Protracted Anemia Associated with Chronic, Relapsing Systemic Inflammation Induced by Arthropathic Peptidoglycan-Polysaccharide Polymers in Rats R

Home , Peptidoglycan

INFECTION AND IMMUNITY, Apr. 1989, p. 1177-1185 Vol. 57, No. 4 0019-9567/89/041177-09$02.00/0 Copyright © 1989, American Society for Microbiology Protracted Anemia Associated with Chronic, Relapsing Systemic Inflammation Induced by Arthropathic Peptidoglycan-Polysaccharide Polymers in Rats R. BALFOUR SARTOR,l.2* SONIA K. ANDERLE,3 NADER RIFAI,4 DAVID A. T. GOO,4 WILLIAM J. CROMARTIE,3 AND JOHN H. SCHWAB2'3 Department of Medicine,1 Department of Microbiology and Immunology,3 Department ofPathology,4 and Core Center for Diarrheal Disease,2 University of North Carolina, Chapel Hill, North Carolina 27599-7080 Received 3 October 1988/Accepted 6 January 1989

Mild hypoproliferative anemia with abnormal iron metabolism frequently accompanies chronic inflammation and infection in humans. To determine whether anemia is associated with chronic relapsing arthritis induced by bacterial cell wall polymers, serial determinations of the hematocrit were measured in rats injected intraperitoneally with sonicated peptidoglycan-polysaccharide fragments from group A streptococci. Acute anemia peaked 5 to 10 days after injection, and chronic, spontaneously relapsing anemia persisted for 309 days. 51Cr labeling demonstrated decreased erythrocyte survival, i.e., a half-life of 8.4 days in cell wall-injected rats versus 11.8 days in controls. Erythrocytes were mildly microcytic, and leukocyte counts were elevated during early spontaneous reactivation of arthritis, 15 days after injection of peptidoglycan-polysaccharide. Bone marrow myeloid/erythroid precursor ratios were elevated in arthritic rats (P < 0.0001). Purified peptidoglycan produced an acute anemia lasting 10 days, while injection of group A streptococcal polysaccharide and mutanolysin-digested cell wall did not affect the hematocrit. The minimal effective dose of peptidoglycan- polysaccharide was 5 ,ug of rhamnose per g (body weight). Serum iron and transferrin levels were decreased in cell wall-injected rats (P < 0.005) and were closely correlated with hematocrit values and joint inflammatory scores. Stainable iron was increased in the liver, spleen, and mesenteric lymph nodes and unchanged in the bone marrow of cell wall-injected rats. Anemia accompanying chronic, relapsing systemic inflammation induced by peptidoglycan-polysaccharide polymers appears to be an excellent animal model of the anemia of chronic disease.

Mild nonprogressive anemia accompanies chronic inflam- group A streptococci induce an acute and chronic systemic mation of diverse etiologies, including chronic bacterial and inflammatory response characterized by hepatic granulomas fungal infections (42), rheumatoid arthritis (43), Crohn's (20, 47) and spontaneously relapsing arthritis (8) that can disease (10), and widespread malignancies (13). The anemia persist at least 1 year after a single intraperitoneal (i.p.) of chronic disease (inflammation) is characterized by mildly injection into susceptible rat strains. Macrophages are acti- decreased circulating erythrocyte (RBC) survival, lack of an vated to secrete increased amounts of interleukin-1 (IL-1) by appropriate bone marrow compensatory response, and dis- both in vitro and in vivo exposure to PG-APS (L. A. Bristol ordered iron kinetics (13, 24, 48). Typically, the serum iron, and J. H. Schwab, submitted for publication). total iron-binding capacity (TIBC), and marrow sideroblasts This study is designed to determine whether a single i.p. are decreased in spite of normal-to-increased bone marrow injection of PG-APS would produce protracted anemia and and tissue iron stores. Recent studies have suggested that to determine whether this well-characterized and reproduc- monokines secreted by activated macrophages may be re- ible chronic, systemic inflammatory response in genetically sponsible for suppression of bone marrow erythroid precur- susceptible rats could serve as a reliable model of the anemia sors (33, 37). Anemia accompanies several experimental of chronic disease. We demonstrate that PG-APS produces a inflammatory states associated with macrophage activation, chronic, spontaneously relapsing anemia that persists at including adjuvant arthritis (25), arthritis induced by viable least 10 months after a single i.p. injection and whose course BCG (26) and Mycoplasma arthritidis (11), and turpentine closely parallels activity of peripheral arthritis. PG is the abscesses (32). These experimental models have been pro- moiety of PG-APS responsible for the anemia and arthritis. posed as animal models of the anemia of chronic inflammation, however, none of the models produce protracted ane- PG-APS-induced anemia resembles human anemia of mia nor exactly reproduce the bone marrow suppression and chronic disease in the following respects: it is mild to disordered iron metabolism found in the anemia of chronic moderate, RBC survival is decreased, RBCs are slightly disease (48). microcytic, serum iron and transferrin concentrations are Bacterial cell wall polymers induce well-characterized decreased, and reticuloendothelial iron stores are normal or acute and chronic granulomatous inflammation in a variety increased. of organs after local and systemic injection (35, 41). Pepti- (A portion of this work was presented as a poster at the doglycan-polysaccharide (PG-APS) polymers derived from 87th Annual Meeting of the American Society for Microbi- ology in Atlanta, Ga., 1 to 6 March 1987, and was published as an abstract [R. B. Sartor, S. K. Anderle, W. J. Cromar- tie, and J. H. Schwab, Abstr. Annu. Meet. Am. Soc. Micro- * Corresponding author. biol. 1987, B-146, p. 49].) 1177 1178 SARTOR ET AL. INFECT. IMMUN.

