Muramic Acid Is Not Generally Present in the Human Spleen as Determined by Gas Chromatography-Tandem Mass Spectrometry

Abstract

It has been hypothesized that bacterial debris may accumulate in tissues of the reticuloendothelial system (RES) serving as an inflammatory stimulus for human disease. In support of this hypothesis, muramic acid (Mur), a component of bacterial peptidoglycan (PG), has previously been reported to be present in culture-negative human spleen. High-performance liquid chromatography (HPLC) was employed in these analyses, and a peak was detected at the retention time of Mur. However, HPLC is best used as a screening technique, and it is vital that these tentative observations be reexamined by the state-of-the-art approach (gas chromatography-tandem mass spectrometry [GC-MS²]). Indeed, in the present work using GC-MS², Mur was not detected in six out of seven human spleens previously examined by HPLC. However, Mur was categorically detected at minute concentrations, 50 ppb, in one spleen. In conclusion, since Mur is not generally found in culture-negative human spleen, in future studies, these tissues can serve as negative controls. The study of Mur levels in inflammation (e.g., reactive arthritis) could prove important in testing the hypothesis that bacterial debris persisting in tissues could serve as a depot inciting diseases of unknown etiology.

Reactive arthritis is defined as a nonpurulent joint inflammation developing after infection elsewhere in the body (10). The inflammation occurs despite organisms rarely being cultured from the joint (10, 18). Forms of reactive arthritis where bacteria have been shown to play a role include Lyme disease (37), rheumatic fever (41), and Reiters sydrome (1, 6, 39).

The etiology of rheumatoid arthritis is not clearly understood. It is widely accepted that the pathogenesis of rheumatoid arthritis includes the presentation of an unknown arthritogenic antigen to helper T cells (40). Activated CD4⁺ T cells are thought to stimulate macrophages to produce pro-inflammatory cytokines (e.g., tumor necrosis factor alpha and interleukin-1) that play a major part in stimulating the inflammatory response (4). Reactive arthritis and rheumatoid arthritis may share a common etiology, and it might be premature to rule out a role for bacteria in the latter.

In experimental models of arthritis, sterile bacterial cell wall fragments injected intraperitoneally into rats cause chronic inflammation in joints where small amounts of bacterial debris accumulate (8, 12, 17, 26, 33, 34, 35, 38). The majority of this material, however, localizes in the mononuclear, phagocytic system and is only slowly eliminated over time (12, 17, 35, 36). Mammalian enzymes, including lysozyme (16) and muramyl-l-alanine amidase, have limited activity on many gram-positive bacterial cell walls (25). However, it has been recently demonstrated using gas chromatography-tandem mass spectrometry (GC-MS²) that Mur is not present in normal rat spleen (23).

Mur (3-O-lactyl-glucosamine) is an amino sugar present in PG backbones of gram-positive and gram-negative bacteria. It is not synthesized by mammalian enzyme systems and therefore serves as a chemical marker to indicate the presence of viable bacteria as well as their nonviable cell wall remnants. These remnants have been shown to play a direct role in the disease process by activating cytokines and promoting acute inflammation (5, 19).

It can be hypothesized that, in patients with an underlying disease process (e.g., colon cancer or Chrohn's disease), inflammation or trauma could allow the translocation of whole or digested bacteria across the gut epithelial barrier (3, 28). Bacterial remnants may accumulate in the tissues of the reticuloendothelial system (RES). These depots of bacterial debris might serve as reservoirs for a persistent antigenic stimulus inducing and perpetuating diseases of unknown etiology.

Mur is readily demonstrated in body fluids in documented infection, including septic arthritis (7, 14, 27), bacterial meningitis (23), and urinary tract infections (2). It has also been reported that Mur is detectable in reactive arthritis. However, in the majority of cases, the levels are so low that detection proves elusive (27). These studies demonstrate the utility of nontraditional, gas chromatography-mass spectrometry (GC-MS) and more advanced GC-MS² techniques in demonstrating the presence of bacteria in human body fluids.

As noted above, Mur is not present in normal rat spleen. However, other studies have reported that Mur is found in human spleen (20, 31). Although these human tissues were culture negative, they were from patients with gastric carcinoma or splenomegaly. In some cases of gastric carcinoma, at certain stages of the disease process, there may be an influx of bacteria or bacterial debris from the gastrointestinal flora. Splenomegaly might result from accumulation of bacteria or bacterial remnants from translocation across the gut epithelium. These studies employed high-performance liquid chromatography (HPLC) with fluorescence detection, which is extremely sensitive but lacks specificity and is best used as a screening technique. Observing a peak at the correct retention time is encouraging but not a definitive identification since one may merely be detecting a coeluting contaminant.

