[CANCER RESEARCH 50. 278-283. January 15.

1990]

Absence of IFNA and IFNB Genes from Human Malignant Glioma Cell Lines and
Lack of Correlation with Cellular Sensitivity to Interferons1
Junji Miyakoshi,2 Kelly D. Dobler, Joan Allalunis-Turner, John D. S. Me Kcan, Kenneth Petruk, Peter B. R. Allen,
Keith N. Aronyk, Bryce Weir, Debbie Huyser-Wierenga, Dorcas Fulton, Raul C. Urtasun, and Rufus S. Day III3
Molecular Genetici and Carcinogenesis Laboratory, Department of Medicine [J. M., K. D., R. S. D.], Radiobiology Laboratory, Department of Radiation Oncology [J.
A-T.], and Department of Radiation Oncology [D. H-W., D. F., R. C. U.), Cross Cancer Institute, Edmonton, Alberta T6G IZ2, and Division of Neurosurgery, Faculty
of Medicine, i'niversity of Alberta, Edmonton, Alhena T6G 2G3, Canada J.D. S. M., K. P., P. B. R. A., K. N. A., B. H'.]

ABSTRACT
We report that 5 of 19 human malignant glioma cell lines have neither
interferon a (IFNA) nor interferon 0 (IFNB) genes that are detectable
by Southern blotting. Of 5 other of these malignant glioma lines that
have a single IFNB gene copy, 3 lack the IFNA genes entirely and two
have one copy. One of the lines that lacks the IFNA genes entirely but
has one copy of the II \H gene has a rearrangement near the IFNB gene
that is most easily interpreted as an insertion of a large segment of DNA
(at least 50 kilobases) the 'end of which is <1.3 kilobases 5' to the
known regulatory sequences of the IFNB gene. In spite of the re
arrangement, IFNB-specic RNA is highly inducible in this line by
poly(I)-poly(C). The ability of interferon a or interferon 0 to inhibit cell
growth does not depend upon the presence or absence of the respective
gene. This Finding adds solid tumors to those tumor cell lines (acute
lymphocytic leukemia, chronic myelogeneous leukemia) previously deter
mined to lack the IFNA and IFNB genes (Diaz et al., Proc. Nati. Acad.
Sci. USA, 85:5259-5263, 1988).

INTRODUCTION

IFN4-or -f) treatment

the absence of the IFNA and IFNB genes in human malignant

glioma cell lines.
Although the molecular basis for cellular growth rate sensi
tivity to IFN-a and -
is unknown, IFNs produce many molec
ular effects. IFNs interact with IFN receptors (genes for which
map on chromosome 21) to induce the synthesis of (2',5')oligoadenylate synthetase, which activates RNase L (an endoribonuclease also induced by the IFN treatment) and the PI/
eIF-2a protein kinase (7), as well as to induce several mRNAs
having products of unknown function (8, 9).
A possible clue to the cellular function of IFNs-a and -
was
the concurrent and homozygous loss of IFNA and IFNB genes
found in five acute lymphoblastic leukemia cell lines (and in
the leukemic cells of 3 of 42 acute lymphocytic leukemia
patients) as well as in one chronic myelogenous leukemia line
(10). IFNs may thus constitute a class of tumor suppressor
genes as proposed (10, 11) and further suggested to us that we
analyse our malignant glioma cell lines for the presence of the
IFNA and IFNB genes.

of IFN-a- or -/3-sensitive human

MATERIALS

tumor cell lines results in dramatic inhibition of growth and

loss of colony-forming ability (1-5). Our previous work showed
that 10 of 10 lines that were able to repair moGua in their
DNA (and were Mer*) were resistant to IFN-a or -,while 6

AND METHODS

Cell Lines. The cell lines used and references to their origins were:
A172, 118MG, T98, U87MG, and H4 (12, 13); A1235 (14) was kindly
provided by Dr. S. A. Aaronson; SAN, CLA, MIL, R1C (15), GRE,
and P4 are all malignant glioma lines generously provided by Drs. Paul
Kornblith (Albert Einstein College of Medicine, Bronx, NY) and Barry
Smith (Cornell Medical College, New York, NY). KD (16) and
GM3314 (Human Genetic Mutant Cell Repository, Camden, NJ) are
normally appearing human fibroblast strains. A431, from an epidermoid carcinoma of the vulva, A875, from a melanoma, and A427, a
lung cancer cell line, are described in Refs. 13 and 14. Continuous
malignant glioma lines that were produced in our laboratory included
M002, M006, M007, M010, M012, M016, and M027.5
Cell Culture. Cells for DNA or RNA extraction were cultured in
DMEM (Gibco) plus 10% fetal calf serum plus a 1/200 dilution of a
stock consisting of 10,000 units penicillin plus 10,000 ^g streptomycin/
ml. For experiments on the effects of IFN-a and -on growth rate,
DMEM plus 5% PCS plus antibiotics (as above) was used. Stock
cultures of tumor cells and normal fibroblasts were transferred once
weekly at 1:6 and 1:3 ratios, respectively. All lines were maintained at
37C,10% CO2, and 80-90% relative humidity.