MATERIALS AND METHODS injection of PG-APS or PBS. Serial determinations of hematocrit (HCT), 51Cr counts per minute, or sedimentation rates Animals. Female inbred Lewis rats were purchased from were obtained at intervals outlined in individual experiments Charles River Breeding Laboratories, Inc., Wilmington, from blood derived from transection of the tip of the tail. A Mass., and were housed 2 or 3 per cage with free access to minimal volume of blood was obtained to prevent iron Purina rat chow and water at all times. Rats weighed 140 to deficiency. Bleeding was controlled by digital pressure on 160 g at the time of injection. the tail above the transection and by cauterization immedi- Cell wall polymer preparation. PG-APS polymers were ately after blood collection. Blood was collected directly isolated under aseptic conditions from the cell walls of strain into D58 of group A streptococci (Streptococcus pyogenes) as two heparinized 75-mm-long microhematocrit capillary previously described in detail (40). Briefly, streptococci tubes (approximately 70 [LI of operational volume; Fisher grown in Todd-Hewitt broth (BBL Microbiology Systems, Scientific, Pittsburgh, Pa.) for duplicate HCT and sedimen- Cockeysville, Md.) were disrupted in a Braun MSK cell tation rate determinations and into heparinized microcentri- homogenizer (Bronwell Scientific, Inc., Rochester, N.Y.), fuge tubes (approximately 200-pL volume) for 51Cr or auto- and the cell wall fraction was separated by filtration and mated cell counting. Serum for iron and transferrin levels centrifugation. The cell walls were extracted three times in was obtained at necropsy by cardiac puncture. All injections 2% sodium dodecyl sulfate (Sigma Chemical Co., St. Louis, and blood collections were done under ether inhalation Mo.) at 56°C and then washed extensively and dialyzed to anesthesia to minimize discomfort, and animals were killed remove the sodium dodecyl sulfate. Purified PG-APS poly- by CO2 overdose. mers (400 mg) were suspended in 20 ml of sterile phosphate- Measurement of laboratory and clinical parameters. HCT buffered saline (PBS) (pH 7.2) and sonicated for 35 min using values were determined in duplicate microhematocrit tubes a 0.5-in. (1.27-cm) probe and a setting of 8 (model 350; immediately after centrifugation at 12,600 x g for 3 min. Branson Sonic Power Co., Danbury, Conn.). The suspen- Spontaneous sedimentation of RBCs was measured in mi- sion was centrifuged at 10,000 x g for 30 min to remove large crohematocrit capillary tubes allowed to stand upright for 30 particles. PG-APS fragments prepared by this method have min after collection at room temperature. The percent molecular weights ranging from 5 x 106 to 5 x 108 (16). plasma relative to total volume was read with a microhema- Sterility was confirmed by culturing 0.1 ml on sheep blood tocrit capillary tube reader (Lancer, Brunswick Co., St. agar. PG-APS concentrations for injections were calculated Louis, Mo.). The amount of 51Cr within 0.1 ml of heparin- on the basis of rhamnose values determined by the method ized blood was determined in a gamma counter (1197 series; of Dische and Shettles (9). Immediately before injection, Automatic Gamma Counting System; Searle Analytic, Inc., PG-APS was diluted to appropriate concentrations in PBS Des Plains, Ill.). Counts per minute at each time point were (pH 7.2) and sonicated for 3 min with a 9-kilocycle Sonic corrected for the spontaneous decay in 51Cr activity (phys- oscillator (Raytheon Co., Waltham, Mass.) to disperse ag- ical half-life, 27.7 days). An automated cell counter (150 gregates. series; Baker Instruments Corp., Allentown, Pa.), pro- Preparation of PG-APS fractions. Group-specific APS and grammed for rat blood cells, was used for hemoglobin, RBC PG were prepared by hot formamide treatment of cell walls. and leukocyte (WBC) counts, and mean corpuscular volume APS was purified from the supernatant by the method of determinations. Transferrin was measured immunoturbidi- Krause (22), as modified by Chetty et al. (4), and contained metrically (31) with the following modifications. Rat sera 47% rhamnose. PG was purified by multiple formamide were diluted 1:25 with PBS. Rabbit anti-rat transferrin extractions by the method of Krause and McCarty (23) and antiserum (Nordic Immunological Reagents, El Toro, Calif.) contained 0.5% rhamnose. was diluted 1:50 with 40 g of polyethylene glycol per liter of Mutanolysin-digested PG-APS was prepared by a modifi- PBS. Within-run imprecision for was cation of the method of Chetty et al. (5). The M-1 fraction of the transferrin assay mutanolysin, an endo-N-acetylmuramidase derived from less than 5%. The iron concentration was determined by Streptomyces globisporus, was obtained from Miles Labo- using Beckman's ferrozine method (Beckman Instruments, ratories, Inc., Elkhart, Ind. Samples (8 ml) of sonicated Inc., Brea, Calif.) by the instructions of the manufacturer. PG-APS (27.4 mg of rhamnose) were diluted with 12 ml of The analytical imprecision for the iron assay was less than PBS containing 0.5 M sodium citrate and 0.05 mM MgCl2 4%. The determinations of transferrin and iron were per- containing 3.29 mg of mutanolysin and then incubated at formed on the Cobas-BIO centrifugal analyzer (Roche Ana- 37°C with constant mixing for 11 h. The optical density lytical Instruments Inc., Nutley, N.J.). Iron stains were measured at 600 nm decreased from 0.492 to 0.135 after performed on femoral bone marrow touch preparations and incubation with mutanolysin. Mutanolysin-digested PG-APS Formalin-fixed liver and spleen sections by the method of has previously been shown to have a molecular weight of <5 Gomori (18). Bone marrow myeloid/erythroid precursor x 106 and to produce only transient edema without arthritis ratios were performed on touch preparations of femoral upon i.p. injection (5). marrows, which were stained with Wright's-Giemsa re- Labeling of erythrocytes with -5Cr. RBCs from Lewis rats agents. Bone marrow cellularity was determined in Forma- were labeled in vitro with 51Cr as previously described (34). lin-fixed decalcified femurs, sternums, and ankle joints 51Cr-labeled RBCs were diluted to 1.07 x 10' cpm/ml in PBS stained with hematoxylin and eosin. Arthritis was measured and injected within 30 min. by using a clinical joint score of 0 to 4+ for each ankle Experimental design. Lewis rats were injected i.p. with 1 (maximum score, 16) (8). We have previously demonstrated ml of PG-APS or cell wall fraction diluted with PBS (pH 7.2) a close correlation of clinical joint scores with radiographic to the appropriate concentration (rhamnose concentrations (6) and histologic (8) assessments of joint inflammation. cited in individual experiments). In each experiment, control Statistical evaluation. Mean values of each experimental rats were injected i.p. with 1 ml of PBS. In the study parameter were calculated for each group of rats and com- measuring RBC survival, 0.5 ml of 51Cr-labeled RBCs (5.35 pared by Student's unpaired t test. Experimental parameters x 105 cpm) was injected intravenously 30 min before i.p. were correlated by linear regression analysis. VOL. 57, 1989 ANEMIA INDUCED BY BACTERIAL CELL WALLS 1179

0 tO 40 .-I +1 7 C wcn 30 6 +1 C S 0 0- E 0 u 0 20 3 o 0 o - 2u E 1 2 C II) Io- 40 80 120 160 200 Days After Injection FIG. 1. Acute and chronic anemia and arthritis induced by PG-APS polymers derived from group A streptococci. Sterile, sonicated PG-APS (20 jig of rhamnose per g) or PBS was injected i.p. into 10 female Lewis rats per group. Serial determinations of HCT were measured from blood collected in duplicate microcapillary tubes after transection of the tail. Joint scores were measured clinically as specified in the text. The joint score of PBS-injected rats was 0 (scores not shown). SEM, Standard error of the mean; SD, standard deviation.

