cleaved in cells undergoing apoptosis. fine mapping of 28s rrna

1995, 15(4):2051. Mol. Cell. Biol.

Gjertsen, M Lanotte and S O DøskelandG Houge, B Robaye, T S Eikhom, J Golstein, G Mellgren, B T cleaved in cells undergoing apoptosis.Fine mapping of 28S rRNA sites specifically

http://mcb.asm.org/content/15/4/2051Updated information and services can be found at:

These include:

CONTENT ALERTS more»cite this article),

Receive: RSS Feeds, eTOCs, free email alerts (when new articles

http://journals.asm.org/site/misc/reprints.xhtmlInformation about commercial reprint orders: http://journals.asm.org/site/subscriptions/To subscribe to to another ASM Journal go to:

on February 20, 2013 by P

EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/cgi/alerts

http://mcb.asm.org/

MOLECULAR AND CELLULAR BIOLOGY, Apr. 1995, p. 2051–2062 Vol. 15, No. 40270-7306/95/$04.0010Copyright q 1995, American Society for Microbiology

Fine Mapping of 28S rRNA Sites Specifically Cleaved in CellsUndergoing Apoptosis

G. HOUGE,1* B. ROBAYE,2 T. S. EIKHOM,3 J. GOLSTEIN,2 G. MELLGREN,1 B. T. GJERTSEN,1

M. LANOTTE,4 AND S. O. DØSKELAND1

Department of Anatomy and Cell Biology1 and Department of Biochemistry and Molecular Biology,3 University of Bergen,N-5009 Bergen, Norway; IRIBHN, Universite Libre de Bruxelles, Campus Erasme, B-1070 Brussels, Belgium2;

and INSERM Unite 301, Centre Hayem, Hopital St. Louis, F-75010 Paris, France4

Received 24 October 1994/Returned for modification 9 December 1994/Accepted 5 January 1995

Bona fide apoptosis in rat and human leukemia cells, rat thymocytes, and bovine endothelial cells wasaccompanied by limited and specific cleavage of polysome-associated and monosome-associated 28S rRNA,with 18S rRNA being spared. Specific 28S rRNA cleavage was observed in all instances of apoptotic deathaccompanied by internucleosomal DNA fragmentation, with cleavage of 28S rRNA and of DNA being linkedtemporally. This indicates that 28S rRNA fragmentation may be as general a feature of apoptosis as internu-cleosomal DNA fragmentation and that concerted specific cleavage of intra- and extranuclear polynucleotidesoccurs in apoptosis. Apoptosis-associated cleavage sites were mapped to the 28S rRNA divergent domains D2,D6 (endothelial cells), and D8. The D2 cuts occurred in hairpin loop junctions considered to be buried in theintact ribosome, suggesting that this rRNA region becomes a target for RNase attack in apoptotic cells. D8 wascleaved in two exposed UU(U) sequences in bulge loops. Treatment with agents causing necrotic cell death oraging of cell lysates failed to produce any detectable limited D2 cleavage but did produce a more generalizedcleavage in the D8 region. Of potential functional interest was the finding that the primary cuts in D2 exactlyflanked a 0.3-kb hypervariable subdomain (D2c), allowing excision of the latter. The implication of hypervari-able rRNA domains in apoptosis represents the first association of any functional process with these enigmaticparts of the ribosomes.

Apoptosis is a suicide-like process that requires the activeparticipation of the dying cell and occurs physiologically duringdevelopment and tissue remodelling (45). Demands for apop-tosis would be expected to increase during phylogeny as or-ganisms become more complex and long-lived, but apoptosisalso occurs in primitive organisms like Caenorhabditis elegans(11). Morphological hallmarks of cellular apoptosis are chro-matin condensation, cell shrinkage accompanied by organellerearrangements, and surface blebbings leading to cell fragmen-tation (5, 54). Considerable efforts have led to the pinpointingof gene products induced during preapoptosis (37, 45), but thebasic machinery required to execute apoptosis is present inmost cells and does not require gene transcription (2, 9, 39). Infact, the first recognized and still the most consistently foundbiochemical aberration in mammalian cell apoptosis is inter-nucleosomal DNA fragmentation (48), which also can occur inisolated nuclei from normal cells (34). The present study fo-cuses on the limited degradation of another abundant polynu-cleotide (28S rRNA), which recently has been reported to bedegraded in cyclic AMP (cAMP)-induced apoptosis of a ratpromyelocytic leukemia cell line (21) and in radiation-induceddeath of human lymphocytes (10).One major aim of the present study was to find whether 28S

rRNA fragmentation occurred generally in apoptotic systems,whether it was dependent on new protein synthesis, andwhether it was correlated with a particular apoptotic inducer,with a particular morphology, and with internucleosomal DNAfragmentation. For this purpose, 28S rRNA fragmentation wasstudied in rat thymocytes treated with glucocorticoids (53) or

with the phosphatase inhibitor okadaic acid, which is known tobe a general inducer of apoptosis (2); in human promyelocyticNB4 cells treated with okadaic acid (23); in primary bovineendothelial cells treated with tumor necrosis factor (TNF),cycloheximide (42), or okadaic acid; and in rat myeloid IPC-81cells treated with cAMP analogs (27), the phosphatase inhib-itor okadaic acid or calyculin A, or cycloheximide or exposedto a cycle of cooling and heating. The specificity of the apop-totic rRNA fragmentation was ensured by studying RNA frag-mentation patterns in models of necrotic cell death (inducedby sodium azide, isopropanol, or hypotonic lysis). It was foundthat specific 28S rRNA fragmentation occurred in one or moretypes of apoptosis in all of the cell systems mentioned and wasalways closely linked to internucleosomal DNA fragmentation.28S rRNA fragmentation also correlated with limited proteindegradation but not with any particular morphological subtypeof apoptosis.As a first step towards understanding the mechanism of the