The purpose of the present study was to attempt to confirm these results using a methodology that provides unequivocal identification. Until the present study, GC-MS² has remained the state of the art. However, in the present report, it proved necessary to employ the next generation technique, GC-MS³, to even further lower detection limits.

MATERIALS AND METHODS

Rat spleens.

Female Sprague-Dawley rats weighing ≈150 to 200 g were sacrificed, and the spleens were removed and weighed. Approximately 10 ml of sterile H₂O was added to each organ for homogenization. Homogenized organs were lyophilized and stored at −70°C until analysis.

Human spleens.

Portions of seven unfixed human spleens were obtained from the Pathology Department, University Hospital, Rotterdam, The Netherlands, immediately after surgery and were kept frozen (−20°C) until use. Spleens were removed for surgical and/or technical reasons from three patients with gastric carcinoma and from four patients because of splenomegaly due to hematologic diseases. All of the spleens were cultured under aerobic conditions on sheep blood agar. No bacterial growth was observed. The spleens analyzed by GC-MS² were the same ones analyzed previously by HPLC (31).

Sample preparation for GC-MS² and GC-MS³.

The alditol acetate derivatization procedure for Mur has been described elsewhere (13, 15). Work from this laboratory has demonstrated the ubiquitous presence of Mur in dust (11, 22, 24). Scrupulous attention was essential to eliminate environmental contamination of the samples. All samples were analyzed at least in duplicate. Samples were first hydrolyzed to release Mur from PG by the addition of 1 to 2 ml of 2 N sulfuric acid to 20 to 40 mg of lyophilized tissue (100 to 200 mg [wet weight]) for 3 h at 100°C. ¹³C-labeled Mur was prepared in advance by hydrolyzing 40 mg of ¹³C-labeled cyanobacteria (Isotec, Miamisburg, Ohio). The bacteria were 0.4% Mur on a dry weight basis. Thirty-four nanograms of ¹³C-labeled Mur was added to each sample as an internal standard. Additionally, 50 μg of glucose was added to each sample as a carrier. External standards consisted of a known amount of Mur and constant amounts (34 ng) of ¹³C-labeled Mur. Blanks consisted of water spiked with ¹³C-labeled Mur. Following hydrolysis, samples were neutralized by mixing with 2 to 4 ml of N,N-dioctylmethylamine (Fluka, Buchs, Switzerland) in chloroform (50:50 [vol/vol]) and centrifuged. The aqueous phase was removed and passed through C-18 columns (J&W Scientific, Folsom, Calif.) and reduced with 5 mg of sodium borohydride. To remove generated borate, methanol/acetic acid (200:1 [vol/vol]) was added continuously while evaporating under nitrogen. The samples were dried under vacuum. The alditols were acetylated at 100°C overnight. Acetic anhydride was decomposed with 0.75 ml of H₂O. One milliliter of chloroform was added, and after mixing, the aqueous phase was discarded. Then, 0.8 ml of ammonium hydroxide in H₂O (80:20 [vol/vol]) was added, the mixture was passed through a Chem Elut column (Varian, Walnut Creek, Calif.), and the organic phase was collected. The samples were evaporated to dryness and reconstituted in 25 to 30 μl of chloroform prior to analysis.

Instrumentation.

Samples were analyzed in the quantitation mode on a VG Quattro 1 triple quadrupole tandem mass spectrometer (Micromass, Boston, Mass.) coupled to a Fisons 8000 GC equipped with an automated sample injector (A200s) and a nonpolar, DB-5ms fused silica capillary column (J&W Scientific). Quantitation was based upon the peak area ratio of Mur to the internal standard (¹³C-labeled Mur) in the sample compared with the peak area ratio in the external standard mixture (containing a known amount of Mur and ¹³C-labeled Mur). Electron impact (EI) ionization was performed with a standard five-coil filament with an electron energy of 35 eV and emission current at 200 μA. Source temperature was maintained at 225°C. Collision-induced disassociation (CID) of precursor ion mass of 402.98 for Mur and a mass of 411.98 for ¹³C-labeled Mur was performed at a collision energy of 10 V. Mur and ¹³C-labeled Mur were detected using the mass transitions 402.98→197.84 and 411.98→204.84, respectively, for quantitation.