RESULTS Time course of anemia and arthritis. Serial HCTs were measured in female Lewis rats injected with PG-APS (20 ,ug of rhamnose per g) or PBS. PG-APS-injected rats developed a transient increase in HCT 17 h after injection, followed by a protracted anemia compared with the PBS-injected control group (Fig. 1). The time course of anemia followed a typical pattern in this and subsequent experiments. An acute drop in HCT occurred by 3 days after injection, with the lowest HCT values in the PG-APS versus 45.1% in the (36.2% group (I) control group [P < 0.0001]) recorded 10 days after injection. +1 Over the next week, the HCT slowly increased but never C 1000 reached control values. Between 3 and 5 weeks after injec- E tion, individual rats developed a spontaneous recurrence of persistent anemia. The peak of the chronic anemia at this 0 dose of PG-APS occurred 49 days after injection (HCT value 0 of 36.4% ± 2.8% versus 46.5% ± 1.1% in the control group, D \ \~~~~~~~~~~* [P < 0.0001]). HCT values in the PG-APS-injected rats were significantly lower than those of PBS-injected controls for 309 days after injection. Arthritis, as measured by the 0 clinical joint score, demonstrated a similar acute phase, C)0.100, which peaked 3 days after PG-APS injection, and a chronic, spontaneously relapsing phase beginning 4 weeks after injection (Fig. 1). Mean values obscure the occasional dra- matic change in the HCT and joint scores of individual rats during the spontaneous relapses and remissions of anemia and arthritis. The sporadic nature of the recurrence of anemia led to relatively high standard deviations of HCT values 3 to 4 weeks after injection (Fig. 1). All rats injected with PG-APS abruptly developed recurrent anemia (HCT of s40% and a decrease in HCT by >3 percentage points 10 20 30 40 50 60 70 within 1 week); 57% (4 of 7) had a reactivation of arthritis Days After Injection within 1 week of the sudden drop in HCT. Recurrent anemia FIG. 2. RBC survival following PG-APS or buffer injection. preceded or accompanied spontaneous reactivation of arthri- Shown are serial measurements of 51Cr activity in heparinized blood tis, which in the Lewis rat persists indefinitely as an anky- following intravenous injection of syngeneic RBCs (5.35 x i05 cpm) losed joint with massive bony proliferation (20). Control rats labeled in vitro 30 min before the i.p. injection of PG-APS (20 jig of injected with PBS developed neither anemia nor arthritis. rhamnose per g) or PBS into groups of 10 female Lewis rats (the RBC survival. Survival of syngeneic RBCs labeled in vitro same rats were used in Fig. 1). Results are corrected for the with 51Cr was compared in rats injected with PG-APS (20 ,ug spontaneous decay of 51Cr activity. SEM, Standard error of the of rhamnose per g [body weight]) or PBS. PG-APS-injected mean. rats had significantly lower counts per minute per 100 RI of blood than the controls did, beginning 7 days after injection 1180 SARTOR ET AL. INFECT. IMMUN.

TABLE 1. Ability of PG-APS constituents to induce anemia Material HCT" (%) (mean ± SEM) on day: injected 1 6 10 14 55 PG-APSb 45.6 ± 0.5* 40.5 ± 0.4** 41.6 ± 0.7* 42.3 ± 0.7* 44.5 ± 2.1 PG-APS digestc 43.7 ± 0.4 43.0 ± 0.6 44.2 ± 0.6 43.2 ± 0.7 46.5 ± 0.3 PGd 41.1 ± 0.6* 38.3 ± 0.9** 42.2 ± 0.8* 43.5 ± 0.8 46.9 ± 0.6 APSe 44.2 ± 0.6 42.6 ± 0.4 45.1 ± 0.6 44.8 ± 1.2 46.8 ± 0.2 PBSf 43.5 ± 0.4 44.0 ± 0.6 45.2 ± 0.7 44.7 ± 0.4 48.1 ± 0.2 a * for P < 0.01 compared with the PBS-injected group; **, P < 0.001 compared with the PBS-injected group. b PG-APS polymers from group A streptococci (30 ,ug of rhamnose; approximately 90 ,ug [dry weight] of PG-APS per g [body weight]) injected i.p. into six female Lewis rats. c PG-APS polymers from group A streptococci (30 p.g of rhamnose per g [body weight]) predigested for 12 h with mutanolysin and then injected i.p. into six rats. d Purified PG (45 p.g [dry weight]/g [body weight]) injected i.p. into six rats. e Formamide-extracted group A streptococcal APS (30 p.g of rhamnose per g [body weight]) injected i.p. into six rats. f PBS injected i.p. into six rats (buffer control).

