Solid-state NMR Studies of KZebsieZZa pneumoniae Grown under Nitrogen-fixing Conditions*

The carbon and nitrogen metabolism of Klebsiella pneumoniae M5al has been characterized using ‘‘C and I5N labeling with detection by cross-polarization magic-angle spinning solid-state NMR. Cells grown on ammonium typically require some 20 h to derepress fully for nitrogenase when transferred to medium de-void of any source of fixed nitrogen. We have estab- lished that during this period some cellular proteins are catabolized with the liberated nitrogen being used for the synthesis of purines needed for formation of ribosomal RNA. The 20-h derepression period can be shortened to 6 h by the introduction of fixed nitrogen in certain specific forms. Serine is the most successful agent we have examined for shortening the derepression period and glycine among the least successful. We attribute this difference to the advantage of serine over glycine in providing both specific and nonspecific car- bon and nitrogen sources for complete purine synthesis. These determinations were made by tracing the metabolism of 13C- and “N-labeled chemical bonds from the 2 amino acids during derepression. Lindsay

Lindsay (1) observed that the 15-20 h diauxic lag period for growth of Klebsiella pneumonine on N2 following exhaustion of supplied fixed nitrogen such as ammonium, was considerably shortened by the addition of certain amino acids. Brill and co-workers (2, 3) subsequently made use of this fact in their studies of the regulation and genetics of nitrogenase and developed a protocol which involved the addition of 0.5 mM L-serine to the growth medium. This reduced the derepression period to 6-8 h.
The role serine plays in hastening derepression has never been totally understood. It might be simply to alleviate insufficient intracellular serine needed for synthesis of nitrogenase during derepression. However, serine is present in only an average amount within nitrogenase (4). In addition, some amino acids more difficult for the organism to synthesize than serine have no comparable influence on shortening the time of derepression (1).
In the early stages of derepression, cells of K. pneumoniae are nitrogen starved and under metabolic stress (5). The effect of serine, and other amino acids able to shorten the derepression period, might be interpreted in terms of relative uptake * 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. and utilization of serine as a general source of nitrogen during the stress period. This does not explain, however, why serine is more effective than, say, aspartate and glutamate (see Miniprint Supplement), both of which are taken up by the cell and are excellent sources of nitrogen via transamination.
To establish the role that serine plays in the derepression of nitrogenase, we have performed solid-state 13C and 15N NMR experiments on intact lyophilized solid samples of K. pneumoniae, grown under N P fixing and nonfixing conditions, in the presence of both specific and nonspecific 13C and 15N labels. The resulting spectra are sufficiently detailed that the total flow of carbon and nitrogen from serine into purines and proteins can be established. These results lead to the conclusion that serine's advantage over other amino acids in shortening derepression lies in its suitability for supplying both carbon and nitrogen for the purines needed for ribosomal RNA production preceding nitrogenase synthesis. ( (6) was used as the basal medium and was supplemented by 1 mg/ml ammonium acetate for growth of cells under nonnitrogen fixing conditions. Derepression experiments were performed by the procedure of Brill et al. (2). These experiments involved growth of cells under nonnitrogen fixing conditions, followed by suspension of the cells in a nitrogenfree medium in an open-atmosphere fermenter continuously sparged with natural-abundance nitrogen gas. After a 1.5-h incubation to ensure that trace ammonium was depleted from the medium, L-serine (0.5 mM) or other amino acids were added and the lag times measured for derepression of nitrogenase and onset of cellular growth.

Chemicals-~-[2-'~C,'~N]Serine
Cells to be used for solid-state NMR analysis were harvested as a function of time after addition of amino acid, washed with 0.025 M potassium phosphate, pH 7.0, centrifuged, frozen in liquid nitrogen, and lyophilized.
Preparation of Crude Extracts and Assay of Nitrogenase Actiuity-Crude extracts were made following the procedure of Brill et al. (2). Nitrogenase activity of crude extracts was monitored using an acetylene reduction assay (7). Specific activity of nitrogenase was defined as micromoles of ethylene formed per min per mg of protein. Protein concentration was determined by a microbiuret method with bovine serum albumin as the standard (8).
Magic-Angle Spinning NMR-I'N NMR spectra were obtained at 20.3 MHz using matched spin-lock cross-polarization transfers with 2-or 5-ms contacts and 35 kHz Hl's (9). The chemical shift scale is in parts per million downfield from external solid ammonium sulfate. The dried samples were contained in a cylindrical double-bearing rotor spinning at 3.2 kHz. A few spectra were obtained at 9.07 MHz with a single-bearing rotor spinning at 1.6 kHz. Technical details of the spinning and cross-polarization procedures are reported elsewhere (10, 11). Fast cross-polarization rates for protonated nitrogens, long proton rotating-frame lifetimes, and high concentrations of protons in these biological samples ensure representative relative NMR intensities for all nitrogens (except ammonium ion) after a 2-ms contact. When required, absolute concentrations of "N can be determined quantitatively by comparisons to "N NMR spectra of a labeled calibration standard like glutamine containing a known amount of "N (11). Intensities of analytical and calibration spectra are easily adjusted for differences in rotating-frame proton relaxation rates by the standard procedure of systematic variation of the contact time (12).
"C NMR spectra were obtained at 50.3 MHz using a 2-ms contact and 50-kHz Hi's. The carbon chemical shift scale is in parts per million downfield from external tetramethylsilane. All spectra were collected under spinning sideband suppression (13) except for those at 9.07 MHz. Double cross-polarization NMR techniques and analyses (14)(15)(16) are described in the Miniprint Supplement.

