Brain Research, 571 (1992) 19-25

19

© Elsevier BRES 17355

Expression of granulocyte colony stimulating factor and granulocytemacrophage colony stimulating factor genes in human astrocytoma cell lines and in glioma specimens Taizo Nitta 1, Kiyoshi Sato, Mark Allegretta, Stefan Brocke, Mae Lim, Dennis J. Mitchell and Lawrence Steinman 1Department of Neurology and Neurological Science, Stanford University School of Medicine, Stanford, CA (U.S.A.) and 2Departmentof Neurosurgery, Juntendo University School of Medicine, Tokyo (Japan) (Accepted 29 August 1991)

Key words: Granulocyte colony stimulating factor; Granulocyte-macrophage colony stimulating factor; Gene; Astrocytoma; Polymerase chain reaction; Enzyme-linked immunosorbent assay

Expression of granulocyte (G) and granulocyte-macrophage (GM) colony stimulating factor (CSF) genes in human cells of astroglial lineage was studied. Primers for CSFs were used to analyze RNA transcripts in 5 cultured human astrocytoma cell lines and 8 fresh brain specimens by polymerase chain reaction. Constitutive expression of mRNA transcripts of GM-CSF could be detected in all astrocytoma and one neuroblastoma cell lines, and two out of 5 unstimulated astrocytomas, U87MG and U138 MG, expressed G-CSF genes. After stimulation with interleukin (IL)-I/3 + tumor necrosis factor (TNF)-a, all cell lines expressed G-CSF. In addition to the cultured cells, we examined gene expression within human malignant astrocytoma, peritumoral brain and autopsied normal brains. The results show that some of the tumor and its surrounding reactive lesions express G- and GM-CSF genes but normal brains do not. The concentration of G- and GM-CSF in supernatants of cultured cells was assessed at the protein level by ELISA. A low level of GM-CSF activity was constitutively present in all astrocytomas. G-CSF was detected in unstimulated U87MG and U138MG and other cell lines could synthesize G-CSF after the stimulation of IL-lj8 and TNF-a at the level of mRNA. Furthermore, the concentration of CSFs increased markedly upon stimulation with IL-l~8 and/or TNF-a in both a time- and dose-dependent fashion. From these results, it is suspected that astrogllal cell-derived CSFs may participate in local immune reactions accompanying infection, degeneration and malignancies in the brain. INTRODUCTION Gliosis, the most remarkable inflammatory reaction in the brain, seems to be closely related with immune regulatory substances. Only recently have we come to understand that many aspects of the immune responses in the central nervous system are governed by these polypeptides produced by immune effector cells 3. The brain has been traditionally viewed as being an immunologically privileged site for m a n y reasons14; (1) the brain is equipped with a b l o o d - b r a i n barrier (BBB), a network of endothelial-astrocyte tight junctions prohibiting iramune surveillance; (2) the brain lacks a lymphatic drainage and is devoid of the immune interaction with the systemic immune networks; (3) the brain cannot completely reject transplants; and (4) the brain lacks major histocompatibility complex ( M H C ) antigens on normal neuroglia cells 9,t°. However, with the advent of cellular and molecular techniques, the astrocyte and other neural cells, prepared from rat brains, can be stimulated to

function in a manner analogous to monocytes/ macrophages, and this has been postulated to act as an immunocompetent cell in the CNS 2t. A m o n g these iramunologic functions, the astrocyte can be induced to express class II M H C antigens after exposure to interferon (WN)-F or viral antigens, and thus can present antigen to T-cell clones in an MHC-restricted fashion 9. Furthermore, some cytokines can regulate M H C expression, and the astrocyte itself is able to synthesize some cytokines directly 7's'11'12. Cytokines have long been recognized as key elements in both normal and aberrant immune cell interactions and also surprisingly, recent reports showed that primary cultured rat or mouse astrocytes can be induced to express some cytokines, such as interleukin (IL)-I, IL-6, tumor necrosis factor (TNF)-a, granulocyte colony stimulating factor (G-CSF) and granulocyte-macrophage colony stimulating factor (GM-CSF) with appropriate stimuli 5'7'HA2,29. Nevertheless, regulation of gene expression in human brains remains a subject of intense investigation. A m o n g the cy-

Correspondence: T. Nitta, Department of Neurology, Stanford University School of Medicine, Stanford, CA 94305-5235, U.S.A. Fax: (1) (415) 725-7459.