of the 8 lines that were unable to repair moGua (and were

Mer~) were sensitive (5). In pursuing this finding, we concen
trated our efforts on the study of human malignant glioma cell
lines, of which about Vifail to repair moGua. Human malignant
gliomas often show breakpoints in chromosome 9p (6), the
location of the IFNA and IFNB genes. In this paper we present
our study of the association of sensitivity to IFN-a and -
with
Received 5/30/89; revised 9/20/89; accepted 10/10/89.
The costs of publication of this article were defrayed in part by the payment
of page charges. This article must therefore be hereby marked advertisement in
accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1This study was supported by an Establishment/Scholarship Grant from the
Alberta Heritage Foundation for Medical Research (AHFMR) to R. Day and
AHFMR and Alberta Cancer Board fellowships to J. Miyakoshi.
1 Present address: Department of Experimental Radiology, Faculty of Medi
cine. Kyoto University, Kyoto 606, Japan.
J To whom requests for reprints should be addressed, at Molecular Genetics
and Carcinogenesis Laboratory. Department of Medicine. Cross Cancer Institute,
11560 University Ave., Edmonton, Alberta T6G 1Z2.
4 The abbreviations used are: IFN. interferon; IFNA, human interferon gene;
IFNB, human interferon gene; Mer* and Mer", able and unable (respectively)
to support normally the growth of adenovirus treated with A/-methyl-A"-nitro-A/nitrosoguanidine; ACTG, human 7-actin gene; ERCC2, human excision repair
cross-complementing gene 2; XRCC1, human X-ray repair cross-complementing
gene \;GLI, human glioma-associated oncogene; ROSI, human gene homologous
to the avian UR2 sarcoma viral oncogene v-ros; JUN, human gene homologous
to avian sarcoma 17 viral oncogene v-jun; KRAS2, human gene homologous to
Kirsten rat sarcoma 2 viral oncogene v-Ki-ras2; ERBAI, human gene homologous
to avian erythroblastic leukemia viral oncogene \-erb-a; EGFR, human epidermal
growth factor receptor gene (homologous to avian erythroblastic leukemia viral
oncogene v-erA-b): PDGFB. human platelet-derived growth factor
polypeptide
(homologous to the simian sarcoma viral oncogene v-si's); MOS, human gene
homologous to Moloney murine sarcoma viral oncogene \-mos; SDS, sodium
dodecyl sulfate; DMEM, Dulbecco's modified Eagle medium; moGua, O6-methylguanine; ATCC, American Type Culture Collection; PCS, fetal calf serum; Ix
SSC, 0.15 M NaCI-0.015 M Na citrate.

Effect of IFNs on Growth Rate. Cell lines were seeded at 50.000/

100-mm plate at three plates/dose of IFN/selected period of growth.
One day later the medium was removed and replaced with medium
containing 0, 100, 300, 1000. and 3000 units/ml IFN-a (>106 units/
mg protein, from human huffy coats; Meloy Laboratories, Richmond,
VA) or IFN-/3 (recombinant IFN-ft Triton Biosciences, Inc., Lot BP227A) in DMEM plus 5% FCS. Cell numbers (3 plates/point) after
about 1, 3, 5, and 7 days of incubation in IFN-/3 were determined, aided
by a Coulter counter. Growth rates (cell divisions/day) were calculated
from linear regression analyses of plots of log (cell number) versus
hours of incubation. The coefficients of determination were usually at
least 0.98 and tended to be greater at increased growth rates. The
growth rates of IFN-treated cultures were calculated relative to the
5 R. S. Day III. et al., unpublished results.

278

IFNA AND IFNB GENES AND IFN SENSITIVITY

growth rate at zero IFN-/J concentration (taken as 100%).

Preparation and Analysis of Cellular Nucleic Acids. We detected the
FNB gene by Southern blotting (17), using the Hind\\\-Xho\\ or
Hindlll-BstEll fragment of pSY2501 (18) (American Type Culture
Collection, Rockville, MD). The IFNA gene was detected using the
865-base pair EcoRl-Pstl fragment of pIFNa503 (19), a kind gift of Dr.
D. Goeddel (Genentech, Inc.). Probes (20) were hybridized to electrophoresed Hindlll-, EcoRl-, BamHl-, Bglll-, or /^il-restricted genomic
DNAs prepared (21) from malignant glioma cell lines. Purified DNAs
were restricted, electrophoresed at 5 Mg/'ane in 0.8% agarose gels (10
x 14 cm) in 80 IHMTris-phosphate, 2 IHMEDTA, pH 8.5 (22), at 25 V
for 16 h, stained with 0.5 Mg/ml ethidium bromide for visualization,
and washed with water. Gels were prepared for Southern blotting by
30-min depurination in 0.25 N HC1, followed by a 0.4 N NaOH
treatment for 10-20 min prior to overnight sponge-assisted transfer to
BioTrace RP nylon membranes (Gelman) with 0.4 N NaOH (17). After
overnight immersion in 0.4 N NaOH (to ensure DNA fixation) and
drying, membranes were washed in O.lx SSC (23) plus 0.1% SDS at
37Cfor 30 min with shaking. Probes were prepared by isolation of
appropriate restriction fragments from 1.0% low melt agarose gels (run
as above), labeling with 32P-dCTP, priming with random hexamers (20),
and separation of probe from unincorporated nucleotides on Sephadex
G50. Specific activities were 1-3 x 10*cpm/Mg. Prehybridization (2-4
h at 42C)and hybridization with the probe (24-48 h at 42Cwere
done in a plastic bag with 14.1 ml of a solution prepared with 7.7 ml
of deionized 100% formamide, 0.77 ml of 100 x Denhardt's solution
(22), 3.8 ml of 3.0 M NaCl, 0.2 M NaH2PO,20 IHMEDTA, pH 7.4
(22), 0.15 ml of 10% SDS, 1.0 ml of 25% dextran sulfate, 0.28 ml of
10% non-fat milk powder, and 0.4 ml of sonicated herring sperm carrier
DNA (denatured at 100'C for 5-10 min) at 10 mg/ml. Membranes
were washed in O.lx SSC and 0.1% SDS at 55Cfor 30-60 min 2
times, again encased in plastic, and autoradiographed using Kodak
XAR film for 3-7 days at -70C.To analyze membranes for the ACTC
gene, membranes were stripped and reprobed with the 2.1-kilobase
BamHl fragment of pHFgA-1 containing the ACTG gene (24). Other
plasmids used for preparing probes included pER2-12 (ERCC2 gene)
and pXRl-30 (XRCC1), both kindly supplied by Dr. Larry Thompson
(Lawrence Livermore National Laboratory, Livermore, CA); pKK36PL
(GLI), through the kindness of Dr. Burt Vogelstein (Johns Hopkins
Oncology Center, Baltimore, MD); pS65BROSl (ROSI), through the
kindness of Dr. Michael Wigler (Cold Spring Harbor Laboratory, Cold
Spring Harbor, NY); phcJ-1 (JUN), through the generosity of Dr.
Michael Karin (Center for Molecular Genetics, School of Medicine,
University of California, San Diego, CA); and p640 (KRAS2) (ATCC
No. 41026), pHE-Al (ERBAI) (ATCC No. 57334), pE7 (EGFR)
(ATCC No. 57347), pSM-1 (PDGFB) (ATCC No. 57050), and pHM2A
(A/OS) (ATCC No. 41004) from the American Type Culture Collection.
Analysis of RNAs Made during poly(I)-poly(C) Induction. Our pro
tocol for poly(I)-poly(C) induction followed that of Enoch et al. (25).
Cultures of DMEM plus 10% FCS were transferred to serum-free
DMEM for poly(I)-poly(C) or sham treatments. Total RNA was puri
fied (26) from control cultures (sham treated with 50 Mg/ml cyclohexmidefor 4 h) and from poly(I)-poly(C)-treated cultures [treated simul
taneously with 100 Mg/ml poIy(I)-poly(C) and 50 Mg/ml cycloheximide
for 1.5 h, followed by a 2.5-h incubation with 50 Mg/ml cycloheximide
alone]. Five Mgcellular RNA were electrophoresed in 1% agarose gels
with 0.66 M formaldehyde, transferred to Gelman Biotrace RP using
lOx SSC, baked, and analyzed as above with the human IFNB probe.
Membranes were stripped and reprobed with the 7-actin probe. [The
method was fully described (27).)

