Scand J Haematol 1986;36:385-393

Key words: leukaemia - glycoconjugate - glycogen - cytochemistry - cytoplasmic granules

Ultrastructural evaluation of periodate-reactive glycoconjugates in human leukaemia cells

Mitsuoki Eguchi, Toshiharu Furukawa, Hitoshi Sakakibara, Kohji Ishikawa, Kenichi Sugita, Hisashi Sakamaki’ & Samuel S. Spice?

Department of Paediatrics and Internal Medicine, Dokkyo University School of Medicine, Mibu, Tochigi, Japan and 2Departrnent of Pathology, Medical University of South Carolina, Charleston, South Carolina, USA

Periodate-reactive glycoconjugates in human leukaemic cells were examined electron microscopically by the periodic acid-thiocarbohydrazide-silver proteinate (PA-TCH- SP) method. Granules in ALL cells were classified into 4 types based on PA-TCH-SP staining features. Abnormal granules containing glycogen were observed only in children with treatment-resistant ALL. Cytoplasmic granules in leukaemic cells of patients with AML and acute monocytic leukaernia exhibited moderate reactivity. The distribution pattern of glycogen in the cytoplasm of leukaemic cells was classified into 3 types, one lacking glycogen, one containing small glycogen particles scattered throughout cytoplasm, and one showing clusters of glycogen particles. Cells with glycogen clusters were observed in ALL cells and in erythroblasts from patients with erythroleukaemia. PA-TCH-SP reactivity was detected in the rough endoplasmic reticulum in acute promyelocytic leukaemia but not in ALL or other types of AML. Megakaryoblasts in megakaryocytic crisis of chronic myelogenous leukaemia exhibited characteristic PA-TCH-SP reactivity similar to that of normal megakaryocytes.

Accepted for publication December 11, 1985

The distribution of complex carbohydrates con- taining hexoses with vincinal glycols can be detected by the light microscopic PAS technique. Since PAS staining was first applied to the examination of leukaemic cells ( l ) , conflicting results have been obtained concerning its signifi- cance in detecting the character of leukaemia cells. Some reports have noted a relationship between PAS positivity and the duration of the initial remission in ALL (2, 3), whereas other studies found no prognostic value in the PAS

reactivity (4, 5). Such differences may have occurred because the distribution of glycoconjug- ates in the cell organellae could not be dis- tinguished by light microscopy due to limited capacity of resolution. To solve this problem, several ultrastructural techniques to observe vici- nal glycol-containing hexoses in glycoconjugates were developed. The periodic acid thiocar- bohydrazide-silver proteinate (PA-TCH-SP) method is reliable for revealing these substances at the ultrastructural level (6). In this study, we

26’

386 EGUCHI ET AL

examined the distribution of periodate-reactive glycoconjugates in leukaemia cells using this method.

Material and methods Bone marrow or buffy coat was obtained from 70 patients with leukaemia. 35 of the patients were diagnosed as having acute lymphocytic leukaemia (ALL) based on cytochemistry, immunocytochemistry and cell morphology of stained smears. 31 of these patients were considered to have non-B and non-T cell type ALL because of negative staining for B- and T-cell markers. 3 patients had T-cell type ALL, and 1 had B-cell type ALL belonging to the Burkitt category. 17 cases of acute myelogenous leukaemia (AML), 4 cases of acute promyelocytic leukaemia (APL), 5 cases of acute monocytic leukaemia (AMoL), 2 cases of erythroleukaemia and 7 cases of chronic myelocytic leukaemia (CML) were included in this study.

The marrow or buffy coat specimens were fixed with 2.5% glutaraldehyde in 0.1 mol/l cacodylate buffer pH 7.2 for 60 min. The specimens were then rinsed in buffer and embedded in epoxy resin without txposure to osmium tetroxide because this fixative impairs the cytochemical reactivity of carbohydrates (7) . The thin sections were mounted on stainless steel grids, oxidized in 1% periodate for 30 min, rinsed in distilled water and treated with thiocarbohydrazide in 20% acetic acid for 40 min. The sections were then rinsed briefly in 10% acetic acid, 5% acetic acid and distilled water in succession. After rinsing, they were exposed

for 30 min to 1% silver proteinate in the dark and rinsed in distilled water (8, 9). The sections were examined at 80 kv in a JEM 100-B electron microscope without heavy metal counterstaining. As a cytochemi- cal control for all specimens, periodate oxidation was omitted from the sequence. Enzymatic digestion of glycogen was performed on thin sections with 0.5% a-amylase or saliva for 3 h.