20 tokines, G- a n d G M - C S F have p r o f o u n d effects u p o n

survival and activation of neutrophils, eosinophils, and

was heated at 95°C for 5 min, then quickly chilled on ice. Thereafter each cDNA was used for PCR amplification.

macrophages in addition to their potent effects on the

PCR

growth a n d d e v e l o p m e n t of m y e l o i d cells 22. T h e r e f o r e , the identification of C S F genes expressed in the CNS is

In order to identify the RNA transcripts of lymphokines expressed in the astrocytes, we used 3 individual sets of 5"- and 3"-oligonucleotide lymphokine primers. The sequence of CSF primers are as follows. G-CSF; 5" AGC TCC CTG CCC CAG AGC T r c CTG, 3" CAT GGC ACC CTG GGT GGG CTG CAG 25, GM-CSF; 5" ACA CTG CTG AGA TGA ATG AAA CAG TAG, 3" TGG ACT GGC TCC CAG CAG TCA AAG GGG ATG6. As a positive control for PCR, fl-actin primers were also used in this study26. These primers were synthesized and column purified by Operon Technologies, Inc. (Alameda, CA). The resulting singlestrand cDNA samples were amplified using a 5" sense with 3" antisense primers at a final concentration of 1 ftM in each reaction. The amplification was performed with 2.5 units of Taq polymerase (Ampli Taq; Perkin Elmer) on a DNA thermal cycler (Perkin Elmer, Cetus). The PCR cycle profile was 95°C denaturation for 1 min, annealing of primers at 60°C for 1 min, extension of primers at 72°C for 1 min for 25 cycles. Some cDNA samples which negatively expressed G- or GM-CSF genes were also subjected to further 10-20 cycles of reamplification in order to confirm the data. PCR products were separated on 3% regular agarose gels. Expression of lymphokine genes was considered positive when a band was visualized with ethidium-bromide staining. The sizes of PCR amplified products are as follows; G-CSF (320 base pairs (bp)), GMCSF (286 bp) and fl-actin (200 bp). Experiments were repeated 3 times per sample.

critical in o r d e r to u n d e r s t a n d the i m m u n e n e t w o r k in various diseases of b r a i n , such as infection, d e g e n e r a t i o n a n d malignancy. H e r e , we investigated expression of C S F genes in h u m a n astroglial cell lines a n d also in fresh b r a i n specimens by using p o l y m e r a s e chain r e a c t i o n ( P C R ) . I n

addition to examining RNA transcripts of G- and GMCSF, we also e x a m i n e d the c o n c e n t r a t i o n of CSFs in s u p e r n a t a n t s of c u l t u r e d glioma cells q u a n t i t a t i v e l y by e n z y m e - l i n k e d i m m u n o s o r b e n t assay ( E L I S A ) . MATERIALS AND METHODS

Cells and tissue specimens In this study, we used 5 malignant human astrocytomas (U87MG, U138MG, U252MG, U373MG and A172), one neuroblastoma (IMR32) and 4 leukemia cell lines (Jurkat and Molt 4; T-cell leukemia, Raji; B-cell and K562; erythroid leukemia, respectively), The cell lines were purchased from the American Type Culture Collection. The astrocytoma cell lines were established from malignant astrocytoma patients and were confirmed as astroeyte origin by their expression of glial fibrillary acidic protein (GFAP). The cells were cultured in serum-free medium, AIM-V (Gibco, Labo, NY) to avoid the contamination of cytokines from bovine serum at the concentration of 5 x 104 cells/nil. After they were cultured with and without rlL-lfl (0, 1, 10, 100, 1000 U/ml) and/or rTNF-a (0, 0.1, 1, 10, 100 ng/ml) (Boehringer Mannheim, IN) at 37°C, 5% CO2 incubator for 24 h, the supernatant was removed and stored at -20°C for ELISA assay. Cells were harvested and washed twice with 0.1 M phosphate-buffered saline and subjected to RNA extraction. As a positive control for PCR, PBLs (1 x 106 eells/ml) were isolated from a healthy donor by means of Ficoll-Hypaque gradient centrifugation and were cultured in the presence of 1/~g/ml of anti-CD3 (anti-Leu 4; Becton-Dickinson; San Jose, CA) for 2 days6. After harvesting, they were subjected to RNA extraction, Four malignant astroeytoma and two normal autopsied brain specimens were also studied. Malignant astrocytomas were histologically diagnosed as glioblastoma multiforme (astroeytoma grade IV) according to the WHO classification (cases 1-4). Furthermore, two tissue specimens adjacent to a malignant astrocytoma, in which astrocytes show a reactive, proliferative appearance, were studied (cases 5 and 6). Case 5 is a tissue specimen adjacent to ease 1 and case 6 is from the surrounding lesion of case 2. Normal brain specimens were obtained from autopsied cases of those who died from non-neurologic diseases. A portion of the specimen removed at operation or autopsy was stored at -70"C and was used to extract total RNA.