and extending until 51Cr was no longer detectable (Fig. 2). 0.5 or 1 p.g of rhamnose per g (body weight) did not develop The calculated half-life of blood cells, based on four half- clinically significant anemia at any time, although sporadic lives, was 8.4 days in the PG-APS-injected rats compared time points showed a small but statistically significant differ- with 11.8 days in controls. There was no evidence of ence in HCT compared with control values. PG-APS in gastrointestinal blood loss by Hemoccult (Smith-Kline Diag- doses between 10 and 40 p.g of rhamnose per g (body weight) nostics, Sunnyvale, Calif.) testing of fecal samples and no induced acute anemia peaking at 4 to 7 days after injection in evidence of bleeding from the cauterized tails. These results a dose-dependent fashion. Rats injected with 5 to 40 ,ug of demonstrate a moderate decrease in RBC survival in PG- rhamnose developed chronic, spontaneously relapsing ane- APS-treated rats which was not due to bleeding. mia with very interesting differences in the timing of the Active moiety of PG-APS. To determine which portion(s) recurrences, depending on the dose injected (Table 3). Rats of the PG-APS polymer was responsible for protracted receiving 5 ,ug/g (body weight) i.p. had no acute change in anemia, we measured HCTs of rats injected with equivalent HCT but developed chronic anemia 35 to 113 days after high doses of purified PG and APS as well as enzymatically injection. Only one rat in this group had spontaneous reac- degraded PG-APS (Table 1). As described above, rats in- tivation of anemia. Rats receiving 10 to 20 I.g of rhamnose jected with nondigested PG-APS polymers developed acute per g (body weight) developed recurrent, chronic anemia and chronic anemia and arthritis after a transient increase in which, in most animals, never returned to normal levels. HCT 1 day after injection. Rats injected with PG had These rats developed a recurrence of moderate (HCT of arthritis and significantly decreased HCTs from 1 to 10 days <40%) anemia 18 to 28 days after PG-APS injection; the after injection, with peak anemia at 6 days, but no evidence second and third recurrences of anemia occurred 50 to 113 of chronic anemia or arthritis. In contrast, rats injected with days after injection. In contrast, HCTs of rats receiving the APS or mutanolysin-digested PG-APS did not demonstrate highest dose of PG-APS (40 jig of rhamnose per g [body decreased HCT values or arthritis at any time point but did weight]) returned to completely normal levels by 10 days develop transient peripheral edema 30 min after injection as after injection and remained no different than control levels previously described (4, 5). PG-injected rats did not develop until 70 days after injection (Tables 2 and 3). Eventually, five edema, but four of six rats had clinically detectable diarrhea, of nine rats injected with 40 jig of rhamnose PG-APS per g and all six had conjunctivitis. (body weight) developed moderate chronic anemia, but none Dose response. Rats were injected with PG-APS in doses had a second relapse during the 135-day observation period. ranging from 0 to 40 p.g of rhamnose per g (body weight) to Iron and transferrin levels. To determine whether iron determine the minimally effective dose that produced acute metabolism was abnormal during chronic inflammation in- and chronic, relapsing anemia (Table 2). Rats injected with duced by PG-APS, serum iron and transferrin levels were

TABLE 2. Anemia with increasing doses of PG-APS polymers PG-APS HCTb (%) (mean ± SEM) on day: dose 7 14 22 50 113 0 44.3 ± 0.3 47.8 ± 0.2 46.0 ± 0.3 48.6 ± 0.3 47.3 ± 0.6 0.5 43.2 ± 0.4 46.5 ± 0.3* 46.0 ± 0.6 48.1 ± 0.4 47.3 ± 0.5 1 43.6 ± 0.5 46.8 ± 0.6 45.8 ± 0.3 47.5 ± 0.2* 46.4 ± 0.3 5 44.4 ± 0.3 46.8 ± 0.5 45.8 ± 0.3 43.9 + 1.9 42.1 ± 1.2* 10 42.9 ± 0.4* 43.7 ± 0.8** 40.2 ± 1.9* 43.9 ± 1.9 42.9 ± 0.9* 15 40.9 ± 0.8** 39.8 ± 1.3** 35.3 ± 1.6** 43.6 ± 1.2* 42.6 ± 0.5** 20 41.9 ± 0.9* 42.2 ± 1.4* 39.0 ± 2.5 43.6 ± 1.2* 43.5 ± 1.1* 40 39.6 ± 0.8** 46.0 + 0.3** 45.9 ± 0.4 48.7 ± 0.3 38.9 ± 1.5** " PG-APS polymers from group A streptococci injected i.p. into six to nine female Lewis rats per group. Values are in micrograms of rhamnose per gram (body weight). b *, P < 0.02 compared with rats given no PG-APS; **, P < 0.001 compared with rats given no PG-APS. VOL. 57, 1989 ANEMIA INDUCED BY BACTERIAL CELL WALLS 1181

TABLE 3. Incidence and timing of severe anemia (HCT of <40%) with increasing doses of PG-APS Chronic anemia PG-APS Acute anemia dose' Recurrence 1 Recurrence 2 No.b Peak" No. Peak No. Peak 0 0 0 0 0.5 0 0 0 1.0 0 0 0 5 0 5/8 35-113 1/8 70 10 0 5/9 22-28 5/9 63 15 4/7 7 7/7 22 4/7 63-70 20 5/9 7-10 7/9 18-22 3/9 63-70 40 5/7 4 5/9 70-113 0 a See footnote a of Table 2. b Number of rats with anemia/number of rats injected. ' Peak occurrence of anemia. Values shown are days after i.p. injection of PG-APS.

s W.f measured 36 days after i.p. injection of PG-APS (15 ,ug of rhamnose per g [body weight]). PG-APS-injected rats had 'A I s; b . significantly lower iron and transferrin levels than buffer- ..\ injected control rats (Table 4), despite one misinjected rat in the PG-APS group which did not develop arthritis, anemia, or abnormal iron and transferrin levels. Bone marrows of rats injected with PG-APS 15 days prior to sacrifice contained normal concentrations of intracellular FIG. 3. Iron-laden macrophages within a hepatic granuloma. A iron by cytochemical staining. By 33 and 65 days after female Lewis rat was injected i.p. with PG-APS polymers (30 ,ug of injection, iron staining within the bone marrow was dimin- rhamnose per g) 65 days before necropsy. This small granuloma in the hepatic parenchyma near a central vein contains macrophages ished compared with iron staining in controls but remained with densely staining iron by the method of Gomori (18). Magnifi- detectable. Iron-containing macrophages were markedly in- cation, x450). creased in livers of rats injected with PG-APS, with intensely staining enlarged macrophages concentrated at the periphery of hepatic granulomas (Fig. 3), portal tracts, and central tween arthritis and delayed onset or recurrent anemia (Fig. veins. Kupffer cells of PG-APS-injected rats frequently 4) as well as between the acute phase of inflammation and contained detectable iron and were randomly distributed. In anemia (Fig. 1). In pooled experiments, 35 of 39 (90%) rats contrast, livers of control rats had rare Kupffer cells con- which developed recurrent anemia had their first reactiva- taining faintly staining iron, which were usually adjacent to tion of arthritis within 1 week of the abrupt drop in HCT. At central veins. Mesenteric lymph nodes and spleens of PG- four time points between 10 and 36 days after injection of APS-injected rats contained increased numbers of iron- PG-APS or PBS, HCT and joint scores were very closely containing macrophages, frequently present in aggregates correlated, with r values ranging from -0.93 to -0.98 and P and giant cells. < 0.0001. Owing to variability in the timing of spontaneously Correlation of experimental parameters. PG-APS-injected reactivating inflammation within an experimental group, rats which developed anemia consistently displayed concur- mean values did not adequately demonstrate the close cor- rent arthritis, increased RBC sedimentation rates, hepatic relation that existed between HCT values and joint scores granulomas, and decreased serum iron and transferrin levels during recurrence of arthritis and anemia in individual rats (Table 4). There was a striking temporal relationship be- (Fig. 4). Lewis rats develop irreversible destructive arthritis

