable us to unravel their molecular structure. Two different glutamate pools are responsible for their biosynthesis, THP(A) carbon skeleton derives from the extracellular ~-['~C]glutamate, whereas THP(B) stems from D-fructose via the intracellular glutamate. During cell growth, THP(A) is synthesized and becomes the major constituent of the intracellular pool. It is consumed after THP(B) is accumulated intracellularly. The onset of THP(A) and -(B) synthesis seems correlated to the time of actinomycin D synthesis. Their high cellular concentrations during actinomycin D synthesis suggest that they may function as nitrogen storage. Other possible functions of THP molecules within the cell are discussed.


The Structure and Biosynthesis of New Tetrahydropyrimidine Derivatives in Actinomycin D Producer Streptomyces parvulus
USE OF 13C-AND 15N-LABELED L-GLUTAMATE AND 13C AND 15N NMR SPECTROSCOPY* (Received for publication, March 4, 1988)

Livia Inbar and Aviva Lapidot$
From the Isotope Department, Weizmann Institute of Science, 76100 Rehouot, Israel Two novel compounds, 2-methyl, 4-carboxy, 5-hydroxy-3,4,5,6-tetrahydropyrimidine (THP(A)) and 2-methyl, 4-carboxy-3,4,5,6-tetrahydropyrimidine (THP(B)) have been identified in the pool of Streptomycesparvulus by in vivo and in vitro studies. 13C and 16N were introduced into the compounds by feeding S. parvulus with "N-and "C-labeled L-glutamate. High resolution 13C and "N NMR have been applied to elucidate their structure and biosynthesis in S. parvulus. The splitting patterns and coupling constants of adjacent nitrogen-carbon molecular fragments enable us to unravel their molecular structure. Two different glutamate pools are responsible for their biosynthesis,

THP(A) carbon skeleton derives from the extracellular ~-['~C]glutamate, whereas THP(B) stems from D-fructose via the intracellular glutamate. During cell growth, THP(A) is synthesized and becomes the major constituent of the intracellular pool. It is consumed after THP(B) is accumulated intracellularly. The onset of THP(A) and -(B) synthesis seems correlated to the time of actinomycin D synthesis. Their high cellular
concentrations during actinomycin D synthesis suggest that they may function as nitrogen storage. Other possible functions of THP molecules within the cell are discussed.
In our recent paper (l), 13C and "N NMR techniques have been used to measure metabolic regulation operating in Streptomyces parvulus during actinomycin D synthesis. We followed the fate of nutrient incorporation intracellularly and the relationship of primary metabolites and actinomycin D synthesis. Two previously unknown metabolites have been found to accumulate intracellularly throughout cell growth and actinomycin D production. Much research has been done on the biosynthesis of actinomycin D in Streptomyces, and most enzymes associated with the antibiotic synthesis have been identified (2)(3)(4)(5)(6)(7). But little is known about the intracellular pool metabolites, their function in the cell life cycle, and the general regulation of antibiotic synthesis. Timing of antibiotic synthesis and rate of its production may also be affected by the availability of intracellular pool precursors (8).
In the present study, the molecular structure of the new metabolites, tetrahydropyrimidine derivative, was under-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed.
taken. In addition, attempts have been made to reveal the pathways leading to the new compounds in order to understand the biogenesis of pyrimidine-like compounds and their association with actinomycin D synthesis in S. parvulus. 13C and 15N were incorporated into the new compounds by culturing S. parvulus with and l6N-1abeled L-glutamate. High resolution 13C and 15N NMR spectroscopy directed toward elucidating their structure and biosynthesis have been performed by in vivo and in vitro studies. 13C and 16N chemical shifts, 'H-13C, 13C-13C, 'H-16N, and 13C-15N splitting patterns and coupling constants enabled us to reveal the molecular structures of the two new metabolites: 2-methyl, 4-carboxy, 6-hydroxy-3,4,5,6-tetrahydropyrimidine (THP(A))' and 2methyl, 4-carboxy-3,4,5,6-tetrahydropyrimidine (THP(B)).