In the identification mode (GC-MS² and GC-MS³), samples were analyzed on a GCQ ion trap mass spectrometer (ThermoFinnigan, San Jose, Calif.) also equipped with an A200s autosampler and a DB-5ms ITD capillary column (J&W Scientific). Using MS², identification was based on the similarity of the product ion spectrum (fingerprint) of a peak at the retention time of Mur in samples in comparison to the spectrum of a Mur standard. The parent molecule had a molecular weight (MW) of 403, and the same major fragments were observed: masses of 361, 301, 258, 240, 198, 156, and 138. The identity of these molecules has been described previously (23). In MS³, characteristic fragment masses of 156 and 138 (parent transition 403.1→198.1) were observed.

The injection port of the gas chromatograph was maintained at 250°C while the source temperature was maintained at 225°C. Initial oven temperature was 160°C for 1 min with the split valve closed and then a ramp of 20°C/min to 270°C and a hold for 1 min. Surge pressure was 75 lb/in² for 1 min. Helium was used as a carrier gas with a constant flow of 40 cm/s maintained with electronic pressure control on. EI ionization was performed with an electron energy of 70 eV and emission current at 250 μA by using the standard autotune settings. For GC-MS², the precursor ion with an m/z of 403 with a notch width of 1.0 mass unit was isolated for 16 ms. Collision energy was set at 1.10 V for 15 ms with a q of 0.225. In GC-MS³, parameters were identical for a precursor with an m/z ratio of 403; however, subsequent isolation of an m/z ratio of 198.1 with a 1.0 mass unit notch width for 16 ms and a collision energy of 0.6 V for 15 ms with a q of 0.225 was used.

RESULTS

The primary focus of this study was to determine whether Mur is present in culture-negative human spleen. This laboratory previously demonstrated categorically that Mur is not present in normal rat spleen by using GC-MS² (23). There have been no previous studies using GC-MS³ from this or any other laboratory. Thus, rat spleens were included as negative controls in all GC-MS² and GC-MS³ experiments.

Analysis for Mur in rat spleens using GC-MS² (quantitation mode).

Normal rat spleens were analyzed by GC-MS² as negative controls. A detailed explanation for interpreting GC-MS² chromatograms (quantitation mode) and mass spectra (identification mode) was provided here (23). In brief, natural [¹²C]Mur from bacterial cell walls is obtained by acid hydrolysis and converted to muramicitol lactam penta acetate (MW, 445). In the first stage of the instrumental analysis, the molecule was isolated essentially intact, with an MW of 403 (M-ketene). In the second stage, only molecules with an MW of 198 were isolated and detected (loss of MW of 145 from a break between C-4 and C-5 and a further loss of acetic acid [loss of 60]). The ¹³C-labeled isotope of Mur (internal standard) is detected by the corresponding 412→205 transition.

In Fig. 1, the top panel in each chromatogram shows the presence of 34 ng of ¹³C-labeled Mur (internal standard). The characteristic twin peaks of the alditol acetate of Mur are clearly evident. The bottom panel shows the absence of [¹²C]Mur. The major peak has been identified as muramicitol lactam pentaacetate (12). The minor peak has not been identified. Figure 1A and B depict normal rat spleens and show no discernible peak at the retention time of the ¹³C-labeled Mur (internal standard). Figure 1C shows the analyses of 100 mg of normal rat spleens spiked with 10 ng of Mur (positive control). Muramic acid is clearly present, illustrating the sensitivity and specificity of the technique. Both the major and the minor peaks for Mur are clearly observed.

FIG. 1.

GC-MS² analysis (quantitation mode) of normal rat spleen (100 mg [wet weight]). Shown are representative chromatograms of two normal rat spleens (A and B) and rat spleen spiked with 10 ng of Mur (C). In each chromatogram, the upper window shows results obtained using 34 ng of ¹³C labeled Mur as an internal standard. The lower window depicts results for normal [¹²C]Mur. There was no discernible peak at the retention time of the internal standard [¹³C]Mur with the normal rat spleen (A and B), but a peak was observed for rat spleen spiked with Mur (C).

Analysis for Mur in human spleen using GC-MS² (quantitation mode).

Spleens removed from seven patients with underlying diseases (gastric carcinoma and hematologic diseases) were analyzed in duplicate. Six of the seven spleens showed no discernible peak for Mur in the quantification mode. In Fig. 2B, the top panels shows the absence of major or minor peak at the correct retention time for Mur. The top panel shows the characteristic twin peaks of the ¹³C-labeled internal standard. In one case, however, Mur was detected confirming the HPLC results (31). Figure 2A depicts the detection of 5 ng of Mur in 100 mg (wet weight) of human spleen. Previous HPLC analysis of these same human spleens suggested the presence of 400 to 600 ng of Mur in this mass of tissue.

FIG. 2.