Extraction of RNA and synthesis of complementary DNA (cDNA) Each tumor specimen was thawed at room temperature. Then, total RNA from glioma tissues was prepared with gnanidinum thiocyanate extraction using RNA zol (Cinna/Biotex, TX)according to the established procedure 19. 0.1/~g of total RNA was used for the synthesis of single-strand cDNA using reverse transciptase26'2s. Briefly, in a final volume of 20/zl 1 x PCR buffer (50 mM KC1, 20 mMTris-Cl, pH8.4, 2.5 mMMgC12), 1 mMofdeoxyribonucleotide triphosphates (Perkin-Elmer), 20 units of placental ribonuelease inhibitor (BRL), 100 pmol of random hexamerie oligonucleotides (Pharmacia, CA) and 200 units of BRL MoMulV reverse transcriptase (BRL) were incubated with RNA (0.1/~g) for.40 min at 42°C according to the described procedure 28. The reaction mixture

ELISA For the quantitative assay of human G- and GM-CSF, a solid phase ELISA, Quantikine (R&D systems, MN) was employed24. Briefly, 200 /A of the supernatant, with 10-fold and/or 100-fold dilution, were added into a 96 well-microplate coated with mouse anti-human G- or GM-CSF mAb and were incubated at RT for 2 h. After washing 3 times, 200 tA of affinity purified HRP-conjugated rabbit anti-IL-G- or GM-CSF antibodies (Abs) was added and incubated at room temperature (RT) for 2 h. The wells were washed 3 times and reacted with the mixture of hydroperoxide and tetramethylbenzide. After a 30-min incubation, the reaction was stopped by adding 50 ~1 of 2 N sulfuric acid, and then the intensity of color was measured by spectrophotometer at 450 nm. A curve was prepared, plotting the OD4s0 versus the concentration of G- or GMCSF in the standard wells. By comparing the OD450 of the samles to this standard curve, the concentration of the CSFs in the samples was determined. The cytokines and growth factors used in this study are as follows; rIL-lfl, rIFN-7, basic fibroblast growth factor (bFGF) (5 ng/ml), epidermal growth factor (EGF) (5 ng/ml), insulin-like growth factor (IGF) II (10 ng/ml) (Boehringer-Mannhelm; IN) and lipopolysaccharide (LPS) (10 ng/ml) (Sigma; M O ) 31.

RESULTS A s Fig. 1A shows, we have efficiently detected expression of transcripts of CSFs studied, using c D N A from a n t i - C D 3 s t i m u l a t e d PBLs as a positive control. T h e n we e x a m i n e d the C S F gene expression in h u m a n a s t r o c y t o m a a n d o t h e r cell lines. A l l cells expressed actin p r o d u c t s as expected. A s Fig. 1B shows, R N A tran-

scripts of G-CSF could be amplified in U87MG and U138MG cells in the absence of any stimulations. Furt h e r m o r e , after 24-h c o - s t i m u l a t i o n with 1000 U / m l of r l L - l f l a n d 100 ng/ml of rTNF-ct, R N A transcripts could be d e t e c t e d in all a s t r o c y t o m a a n d n e u r o b l a s t o m a cell

