ORIGINAL ARTICLE Serglycin Proteoglycan in Hematologic

Serglycin proteoglycan in hematologic malignancies: a marker of acute myeloid leukemia

CU Niemann1, L Kjeldsen1, E Ralfkiaer2, MK Jensen2 and N Borregaard1

1Department of Hematology, Rigshospitalet, Copenhagen, Denmark and 2Department of Pathology, Rigshospitalet, Copenhagen, Denmark

Serglycin is the major cell-associated proteoglycan of hemato- characterize the expression of serglycin in the malignant poietic cells. Previous work has demonstrated that serglycin may counterparts of other hematopoietic cell types that have been be involved in targeting some proteins to granules of cytotoxic lymphocytes, mast cells and neutrophils. We characterized the reported to express serglycin, for example, anaplastic large cell expression of serglycin in various hematologic malignancies by lymphoma (ALCL), natural killer cell (NK cell) lymphoma and immunohistochemistry and ELISA. Serglycin expression was acute lymphoblastic leukemia (ALL). The level of serglycin in found to distinguish acute myeloid leukemia (AML) from acute plasma samples from patients with acute leukemia and lymphoblastic leukemia. In contrast to myeloperoxidase, sergly- Philadelphia chromosome (Ph1)-negative chronic myeloproli- cin was found to be a selective marker for immature myeloid ferative disorders was also investigated. cells, distinguishing AML from Philadelphia chromosome-negative chronic myeloproliferative disorders. Leukemia (2007) 21, 2406–2410; doi:10.1038/sj.leu.2404975; published online 11 October 2007 Materials and methods Keywords: neutrophil elastase; serglycin proteoglycan; granules; AML; ALL Patient samples Plasma was obtained from patients at diagnosis (patients with acute leukemia) or during treatment and follow up (patients with Introduction chronic myeloproliferative disorders) and from normal volun- teers. AML, n ¼ 35; ALL, n ¼ 6; controls, n ¼ 22; idiopathic Serglycin is the major intracellular proteoglycan of hematopoie- myelofibrosis, n ¼ 14; polycythemia vera, n ¼ 13; essential tic cells.1–3 Serglycin mRNA expression has been demonstrated thrombocytosis, n ¼ 11. Formalin-fixed, paraffin-embedded spe- in several myeloid and lymphoid cell lines,4–7 and serglycin cimens of bone marrow were obtained from 18 patients with immunoreactivity has been observed in myeloid cells and cell AML (M1, n ¼ 5; M2, n ¼ 8; M3, n ¼ 1; M4, n ¼ 2; unclassified, lines8,9 as well as in cell lines of lymphoid, mast cell, n ¼ 2), 4 with T-ALL, 8 with B-ALL and from 1 patient without plasmacytoid and myelomonocytic origin.10 The function of hematologic disease. In addition, samples from tumors were serglycin in the different types of cells is slowly becoming obtained from three patients with extranodal T/NK-cell lym- apparent. Initially, sulfation of the heparin side chains of phoma, nasal-type, and seven with ALCL (three anaplastic serglycin was shown to be essential for localization of several lymphoma kinase (ALK) negative, and four ALK positive). The mast cell proteases to granules.11,12 More recently, serglycin samples were from the archives at the Department of Pathology, proved to be essential both for maturation of mast cell granules Rigshospitalet, University of Copenhagen, Denmark. All sam- and for localization of granule proteins.13 Furthermore, gran- ples were reviewed by histology and immunohistology using zyme B was shown to be delivered from cytotoxic lymphocytes standard panels of antibodies (CD34, CD117, myeloperoxidase to target cells as a macromolecular complex with serglycin.14 (MPO), CD68 (PGM1), CD20, CD79a, CD10, TdT, CD3, CD5, The amount of granzyme B but not granzyme A nor perforin CD4, CD8, CD56, TIA-1, Granzyme-B, CD30, ALK1) for in granules is reduced in T cells lacking serglycin and classification in accordance with WHO guidelines (2001). the ultrastructure of granules is slightly altered.15 We recently Informed consent was obtained according to the permission showed that targeting of neutrophil elastase to azurophil and guidelines from the Local Ethics Committee. granules depends on serglycin, whereas this is not the case for other granule proteins tested.3 ELISA During differentiation of neutrophils in human bone marrow, The amount of serglycin in plasma and bone marrow plasma serglycin mRNA expression and immunoreactivity is present samples was determined by a previously described ELISA.9 The only at the promyelocyte and immature myelocyte stage, polyclonal rabbit antibody used for the ELISA was generated whereas mature neutrophils are devoid of serglycin immuno- toward full-length, nonglycosylated, recombinant human ser- reactivity and mRNA.6,9 This prompted us to investigate the glycin. The ELISA has the same sensitivity for glycosylated and expression of serglycin in the malignant counterpart of immature nonglycosylated serglycin.9 Neutrophil elastase was measured myeloid cells, the myeloblasts, in bone marrow samples from by a commercially available ELISA (HyCult biotechnology, patients with acute myeloid leukemia (AML). Furthermore, we Uden, the Netherlands). MPO and Lysozyme levels were measured by ELISA as described previously.16,17 Correspondence: Dr CU Niemann, Rigshospitalet, Department of Hematology, The Granulocyte Research laboratory, Building 9322, Blegdamsvej 9, Copenhagen Ø DK-2100, Copenhagen, Denmark. E-mail: [email protected] Immunohistochemistry Received 9 July 2007; revised 2 September 2007; accepted 3 Sections of formalin-fixed, paraffin-embedded biopsies were September 2007; published online 11 October 2007 heated in a microwave oven (Milestone Micromed, Hacker Serglycin in hematologic malignancies CU Niemann et al 2407 Instruments & Industries Inc., NJ, USA) in TEG buffer (pH 9) for 15 min and stained with a 1:300 dilution of affinity-purified polyclonal rabbit anti-serglycin immunoglobulin G9 for 60 min at room temperature. Further processing was with the Techmate 500 Immunostainer (DAKO, Glostrup, Denmark), using the DAKO Envision K5007 kit.