Resu I ts The PA-TCH-SP method stained cytoplasmic granules, Auer bodies, Golgi cisternae and glycogen in leukaemic cells. The method also demonstrated periodate-reactive glycoconjugates in rough endoplasmic reticulum and on the outer surface of the plasmalemma in some cases (Table 1). Staining varied with the different clinical conditions as described in the following sections.

Acute lymphocytic leukaemia The lymphoblastoid cells without T- or B-cell markers disclosed no consistent reaction pattern, whereas those with T-cell or B-cell characters revealed PA-TCH-SP reactivity resembling that of normal T or B lymphocytes (8).

Non B, non T type ALL Lymphoblastoid cells varied markedly among the cases examined in the distribution of cytoplasmic granules (Figures 1-4). In some individuals a

TABLE 1 Periodic acid-thiocarbohydraride-silver proteinate reactivity of leukaemic cells

Number of cases

In which the In which the of leukaemic cells leukaemic cells show leukaemic cells show

lack glycogen clustered glycogen positive RER

Total In which a majority cases of leukaemic cells In which a majority

contain stained glycogen particles

ALL AML APL AMoL Ery. L JCML" ACML'* Total

35 17 4 5 2 3 4

70

25 (71 %) 14 (82%) 4 (100%) 4 (80%) 2 (100%) 3 (100%) 4 (l00%)

56 (80%)

10 (29%) 3 ( I S % ) 0 1 (20%) 0 0 0

14 (20%)

12 (34%) 0 0 0 0 '3

0 0

12 (17%)

0 0 3 (75%) 0 0 0 0 3 (4%)

*I JCML = Juvenile type CML. '2 ACML = Adult type CML. Blast crisis was excluded from the table. - 3 Erythroblasts were excluded.

EM OF GLYCOCONJUGATES IN LEUKAEMIA 387

Figure 1 . An acute lymphocytic leukaemia (ALL) cell lacking T or B surface markers (non-T, non-B cell). The cytoplasm contains PA-TCH-SP-positive clustered cytoplasmic granules and small evenly dispersed glycogen particles. x 14900. Figure 2. A non-T non-B ALL cell. Scattered cytoplasmic granules and dispersed glycogen are evident. x 11300. Figure 3. An ALL cell without T or B surface markers contains clustered glycogen and a giant granule 0.6 bm in diameter (arrow). x 16200. Figure 4. Cytoplasmic granules abnormally enclosing glycogen are present in a lymphoid cell from a child with treatment-resistant ALL. x 36200. Figure 5 . Cell from a patient with acute T cell leukaemia shows clustered cytoplasmic granules and lack of glycogen which are frequently observed in normal T cells. M: Mitochondria. x 30000.

388 EGUCHI ET AL

high proportion of cells exhibited clustering of PA-TCH-SP-positive granules in the notch of the nucleus or near the Golgi zone. The clusters resembled those seen in normal T lymphocytes, and consisted of 3 to 10 or more granules (Figure 1). Cells of other individuals revealed single granules scattered throughout the cytoplasm and not clustered (Figure 2) . Cytoplasmic granules also varied among the individuals studied rang- ing from nearly absent to an average of 5 granules per cell.

The structure of the cytoplasmic granules in the lymphoblastoid cells also varied between cases. Granules in cells of many individuals were fairly uniform and somewhat elongated, measur- ing up to 0.5 pm and were stained at a faint to moderate intensity throughout their matrix, showing stronger reactivity in the rim in some cases.

Granules in cells from 4 children had one or more foci with very intense reactivity often of a particulate nature similar to that of glycogen particles in the cytoplasm (Figure 4). Because of

its staining property and structure, this material, which was distributed unevenly in the granules, was assumed to be glycogen incorporated by the lysosomal cytoplasmic granules.

The last type of PA-TCH-SP-positive granule observed was large, measuring more than 0.5 pm diameter, and was scattered in the cytoplasm separate from the other granules (Figure 3). This type of granule was not observed frequently in ALL.

Glycogen in the cytoplasm of ALL cells showed 2 different patterns of distribution. In one type, the small glycogen particles measur- ing about 10 nm were scattered evenly in the cytoplasm (Figures 1 and 2) . In the other type, glycogen particles formed large clusters (Figure 3). ALL cases lacking T and B markers could thus be classified into 3 groups, (a) containing clustered glycogen, (b) contianing scattered glycogen, and (c) lacking glycogen.