TABLE 4. Correlation of experimental parameters in cell wall-injected and control rats Mean" + SEM Material injected or correlation Serum RBC coefficient HCT (%) . Serum Iron transferrin sedimentation Joint score (jig/dI) ~~(mgldl)(% Material injected PG-APSb 38.8 ± 2.7* 149 ± 49* 30.4 ± 12.9* 3.8 ± 1.3** 6.4 ± 1.6* PBSC 47.7 ± 0.2 358 19 79.0 ± 1.5 0.2 ± 0.1 0 0 Correlation coefficient HCT 0.97 0.97 0.87 0.98 Joint score 0.98 0.97 0.98 0.86 Transferrin 0.97 0.95 0.68 0.98 a *, p < 0.005 compared with PBS-injected rats; **, P = 0.01 compared with PBS-injected rats. b PG-APS polymers (15 ,ug of rhamnose per g [body weight]) injected i.p. into five rats 36 days before sacrifice. ' PBS injected i.p. into five rats 36 days before sacrifice. 1182 SARTOR ET AL. INFECT. IMMUN.

50 r DISCUSSION

Anemia associated with acute and chronic systemic in- 45 flammation induced by PG-APS has several unique features Io-1- 0 which make it a useful model to study the anemia of chronic

E inflammation. PG-APS-induced anemia was protracted, as it 40( persisted for at least 10 months after a single aqueous i.p. E0 injection, and spontaneously relapsed and remitted, with its a) clinical course correlating with exacerbations and remissions 35 F 8 4 of arthritis. Anemia following bacterial lipopolysaccharide C)T(L) 0 (endotoxin) injection is transient and very mild (17). Forma- U 4(n lin-killed Corynebacterium parvum injected into mice (27), 2 -C infection with M. arthritides (11) or BCG (26), and footpad -0 injection of complete Freund adjuvant (25) induce anemia - -1 . li 1. 20 60 80 100 120 which is monophasic, with no evidence of spontaneous Days After Injection recurrence, and is limited to at most 8-weeks duration (26). FIG. 4. Temporal relationship between delayed onset of anemia PG-APS is resistant to biodegradation (39) and persists and arthritis in a female Lewis rat injected i.p. with PG-APS within macrophages for at least 4 months after injection (8, polymers (5 ,ug of rhamnose per g). This rat did not develop anemia 35). Spontaneous reactivation of inflammation following a or arthritis until 42 and 50 days, respectively, after injection of a low single local or systemic injection of PG-APS is a character- dose of PG-APS. The anemia was maximal at the onset of arthritis. istic feature of inflammation induced by these phlogistic complexes and may be due to slow intracellular degradation with bony proliferation and ankylosis, leading to persistently of polymers with episodic release from tissue stores (14). elevated joint scores because of ankle enlargement due to In the anemia of chronic disease, serum iron, transferrin, new bone formation, but no gross evidence of active inflam- and TIBC are low despite normal or increased bone marrow mation was detected by edema and erythema. Anemia was and tissue iron stores (13, 24, 48). Rats injected with PG-APS better correlated with active joint inflammation than rela- had significantly lower serum iron and transferrin concen- tively inactive chronic joint destruction (Fig. 1) and usually trations in the chronic phase of inflammation compared with immediately preceded reactivation of arthritis in individual those of control rats, and there was a strong correlation rats (Fig. 4). Similarly, increased spontaneous sedimentation between iron and transferrin levels and the activity of of RBCs was associated with recurrence of anemia and arthritis. Macrophages within the liver, spleen, and mesen- arthritis (Table 5) and returned to normal values during teric lymph nodes had increased stainable iron, particularly remission of active inflammation. within giant cells and granulomas. Fibrotic, more-advanced Rats studied during the first spontaneous reactivation of hepatic granulomas had more densely staining iron than arthritis, 15 days after PG-APS injection, developed severe recently formed, active granulomas with necrotic centers, anemia which was accompanied by significantly decreased suggesting increased sequestration of iron with chronic hemoglobin and RBC counts (P - 0.001, Table 6). The mean inflammation. Mice injected with Freund adjuvant have corpuscular volume of the PG-APS-injected rats was slightly slightly decreased plasma iron, but decreased TIBC and lower than that for control rats. PG-APS-injected rats also transferrin saturation develop only late in the course of had elevated WBC counts, which were closely correlated anemia (25) and splenic iron is diminished, rather than with HCT values (r = 0.93) and joint scores (r = 0.94). The increased, after repeated injections of adjuvant (28). The bone marrows of these rats had decreased numbers of RBC anemia secondary to BCG infection is not associated with precursors relative to WBC precursors, as demonstrated by hypoferremia (26). Injection of C. parvum into rats leads to their significantly elevated bone marrow myeloid precursors/ a prompt decrease in serum iron and retention of iron within erythroid precursor ratios (Table 6). Bone marrow histology mononuclear phagocytes but an increased plasma TIBC (3). revealed normal-to-increased cellularity, with increased Similarly, mice with turpentine-induced inflammation (2) or numbers of megakaryocytes and granulocyte precursors, in infected with M. arthritidis (11) have a prompt decrease in the PG-APS-injected rats compared with PBS-injected con- plasma iron but an increase in transferrin and TIBC. Thus, trols. PG-APS-induced chronic anemia in rats more closely mimics

TABLE 5. HCTs, sedimentation rates, and joint scores during reactivation of PG-APS-induced inflammation PG-APS-injected rats" PBS-injected rats Rat no. or mean (%) Sedimentation Sedimentation HCT ratebJoint score HCT % ~~~~~~~~~~~~~~~~rate Joint score 1 36 15 8 47.5 0.5 0 2 47.5 0.5 0 48 0.5 0 3 28.5 10 8 49 1 0 4 31.5 24 4.5 48.5 0.5 0 5 30 36 7 48.5 0.5 0 Mean + SEM 34.7 ± 3.4' 17.1 + 6.1" 5.5 ± 1.5' 48.3 ± 0.3 0.6 ± 0.1 0 a Values measured 21 days after i.p. injection of PG-APS (15 ,ug of rhamnose per g [body weight]). b Spontaneous sedimentation of RBC measured as percentage of total capillary tube volume 30 min after collection. pP < 0.003 compared with PBS-injected rats. d p = 0.02 compared with PBS-injected rats. VOL. 57, 1989 ANEMIA INDUCED BY BACTERIAL CELL WALLS 1183