Statistics All statistics were performed with the SPSS 13.0 software (SPSS Inc., Chicago, IL, USA). Data were log10 transformed for the independent t-test, untransformed for multiple regression analysis.

Results

Serglycin is localized to leukemic blasts in bone marrow from patients with AML In normal human bone marrow, serglycin is detectable only by immunohistochemistry at the promyelocyte and myelocyte stage, whereas mature neutrophils stain negative for serglycin (Figures 1g and h,9). In bone marrow specimens from all 18 patients with AML, the vast majority of the abnormal leukemic blast cells showed cytoplasmic immunostaining for serglycin with a predominantly perinuclear, dot-like staining pattern, resembling that seen in promyelocytes from normal bone marrow (Figures 1a–f). This staining corresponds to localization in the Golgi.9 No differences could be detected in the pattern of serglycin immunoreactivity between different French–Ameri- can–British (FAB) classes of AML (compare Figures 1a and b (M1) with c and d (M3) and e and f (M4)).

ALL cells are devoid of serglycin immunoreactivity We next looked at the expression and localization of serglycin in bone marrow from 12 patients with ALL. No serglycin immunoreactivity was detected in bone marrow specimens from patients with either B-ALL or T-ALL except from a few scattered residual myeloid precursor cells (Figures 1i–l).