To determine whether PAS reactivity was cor- related to the prognosis of ALL in children, children with non-T, non-B ALL were followed

TABLE 2 Periodate-reactive complex carbohydrates in childhood ALL without surface markers

PA-TCH-SP reactive sites in leukemic cells

Glycogen Granules

Case PAS Cell with clu- Cell with Cell with gly- Cell with clu- Cell Abnormal Giant cogen parti- stered glyco- without gly- stered scattered granules with granules

gen glycogen > 0.5 p cogen plasmic cytoplasmic cles

granules granules

1 + 100% 53 % 0 vo 1 I % 16% 0 To 0 vo 2 f 100 I5 0 0 0 0 0 3 + 93 50 7 29 14 0 0 4 - t o * 100 10 0 50 30 0 0 5 + 70 46 30 54 46 0 8

A' 6 - t o + 100 0 0 14 43 0 0 7 0 0 100 78 I I 0 0 8 18 0 82 9 73 0 0 9 - t o * 100 0 0 50 40 0 0 10 - to + 0 0 100 5 1 14 0 14 11 0 0 100 20 40 0 0 12 100 0 0 7 33 0 7 13 N.D. 100 0 0 18 95 95 0 14 83 0 17 0 35 4 13

B" I5 + 100 0 0 38 62 62 11 16 f to + 67 0 33 I 1 56 33 I I 17 79 0 21 36 51 0 0

-

* Group A = Treatment-responsive group (No relapse more than 2 yr). * * Group B = Poor prognostic group (No remission or relapse in 2 yr).

390 EGUCHI ET AL

up to 2 or more yr, and the results were compared with the findings on PAS reactivity (Table 2). All patients having abnormal granules with glycogenoid substances in the granule (Cases 13-16) had a poor prognosis; those totally lacking glycogen in the cytoplasm (Cases 7, 10 and 11) responded to treatment; and the patients having lymphoblastoid cells containing clustered glycogen (Cases 1-5) also tended to be responsive to treatment.

ALL with T-cell characteristics The lymphoblastoid cells lacked glycogen parti- cles in the cytoplasm except for a few cells that had only a few particles. This staining pattern resembled that in normal T cells. However, the leukaemic T cells differed from normal T cells in that they contained fewer clustered cytoplasmic granules with PA-TCH-SP reactivity (Figure 5 ) .

ALL with B-cell characteristics The leukaemic cells in Burkitt-type ALL con- tained a moderate abundance of glycogen parti- cles scattered in the cytoplasm as in normal B cells (8). The PA-TCH-SP-positive, clustered granules were more frequent in Burkitt-type than in normal B cells (Figure 6).

Acute myelocytic, promyelocytic, and monocytic leukaemias In AML and APL the cytoplasmic granules were PA-TCH-SP-positive, and thus resembled imma- ture primary granules in normal early neutrophils (Figures 7-9). The masking of PA-TCH-SP- reactivity that occurs during maturation of nor- mal primary granules (6) was not observed in the primary granules of these leukaemic cells. Auer bodies exhibited striated PA-TCH-SP-reactivity in the matrix of the body, and the intensity of the reactivity resembled that of the neighbouring cytoplasmic granules (Figure 8). PA-TCH-SP reactivity was observed in cisternae of the rough endoplasmic reticulum in 3 out of 4 APL cases (Figure 9). PA-TCH-SP staining of the granular reticulum was not observed in ALL, other types of AML or acute monocytic leukaemia (Table 1).

Cells in acute monocytic leukaemia showed

PA-TCH-SP reactivity in small granules. Golgi cisternae in these cells and those of AML also reacted positively (Figure 10). 3 cases of AML and 1 of acute monocytic leukaemia totally lacked glycogen particles in the cytoplasm of the leukaemic cells.

Eryt hroleukaemia The PA-TCH-SP-positive glycogen particles appeared to be increased in size and in number in the erythroblasts of a patient with erythro- leukaemia (Figure 11). The PAS reactivity of these erythroblasts corresponded with the in- creased number and size of glycogen particles that were observed ultrastructurally.

Chronic myelocytic leukaemia Granulocytes of CML resembled normal gran- ulocytes in PA-TCH-SP reactivity. Distinctive staining was noted, however, in the crisis stage. Small leukaemic megakaryocytes in peripheral blood in the megakaryocytic crisis in adult CML exhibited PA-TCH-SP reactivity similar to that characteristic of normal megakaryocytes (10). The granules in these cells showed prominent PA-TCH-SP staining in nucleoids that resembled those showing such reactivity in normal platelet granules. The PA-TCH-SP reactivity of the cell surface of leukaemic megakaryocytes was strong- er than that of the surface glycocalyx of all other types of leukaemic cells. This glycocalyx staining corresponded with the stronger staining of the surface of normal megakaryocytes compared to that of other blood cells. Demarcation mem- branes were strongly positive (Figure 12).