21 lines as is clearly shown in Fig. lB. As we will show later from the time kinetic study, the astrocytomacells showed a maximum G- and GM-CSF activity after 24-h stimulition. Therefore, we used 24-h incubation with IL-lfl and/or TNF-a to stimulate cells. Four leukemic cell lines did not show RNA transcripts of G-CSF even with the stimulation. GM-CSF genes were amplified in all unstimu l a t e d a s t r o c y t o m a a n d n e u r o b l a s t o m a cell lines (Fig. 1C). T h e c u l t u r e d a s t r o c y t o m a cells might have different p r o p e r t i e s c o m p a r e d to fresh t u m o r s p e c i m e n s a n d norreal astrocytes b e c a u s e of in vitro artifact. It is essential

A

I.I. -," .~ ~, ! X ~. da. 603~

I

310 ~ 281 234 ~

B t~

~ ~

~

~ ~

~ ~

,,, ~" ~

~,. ~

a b a ba b a ba b a b a b i ~ ~

~

1 2 3 4 5 6 7 8 a b a ba b a b a ba b a b a b _ ~

~ 603 310 281 -

Fig. 2. G- and GM.CSF gene expression in human brain specimens. Location of samples are as follows; center of tumors (1-4), the marginal reactive lesion adjacent to tumors (5, 6) and normal autopsied brains (7, 8). Lane a, G-CSF; lane b, GM-CSE PCR condition was the same as Fig. 1. The left lane shows the molecular sizes (bp).

to examine the gene expression of CSFs within native tissue specimens. In order to determine differences in gene expression, care was taken to isolate aliquots from the center of tumors (cases 1-4) and the surrounding reactive lesion (cases 5 and 6). From clinical and histological findings, we could not know the difference among these samples. Single-stranded eDNA was prepared from messenger RNA (mRNA) transcripts isolated from tumor specimens. Location of samples was ascertained with frozen and permanent histologic sections. Normal brain specimens obtained from autopsy were also subjected to this study (cases7and 8). As Fig. 2 shows, two out of 4 astrocytoma specimens and one peritumoral reactive lesion expressed G-CSF genes and also expression of GMCSF genes was detected in 3 of tumors and both two peritumoral lesions. However, from all normal brain specimens, mRNA transcripts of G- and GM-CSF could not be obtained. From eDNA samples which did not

~O000

C

I

I~ 03 tO I~

"~eOoo

~I ~ 0

603 310~

,~o~ t~

281"--"

~'

2341

~

Fig. 1. A: amplified products from anti-CD3.activated human PBLs are shown after 25 cycles of PCR with an individual set of fl-actin, G-CSF and GM-CSF primers. B: PCR amplified products from 5 human malignant astrocytomas (U87MG, U138MG, U251MG, U373MG, A172), one neuroblastoma (IMR32) and T-cell leukemia (Molt 4) with primers for G-CSF without (a) and with (b) the stimulation of rTNF-a (100 ng/ml) and IL-lfl (1000 U/nil). PCR condition was the same as A. C: PCR amplified products of GMCSF from unstimulated cells as was described in B. The left lane shows the molecular sizes (bp) (Phi X 174; New England Biolabs, MA).

1~'ooo

~100oo

.~~000 "4000

~0' 0'.

~0

~,~. ~ IL-lp (U/ml)o

,

o

~

' ~'>

TNF-ct(ng/ml)

Fig. 3. The effect of IL-1/3 and TNF-a upon G-CSF production by astrocytomas. U138MG cells were placed in culture medium with various concentrations of IL-lfl (U/ml) and TNF-a (ng/ml), following 24-h stimulation, G-CSF activity of each supernatant was subjected to study with ELISA.

22 TABLE I

TABLE III

G-CSF production from neuroglial cell lines (pg/ml)

G- and GM-CSF production by neuroglial cell lines in response to cytokines (pg/ml)

Each cell line (5x104/ml) was pretreated with rlL-lfl (1000 U/ml), rTNF-a (100 ng/ml) or their combinations for 24 h, the supernatants harvested and was subjected to ELISA. Cells

Stimulated with (-)

U87 U138 U251 U373 A172 IMR32 Molt 4 Jurkat Raji K562

Each cell line (5×104/ml) was pretreated with a different growth factor or cytokine for 24 h, the supernatants harvested and subjected to ELISA.

2000 50