RESULTS
A total of 16 different blots of two sets of genomic DNAs (as
in Fig. 1) prepared after restriction with EcoRl, Pstl, Bglll,
BamHl, and Hindlll were probed to detect the presence or
absence of the IFNB gene. Five of the 18 malignant glioma cell
lines lack the IFNB gene entirely. Fig. la, in which BamHlrestricted genomic DNAs were probed, shows that 4 of 11

malignant glioma lines, GRE, 118MG, SAN, and U87MG,

gave no band corresponding to the IFNB gene, whereas the
other malignant glioma lines and one normal fibroblast strain,
KD, showed the expected (28, 29) 15-kilobase band. A similar
loss of both gene copies was detected when genomic DNAs
were restricted with either EcoRl, Bglll, Hindlll, or Pstl,
suggesting that the absence of the band was not likely due to
an anomalous restriction fragment length polymorphism pro
ducing /F/VA-containing fragments small enough to run off the
gel. The absence of the gene from the lines was confirmed in
DNA dot blots (data not shown). Comparison of densitometer
tracings of this autoradiograph with an autoradiograph pre
pared from the same membrane stripped and rehybridized with
an ERBAI probe shows that H4, RIC, and P4 likely have only
one copy of the IFNB gene relative to an assumed ERBAI (or
relative to both JUN and XRCC1 in the case of H4) copy
number of 2. Fig. Ib shows an autoradiograph from the same
membrane as in Fig. la, stripped and rehybridized with a ACTG
probe (24). Restriction fragment length polymorphisms are
seen, and the figure shows that DNA is indeed present in the
lanes in which DNA from lines without the IFNB gene was
electrophoresed. In addition, hybridization of probes for EGFR,
PDGFB, GLI, ERBAI, JUN, ROSI, ERCC2, XRCCI, and
KRAS2 to similarly prepared membranes (from which the IFNB
gene was also determined to be missing) showed that these nonIFN genes were indeed present in all the lines.
Fig. le shows data obtained with eight more malignant
glioma lines, GM3314 normally appearing fibroblasts, A875
melanoma, A427 lung carcinoma, and A431 epidermoid carci
noma. Two of the eight glioma lines show clear IFNB gene
alterations; in M007 the gene is absent and A1235 shows a
BamHl band of >40 kilobases, possibly due to a loss or a gain
of a large DNA segment in the area of the gene. The size of
this band is 40-50 kilobases, as judged by field inversion gel
electrophoresis using Aoligomers as markers (data not shown).
In membranes prepared with either EcoRl- or ft/I-restricted
genomic DNAs, the A1235 line showed normal length frag
ments (1.8 or 2.4 kilobases, respectively). But hybridization of
the IFNB probe to membranes prepared with Hindlll genomic
digests showed a band of 3.3 kilobases (Fig. Id), 6.7 kilobases
less than the 10-kilobase Hindlll fragments shown by the other
7F7V/?-containing lines and obtained previously by others (28,
29). In two membranes prepared by transfer of Bglll genomic
digests, A1235 DNA showed a 2.2-kilobase IFNB fragment, in
contrast to the normal 7.5-kilobase fragments observed in di
gests of the other lines than contained the gene (data not
shown). The simplest interpretation is that a piece of DNA has
been translocated near the IFNB gene control region without
interrupting the IFNB coding sequence (Fig. 2). A comparison
of our results with published data (28, 29) would place the
putative breakpoint between 0 and 1.3 kilobases 5' to the
regulatory regions controlling the expression of IFNB (30).
MO16 and the melanoma line A875 were each judged to have
one IFNB gene copy, by comparing densitometer tracings of an
autoradiograph from a membrane to which ///m/III-digested
genomic DNA was transferred and then hybridized both with
the IFNB probe and with a c-MOS probe (data not shown).
Two copies of the c-MOS gene were assumed to exist in the
DNA of each line. Beside the ACTG (Fig. le) and the c-MOS
genes, all lines in Fig. Id were shown to contain the EGFR,
ERBB2, GLI, PDGFB, JUN, ROSI, ERCC2, XRCCI, and
KRAS2 genes (data not shown).
If inactivation of the IFNB gene were an important factor in
tumor etiology, then the translocation near the control region