Abnormal erythroblasts of juvenile CML con- tained numerous glycogen particles (Figure 13).

Cytochemical controls The cytochemical controls in which the pe- riodate-oxidation was omitted from the sequence of PA-TCH-SP showed no densification in the cytoplasmic granules, glycogen, Golgi apparatus or rough endoplasmic reticulum. The hetero- chromatin of the nucleus revealed non-specific density (Figure 14). Enzymatic digestion with a-amylase or saliva decreased the reactivity of

EM OF GLYCOCONJUGATES IN LEUKAEMIA 391

Figure 12. Megakaryoblast in megakaryocytic crisis of CML. Note the stained cell surface and demarcation membranes (D). The granules exhibit an eccentric focus of reactivity corresponding to the nucleoid of normal alpha granules. x 27400. Figure 13. Erythroblast in erthroblastosis of juvenile CML. Glycogen particles appear increased in number and size. x 8 100. Figure 14. Cytochemical control section treated with the same sequence except for omitting periodate-oxidation. AML cell. The granules (G) exhibit no densification. The electron opacity in the nucleus is attributed to a non-specific reaction. M: Mitochondria. x 5100.

glycogen particles, but some glycogen staining persisted after digestion, presumably because of poor penetration of these agents into the epoxy sections (6).

Discussion Although previous works have established that the PA-TCH-SP procedure has the same cy- tochemical significance in visualizing complex carbohydrates with vicinal glycol groups as does the PAS method (6, 11) the more sensitive PA- TCH-SP method offers the advantage of demon- strating smaller amounts of glycogen, showing

varying patterns of its cytosolic distribution in different cases and localizing complex car- bohydrates to cell organellae (8). By using the PA-TCH-SP method, we could classify the PAS- negative leukaemic cells into the following 3 groups; (a) totally lacking periodate-reactive complex carbohydrates, (b) with a small amount of glycogen, and (c) with a small amount of PA- TCH-SP-positive cytoplasmic granules but no glycogen. Although PAS staining is negative in some forms of leukaemia, the ultrastructural PA-TCH-SP method can be utilized to obtain further information for classification of leuka- emias.

392 EGUCHI ET AL

It is not known whether differences exist in the origin and biologic nature of the 4 types of cytoplasmic granule classified in this study according to the pattern of PA-TCH-SP reac- tivity. However, in addition to morphological differences between the cell types, the carbohydr- ate content of clustered cytoplasmic granules also differs from that of the scattered cytoplasmic granules. For example, our recent studies have revealed that the dialyzed iron reaction for localizing acid mucosubstances was positive in scattered cytoplasmic granules, in contrast to the lack of dialyzed iron reactivity in clustered cytoplasmic granules (12). This difference indi- cated that these 2 types of granule differ in content and, hence, in origin and perhaps func- tional activity. The cytochemical difference could also reflect a difference in degree of cell or granule maturation, but this seems less likely because of the absence of intermediate stages.

The giant granules frequently observed in most of the patients unresponsive to treatment and a minority of the responsive group were widely separated in the cytoplasm, distributed as in the small scattered cytoplasmic granules. These gran- ules were interpreted as resulting from fusion of scattered small granules into atypically enlarged granules. Such a fusion process, if it underlines the genesis of the giant granules, would resemble the process leading to giant granules in leuco- cytes of patients with the Chediak-Higashi syn- drome.

The abnormal granules with non-uniform staining in ALL cells were only detectable by electron microscopic cytochemistry. Since these granules occurred only in the group unresponsive to treatment, they were interpreted as a sign of poor prognosis. The density and particulate nature of the material with PA-TCH-SP reac- tivity in the unevenly stained granules closely resembled that of the cytoplasmic glycogen which was the only component as yet observed with such structure and staining intensity. This similarity indicated that the stained granule substance is glycogen.

Uptake of glycogen into lysosomes occurs in Pompe’s glycogenesis (13) and appears to reflect

an abnormality of lysosomal function in ALL cells. The abnormal uptake of cytoplasmic glycogen and abnormal production of giant lysosomes in ALL point to alterations in the lysosomes limiting membrane as manifestations of the leukaemic state.

Two distribution patterns of glycogen were noted with PA-TCH-SP staining in ALL cells. In one pattern, which corresponded to diffuse PAS staining seen in light microscopy, glycogen parti- cles appeared scattered evenly throughout the cytoplasm. However, very large amounts of glycogen, as often seen in mature neutrophils, were rarely observed in ALL cells. The second distribution pattern, consisting of clusters of glycogen, appeared to correspond to the well- known clustered PAS reactivity of ALL. The latter aggregates, with strong PAS staining in ALL cells, may not be the same as the clustered cytoplasmic granules with PA-TCH-SP positi- vity, because the cells in case 10, as an example (Table 2), lacked glycogen and contained clus- tered cytoplasmic granules but showed only faint ‘clustered’ PAS reactivity. Thus, the PAS reac- tivity of ALL cells when compared to PA- TCH-SP staining for electron microscopy repre- sents different reactive intracellular sites in dif- ferent cases.