TABLE 6. Blood and bone marrow parameters during the first reactivation of PG-APS-induced arthritis Mean ± SEM" Material injected" HCT t Hemoglobin No. of RBC MCV' No. of WBC Bone marrow myeloid (g/dl) (106/mm3) (tim3) (103/mm3) precursor/erythroid precursor ratio PG-APS 33.7 ± 2.9* 11.5 ± 0.9* 5.73 + 0.42* 58.7 ± 1.5** 38.5 ± 6.0* 3.4 ± 0.5* PBS 47.7 ± 0.7 16.2 ± 0.2 7.65 ± 0.12 62.4 ± 0.7 6.9 ± 0.7 0.8 ± 0.2 a Values measured 15 days after i.p. injection of PG-APS (15 ,ug of rhamnose per g [body weight]) or PBS (1 ml) during the first spontaneous reactivation of arthritis. b Values shown are the mean for five rats per group. *, P < 0.001 compared with PBS-injected rats; **, P = 0.03 compared with PBS-injected rats. ' Mean corpuscular volume, calculated by HCT/RBC. the clinical abnormalities of iron metabolism found in the iron is sequestration of lactoferrin-iron complexes after anemia of chronic disease than previously described models liberation of lactoferrin from the secondary granules of of chronic inflammation or infection. PG-APS-activated polymorphonuclear leukocytes. The demonstration that the PG fraction induced anemia is The primary defect in the anemia of chronic disease consistent with previous observations that polymeric PG is appears to be the lack of an appropriate bone marrow the active moiety responsible for most inflammatory reac- response to the slight decrease in RBC survival (13, 48). tions induced by PG-APS fragments from a variety of Figure 2 shows only a slightly increased destruction of RBCs bacterial species (20, 41). Digestion of PG by the murami- during the relatively abrupt recurrence of anemia 3 to 5 dase, mutanolysin, abrogates acute and chronic arthritis (20) weeks after PG-APS injection, indicating variable suppres- and prevents acute vascular permeability changes in the sion of RBC synthesis during different phases of experimen- gut-associated lymphoid tissue induced by PG-APS (R. B. tal inflammation. Activated macrophages have been shown Sartor, unpublished data). Purified PG is readily degraded by to suppress erythropoiesis both during stimulation with LPS lysozyme, in contrast to PG covalently bound to APS (1). (36, 37) or in disseminated histoplasmosis infection (50). This observation is consistent with the transient anemia and Mouse peritoneal macrophages activated by human recom- arthritis which resolved by 14 days after injection of isolated binant tumor necrosis factor or Cryptococcus neoformans PG with no evidence of recurrence, while PG-APS produced infection suppress erythroid progenitor cell growth (33). Lee protracted, spontaneously relapsing anemia and arthritis. (24) speculated that IL-1 mediates the anemia of chronic Purified group A streptococcal APS produced transient disease, but it is apparent that IL-1 has no direct erythroid peripheral edema, as previously described (4), but no anemia inhibitory effect (21, 33, 37), although this monokine acti- or chronic arthritis. vates polymorphonuclear leukocytes to release lactoferrin, The mechanism of anemia induced by PG-APS is un- causes a decrease in serum iron (24), and is produced by known but probably involves both decreased RBC survival scattered bone marrow cells (44). IL-1 has indirect regula- and synthesis. The rapid decrease in HCT during the acute tory effects on hematopoiesis by stimulating bone marrow phase and spontaneous recurrence of anemia, combined stromal cells (15), T lymphocytes (19), fibroblasts (21), and with the slightly decreased RBC half-life measured by 5"Cr endothelial cells (38) to release multilineage hematopoietic labeling, indicates that RBC survival is diminished. There growth factors and activating fibroblasts to release soluble was no evidence ofgastrointestinal or external loss of RBCs, factors that regulate early hematopoietic progenitor cells and the purified PG-APS preparation used for these studies (51). Thus, IL-1 may be responsible for the leukocytosis did not contain streptococcal hemolysin. Patients with rheu- observed after PG-APS injection. Recently, Tracey et al. matoid arthritis have a 20% decrease in RBC survival (43), (46) demonstrated that twice-daily administration of tumor and rats with adjuvant arthritis have a circulating 51Cr-RBC necrosis factor to rats for 8 days induced a progressive half-life of 10.2 days compared with 13.4 days in controls anemia which appeared within 3 days and was associated (28). One mechanism of shortened RBC survival is phago- with a leukocytosis. We postulate that the chronic anemia cytosis of RBCs by reticuloendothelial cells, which occurs in induced by PG-APS is a result of peptidoglycan activation of C. parvum-treated mice (27). Splenic macrophages normally macrophages which liberate monokines, including tumor clear senescent RBCs from the circtilation. Macrophages in necrosis factor (45) and IL-1 (Bristol and Schwab, submit- the spleen, liver, bone marrow, and lymph nodes can phago- ted), and soluble inflammatory mediators, such as prosta- cytose R13Cs that have subtle membrane damage or are glandin E2 (49), which may suppress erythropoiesis directly coated with antibody or complement (7). Activation of or indirectly. Alternatively, stimulated T cells, which have macrophages is a common feature of many types of inflam- been implicated in the pathogenesis of PG-APS-induced mation and has been postulated to be important in the inflammation (47), may suppress the bone marrow, as has pathogenesis of the anemia of chronic disease (24, 33, 37). been demonstrated in rheumatoid arthritis (43) and BCG PG-APS activates macrophages in vitro and in vivo (39; infection in mice (29). Bristol and Schwab, submitted), and intravenous injection of PG-APS-induced inflammation in rats provides an excel- PG-APS leads to phagocytosis of RBCs by macrophages lent model for examining protracted, spontaneously relaps- within mesenteric lymph nodes (34). PG activates comple- ing anemia with modestly decreased RBC survival and ment (12), and Pryzwansky et al. (30) have demonstrated in altered iron metabolism associated with well-characterized vitro aggregation of RBCs and phagocytic cells via PG- arthritis. PG-APS is chemically and antigenically defined and APS-complement (C3b) bridging. Erythrophagocytosis by can activate macrophages in vivo or in vitro, permitting activated reticuloendothelial cells could explain increased more thorough evaluation of mechanisms of bone marrow stainable iron within macrophages of the liver, spleen, and suppression. Activation of macrophages by PG-PS may be a mesenteric lymph nodes of PG-APS-injected rats with ane- common mechanism of the anemia accompanying chronic mia. Another possible explanation of increased macrophage infection by many bacterial species. 1184 SARTOR ET AL. INFECT. IMMUN.