279

IFNA AND IFNB GENES AND IFN SENSITIVITY

Si^OOOOOOO

Fig. 1. IFNB and ACTG genes in human

malignant glioma lines, a. genomic DNAs
from 11 malignant glioma lines digested with
BamHt and hybridized with the Hindlll-Xholl
fragment of pSY2501 containing the coding
sequence of the IFNB gene, b, autoradiograph
from the same blot as in a but stripped by
placing the membrane in water (100'C but
removed from heat) to which, just prior to the
blot, had been added 0.1 % SDS and 0.1 x SSC
(final concentrations) and allowing it to cool
at room temperature (3-4 h with gentle shak
ing). The membrane was then hybridized with
the 2.1-kilobase BamHl fragment of pHFgA1 containing the 7-actin gene, c, genomic
DNAs from 8 malignant glioma lines digested
with BamHl and hybridized with the Hindlll-VA0II fragment of pSY250l containing the
coding sequence from the IFNB gene. d. as in
c but genomic DNAs were digested with
Hind\\\. e, same blot as in c but stripped (as
for A) and hybridized with the 2.1-kilobase
BamHl fragment of pHFgA-1 containing the
-. ;utin gene. Strain designations on the left
apply to a and b; designations on the right to
c-e.

3.3
[-actin,BamH i

*!tiffffftf!
[y-actin.BamH

-12.2
-10 2
- 8.1

l]

IFN-REGION
NORMAL
Ba

A-1235

Bg

E PBg

HBg

HE

E P Bg HE

Ba

a.

Fig. 2. Comparison of IFNB regions of normal and A1235 cells. Distance is

in kilobases. Ba, BamH\; H, Hindl\l: Bg, Bgl\l; E, EcoRl; and P, Pstl. The region
marked IFN-g is both the coding sequence of the IFNB gene and the region
detected by the pSY2S01 probe (18). RE refers to regions important in regulation
of the IFNB gene (29, 33). The numbering convention follows that in Ref. 29.

kb
2.4-

1.4-

of the IFNB gene in A1235 may have blocked its inducibility.

To determine whether this was so, electrophoresed RNAs from
[IFN-/3]
poly(I)-poly(C)-treated (100 ig/ml)and sham-treated A1235
(and five other lines serving as controls) were probed for IFNB4.4specific transcripts. Similar results were obtained in two exper
iments. Fig. 3a compares the poly(I)-poly(C) induction of the
IFNB gene of A1235 with that in other lines having zero, one,
2.4or two copies of the gene. The A1235 IFNB gene was as
inducible and made a mRNA of the same size as the other lines
that contain the gene. The absence of any /fW-specific mRNA
1.4in M007 and U87MG is consistent with the absence of the
IFNB gene from these lines. Reprobing the same membrane
with the 7-actin probe showed RNA to be present in the lanes
[7-actin]
in which no /fWA-specific transcripts were detected (Fig. 3>).
Fig. 3. Analysis of /F/VS-specific mRNA from poIy(I)-poly(C)-treated human
We note that poly(I)-poly(C) plus cycloheximide results in
malignant glioma cell lines. P, cell lines eretreated with 100 /ig/ml poly(I)/ICrG-specific sequences larger than the mature 2.4-kilobase
poly(C) plus 50 ig/mlcycloheximide for 1.5 h, followed by 2.5 h with cyclohex
size (31), possibly blocking mature 7-actin mRNA production
imide alone; C, cells were treated with cycloheximide for 4 h. RNAs were
extracted, electrophoresed, blotted and hybridized with the Il-'N'-.icoding sequence
in cells containing no IFNB gene, a finding we are pursuing
from pSY2501 (a) or hybridized with the ACTG gene probe (b).
further. In cells receiving no cycloheximide, only the 2.4-kilo
base 7-actin mRNA was observed, in agreement with others
treatment. Table 1 shows the lines in order of increasing sen
(31) (data not shown).
sitivity to IFN-0 treatment at 3000 units/ml, the doses of IFNWe wished to determine whether the absence of the IFNB
gene might be a predictor of cellular response to treatment with applied, and the calculated growth rates (generations/day).
IFN-/3. Cells were seeded at 50,000/plate, treated with IFNs,
Resistant lines (MO 10 and 118MG) show little (but doseand assayed for cell number on day 1, 3, 5, and 7 of IFNdependent) slowing of growth with increasing dose, whereas the

b.

280

IFNA AND IF.\B GENES AND IFN SENSITIVITY

Table 1 Sensitivity of malignant glioma cell lines to growth inhibition by IFK-rf

Growth rates in generations/day are followed in parentheses by the percentage
of the growth rate obtained in the absence of IFN-rf
rateLine

Growth

0MO

10000.216(77)

example Ref. 32). Both size classes map on chromosome 9 (33).

The pIFNa503 probe we used is for 1 of the 166 amino acid
IFNA genes (19). Membranes bearing 5 ng EcoRl-, Hindlll-,
or Psfl-treated genomic DNAs probed under less stringent
conditions showed 6, 7, and 10 bands, respectively, from 2.0 to
(79)0.527
1
>20 kilobases, which presumably represent other IFNA genes,
(69)0.559(55)0.266
all of which were absent from the lines that lack the IFNA gene
in Fig. 4 (data not shown). Thus the deletions in the region of
(50)0.583
(48)0.368
the IFNA genes from the eight lines are extensive.
(42)0.222
IFN-iv growth inhibition studies were carried out on eight of
(40)0.226
the lines, as for IFN-/3. The results are summarized in Table 2.
Again, no clear association of IFN-a growth rate sensitivity to
IFNA copy number was obtained.