Strain and Culture-8. paruulus (ATCC 12434) (kept on soil culture at 4 "C) was grown on NZ amine medium for 2 days at 30 "C in a gyratory shaking incubator. After centrifugation and washing twice with 0.2% KCl, a suspension of the mycelium (3%) served as inoculum for the chemically defined growth medium (GF) (11) which consisted of 40 g D-fructose, 1.0 g of KZHPO,, 25 mg ZnSOl. 7H20; 25 mg of CaC12. 2H20; 25 mg of MgS04.7H20,25 mg of FeS04. 7H20, and 2.1 g of L-glutamic acid/1000 ml of deionized water at pH 7.1 (growth medium 1). Medium 2 is as medium 1 in which L-glutamate was replaced by L-[13C]glutamate. Medium 3 is as medium 1 in which L-glutamate was replaced by ~-['~N]glutamate.
Cell and Cell Extract Preparations for N M R Measurement-Cell extracts of nonlabeled S. paruulus cells were prepared from 1 liter of S. paruulus growth culture. Cells were harvested by centrifugation (at 5,000 X g for 5 min) at -4 "C. Cells were washed twice with 0.2% KC1 solution to remove traces of culture medium. Intracellular extracts were obtained either by suspending washed cell pellets in 10 ml of water and heating them for 15 min at 100 "C as previously described (12) or by the perchloric acid procedure. The supernatant was sepa-'The abbreviations used are: THP(A), 2-methyl, 4-carboxy, 5hydroxy-3,4,5,6-tetrahydropyrimidine (metabolite A); THP(B), 2methyl, 4-carboxy-3,4,5,6-tetrahydropyrimidine (metabolite B); GC-MS, gas chromatography mass spectroscopy; NOE, nuclear Overhauser effect; TMS, tetramethylsilane. rated by centrifugation at 15,000 X g for 15 min and concentrated under reduced pressure to 1 ml. No significant changes of cell extracts components have been noted by the NMR or other analytical methods using the two procedures for cell extraction. For NMR studies of isotopically labeled cells, S. paruulus cells were incubated at 30 "C for 72 h (mid-log phase) in growth medium containing ~-['~C]glutamate (medium 2). Three-ml cell culture samples were harvested by centrifugation and washed twice with 0.2% KC1 before transferring to the NMR tube. For 15N NMR studies, 200 ml of cell culture were harvested after 48 h of incubation with L-["N] glutamate (medium 3). Cell extracts were prepared as described above and transferred to the 10-mm NMR tube. The NMR measurements were carried out at 7 "C. Separation of THP(A) and -(B) from S. paruulus Cell Extracts-Cell extract samples were mixed with 4 volumes of 1 M acetic acid before separation by Dowex 50 W chromatography. The columns were washed with 10 ml of deionized water to remove carbohydrates and polyols. The amino acids were eluted with 5 ml of 3 M NH40H, this fraction contained also the two new compounds THP(A) and -(B). Ammonia was evaporated to dryness, and the residue was brought to pH 5 and separated on Dowex 1 anion-exchange chromatography. Acidic amino acids such as glutamic acid remained attached to the anion exchange. THP(A) and -(B) were eluted with 10 ml of water, as was shown by I3C NMR spectroscopy. Glutamic acid was eluted by 10 ml of 1 N HCl.
Basic Hydrolysis of THP(A) and -(B)-The water fraction eluted from Dowex 1 anion-exchange column, which contains THP(A) and -(B), was concentrated to 1 ml; 1 ml of 20% KOH was added, and the mixture was heated in sealed tubes at 110 "C for 3 h. After cooling the mixture was analyzed by NMR, GC-MS, and amino acid analyzer.