The PA-TCH-SP reactivity of AML granules resembled that of normal primary granules of neutrophils, except that the staining persisted in granules of leukaemic cells in peripheral blood. The leukaemic cells and their primary granules do not seem to undergo normal maturation (6 , 14), which involves loss of the granules PA- TCH-SP reactivity. Leukaemic granules were stained even in cells in which a degree of cell maturation was evidenced by prominent mar- gination of nuclear heterochromatin. This indi- cated a discrepancy of maturation between the cytoplasm and nucleus in AML cells. Most profiles of leukaemic cells reveal some cisternae of rough endoplasmic reticulum, but PA-TCH- SP-positive material in cisternae of granular reticulum was observed only in APL. This positively-stained matter may be related to the increased synthesis of glycoprotein, which is the

EM OF GLYCOCONJUGATES IN LEUKAEMIA 393

main source of PA-TCH-SP-reactive substance apart from glycogen (14). This selective staining of granular reticulum in APL cells may, there- fore, serve as a means of diagnosing APL.

Acknowledgement The authors wish to express their appreciation to Dr A. Komiyama at Shinshu University, Dr T. Suda at Jichi Medical School, Dr M. Tsuchida at Toho University for the supply of bone marrow specimens.

References I . Gibb RP, Stowell RE. Glycogen in human blood cells.

Blood 1949;4:569-79. 2. Ascari E, Marini G, Invernizzi R. On the usefulness of

PAS reaction for the prognosis of acute lymphoblastic leukemia. Hematologica 1975;60:300-7.

3. Felges AJ, Aur RJA, Verosa MS, Daniels S . Periodic acid Schiff reaction, a useful index of duration of complete remission in acute childhood lymphocytic leukaemia. Acta Haematol 1974;52:8- 13.

4. Humphrey GB, Nesbit ME, Brunning RD. Prognostic value of the periodic acid Schiff (PAS) reaction in ALL. Am J Clin Pathol 1974;61:393-7.

5 . Shaw MT, Humphrey GB, Lawrence R, Fischer DB. Lack of prognostic value of the periodic acid-Schiff reaction and blast cell size in childhood acute lymphocytic leukemia. Am J Hematol 1977;2:237-43.

6 . Parmley RT, Eguchi M, Spicer SS, Alavarez CJ, Austin RL. Ultrastructural cytochemistry and radioautography of complex carbohydrates in heterophil granulocytes from rabbit bone marrow. J Histochem Cytochem 1980;28: 1067-80.

7. Simson JAV. The influence of fixation on the carbohydr- ate cytochernistry of rat salivery gland secretory granules. Histochem J 1977;9:64-57.

8. Eguchi M, Sugita K, Sugiyama S , Furukawa T. Electron microscopic detection of periodate reactive complex car- bohydrates in human T and B lymphocytes. Scand J Haematol 1982;29:97- 104.

9. ThiCry JP. Mise en evidence des polysaccharides sur coupes fines en microscopie electronique. I Microscopie 1967;6:987-1018.

0. Eguchi M, Furukawa T, Yamaguchi K. Ultracytochemical observation of periodic acid-reactive complex carbohydr- ates in human blood cell. Erythrocyte series, mega- karyocytes and platelets. J Clin Electron Microscopy 1980; 13 :697-8.

1. Jewel1 GG, Saxton CA. The ultrastructural demonstration of compounds containing 1-2-glycol group in plant cell wall. Histochem J 1970;2:17-28.

12. Eguchi M, Furukawa T, Sugita K, Sugiyama S . Ultra- cytochemistry of the cell line from human T cell lympho- ma. Comparison with EB tranformed B cell line. Cell 1982; 14:68-72.

13. Martin JJ, Barsy T, Van Hoof F, Palladini G. Pompe’s disease. An inborn lysosornal disorder with storage of glycogen. Acta Neuropathol 1973;23:229-39.

14. Parmley RT, Eguchi M, Spicer S S . Ultrastructural cytochemistry of complex carbohydrates in leukocyte granules. J Histochem Cytochem 1979;27: 1167-70.

Correspondence to: D:. Mitsuoki Eguchi Department of Paediatrics Dokkyo University School of Medicine Mibu-machi, Tochigi-ken 321-02 Japan