0.280(100)118MG
10
(96)0.686(91)0.927
0.200(71)0.744(98)
0.758(100)A1235
0.650(86)0.944(93)
1.01(100)M002
(93)0.428(81)1.13(93)0.843
0.816(81)0.377(71)
0.527(100)H4
0.306(58)0.996(82)
(100)U87MG1.21
0.734(61)0.771(88)
0.877(100)M007
(96)0.366
0.603(69)0.307(34)
0.555(100)CLA
(66)0.480
0.229(41)0.404
0.627(100)M027
(76)0.353
(39)0.323(56)
(64) 0.247
(36)0.152(26)0.119(25)0.102(22)0.063(10)
0.574(100)P4
(62)0.212(46)0.316(69)0.385(61)300
0.210(37)0.193(42)
0.459(100)GRE
0.120(26)0.202(44)
0.455(100)SAN
0.157(34)0.248(39)
0.630(100)1000.268
0.133(21)30000.22
Concentration
(units/ml).A431of IFN-(i

DISCUSSION
In summary, 10 of 19 malignant glioma lines had genomic
alterations in the IFNA-IFNB region. Five lines had homozyIM002M006M007M010M012M016M027A1235A427A875IGM3314
gous deletions of both the IFNA and IFNB genes. Three lines
had homozygous deletions of the IFNA gene but retained a
copy of the IFNB gene. One of these lines had a rearrangement
of DNA near the IFNB gene. Two malignant glioma lines (and
the one melanoma line in this study) had one copy each of the
IFNA and IFNB genes. No alterations were observed in the nine
remaining malignant glioma lines, two normal fibroblast
strains, one lung tumor line, and one epidermoid carcinoma
line (see Table 3). Presence or absence of the genes was a
predictor neither of cellular resistance nor of sensitivity to
growth inhibition by either IFN.
Table 2 Sensitivity of malignant glioma cell lines to growth inhibition by IFN-a
Growth rates in generations/day are followed in parentheses by the percentage
of the growth rate obtained in the absence of IFN-

|i
i<M
CMcJ

rate3000.931

iM<N
i

LineT98
01T"COA172T98U87MGSANR1C118MGP4GRECLAM1LH4KD1i*i
o
T-.evi

Fig. 4. Absence of the IFNA gene from human brain tumor lines. Five fig of
genomic DNA were treated with BamHl. electrophoresed. blotted, and probed
with the 865-base pair fcoRI-ftfl fragment of pIFNa503.

more sensitive lines tend to show at least two components, an

initial rapid drop in growth rate followed by a more resistant
behavior.
To assay the lines for the presence of IFNA genes, genomic
DNAs were treated with either Pstl, BamHl, Hindlll, or EcoRl,
electrophoresed, transferred to membranes, and probed. All of
the 12 membranes analyzed showed the absence of the IFNA
gene that is detected by the EcoRl-Pstl fragment of pIFNa503.
Fig. 4 shows the IFNA gene to be absent from 8 of the 19
malignant glioma lines. We showed that DNA was indeed
present in the lanes by stripping and reprobing 5 of the mem
branes for the presence of other genes, namely JN(chromo
some 1), ERCC2 and XRCC1 (chromosome 19), ROSI (chro
mosome 6), and KRAS2 (chromosome 12; data not shown). All
lines that had no IFNB gene also lacked the IFNA gene; in
addition, P4, H4, and A1235, which have one copy of the IFNB
gene, lack the IFNA gene entirely. To determine IFNA gene
copy number in the lines, membranes similar to those that gave
rise to Fig. 4 were reprobed for the ERCC2 or the JUN genes,
respectively. By densitometric analysis of the autoradiographs
(data not shown) M016, RIC, and the melanoma line A875
were each judged to have one IFNA gene copy.
There are at least 23 human IFNA gene loci which encode
IFNs of two size classes, 166 and 172 amino acids (see for
281

(95)
MILA1235 0.419(100)
0.373 (89)
0.343
1.140(100)
1.088(95)
0.990
U87MG
1.041 (100) 0.777 (75)
0.681
A172
0.742 (100) 0.616(83)
0.520
H4
0.897(100)
0.925 (103) 0.593
CLA
0.508(100)
0.359(72)
0.304
0.544(100)1000.992
SAN0"1.046(100)
0.351 (65)Growth
0.284
" Concentration of IFN-a (units/ml).

(89)
(76)
(83)
(82) 0.280 (67) 0.282 (67)
(87) 0.771 (67) 0.703 (62)
(65) 0.640 (62) 0.624 (60)
(70) 0.486 (66) 0.419(56)
(66) 0.555 (62) 0.395 (44)
(60) 0.202 (40) 0.142(28)
(52)10000.799
0.251 (46)30000.871
0.134(25)

Table 3 Summary of malignant glioma cell lines in terms of growth inhibition by

Interferon, Mer phenotype, and number of IFN gene copies
ofLine

Copies
(%)Mpr

ph<-notype

geneM010118MGA
IFNA

rate in
hibition at 3000
units/ml

IFN-oND+
gene2012100221002212212Growth

ND38+
1235M002H4U87MGM007CLAM027P4GRESANM006M012M016A
ND+
5640+
ND72NDND+

ND+
75ND+
ND+
ND44+
172T98RICMIL2001000220002212212IFNB
17ND+

33IFN-ff21314550525860644074789
' ND. not determined.