NMR Spectr~scopy-~~C NMR spectra at 67.89 MHz were obtained with a Bruker WH-270 MHz spectrometer operating in the FT mode (at 7 "C). Proton broad band-decoupled 13C NMR spectra were obtained with the following spectrometer conditions: 60" pulses, 12.5 kHz spectral width, 2-s repetition time which avoids signal saturation of protonated carbons, and 16K Fourier data transform. Field stabilization was accomplished by locking on D20. 13C chemical shifts were measured with respect to external tetramethysilane. Signal areas were integrated manually and with the aids of the Bruker software.
High resolution I3C NMR spectra obtained with a Bruker AM-400 spectrometer operating at 100.62 MHz were operated with powergated proton decoupling to reduce effects from dielectric heating and to maintain the sample temperature at about 10 "C. Proton-decoupled 13C NMR spectra and gated 'H decoupling with full nuclear Overhauser effect (NOE) were obtained, with the following spectrometer conditions: 60" pulses, 23.8 KHz spectral width, 2-s repetition time, and 16K Fourier data transform. Sample tube of 10 mm outer diameter were used with both instruments. I5N spectra were obtained at 27.37 MHz with a Bruker WH 270 Spectrometer. Proton-decoupled 15N NMR spectra and gated 'H decoupling with full NOE were obtained with pulse width 60 p s , 6 KHz spectral width, 16K data points, and a recycle time of 6.3 s. These NMR conditions did not result in saturation or in differential NOES of most of the resonances.

GC-MS Measurements-GC-MS analyses were performed on a
Finnigan 4500 quadrupole GC-MS interfaced to an INCOS data system. The mass spectrometer was operated in the chemical ionization mode with isobutane as reactant gas. Samples were introduced through the GC-MS inlet system. Measurements of isotopic abundance were made using computer-controlled selected ion monitoring. Purified cell extracts were hydrolyzed for 3 h in 10% KOH solution at 110 "C in sealed tubes. Samples were evaporated to dryness under a stream of nitrogen gas. Last traces of water were removed by azeotropic distillations of methylene chloride. The dry samples were derivatized to N-trifluoroacetyl-n-butyl esters as described previously (13). The mixture of trifluoroacetate-n-butyl esters was injected into a glass column packed with Tabsorb Hac or into a 30 m fused silica capillary SE 54 column (J & W Scientific Inc.). The separation conditions of the GC were: pressure, 12 p.s.i.; split, 15 ml/min (ratio 1:25); injection temperature, 220 "C; temperature program, 100 "C isothermal for 1 min, then 100-200 "C at 5 "C/min. MS conditions were: transfer line temperature, 250 "C; ion source temperature, 150 'C; manifold temperature, 100 "C; ionizing energy, 70 eV; multiple ion voltage supply, 1.5 kv; and emission current, 0.5 mA. Calculations of isotopically abundance were made as described previously (14) and are presented as atom % excess.
Amino Acid Analysis-Intracellular pool of S. paruulus grown on different media (1, 2, and 3) were measured for their amino acids contents. Cell extract hydrolysates were also analyzed by amino acid analyzer (Dionex D-500).  a I3C chemical shifts are in uarts/million down-field from tetramethvlsilane. "N chemical shifts are in Darts/ million up-field from HN03. -.
Not separated.

RESULTS
out adding any isotopically labeled precursor, was analyzed by I 3 C NMR spectroscopy. The natural abundance proton-Natural Abundance 13C NMR Spectrum of s. parvulus Cell decoupled spectrum of cell extract consists of numerous sharp Extract-The intracellular pool of S. parvulus cell culture, resonances that have been assigned on the basis of chemical grown in chemically defined medium (GF) (medium 11, with-shifts previously reported (1, 12, 15). The spectrum reflects Amino acid analysis confirmed our results that glutamic acid constitutes >80% and alanine -15% of the total amino acids pool level after 48 h growth of S. parvulus in GF medium.