GC-MS² analysis (quantitation mode) of diseased, culture-negative human spleen (100 mg [wet weight]) Shown are the detection of 5 ng of Mur in a human spleen (A) and a human spleen with no detectable Mur (B). In each chromatogram, the upper window depicts the addition of 34 ng of ¹³C-labeled Mur as an internal standard. The bottom window depicts results for natural [¹²C]Mur.

Analysis for Mur in human spleen by using GC-MS² and GC-MS³ (identification mode).

In GC-MS² identification mode (Fig. 3), the major fragments observed during the second step of the instrumental analysis from mass 403 were 361, 301, 258, 240, 198, 156, and 138 (Fig. 3A). These characteristic fragment masses provide a chemical fingerprint for Mur. Ten nanograms of Mur spiked in human spleen was readily identified (Fig. 3B). The characteristic masses of 361, 301, 258, 240, 198, 156, and 138 are clearly visible. As noted above, Mur was detected in one human spleen without spiking at 5 ng/100 mg (wet weight) by GC-MS² (quantitation mode). However, the presence of Mur could not be confirmed in GC-MS² identification mode. Although some of the characteristic masses (258, 240, and 198) were observed, background present in the spectrum masked detection (Fig. 3C). A fingerprint of a typical human spleen is shown in Fig. 3D; note that none of the characteristic masses were observed.

FIG. 3.

GC-MS² analysis (identification mode) of diseased human spleen (100 mg [wet weight]). Shown are an authentic Mur standard (A), a human spleen spiked with 10 ng of Mur (B), unspiked human spleen with 5 ng Mur detected (C), and unspiked human spleen with no detected Mur (D). Compare the characteristic masses observed, 138, 156, 198, 240, 258, 301, 361, and 385, in panels A and B with those in panels C and D. Some of the characteristic fragments, masses 198, 240, and 258, are present in panel C; however, none of the characteristic fragment masses are present in panel D.

To further reduce background, GC-MS³ was employed. MS³ employs a three-step instrumental analysis. The first two steps of the analysis were the same as in MS². In the first stage, only molecules with a mass of 403 were isolated. In the second stage, these molecules were fragmented and only molecules with a mass of 198 were isolated. In the third stage, molecules of mass 198 were isolated and were further fragmented. In GC-MS³ analysis, the characteristic fragment masses from 198 were MW 156 (minus 42 due to loss of ketene) and MW 138 (minus 60 due to loss of acetic acid).

Figure 4A and B show the characteristic fingerprints of a Mur standard and a spiked human spleen in the GC-MS³ identification mode and are essentially identical. Figure 4C shows the fingerprint of the unspiked human spleen. The characteristic fragments of MW 156 and 138 are unmistakable and provide categorical identification of Mur in a culture-negative human spleen. Figure 4D depicts the GC-MS³ fingerprint of a human spleen that did not contain Mur. Neither characteristic fragment masses were observed. Normal rat spleens also did not contain Mur, as shown by GC-MS³ analysis, and were identical to the other six culture-negative human spleens in lacking Mur.

FIG. 4.

GC-MS³ analysis (identification mode) of diseased human spleen (100 mg [wet weight]). Shown are an authentic Mur standard (A), a human spleen spiked with 10 ng of Mur (B), human spleen with 5 ng of Mur detected (C), and human spleen with no detectable Mur (D). Note that the characteristic masses of 138 and 156 are present with no background in panels A, B, and C.

DISCUSSION

Bacterial debris preferentially localizes in the spleen in experimental rat models of arthritis (9, 12, 17). By extrapolation, it is reasonable to assume that in normal, healthy animals, bacterial debris processed by the host immune system might eventually find its way into tissues of the RES. However, in previous studies, it was demonstrated that Mur is not detectable in normal rat spleen (23, 35). It is conceivable that in an underlying disease process, e.g., gastric carcinoma, large amounts of bacterial debris might pass through the gut into the circulation and localize in tissues (e.g., the spleen).

In support of this, it was previously reported that Mur was present in seven spleens of patients with gastric carcinoma or hepatosplenomegaly due to hematologic disease. The spleens were analyzed using HPLC and Mur was detected fluorospectrometrically as a dansyl derivative, which detects amino sugars, amino acids, and other compounds containing amino groups. The levels of Mur reported in the spleens were 400 to 600 ng/100 mg of spleen (wet weight) (31). In the present study, GC-MS² in the quantitation mode suggested the presence of Mur in only one of these spleens (5 ng Mur/100 mg [wet weight]). This level is 2 orders of magnitude lower than previously reported. It was not possible to use GC-MS² in the identification mode to provide confirmation due to excessive background peaks at these minute concentrations (ppb). The next generation technique of GC-MS³ was required for categorical identification. It is probable that the levels of Mur in human spleen that were previously reported were spuriously elevated due to coeluting contaminants.