Plasma serglycin is significantly higher in AML vs ALL patients, and no difference is seen between Ph1- negative chronic myeloproliferative disorders and normal controls In order to assess whether the serglycin abundantly expressed in Figure 1 Immunohistochemistry for serglycin. Staining for serglycin AML blasts is reflected in plasma, we quantitated the amount of in bone marrow specimens from patients with AML M1 (a, b), AML M3 serglycin in plasma from patients with AML and ALL. The (c, d) and AML M4 (e, f) compared with normal bone marrow (g, h) samples were blinded, that is tested without knowledge of type and bone marrow with B-ALL (i, j) or T-ALL (k, l). Staining is confined of leukemia; all samples were tested in parallel. The mean to AML blasts and myeloid precursors in a perinuclear pattern. (a, c, e, g, i, k) Original enlargement 10, bar 200 mm; (b, d, f, h, j, l) original concentration of serglycin in plasma from patients with AML enlargement 100, bar 20 mm. Counterstaining with hematoxylin. (N ¼ 31) was more than 15 times higher than the mean for ALL AML, acute myeloid leukemia; ALL, acute lymphoblastic leukaemia. patients (N ¼ 6) (Figure 2a). The difference in plasma concentration of serglycin was statistically significant (Po0.001). These results are in accordance with the pattern of serglycin immunoreactivity in bone marrow specimens from patients types of AML (Figures 1a–f). The plasma level of serglycin with acute leukemia. covered a wide range in patients with AML (Figure 2a). Since As the stage of maturation arrest in AML is reflected by the serglycin is expressed by immature myeloid cells, we hypothe- FAB classification,18 we also tested the correlation between the sized that plasma serglycin would correlate with granule plasma level of serglycin and FAB classification, as serglycin is proteins expressed in myeloblasts and promyelocytes. Multiple expressed in a differentiation-specific manner. However, we regression was performed to test the relation between plasma could not detect any significant difference in serglycin level serglycin and lysozyme (previously reported as a marker of among the different FAB groups (data not shown). This is in myelopoiesis19), MPO (a marker of azurophil granules20) and agreement with the uniform level of serglycin immunoreactivity neutrophil elastase (recently shown by us to be dependent on among bone marrow specimens from patients with different FAB serglycin for granule localization in neutrophils3). As seen in

Leukemia Serglycin in hematologic malignancies CU Niemann et al 2408 lymphocytes,21 known to harbor granzyme B in a macromolecular complex with serglycin in their granules.14 We could not detect immunoreactivity for serglycin in the malignant cells in any of these samples (Figures 3a and c). By contrast, granzyme B immunoreactivity was detected in the vast majority of malignant cells in both NK-cell lymphoma specimens and ALCL specimens, irrespective of ALK status (Figures 3b and d).