ACKNOWLEDGMENTS Mertelsmann. 1988. Interleukin 1 stimulates T lymphocytes to We thank Ann K. Schwab, Julia Blevins, Joseph J. White, and produce granulocyte-monocyte colony-stimulating factor. J. Roger R. Brown for expert technical assistance; Shirley Willard for Clin. Invest. 81:1415-1418. typing the manuscript; and Eugene Orringer, Stephen A. Stimpson, 20. Janusz, M. J., C. Chetty, R. A. Eisenberg, W. J. Cromartie, and and Kevin Connally for reviewing the manuscript. J. H. Schwab. 1984. Treatment of experimental erosive arthritis This work was supported by Public Health Service grants AM in rats by injection of the muralytic enzyme mutanolysin. J. 01112, AR 25733 and DK 34987 from the National Institutes of Exp. Med. 160:1360-1374. Health and the National Foundation for Ileitis and Colitis. 21. Kaushansky, K., N. Lin, and J. W. Adamson. 1988. Interleukin 1 stimulates fibroblasts to synthesize granulocyte-macrophage LITERATURE CITED and granulocyte colony-stimulating factors. J. Clin. Invest. 1. Abdulla, E. M., and J. H. Schwab. 1966. Biological properties of 81:92-97. streptococcal cell wall particles. III. Dermonecrotic reaction to 22. Krause, R. M. 1967. Preparation of cell wall antigens from cell wall mucropeptides. J. Bacteriol. 91:374-383. gram-positive bacteria, p. 34-40. In C. A. Williams and M. W. 2. Beaumier, D. I., M. A. Caldwell, and B. E. Holbein. 1984. Chase (ed.), Methods in immunology and immunochemistry, Inflammation triggers hypoferremia and de novo synthesis of vol. 1. Academic Press, Inc., New York. serum transferrin and ceruloplasmin in mice. Infect. Immun. 23. Krause, R. M., and M. McCarty. 1961. Studies on the chemical 46:489-494. structure of the streptococcal cell wall. 1. The identification of a 3. Brozovic, B., V. S. Sljivic, and G. W. Warr. 1975. Haematolog- macropeptide in the cell walls of group A and A-variant strep- ical changes and iron metabolism in rats after administration of tococci. J. Exp. Med. 114:127-140. Corynebacterium parvum. Br. J. Exp. Pathol. 56:183-192. 24. Lee, G. R. 1983. The anemia of chronic disease. Semin. Hema- 4. Chetty, C., R. R. Brown, and J. H. Schwab. 1983. Edema- tol. 20:61-80. producing activity of group A streptococcal polysaccharide and 25. Lukens, J. N., G. E. Cartwright, and M. M. Wintrobe. 1967. its possible role in the pathogenesis of cell wall-induced polyar- Anemia of adjuvant-induced inflammation in rats. Proc. Soc. thritis. J. Exp. Med. 157:1089-1100. Exp. Biol. Med. 126:346-349. 5. Chetty, C., D. G. Klapper, and J. H. Schwab. 1982. Soluble 26. Marchal, G., and G. Milon. 1981. Decreased erythropoiesis: the peptidoglycan-polysaccharide fragments of the bacterial cell origin of the BCG induced anaemia in mice. Br. J. Haematol. wall induce acute inflammation. Infect. Immun. 38:1010-1019. 48:551-560. 6. Clark, R. L., J. T. Cutting, S. K. Anderle, W. J. Cromartie, and 27. McBride, W. H., J. T. Jones, and D. M. Weir. 1977. Increased J. H. Schwab. 1979. Radiologic analysis of arthritis in rats after phagocytic cell activity and anaemia in Corynebacterium par- systemic injection of streptococcal cell walls. Arthritis Rheum. vum treated nmice. Br. J. Exp. Pathol. 55:38-46. 22:25-35. 28. Mikolajew, M., Z. Kuratowska, M. Kossakowaska, M. 7. Cooper, R. A., and J. H. Jandl. 1984. Destruction of erythro- Plachecka, and M. Kopec. 1969. Haematological changes in cytes, p. 377-388. In W. J. Williams, E. Beutler, A. J. Eislev, adjuvant disease in the rat. Ann. Rheum. Dis. 28:172-178. and M. A. Lichtman (ed.), Hematology. McGraw-Hill Book 29. Milon, G., G. Marchal, and M. Lebastard. 1984. BCG infection Co., New York. in mice: a model for in vivo analysis of T lymphocytes as 8. Cromartie, W. J., J. G. Craddock, J. H. Schwab, S. K. Anderle, regulators of haemopoiesis. Ann. Immunol. (Paris) 135C:291- and C. Yang. 1977. Arthritis in rats after systemic injection of 294. streptococcal cells or cell walls. J. Exp. Med. 146:1585-1602. 30. Pryzwansky, K. B., J. D. Lambris, E. D. MacRae, and J. H. 9. Dische, A., and L. R. Shettles. 1948. A specific color reaction of Schwab. 1985. Opsonized streptoccal cell walls cross-link hu- methyl pentose and spectrophotometric micromethod for their man leukocytes and erythrocytes by complement receptors. determination. J. Biol. Chem. 175:595-603. Infect. Immun. 49:550-556. 10. Dyer, N. H., J. A. Child, D. L. Mollin, and A. M. Dawson. 1972. 31. Rifai, N., M. E. King, A. Malekpour, J. Smith, and J. Lawson. Anaemia in Crohn's disease. Q. J. Med. 41:419-436. 1986. Immunoturbidimetric assay of transferrin: effect of iron 11. Edwards, C. O., A. Deiss, B. C. Cole, and J. R. Ward. 1975. and need for serum blanks. Clin. Biochem. 19:31-34. Hematologic changes in chronic arthritis of mice induced by 32. Rigby, P. G., H. Strasser, C. P. Emerson, A. Betts, and G. H. Mycoplasma arthritidis (39101). Proc. Soc. Exp. Biol. Med. Friedell. 1962. Studies in the anemia of inflammatory states. 1. 150:664-668. Erythrocyte survival in dogs with acute and chronic turpentine 12. Eisenberg, R. A., and J. H. Schwab. 1986. Arthropathic group A abscesses. J. Lab. Clin. Med. 59:244-248. streptococcal cell walls require specific antibody for activation 33. Roodman, G. D. 1987. Mechanisms of erythroid suppression in of human complement by both the classical and alternative the anemia of chronic disease. Blood Cells 13:171-184. pathways. Infect. Immun. 53:324-330. 34. Sartor, R. B., S. K. Anderle, W. J. Cromartie, and J. H. 13. Erslev, A. J. 1984. Anemia of chronic disorders, p. 522-528. In Schwab. 1986. Localized gut-associated lymphoid tissue hemor- W. J. Williams, E. Beutler, A. J. Erslev, and M. A. Lichtman rhage induced by intravenous peptidoglycan-polysaccharide (ed.), Hematology. McGraw-Hill Book Co., New York. polymers. Infect. Immun. 51:521-528. 14. Esser, R. E., S. A. Stimpson, W. S. Cromartie, and J. H. 35. Sartor, R. B., W. J. Cromartie, D. W. Powell, and J. H. Schwab. Schwab. 1985. Reactivation of streptococcal cell wall-induced 1985. Granulomatous enterocolitis induced in rats by purified arthritis by homologous and heterologous cell wall polymers. bacterial cell wall fragments. Gastroenterology 89:587-596. Arthritis Rheum. 28:1402-1411. 36. Sassa, S., M. Kawakami, and A. Cerami. 1983. Inhibition of the 15. Fibbe, W. E., J. van Damme, A. Billiau, H. M. Goselink, P. J. growth and differentiation of erythroid precursor cells by an Voogt, G. van Eeden, P. Ralph, B. W. Altrock, and J. H. F. endotoxin-induced mediator from peritoneal macrophages. Falkenburg. 1988. Interleukin 1 induces human marrow stromal Proc. Natl. Acad. Sci. USA 80:1717-1720. cells-in long-term culture to produce granulocyte colony-stimu- 37. Sassa, S., S. Wolpe, and A. Cerami. 1987. Inhibition of erythroid lating factor. Blood 71:430-435. differentiation of mouse erythroleukemia cells by a macrophage 16. Fox, A., R. R. Brown, S. K. Anderle, C. Chetty, W. J. Cromar- product(s). Blood Cells 13:161-169. tie, H. Gooder, and J. H. Schwab. 1982. Arthropathic properties 38. Segal, G. M., E. McCall, T. Stueve, and G. C. Bagby, Jr. 1987. related to the molecular weight of peptidoglycan-polysaccharide Interleukin 1 stimulates endothelial cells to release multilineage polymers of streptococcal cell walls. Infect. Immun. 35:1003- human colony-stimulating activity. J. Immunol. 138:1772-1778. 1010. 39. Smialowicz, R., and J. H. Schwab. 1977. Cytotoxicity of rat 17. Fruhman, G. J. 1966. Bacterial endotoxin: effects on erythro- macrophages activated by persistent or biodegradable bacterial poiesis. Blood 27:363-370. cell walls. Infect. Immun. 17:591-598. 18. Gomori, G. 1936. Microtechnical demonstration of iron. Am. J. 40. Stimpson, S. A., R. R. Brown, S. K. Anderle, D. G. Klapper, Pathol. 12:655-663. R. L. Clark, W. J. Cromartie, and J. H. Schwab. 1986. Arthro- 19. Herrmann, F., W. Oster, S. C. Meuer, A. Lindemann, and R. H. pathic properties of cell wall polymers from normal flora bacte- VOL. 57, 1989 ANEMIA INDUCED BY BACTERIAL CELL WALLS 1185