RESULTS A N D DISCUSSION'
Nitrogen from Proteins-The CPMAS' "N NMR spectrum of uniformly "N-labeled intact K. pneumoniue is shown in Fig. 1 (left). The bacterium was grown in the presence of spectrum (obtained under low-speed spinning conditions which produce the mechanical spinning sidebands observed at the extremes of the spectrum) shows label in purine and peptide nitrogen, as well as protein side chain nitrogen in histidyl, arginyl, and lysyl residues (17,18).
Cells produced in the above manner were then transferred into fresh nitrogen-free medium contained within a closedatmosphere fermenter built for long term I5N2 labeling experiments (see Miniprint Supplement for details). The cells were derepressed for nitrogenase under an anaerobic atmosphere composed of 30% 14N2 and 70% helium, with the addition of 0.5 mM L-serine 1.5 h after transfer. Under these conditions, cells harvested at the end of one doubling period contain only half as much I5N on a per mg cellular dry weight basis as the starting cells. Thus the I5N NMR spectrum of a sample of final cells with twice the weight of a sample of starting cells permits direct comparisons to be made (Fig. 1, left and middle). We observe a diminution in the peptide-peak intensity, along with a corresponding increase in purine-peak intensity for the final cells relative to those of the starting cells. The complementary experiment, using an I4N inoculum and performing 15N2 fixation within the closed-atmosphere fermenter, leads to an I5N NMR spectrum which shows a smaller purine signal than that observed under uniform labeling conditions (Fig. 1, left and right).
We interpret the above results as evidence of the need for enhanced ribosomal RNA synthesis by K. pneumoniae during derepression of nitrogenase. With limited fixed nitrogen available from the medium, this synthesis depends on relatively slow catabolism of existing proteins. Not only must the cells synthesize ribosomal RNA for the nitrogenase needed to fix nitrogen and restore depleted nitrogen pools, they must also increase ribosome production to begin rapid growth (19,20). By the time nitrogen fixation begins to provide an appreciable source of nitrogen, a smaller percentage of newly fixed nitrogen is needed for additional purine synthesis than at earlier times in the adaptive period because existing proteins have been utilized for this purpose.
Nitrogen from Serine and Glycine-The "N NMR spectrum of K. pneumoniae derepressed for 3 h in the presence of ~-[ 2 -13C,'5N]serine (Fig. 2, bottom left) appears much like that from the nonspecific label (Fig. 1, left). This shows that most serine nitrogen is scrambled and used as a general source of nitrogen. However, the appearance of two DCPMAS I5N difference peaks (Fig. 2, top left) indicates that some serine is used for products which retain the original labeled "'C-lSN chemical bond of serine. (Details of the DCP experiment and its interpretation may be found in the Miniprint Supplement.) Based on a previous I5N NMR study of glycine metabolism (21), we assign the 220-ppm DCP signal to the N(7)-C(5) bond in purines, and the 90-ppm signal to a(C)-amide-N bonds of glycyl and seryl residues within proteins.
When both [2-13C,'5N]glycine and natural-abundance Lserine are present during derepression, the resulting 15N NMR spectrum of K. pneumoniae (Fig. 2, bottom right) shows less incorporation of label into cells than observed previously for labeled serine. Some 15N label appears in lysyl residues (12 ppm) and in amide groups of proteins (100 ppm). Much of the glycine transported into the cell is directly incorporated into either proteins or purines, giving rise to strong DCP signals at 220 and 90 ppm (Fig. 2, top right).
We interpret the results of the previous three paragraphs in terms of standard serine and glycine metabolic pathways (Miniprint Supplement, Fig. S6). Serine can either enter protein directly, or it can be used to produce purines after conversion to glycine. The latter path labels the C(5)-N(7) purine ring position. The observed DCP signals show that both paths are active in K. pneumoniae. Nitrogen labels from serine can also enter proteins and purines nonspecifically by conversion of serine to pyruvate, catalyzed by serine dehydratase, releasing labeled ammonium. The weak DCP signals in the spectrum of K. pneurnoniae grown on double-labeled serine, and the average peptide chemical shift of about 100 ppm, confirm that while some serine is converted to glycine and used to make purines, most is catabolized for its nitrogen.
We claim that the combination of metabolic pathways available to serine is the explanation for serine's effectiveness in shortening derepression time. Serine is a ready source of all the nitrogen for ribosomal RNA synthesis which, from the results of the previous section, we know is taking place, and which is known generally to be a necessary requirement for increased protein synthesis and rapid growth (19,20). Purine N-7 is derived from the glycine pool, while purine N-1, N-3, and N-9 are derived from the general nitrogen pool. Serine supplies nitrogen to both pools. We believe that glutamate and aspartate are not quite as effective as serine in shortening the derepression time (Miniprint Supplement, Table I) because complete purine synthesis still depends on a ready source of glycine. Glycine itself does not shorten the derepression time because glycine is not an adequate source of general nitrogen.
Carbons from Serine: Difference Spectra-The CPMAS 13C NMR spectrum of K. pneumoniae derepressed for nitrogenase in the presence of ~-[2"~C,"N]serine is shown in Fig. 3  (bottom left). The spectrum arises from natural-abundance 13C of the cells, as well as 13C due to uptake and metabolism of the labeled serine. The difference spectrum, generated by subtracting the natural-abundance background from the labeled-serine spectrum, is considerably simpler (Fig. 3, top  left). This indicates specific incorporation of label. Line assignments can be made using known chemical shifts (21). Label from the C-2 carbon of serine ends up primarily in three sites. Labeled serine is incorporated directly into proteins (62 ppm), or is converted first to glycine, with label then appearing in glycyl residues of proteins (45 ppm) and in purines (119 ppm) synthesized from glycine (see Miniprint Supplement, Fig. S6). A minor negative-going peak appears in the difference spectrum near 80 ppm; we attribute this peak to a slightly different level of carbohydrates in the natural-abundance control sample compared to the labeled analytical sample.
The corresponding difference spectrum of K. pneumoniae grown on ~-[3-'~C]serine shows a totally different specific labeling pattern (Fig. 3, top right). The labeling patterns for the above two labeled-serine experiments are consistent with seryl and glycyl fragment insertions into proteins and purines. Conversion of endogenous serine to glycine involves transfer of the C-3 carbon of serine into the tetrahydrofolate cofactor system, where it is routed into a number of metabolites (