It is possible that Mur could have been degraded over time so that it was present at the time of the HPLC analyses and absent at the time of the GC-MS² analyses. However, samples were stored frozen under sterile conditions and there have been no mammalian enzymes described that degrade Mur. Furthermore, we have performed numerous Mur analyses on mammalian tissues stored for long time periods (e.g., see references 10 and 17). Disappearance of Mur on storage has never been found. It is also conceivable that Mur could have been unevenly distributed in spleens so that parts analyzed by HPLC contained Mur and parts analyzed by GC-MS² did not contain Mur. However, all spleens were analyzed by GC-MS² in duplicate and often in quadruplicate. The replicates yielded the same results.

Another possible explanation for the discrepancy between the two studies is that the HPLC method used was more sensitive than GC-MS². Thus, positive results by HPLC were below the limit of detection of GC-MS². For a pure standard, it is probable that HPLC is comparable in sensitivity to GC-MS². However, for complex biological samples, background from other components of the sample matrix determines the detection limit. MS² as a chromatographic detector dramatically lowers detection limits after GC separation. Fluorescence detection after HPLC detects any compound that contains a derivatized amino group (e.g., amino acids and sugars). There is a great deal of background with HPLC. However, it is worthy of note that the levels reported by others were 400 ng of Mur/100 mg (wet weight) of spleen. This is 2 orders of magnitude higher than our limit of detection (1 to 2 ng/100 mg). Thus, the lack of sensitivity of GC-MS² cannot explain the differences between the two studies. It would have been of interest to know whether the one GC-MS²-positive spleen was the one that gave the highest results for Mur in the HPLC method. Unfortunately, HPLC results in these studies were qualitative, while GC-MS² results were quantitative.

A solid-phase immunoassay has been used to quantitate the levels of bacterial antigens in rats developing arthritis after a single injection of streptococcal cell walls. Using an antibody recognizing the d-alanine-d-alanine (d-Ala-d-Ala) determinant, PG was readily detected in all experimental tissues surveyed, including joints, with the largest amounts in the liver and spleen (9). However, tissues from control rats (including joints and spleen) gave background values. These results are in agreement with results obtained by mass spectrometry analysis, demonstrating that the Mur portion of PG is not present in normal mammalian tissues as presented here.

More recently, an antibody recognizing the glycan backbone of PG (monoclonal antibody 2E9) was found to react with culture-negative human spleen (31). This antibody also reacts with synovial cells from patients with rheumatoid arthritis, osteoarthritis, crystal arthritis, and joint trauma (32, 42). Bacteria are not involved in the etiology of osteoarthritis, crystal arthritis, or joint trauma. It is possible that a PG moiety other than d-Ala-d-Ala or Mur is present in normal mammalian tissues (e.g., from prior bacterial infection). Alternatively, monoclonal antibody 2E9 staining of lymphoid cells may result from cross-reactivity between human glycoproteins or proteoglycans with the glycan backbone of PG. All of these antigens contain large amounts of glucosamine. Animals experimentally injected with cell walls from group A variant streptococci produce antibodies not only against the d-Ala-d-Ala determinant of PG but also against the polysaccharide backbone with N-acetylglucosamine as the immunodominant sugar (21, 29, 30). Alternatively, when PG from Micrococcus lysodeikticus is used as the immunizing agent, antibodies directed against N-acetylmuramic acid are obtained. This is consistent with the structure of the M. lysodeikticus PG since only 40% of the N-acetylmuramic acid moieties are substituted by peptide subunits (43).

The chronic inflammation in rheumatoid arthritis is in response to an undefined antigen. One possibility is the persistence of nonviable bacterial debris (including PG) stimulating inflammation in the rheumatoid joint. However, it is also possible that bacterial infections might lead to the induction of autoimmunity (e.g., from an immune response to the glycan backbone of PG).

In conclusion, Mur is not generally present in culture-negative human spleen or normal rat spleen. However, Mur was categorically detected in one of seven human spleens. Even in this spleen the levels of Mur were considerably lower than first reported (31). For the first time, GC-MS³ has been used to confirm the presence of Mur (or any microbial cell wall constituent) in a mammalian tissue. The fact that one of the spleens contained trace levels of Mur encourages further work to test the hypothesis that bacterial debris persisting in antigen-presenting cells could serve as a depot stimulating inflammatory diseases of unknown etiology.