ria. Infect. Immun. 51:240-249. cachexia, anemia, and inflammation. J. Exp. Med. 167:1211- 41. Stimpson, S. A., J. H. Schwab, M. J. Janusz, S. K. Anderle, 1227. R. R. Brown, and W. J. Cromartie. 1986. Acute and chronic 47. Wahl, S. M., D. A. Hunt, J. B. Allen, R. L. Wilder, L. Paglia, inflammation induced by peptidoglycan structures and polysac- and A. R. Hand. 1986. Bacterial cell wall-induced hepatic gran- charide complexes, p. 273-290. In H. P. Seidl and K. H. ulomas. An in vivo model of T cell-dependent fibrosis. J. Exp. Schleifer (ed.), Biological properties of peptidoglycan. Walter Med. 163:884-902. de Gruyter, Berlin. 48. Wallner, S. F. 1983. The anemia of chronic disorders: clinical 42. Strausbaugh, L. J. 1987. Hematologic manifestations of bacte- and pathological features, p. 209-231. In C. D. R. Dunn (ed.), rial and fungal infections. Hematol. Oncol. Clin. N.A. 1:185- Current concepts in erythropoiesis. John Wiley & Sons, Inc., 206. New York. 43. Sugimoto, M., Y. Wakabayaski, S. Hirose, and F. Takaku. 1987. arthritis. 49. Yoshino, S., W. J. Cromartie, and J. H. Schwab. 1985. Inflam- Immunological aspects of the anemia of rheumatoid in the rat air Am. J. Hematol. 15:1-11. mation induced by bacterial cell wall fragments 44. E. J. Kovacs, M. R. Smith, H. A. Young, and S. K. pouch. Comparison of rat strains and measurement of arachi- Takacs, L., metabolites. Am. J. Pathol. 121:327-336. Durum. 1988. Detection of IL-la and IL-lb gene expression by donic acid in situ hybridization. Tissue localization of IL-1 mRNA in the 50. Zanjani, E. D., P. B. McGlave, S. F. Davies, M. Banisadre, normal C57BL/6 mouse. J. Immunol. 141:3081-3095. M. E. Kaplan, and G. A. Sarosi. 1982. In vitro suppression of 45. Tiegs, S. L., J. L. Rothstein, T. J. Lehman, and D. B. Magilavy. erythropoiesis by bone marrow adherent cells from some pa- 1987. Tumor necrosis factor in Lactobacillus casei induced tients with fungal infection. Brit. J. Haematol. 50:479-490. coronary arteritis: a murine model of Kawasaki syndrome. J. 51. Zucali, J. R., H. E. Broxmeyer, C. A. Dinarello, M. A. Gross, Leukocyte Biol. 42:553. and R. S. Weiner. 1987. Regulation of early human hematopoi- 46. Tracey, K. J., H. Wei, K. R. Manogue, Y. Fong, D. G. Hesse, etic (BFU-E and CFU-GEMM) progenitor cells in vitro by H. T. Nguyen, G. C. Kuo, B. Beutler, R. S. Cotran, A. Cerami, interleukin 1-induced fibroblast conditioned medium. Blood and S. F. Lowry. 1988. Cachectin/tumor necrosis factor induces 69:33-37.