IFNA AND IFNB GENES AND IFN SENSITIVITY

It is perhaps not entirely surprising, on the basis of previous

cytological evidence, that a high percentage of malignant glioma
cell lines shows the absence of one copy of the IFN genes. For
example, of the 49 low passage malignant gliomas presented in
Ref. 6, 15 had a deletion of part of, a missing, or a translocated
chromosome 9. Most breakage events affected the 9p arm, the
region of the IFNA and IFNB genes. While predictive of the
possibility that one copy of an IFN gene is missing, banding
evidence alone cannot distinguish with certainty whether the
genes are homozygously absent. One way to account for the
frequent loss of chromosome 9p material is to suppose that the
IFNA and IFNB genes act as tumor suppressor genes ( 10). To
view mechanisms by which IFNs may act in this capacity, it is
helpful to review what is known about regulation of the IFN
genes, particularly of the IFNB gene, since more is known about
its regulation than about the regulation of the IFNA genes.
Regulation is accomplished in part by a 77-base pair element
centered 270 nucleotides 5' from the beginning of the coding
sequence (34) (see Fig. 3). In addition, there is a 40-base pair
IFNB gene regulatory element (centered 195 nucleotides 5' to
the 5' terminus of the coding sequence) which contains a

EGFR, ERBB2, GLI, PDGFB, KRAS2, JUN, and ROSI genes,

whose role in the survival of the cell lines studied here is not
yet clear, is present in all the cell lines, (c) The chromosomal
location of the IFNA and IFNB genes is a potential site of loss
of DNA in cell lines produced from malignant glioma. The
chromosome 9p region is known to be one of two or three
chromosomal sites in short term cultures of malignant glioma
associated with an elevated number of cytogenetically detectable
breakpoints (6). The remainder of the genome is comparatively
free of such breakpoints and presumably less likely to undergo
allelic loss in malignant glioma, (d) The IFNA gene, which was
absent in 8 of the 19 malignant glioma lines studied here, was
present in all of 7 kidney and bladder tumor cell lines and 12
SV40-transformed human fibroblast lines6 [hybridization done

under conditions of high stringency (as for Fig. 4), apparently

detecting a single gene]. Thus is is unlikely that the cellculturing process alone generates biallelic loss of this gene. An
argument that cell culture may affect genetic constitution may
be made on the basis of data of James et al. (38). These workers
showed loss of heterozygosity in chromosomes 10, 13, 17, and
22 in noncultured biopsies from malignant glioma but did not
regulable enhancer with two positive regulatory domains and detect allelic loss in chromosome 9. The only chromosome 9
one negative regulatory domain to which regulatory proteins
probe used was D9S1. The region detected by D9S1 has been
may bind (29). Because induction by virus or by poly(I)-poly(C)
mapped by a cell fusion and molecular hybridization study to
of the IFNB gene is blocked by a protein kinase inhibitor, 2- chromosome 9 pter-qll (39) and more recently by restriction
aminopurine, it has been inferred that a protein kinase may be fragment length polymorphism linkage studies to a region
involved in activating IFN-/3 synthesis (35). Thus, one way to between GALT and ASSP3 (40). Thus the chromosome 9 site
envision IFNs as tumor suppressors is the following. Neoplastic
detected by D9S1 is likely to be in the piigli
region and more
transformation of a cell may involve activating an oncogene
proximal to the centromere than are the IFNA and IFNB genes.
that increases the activity of a (nuclear-acting) protein kinase
Thus the chromosomal sequence corresponding to D9S1 would
which, in turn, would induce IFN synthesis. Were this cell an be less likely than the FNgenes to be lost by a single breakpoint
IFN-sensitive one it would destroy itself, and tumorigenesis
deletion event, possibly explaining the lack of detection by
would not occur. Tumors would arise from those cells that
James et al. of losses of information in the chromosome 9p
either fail to respond to IFNs or fail to synthesize IFNs in region in human malignant glioma.
response to oncogene activation. One class of such tumor cells
The sensitivity of cell lines to growth inhibition by IFN-a
would be those from which the IFN genes have been deleted.
and -
observed here is similar to that observed previously (5).
In addition to lines from malignant glioma, lines from both
Sensitive lines show a rapid drop in growth rate between 0 and
acute lymphocytic leukemia and chronic myelogenous leukemia
300 units IFN-/3/ml, followed by a more resistant behavior at
show homozygous loss of the IFNA and IFNB genes (10). It is doses up to 3000 units/ml. Table 3 summarizes the lines in
interesting that 9p material should be absent from both solid order of increasing sensitivity to growth inhibition to IFN-/3
tumors and tumors of the blood. It may be that loss of the same
and shows the sensitivity to IFN-,the number of copies of the
tumor suppressor gene from chromosome 9p plays a major role IFN genes, and the Mer phenotype of the lines. Clearly there
in the etiology of the two tumor classes. The putative tumor
is no simple dependence of growth rate inhibition by IFNs upon
suppressor gene may be the IFNA genes, because, as suggested
the presence or absence of the gene. The lines most resistant to
(10), a large deletion would be much more likely than a point
IFN-,MO10 and 118MG, have two and zero copies of the
mutation to inactivate all of the IFNA genes. The putative
gene, respectively; in the line most sensitive to either IFN
tumor suppressor gene may be a gene other than the IFNB gene (SAN), both genes are entirely absent. The observation was
because the chromosomal event that occurred in A1235 did not made previously (5) that 10 of 10 Mer+ lines were resistant to
affect the poly(I)-poly(C) inducibility of the IFNB gene. (We do IFN-/3 while 6 of 8 Mer" lines were sensitive. Here we have
not yet know whether the IFN-/3 produced by A1235 is func
found two Mer+ lines to be sensitive, SAN and GRE. There
tionally active.)
may be an association between sensitivity to IFN-/3 and the Mer
Several lines of evidence suggest that the observed loss of the phenotype, but it is not so direct as previously believed (5). It
IFN genes is not due to random loss of cellular DNA. (a) The
can be expected that other genes, such as the IFN receptor
incidence of allelic loss is so high as to suggest specificity. Of genes on chromosome 21, or the IFN-inducible mRNAs (8, 9)
76 possible al-eles(19 cell lines x 2 genes/line x 2 al-eles/gene) play a role in determining cellular sensitivity to IFN-and -.
of the IFNA and IFNB genes, 33 al-eles(43%) are absent. (For The observation that poly(I)-poly(C) plus cycloheximide treat
the calculation we have taken the number of IFNA genes/cell
ment produces 7-actin-specific sequences larger than the ~2.4line as 1.) Random lack of 43% of cell al-eleswould likely be kilobase mature ACTG (7-actin) mRNA may be interpreted as
incompatible with survival, (b) Other genes are not absent.
either inhibition of mRNA processing or suppression of mRNA
Among the 38 IFN allelic pairs there were 13 cases of biallelic
termination. That the effect seems greater in cells in which
loss (8 of IFNA and 5 of IFNB). By contrast, at least one al-ele
chromosome 9p material (including the IFNB gene) is absent
of the ERCC2 and XRCC1 genes, whose DNA repair functions
suggests a possible function for IFN-/3; it may affect processing
are not essential for cell survival (36, 37), was present in all of
* J. Miyakoshi and R. S. Day III. unpublished observations.
these 19 cell lines. Similarly, at least one al-eleof the MOS,
282