Discussion

Serglycin proteoglycan is implicated in granule formation and localization of granule proteins in several different hematopoietic cell types.3,11–13,15 We have shown that serglycin is present in promyelocytes and disappears as cells mature along the granulocytic lineage during normal myelopoiesis.9 This prompted us to explore the expression of serglycin proteoglycan in bone marrow of AML patients and to compare with ALL patients and normal controls. We found strong immunoreactivity for serglycin in myeloblasts from untreated AML patients. In contrast, no ALL lymphoblasts were positive for serglycin. In normal bone marrow, serglycin was expressed predominantly in cells from the early stages of myelopoiesis, in particular promyelocytes (Figure 1). This is consistent with serglycin being expressed predominantly in immature myeloid cells whether these are AML myeloblasts or cells from the early steps of normal myelopoiesis. We hypothesized that plasma serglycin reflects the expression Figure 2 Plasma level of serglycin and correlation with plasma neutrophil elastase, MPO and lysozyme. Patients with AML and ALL of serglycin in bone marrow cells in acute leukemia patients and (a), and patients with Ph1-negative chronic myeloproliferative might be used to discriminate between AML and ALL. We found disorders compared to normal controls (b) were tested for serglycin statistically significant higher levels of serglycin in plasma from content by ELISA. Boxes indicate quartiles with a line for the median, patients with AML compared with plasma from patients with whiskers indicate range; extreme values are indicated by dots. ALL (Figure 2a). When an arbitrary cut-off at 0.5 (arbitrary units * Signifies statistically significant difference, Po0.001, independent of the ELISA) was used, 32 out of 35 plasma samples from AML samples t-test. AML, n ¼ 35; ALL, n ¼ 6; controls, n ¼ 22; IMF, n ¼ 14; PCV, n ¼ 13; ET, n ¼ 11. (c) The level of serglycin in plasma samples patients were above this level and all 6 out of 6 samples from from patients with AML is plotted against the plasma level of ALL patients were at or below this level. This gives a positive neutrophil elastase for the same patients. The regression line is predictive value of 100% and a negative predictive value of indicated. (d) The outcome of multiple regression on the plasma level 67%, while the specificity in our population will be 100% and of serglycin with plasma levels of Lys, MPO and NE as explanatory the sensitivity 91% in differentiating between patients with AML ¼ variables. No other variables were included in the analysis; n 35. and ALL. The correlation between serglycin expression in bone MPO, myeloperoxidase; AML, acute myeloid leukemia; ALL, acute lymphoblastic leukemia; Ph1, Philadelphia chromosome; IMF, idio- marrow specimens from AML patients and the plasma level of pathic myelofibrosis; PCV, polycythemia vera; ET, essential thrombo- serglycin in AML patients indicates that plasma serglycin cytosis; Lys, lysozyme; NE, neutrophil elastase. originates predominantly from myeloid cells. Chronic myeloproliferative disorders represent states with increased numbers of mature myeloid cells.22 If only immature Figures 2c and d, only neutrophil elastase correlated statistically myeloid cells are positive for serglycin and if the plasma level of significant with the plasma level of serglycin. serglycin reflects the expression in immature myeloid cells, The level of serglycin in plasma from patients with Ph1- plasma serglycin in patients with Ph1-negative chronic myelo- negative chronic myeloproliferative disorders did not differ from proliferative disorders would equal that in normal controls. Our that of normal controls (Figure 2b). This was true whether we data from patients with idiopathic myelofibrosis, polycythemia analyzed the samples as a single group or divided into groups of vera and essential thrombocytosis are in agreement with this patients with idiopathic myelofibrosis, polycythemia vera and (Figure 2b). The few patients with higher plasma levels of essential thrombocytosis. Multiple regression correction for serglycin may represent patients with extramedullary hemato- confounders such as length of disease or previous treatment poiesis or with increased immature myeloid precursors in the did not reveal any differences among the patients (data not bone marrow but we have no available data to substantiate this shown). assumption. These findings indicate that the increase in numbers of mature myeloid cells in Ph1-negative chronic myeloproliferative disorders does not result in elevated plasma serglycin. Serglycin is absent from granzyme B-positive T- and This underscores that serglycin is expressed predominantly in NK-cell lymphomas immature myeloid cells and that these are the predominant We next examined serglycin expression in specimens from source of serglycin in plasma. The fate of serglycin during patients with cytotoxic T/NK cell lymphomas, including three maturation of neutrophils has not been determined. However, with extranodal T/NK-cell lymphoma, nasal-type and seven with the previously described protease resistance of the serglycin ALCL, four of which were positive for ALK. These types of protein core23 has been questioned by several reports.9,24–26 lymphoma represent the malignant counterparts of cytotoxic Thus, degradation of serglycin inside the neutrophil granules,