13C NMR of the Unknown Metabolites A and B-The natural abundance proton-decoupled spectrum of partially purified metabolites A and 3 is shown in Fig. lb. Resonances of alanine (Ala C-3 and Ala C-2) and choline derivative group -N+(CH& (PC) are also observed. However, the intense resonances of carbohydrate F and M carbons at .the region 63-80 ppm, and of glutamate observed in Fig. la, are not seen in the 13C NMR spectrum obtained after removing carbohydrates (by cation-exchange chromatography) and glutamate (by anion-exchange chromatography) (Fig. Ib). In both spectra the intracellular level of metabolite B is %fold higher than that of metabolite A. The different chemical shifts which reflect the electronic distribution surrounding the observed nucleus are used for structural characterization of the measured molecule.
The use of gated-decoupling technique (['HI on between data acquisitions) can regain some sensitivity due to the NOE and in the same time give information on 13C-'H coupling constants as observed in 13C NMR experiments done without 'H decoupling. From the signal multiplicities observed in the gated-decoupled spectrum, it is possible to discriminate between methyl (quartet), methylene (triplet), methine (doublet), and quaternary carbon (singlet) resonances. Fig. 2 is the 13C NMR of the power-gated (Fig. 2a) and gated-decoupled spectra (Fig. 2b) (at 100.6 MHz), of S. parvulus cell extract, partly purified metabolites A and B. Six different carbon species are summarized in Table I for  Proton-decoupled 13C NMR Spectra of S. parvulus Cells Grown on ~-'~C-Labeled Glutamate-The 13C-13C splittingpatterns of enriched products can be used to assign the carbon resonances by the multiplet pattern produced. In the present study highly enriched ~-['~C]glutamate was used as the labeled precursor (medium 2). The I3C spectrum obtained from intact S. parvulus cells grown in the presence of %-labeled glutamate and unlabeled D-fructose is shown in Fig. 3 Fig. 3b). The weak resonances arising from trehalose do not show 13C-'3C coupling, indicating that they are not originating from adjacent 13C-13C fraction of the 13C-labeled glutamate. The 13C-13C splitting pattern provides information on the carbon fragment condensation (9, 16). Each carbon is represented by several peaks which are the super position of spectra from all the 13C isotopomers present. As a result of spin-spin coupling, I3C nuclei that are directly bonded to each other appear as doublets; 13C nuclei that are bonded only to 12C nuclei are single peaks. The same 13C-13C coupling constants appear twice, thus allowing the identification of two adjacent carbons. Fig. 3a, (Table I).
From the above results, derived from 13C and 15N NMR spectroscopies, proton-coupled and -decoupled techniques, 13C and 15N chemical shifts, 'H-13C; 'H-"N, 13C-'3C, and 13C-15N coupling constants, the molecular structure of the two stable metabolites A and B could be elucidated. 13C NMR spectroscopy clearly reveals six carbons in each metabolite. The high resolution 13C proton-coupled natural abundance spectrum is consistent with six different carbon groups: methyl ( C R ) , methylene (CH2), two methane (CH), and two quaternary carbon species for metabolite A and six different carbon species, methyl, two methylene, methane, and two quaternary carbons, for metabolite B. 13C NMR of 13C-enriched metabolite A revealed that the methyl carbon resonate at 19 ppm is coupled to the quaternary carbon at 161.5 ppm. 15N chemical shifts and proton-coupled 15N-'H spectrum confirm the existence of two NH groups in each one of the new metabolites. The availability to identify adjacent carbons to nitrogenous group, from their spin-spin coupling patterns and coupling contants, revealed that C-6 (at 43.5 ppm) and C-4 (at 60.6 ppm) of metabolite A are adjacent to different nitrogenous groups. In metabolite B, C-6 (at 38.0 ppm) and C-4 (at 53.6 ppm) are adjacent to two different nitrogenous groups. The quaternary carbons at 161.5 ppm, in both metabolites A and B are coupled to the two nitrogens in an amidine bond (N=CNH). Summary of 13C and I5N chemical shifts, 13C-'H; 15N-'H, l3C-I3C, and l3C-I5N coupling constants of the two tetrahydropyrimidine derivatives are summarized in Table I. The suggested molecular structure for metabolites A is 2-methyl, 4-carboxy, 5-hydroxy-3,4,5,6-tetrahydropyrimidine.