IFNA AND IFNB GENES AND IFN SENSITIVITY

or termination of certain mRNAs in poly(I)-poly(C)-treated

cells. We did not observe the low level of induction of IFN-/3
mRNA by cycloheximide alone, as has been reported in some
cell lines by others (25, 41).

20.
21.

REFERENCES
22.
1. Cook, A. W., Carter, W. A., Nidzgorski, F., and Akhtar, L. Human brain
tumor-derived cell lines: growth rate reduced by human fibroblast interferon.
Science (Wash. DC), 219: 881-883. 1983.
2. Creasey, A. A., Eppstein, D. A., Marsh, Y. V., Khan. Z., and Merigan, T. C.
Growth regulation of melanoma cells by interferon and (2'-5') oligoadenylate
synthetase. Mol. Cell. Biol., 3: 780-786. 1983.
3. Creasey, A. A., Bartholomew. J. C., and Merigan, T. C. RoleofGo-Gl arrest
in the inhibition of tumor cell growth by interferon. Proc. Nati. Acad. Sci.
USA. 77: 1471-1475, 1980.
4. Gomi, K., Morimoto, M., and Nakamizo, N. Growth-inhibitory activity of
recombinant human interferon-/i against cultured human cells. Gann, 74:
737-742, 1983.
5. Yarosh, D. B.. Scudiere, D. A., Yagi, T., and Day, R. S., III. Human tumor
cell strains both unable to repair 06-methylguanine and hypersensitive to
killing by human a and interferons. Carcinogenesis (Lond.), 6: 883-886.
1985.
6. Bigner, S. H., Mark, J., Burger. R. C, Mahaley, M. S.. Jr., Bullard, D. E.,
Muhlbaier, L. H., and Bigner. D. D. Specific chromosomal abnormalities in
malignant human gliomas. Cancer Res., 48:405-411, 1988.
7. Pestka. S., Langer, J. A., Zoon, K. C., and Samuel, C. E. Interferons and
their actions. Annu. Rev. Biochem., 56: Ill-Ill,
1987.
8. Kelly, J. M., Gilbert, C S., Stark, G. R., and Kerr, I. M. Differential
regulation of interferon-induced mRNAs and c-myc mRNA by a- and yinterferons. Eur. J. Biochem., 153: 367-371, 1985.
9. Tiwari, R. K., Kusari, J., Kumar, R., and Sen, G. C. Gene induction by
interferons and double-stranded RNA: selective inhibition by 2-aminopurine.
Mol. Cell Biol., 8: 4289-4294, 1988.
10. Diaz. M. O., Zieman, S., Le Beau, M., Pitha, P., Smith. S. D.. Chilcote. R.
R., and Rowley, J. D. Homozygous deletion of the a- and rfl-interferon genes
in human leukemia and derived cell lines. Proc. Nati. Acad. Sci. USA, 85:
5259-5263. 1988.
11. Resnitzky, D., Yarden, A., Zipori. D., and Kimchi, A. Autocrine /i-related
interferon controls c-myc suppression and growth arrest during hematopoietic cell differentiation. Cell, 46: 31-40, 1986.
12. Day, R. S., Ill, and Ziolkowski, C. H. J. Human brain tumour cell strains
with deficient host-cell reactivation of A'-methyl-W-nitro-jV-nitrosoguanidine-damaged adenovirus 5. Nature (Lond.), 279: 797-799, 1980.
13. Giard, D. J., Aaronson, S. A., Todaro, G. J., Arnstein. P.. Kersey, J. H.,
Dosik, H.. and Parks, W. P. In vitro cultivation of human tumors: establish
ment of cell lines derived from a series of solid tumors. J. Nati. Cancer Inst..
SI: 1417-1423, 1973.
14. Day, R. S., Ill, Ziolkowski. C. H. J., Scudiere, D. A., Meyer, S. A.,
Lubiniecki, A. S., Girardi. A. J., Galloway, S. M., and Bynum, G. D.
Defective repair of alkylated DNA by human tumour and SV40-transformed
human cell strains. Nature (Lond.). 288: 724-727, 1980.
15. Sariban, E., Kohn, K. W., Zlotogorski, F., Laurent, G.. D'Incaici, M., Day,
R. S., Ill, Smith, B. H., Kornblith, P. L., and Erickson, L. E. DNA crosslinking responses of human malignant glioma cell strains to chloroethylnitrosoureas, cisplatin, and diaziquone. Cancer Res., 47: 3988-2994. 1987.
16. Day, R. S., III. Studies on repair of adenovirus 2 by human fibroblasts using
normal, xeroderma pigmentosum, and xeroderma pigmentosum heterozy
gous strains. Cancer Res., 34: 1965-1970, 1974.
17. Reed, K. C., and Mann, D. A. Rapid transfer of DNA from agarose gels to
nylon membranes. Nucleic Acids. Res., 13: 7207-7221, 1985.
18. Mark, D. F., Lu, S. D., Creasey, A. A., Yamamoto, R., and Lin. L. S. Sitespecific mutagenesis of the human fibroblast interferon gene. Proc. Nati.
Acad. Sei. USA, 81: 5662-5666. 1984.
19. Goeddel, D. V., Yelverton, E., Ullrich, A., Heyneker, H. L., Miozzari, G.,
Holmes, W.. Seeburg, P. H., Dull, T., May, L., Stebbing. N., Crea, R.,
Maeda, S.. McCandliss. R., Sloma. A.. Tabor. J. M., Gross, M.. Familletti.