ppm
peaks cannot be compared directly to those resulting from the serine to glycine pathway since general metabolism of [2-13C] pyruvate can be expected to lead to loss of label to COB The use of the carbons of serine for both specific and general synthesis is fully consistent with the analysis of the nitrogen metabolism of serine discussed in the previous section.
Carbons from Serine: Double Cross-polarization Spectra-The DCPMAS 13C NMR spectra of K. pneumoniae grown on ~-[Z-'~C,'~N]serine (Fig. 4) support our analysis of the difference spectra. In the DCP experiment only those carbons directly bonded to 15N's generate a signal. The three I3C DCP signals are assigned to labeled seryl and glycyl residues of proteins and to purines synthesized from labeled glycine. The observed DCP signals are therefore consistent with intact incorporation of labeled bonds into purines and proteins.
In addition, a minor DCP peak is observed in the carbonyl region (Fig. 4, top left). However, this particular signal is not due to incorporation of an intact labeled chemical bond, but rather to scrambled carbon and nitrogen levels forming a labeled bond within newly synthesized proteins of the cell. Since the general nitrogen pool is small during derepression, it becomes virtually 100% 15N enriched as a result of the active nitrogen scrambling of serine, established earlier. Thus, single-carbon I3C incorporations can produce I3C-l5N bonds. We confirm this interpretation of the carbonyl-carbon DCP peak by observation of a marked diminution in the peak's intensity in the spectrum of K. pneumoniae grown in the presence of double-labeled serine plus natural-abundance ammonium (Fig. 4, top center). The remaining three DCP peaks retain most of their intensity under these conditions, consistent with labeled bond incorporations.
The DCPMAS 13C NMR spectra of K. pneumoniae derepressed for 3 h in the presence of ~-[2-'~C,'~N]serine plus natural-abundance glycine (Fig. 4, top right) characterizes the changes that can occur in carbon routing by an alteration in the levels of intermediates. For instance, in the presence of exogenous glycine, cells transport enough glycine to activate a glycine-cleavage reaction pathway (22,23). Double-labeled serine is used sparingly as a precursor for glycyl residue insertions into proteins (Fig. 4, top right, 45 ppm). Instead, the glycine derived from serine is primarily directed into the tetrahydrofolate cofactor system pathway, resulting in label  15N]serine plus natural-abundance glycine has significant intensity only at 120, 60, and 45 ppm. Only three peaks are observed because at this stage of growth I4N2 fixation has sufficiently diluted the general 15N pool to a degree that singlecarbon incorporations no longer generate 13Cand "N-labeled bonds. (A discussion of the fate of double-labeled glycine under these conditions is presented in the Miniprint Supplement.) Even though K. pneumoniae was grown in media containing 2% sucrose, the carbons of labeled serine are heavily used, producing specific enrichments of 25-50% for some purine ring carbons. We attribute the heavy use of labeled carbon to pressure for the nitrogen of serine for purine synthesis during derepression. The fact that the C-3 carbon of serine can be readily used in purine synthesis probably contributes to the effectiveness of serine in shortening the time for derepression.