4'
Confirmation of the Molecular Structure of the Tetrahydropyrimidine Derivatives by Basic Hydrolysis, NMR, and GC-MS Studies-The suggested molecular structures of the two new metabolites are consistent with the stability of cyclic amidines to acidic condition. Hydrolysis at a significant rate appears only at high pH, as in shown in Scheme 1. Indeed, 10% KOH (at 100 "C for 3 h), they hydrolyzed to amino acids. Their times of elution, obtained by amino acid analyzer, were similar to the expected elution time of basic amino acids, such as lysine and ornithine. The molecular structures of the two hydrolysates were confirmed by I3C, 15N NMR, and molecular weights were determined from mass spectroscopy analysis. The 13C NMR spectrum shown in (Fig. 6b) is significantly different from the THP(A) and -(B) spectrum (Fig. 6a). New resonances at 24.3 and 182.0 ppm arise from acetate carbons as a result of ring opening and hydrolysis of the tetrahydropyrimidine derivatives (Scheme 1).
The carbon 13 resonances of the hydrolysates A' and B' are in agreement with the structures suggested for A' and B': 2,4 diamino-3-hydroxybutyric acid (A') and 2,4-diaminobutyric acid (B') (Table 111) The 15N NMR spectrum of the amino acids A' and B' is shown in Fig. 7. The 15N chemical shifts at 334.9 and 347.8 ppm and those at 335.5 and 342.4 ppm correspond to free amino groups NH2, two for each of the hydrolysate products A' and B', respectively.
Chemical ionization GC-MS was carried out to determine the molecular weights of the basic hydrolysate samples. They were derivatized to N-trifluoroacetyl-n-butyl esters before injection to the GC-MS. Chemical ionization GC-MS revealed molecular ions (M + 1) m/z 383 and (M + 1) m/z 366 of the N-trifluoroacetyl-n-butyl ester derivatives 2,4-diamino-3-hydroxybutyric acid and 2,4-diaminobutyric acid, respectively.
Analysis of the Labeling Pattern of THP(A) from L-PCI Glutamate-Analysis of 13C multiplets of a complex pattern  a I3C enrichments were determined from relative peak areas of the 13C resonances in comparison to glutamate C-3 and C-4 in Fig. 3a. Glutamate C-3 and C-4 13C enrichments were analyzed by GC-MS as previously described (9).
The attenuated signal intensity was corrected by normalizing to natural abundance compound.
Not separated. Ac, acetate derived from the basic hydrolysis. Summary of the chemical shifts are given in Table 11.

C-2 I
originating from labeled metabolites provide means for quantifying the proportion of each labeled species in the mixture of isotopomers (9,16,18). Adjacent carbons that have predominant multiplet pattern resonances are originated from the same precursor fragment. The 13C-13C splitting pattern enables one to follow if the precursor fragment is incorporated intact. The pattern at a specific site give information from which one can determine the relative amount of labeled and unlabeled carbons at the neighboring position and to identify nonrandom labeled distribution. NMR peak intensities can be used to determine the relative 13C enrichment in various carbons. The 13C enrichment of THP(A) carbons were determined by comparison of their peak areas with those of intracellular glutamate C-3 (and/or C-4). The 13C enrichments of glutamate carbons have been analyzed by chemical and electron ionization GC-MS measurements as previously described (9). THP(A) 13C enrichments are summarized in Table 111. The glutamate used in this study consists of three adjacent carbons of equal carbon-13 enrichments; C-2, C-3, and C-4, 70% enriched; C-l,35% enriched; and C-5 is not 13C-enriched. As seen in Table I11 the most enriched THP(A) carbons are C-4 and C-5 (each one is 23% W-enriched). These carbons may be derived from glutamate C-3 and C-4 of equal 13C enrichments. The decreased 13C enrichment of C-6 may arise from glutamate C-5 (zero labeled) and C-2 after randomization via malate-furmarate in trichloroacetic acid cycle. Indeed its 13C enrichment is about half of that of C-4 and C-5, and its doublet to singlet ratio is only 2.4. Whereas the doublet to singlet ratio of the pair C-2 and C-2' is 4.0 (Table 111), indicating that both carbons are derived from intact glutamate C-3 and C-4 carbons. The significant lower 13C enrichment of this fragment (17%) suggests that THP(A) C-2, C-2' fragment, stems from pyruvate or acetyl-coA carbons derived from glutamate C-3, C-4 and is further diluted by unlabeled pool of pyruvate (or acetyl-coA) originating from the unlabeled D-fructose catabolism. DISCUSSION In the present investigation we have shown that S. paruulus cells have the capacity to maintain high levels of previously unknown pyrimidine derivative.