283

23.
24.

25.
26.
27.
28.
29.

30.
31.

32.
33.
34.
35.
36.

37.
38.
39.
40.

41.

P. C., and Pestka, S. Human leukocyte interferon produced by E. coli is

biologically active. Nature (Lond.), 287:411-416, 1980.
Feinberg, A. P.. and Vogelstein, B. A technique for radiolabeling DNA
restriction endonuclease fragments to high specific activity. Anal. Biochem.,
132.-6-3,1983.
Scudiero. D. A., Meyer, S. A., Clatterbuck, B. E.. Mattern. M. R.. Ziolkowski,
C. H. J., and Day, R. S., III. Relationship of DNA repair phenotypes of
human fibroblast and tumor strains to killing by iV-methyl-A"-nitro-iV-nitrosoguanidine. Cancer Res., 44: 961-969, 1984.
Maniatis, T., Fritsch, E. F., and Sambrook, J. Molecular Cloning: A Labo
ratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory,
1982.
Marmur, J. A procedure for the isolation of deoxyribonucleic acid from
micro-organisms. J. Mol. Biol., 3: 208-218, 1961.
Gunning, P., Ponte. P., Okayama, H., Engel, J., Blau, H., and Kedes, L.
Isolation and characterization of full-length cDNA clones for human a, 13and y-actin mRNAs: skeletal but not cytoplasmic actins have an aminoterminal cysteine that is subsequently removed. Mol. Cell. Biol.. 3:787-795,
1983.
Enoch, T.. Zinn, K., and Maniatis, T. Activation of the human /<-interferon
gene requires an interferon-inducible factor. Mol. Cell. Biol.. 6: 801-810,
1986.
Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J.. and Ruttcr, W. J.
Isolation of biologically active ribonucleic acid from sources enriched in
ribonuclease. Biochemistry, 18: 5294-5299, 1979.
Fourney. R. M.. Miyakoshi. J., Day, R. S., Ill, and Paterson. M. C. Northern
blotting: efficient RNA staining and transfer. Focus, 10: 5-7. 1988.
Ohno, S.. and Taniguchi, T. Structure of a chromosomal gene for human
interferon .
Proc. Nati. Acad. Sci. USA, 78: 5305-5309, 1981.
Gross, G., Mayr, U., Bruns, W., Grosveld, F., Dahl, H-H. M., and Collins,
J. The structure of a thirty-six kilobase region of the human chromosome
including the fibroblast interferon gene IFN-/3. Nucleic Acids Res., 9: 24952507, 1981.
Keller, A. D., and Maniatis, T. Identification of an inducible factor that binds
to a positive regulatory element of the human /j-interferon gene. Proc. Nati.
Acad. Sci. USA. 85: 3309-3313, 1988.
Erba, H. P., Eddy, R., Shows. T., Kedes, L.. and Gunning, P. Structure,
chromosome location, and expression of the human 7-actin gene: differential
evolution, location, and expression of the cytoskeletal and 7-actin genes.
Mol. Cell. Biol.. 8: 1775-1789, 1988.
Pestka. S., Langer, J. A.. Zoon, K. C.. and Samuel, C. E. Interferons and
their actions. Annu. Rev. Biochem., 56:121-111, 1987.
Shows, T. B., Sakaguchi, A. Y., Naylor, S. L., Goeddel, D. V., and Lawn, R.
M. Clustering of leukocyte and fibroblast interferon genes on human chro
mosome 9. Science (Wash. DC), 218: 373-374, 1982.
Zinn, K., and Maniatis. T. Detection of factors that interact with the human
/i-interferon regulatory region in vivo by DNAase I footprinting. Cell, 45:
611-618, 1986.
Zinn. K.. Keller, A., Whittemore, L-A., and Maniatis, T. 2-Aminopurine
selectively inhibits the induction of iJ-interferon, c-fos, and c-myc gene expres
sion. Science (Wash. DC), 240: 210-213, 1988.
Weber. C. A., Salazar, E. P., Stewart, S. A., and Thompson, L. H. Molecular
cloning and biological characterization of a human gene, ERCC2, that
corrects the nucleotide excision repair defect in CHO UV5 cells. Mol. Cell.
Biol..: 1137-1146, 1988.
Siciliano, M. J., Carrano, A. V.. and Thompson, L. H. Assignment of a
human DNA-repair gene associated with sister-chromatid exchange to chro
mosome 19. Mutt.Res., 174: 303-308, 1986.
James, C. D., Carlbom, E., Dumanski. J. P., Hansen, M.. Nordenskjold, M.,
Collins, V. P., and Cavenee, W. K. Clonal genomic alterations in glioma
malignancy stages. Cancer Res., 48: 5546-5551, 1988.
Naylor, S. L., Sakaguchi, A. Y., Barker, D., White, R., and Shows, T. B.
DNA polymorphic loci mapped to human chromosomes 3, 5, 9, II, 17, 18
and 22. Proc. Nati. Acad. Sci. USA, 81: 2447-2451, 1984.
Lathrop, G. M., Leppert, M.. O'Connell, P., Nakamura, Y.. Cartwright. P..
Lalouel, J. M.. and White, R. A primary' genetic linkage map of chromosome
9. Cytogenet. Cell Genet., 46: 644. 1987.
Ringold.G. M., Dieckmann, B.. Vannice, J. L.,Trahey, M., and McCormick.
F. Inhibition of protein synthesis stimulates the transcription of human ftinterferon genes in Chinese hamster ovary cells. Proc. Nati. Acad. Sci. USA,
81: 3964-3968. 1984.