Leukemia Serglycin in hematologic malignancies CU Niemann et al 2409

Figure 3 Immunohistochemistry of lymphoma specimens. Extranodal T/NK-cell lymphoma (a, b) and ALK-positive ALCL (c, d) strongly positive for granzyme B (b, d) but negative for serglycin (a, c). (a–c) Stained with horseradish peroxidase and diaminobenzidine as substrate; (d) stained with alkaline phosphatase and fast red as substrate, counterstaining with hematoxylin. Original enlargement 40. NK, natural killer; ALK, anaplastic lymphoma kinase; ALCL, anaplastic large cell lymphoma. rich in proteases, is a likely explanation for lack of serglycin in serglycin. This shows that serglycin cannot be essential for granules of mature neutrophils. Patients with chronic myeloid packaging of granzyme B in granules in these tumor cells but it leukemia were not included, since the level of promyelocytes in does not rule out serglycin as essential for routing granzyme B to bone marrow and peripheral blood is often significantly granules. However, this would require that tumor stem cells that elevated at diagnosis. express serglycin exist. The previous report on granzyme B Serglycin is thus a marker for immature myeloid cells, in dependency on serglycin for granule localization is from a contrast to MPO, which identifies all stages of myeloid cells murine model. The discrepancy may therefore also be due to from promyelocytes to mature neutrophils. This difference is differences between the species. emphasized by the lack of correlation between plasma serglycin In conclusion, serglycin may be a useful marker in the and MPO in AML patients. Serglycin is not a marker of diagnosis of AML, either as a supplement/alternative to classic myelopoiesis per se, as seen by the low plasma serglycin in markers such as MPO used in characterization of bone marrow patients with Ph1-negative chronic myeloproliferative disorders, specimens or as a marker protein measured in plasma. and from the lack of correlation between plasma serglycin and lysozyme.19 However, a statistically significant correlation between plasma serglycin and neutrophil elastase was found Acknowledgements (Figures 2c and d). This may reflect the role of serglycin in targeting of neutrophil elastase to granules.3 We thank Lone Svendstrup for expert technical assistance. This Serglycin is previously reported to be present in granules of study was supported by grants from The Danish Medical Research cytotoxic lymphocytes in complex with granzyme B and Council to NB, CUN was supported by a grant from H:S, perforin,14 and localization of granzyme B in granules of Copenhagen University Hospital, ER was supported by grants cytotoxic lymphocytes is reported to be dependent on sergly- from the Novo Nordisk Foundation and the HS Research Council. cin.15 We were unable to detect immunoreactivity for serglycin in bone marrow specimens from patients with B- or T-ALL References (Figures 1i–l). To further investigate whether the malignant counterparts of cytotoxic lymphocytes were positive for 1 Kolset SO, Gallagher JT. Proteoglycans in haemopoietic cells. serglycin immunoreactivity, we tested specimens from NK-cell Biochim Biophys Acta 1990; 1032: 191–211. lymphomas and ALCL21 for serglycin by immunohistochemistry. 2 Kolset SO, Prydz K, Pejler G. Intracellular proteoglycans. Biochem No staining for serglycin was detectable (Figure 3). Still, these J 2004; 379 (Part 2): 217–227. specimens were positive for granzyme B, which has been 3 Niemann CU, Abrink M, Pejler G, Fischer RL, Christensen EI, reported to be dependent on serglycin for localization to Knight SD et al. Neutrophil elastase depends on serglycin 15 proteoglycan for localization in granules. Blood 2007; 109: granules. We cannot rule out that the granzyme B level is 4478–4486. reduced in these tumor cells compared with their normal 4 Stevens RL, Avraham S, Gartner MC, Bruns GA, Austen KF, Weis counterparts but clearly granzyme B is present in the absence of JH. Isolation and characterization of a cDNA that encodes the