The goal of this study was to define the molecular structure of the new metabolites discovered in the process of actinomycin D synthesis. The molecular structure of the THP(A) and -(B) have been deduced from 13C and 15N NMR chemical shifts, 13C-lH, 15N-'H, and 13C-13C spin-spin splitting patterns and coupling constants. But only the results derived from 13C-15N spin-spin splitting pattern and coupling constants allowed us to eliminate other possible molecular structure than those suggested for THP(A) and THP(B): 2-methyl, 4-carboxy, 5hydroxy-3,4,5,6-tetrahydropyrimidine and 2-methyl, 4-carboxy, 3,4,5,6-tetrahydropyrimidine, respectively. The molecular weights of THP(A) and -(B) hydrolysates, measured by chemical ionization GC-MS technique, substantiate the suggested structures for THP(A) and -(B). The stereochemistry of THP(A) and -(B) are currently being studied.
The labeling experiments here described confirm our recent observations that intracellular glutamate pool did not originate from culture medium glutamate (1 (Fig. 3). is originating from the new intracellular glutamate pool, derived from D-fructose catabolism.
Significant difference of intracellular mobilities of the new THP molecules are noted. THP(B) NMR resonances are attenuated and broadened beyond detection in the intact cells (1). Only following cell membrane and/or cell wall rupture, during intracellular cell extraction, this metabolite is released and can be detected by 13C and 15N NMR spectroscopy.
Times of synthesis and consuming of THP(A) and -(B) during cell life cycle were followed by natural abundance NMR of cell extracts. The extracellular L-glutamate is consumed by the cell during the first 30 h of cell growth, during this period we have found that THP(A) is synthesized and becomes the major constituent of the intracellular pool. THP(A) is consumed after THP(B) is accumulated intracellularly. The onset of THP(B) synthesis seems correlated to the time of onset of actinomycin D synthesis, after the release from exogenous L-glutamate metabolic repression (Fig. 8).
THP(A) and -(B), as well as intracellular glutamate, are consumed during actinomycin D synthesis (Fig. 8). They are not excreted to the medium-like actinomycin D. The high cellular concentration of THP(A) and -(B) suggests that they could function in nitrogen storage. Their nitrogen pool is at least twice that of the endogenous glutamate pool during actinomycin D synthesis (Fig. 8). They are slowly catabolized during actinomycin D synthesis. Their role might be similar to that of trehalose, a carbohydrate storage material, found in this bacteria and in other microorganisms (1, 12, 19-21).
THP(A) and -(B) could be also involved in metabolic regulation of actinomycin D synthesis. It is known that cyclic AMP (CAMP) is involved in catabolic repression in several organisms. Foster and Katz (4) concluded that neither CAMP nor cyclic GMP had any effect on relieving glutamate repression in S. parvulus. It is likely that some other nucleotides take the place of CAMP. THP molecules, found in the present study, could be involved in relieving glutamate repression and regulating actinomycin D synthesis in S. parvulus. Other possible function of THP molecules within the cell may be R defensive role for the organism. Further studies are planned to ascertain the role of these new metabolites.
We anticipate that knowledge of the structures and biosynthesis of new pyrimidine derivatives will contribute to better understanding of the control mechanisms of antibiotic biosynthesis.