Leukemia Serglycin in hematologic malignancies CU Niemann et al 2410 peptide core of the secretory granule proteoglycan of human secretory granule maturation and granzyme B storage. J Biol Chem promyelocytic leukemia HL-60 cells. J Biol Chem 1988; 263: 2005; 280: 33411–33418. 7287–7291. 16 Kjeldsen L, Bainton DF, Sengelov H, Borregaard N. Identification 5 Maillet P, Alliel PM, Mitjavila MT, Perin JP, Jolles P, Bonnet F. of neutrophil gelatinase-associated lipocalin as a novel matrix Expression of the serglycin gene in human leukemic cell lines. protein of specific granules in human neutrophils. Blood 1994; 83: Leukemia 1992; 6: 1143–1147. 799–807. 6 Stellrecht CM, Mars WM, Miwa H, Beran M, Saunders GF. 17 Lollike K, Kjeldsen L, Sengelov H, Borregaard N. Purification of Expression pattern of a hematopoietic proteoglycan core lysozyme from human neutrophils, and development of an ELISA for protein gene during human hematopoiesis. Differentiation 1991; quantification in cells and plasma. Leukemia 1995; 9: 206–209. 48: 127–135. 18 Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DA, 7 Cowland JB, Borregaard N. The individual regulation of granule Gralnick HR et al. Proposals for the classification of the acute protein mRNA levels during neutrophil maturation explains leukaemias. French–American–British (FAB) co-operative group. the heterogeneity of neutrophil granules. J Leukoc Biol 1999; 66: Br J Haematol 1976; 33: 451–458. 989–995. 19 Lollike K, Kjeldsen L, Sengelov H, Borregaard N. Lysozyme in 8 Schick BP, Gradowski JF, San Antonio JD. Synthesis, secretion, and human neutrophils and plasma. A parameter of myelopoietic subcellular localization of serglycin proteoglycan in human activity. Leukemia 1995; 9: 159–164. endothelial cells. Blood 2001; 97: 449–458. 20 Faurschou M, Borregaard N. Neutrophil granules and secretory 9 Niemann CU, Cowland JB, Klausen P, Askaa J, Calafat J, vesicles in inflammation. Microbes Infect 2003; 5: 1317–1327. Borregaard N. Localization of serglycin in human neutrophil 21 Jaffe ES, Krenacs L, Raffeld M. Classification of cytotoxic T-cell and granulocytes and their precursors. J Leukoc Biol 2004; 76: natural killer cell lymphomas. Semin Hematol 2003; 40: 175–184. 406–415. 22 Baxter EJ, Scott LM, Campbell PJ, East C, Fourouclas N, Swanton S 10 Toyama-Sorimachi N, Kitamura F, Habuchi H, Tobita Y, Kimata K, et al. Acquired mutation of the tyrosine kinase JAK2 in human Miyasaka M. Widespread expression of chondroitin sulfate-type myeloproliferative disorders. Lancet 2005; 365: 1054–1061. serglycins with CD44 binding ability in hematopoietic cells. J Biol 23 Stevens RL, Otsu K, Austen KF. Purification and analysis of the core Chem 1997; 272: 26714–26719. protein of the protease-resistant intracellular chondroitin sulfate E 11 Humphries DE, Wong GW, Friend DS, Gurish MF, Qiu WT, proteoglycan from the interleukin 3-dependent mouse mast cell. Huang C et al. Heparin is essential for the storage of specific J Biol Chem 1985; 260: 14194–14200. granule proteases in mast cells. Nature 1999; 400: 769–772. 24 Stevens RL, Humphries DE, Wong GW. Serglycin proteoglycans: 12 Forsberg E, Pejler G, Ringvall M, Lunderius C, Tomasini-Johansson the family of proteoglycans stored in the secretory granules of B, Kusche-Gullberg M et al. Abnormal mast cells in mice deficient certain effector cells of the immune system. In: Iozzo RV (ed). in a heparin-synthesizing enzyme. Nature 1999; 400: 773–776. Proteoglycans Structure, Biology, and Molecular Interactions. 13 Abrink M, Grujic M, Pejler G. Serglycin is essential for Marcel Dekker Inc.: New York, NY, 2000, pp 177–200. maturation of mast cell secretory granule. J Biol Chem 2004; 25 Nicodemus CF, Avraham S, Austen KF, Purdy S, Jablonski J, 279: 40897–40905. Stevens RL. Characterization of the human gene that encodes the 14 Raja SM, Wang B, Dantuluri M, Desai UR, Demeler B, Spiegel K peptide core of secretory granule proteoglycans in promyelocytic et al. Cytotoxic cell granule-mediated apoptosis. Characterization leukemia HL-60 cells and analysis of the translated product. J Biol of the macromolecular complex of granzyme B with serglycin. Chem 1990; 265: 5889–5896. J Biol Chem 2002; 277: 49523–49530. 26 Lohmander LS, Arnljots K, Yanagishita M. Structure and synthesis 15 Grujic M, Braga T, Lukinius A, Eloranta ML, Knight SD, Pejler G of intracellular proteoglycan in HL-60 human leukemic promye- et al. Serglycin-deficient cytotoxic T-lymphocytes display defective locytes. J Biol Chem 1990; 265: 5802–5808.

Leukemia