apoptosis-associated cuts in 28S rRNA, it was established thatthey occurred in polysome-associated ribosomes as well as inmonosomes. A second step was to probe (by sequencing andextensive hybridizations) the specificity of the cleavage reac-tions in 28S rRNA from rat leukemia cells treated with cAMP.Two closely spaced cuts occurred in the 39 region (between U’s3257 and 3258 and at U’s 3288 to 3290 in the D8 domain), andtwo occurred in the 59 region (one at about base 890 and onebetween A-1192 and U-1193 in the D2 domain). 28S rRNAconsists of highly conserved domains as well as so-called di-vergent domains (D1 to D12). The latter represent RNA whichhas expanded in both length and diversity during eukaryoticevolution, constituting nearly half of the 28S rRNA in highereukaryotes (17). D domains have no known function (17),being apparently dispensable for protein synthesis (40). How-ever, a function is suggested by the fact that certain subdo-

* Corresponding author. Mailing address: Department of Anatomyand Cell Biology, Årstadveien 19, N-5009 Bergen, Norway. Phone: 4755 20 63 75. Fax: 47 55 20 63 60. Electronic mail address: [email protected].

2051


EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

mains (particularly D2c and D8b) diverge faster during phy-logeny than expected from random mutations (16), i.e., that anevolutionary pressure for variation within given frames forsecondary structure may exist. Interestingly, the apoptosis-as-sociated cuts in rat and human leukemia cells and rat thymo-cytes were mapped to subdomains D2c and D8b.To determine whether the apoptosis-associated cleavages

were due only to general RNase activation, we studied whethercell extracts from apoptotic cells could induce apoptosis-spe-cific cuts in normal ribosomes. Neither such extracts nor se-lected added RNases could reproduce the apoptosis-specificcuts in D2, both of which occur in regions recently shown to beinaccessible to chemical modification in the isolated ribosome(18, 20). These findings indicate the existence of apoptosis-associated changes of ribosome conformation. The significanceof concerted cleavage of rRNA and DNA in certain types ofapoptosis is discussed here in relation to a possible role indefending the organism against cells harboring potentiallyharmful DNA or RNA molecules.

MATERIALS AND METHODS

Model systems for apoptosis. Four different model systems for apoptosis wereinvestigated. The major model system was the IPC-81 rat myeloid leukemia cellline, which was carefully cultured at densities of around 6 3 105 cells per ml inDulbecco modified Eagle medium or RPMI supplemented with 8% horse serum(Gibco, Grand Island, N.Y.) (25). Fulminant apoptotic cell death was observedapproximately 3 h after addition of 0.2 mM 8-(4-chlorophenylthio)-cAMP (8-CPT-cAMP) (Sigma, St. Louis, Mo.) to the cultures (27). For comparison, agentssuch as cycloheximide, actinomycin D, 7-deaza-adenosine (tubercidine), and thephosphatase inhibitor calyculin A were used as cell death inducers. Cells werealso killed by heat shock or cold shock or by generally toxic agents such asisopropanol, sodium azide, and hydrogen peroxide. Cell death of the humanpromyelocytic NB4 leukemia cell line was also investigated (26). The cells be-came apoptotic approximately 6 h after addition of a 316 nM concentration ofthe phosphatase inhibitor okadaic acid (Calbiochem, La Jolla, Calif.) to thecultures. The third model system was rat thymocytes, prepared from male Spra-gue-Dawley rats as described previously (46). In this case cell death was inducedby adding 10 mM 6a-methylprednisolone 21-hemisuccinate (Sigma) to thymocyteprimary cultures (53). The fourth model system was primary bovine aortic en-dothelial cells, which were prepared and cultured as described previously (42).Apoptosis was usually induced after addition of recombinant human TNF (30ng/ml) combined with cycloheximide (2 mg/ml) to the cell cultures and could beonly partly inhibited by the protease inhibitor calpain inhibitor I (25 mM)(Boehringer). In addition, cell death was induced in all of these in vitro modelsystems by adding 1 mM (or less) phosphatase inhibitor okadaic acid to the

cultures. Okadaic acid, given at optimal concentrations, can induce apoptoticchanges in most cell types (2).During experiments, the morphology of the cells was assessed by ordinary light

microscopy (Zeiss Axiovert microscope). As previously described (2), cells werealso processed for electron microscopy after being mixed with 9 volumes ofprewarmed (378C) glutaraldehyde (1.5%, vol/vol) in 0.1 M Na cacodylate buffer(pH 7.4).Preparation and analysis of RNA and DNA. Total RNA was prepared from

cell cultures by the acid guanidinium thiocyanate (GTC)-phenol-chloroformextraction method (4). Cells were immediately lysed in 25 mM sodium citratebuffer (pH 7) containing 4 M GTC, 0.5% (wt/vol) sodium lauroyl sarcosine, and1% 2-mercaptoethanol. This was done either by dissolving cell pellets in GTCsolution (for IPC-81 cells, NB4 cells, and thymocytes) or by lysing cell monolay-ers directly in GTC solution (for endothelial cells). Denaturing gel electrophore-sis of RNA and capillary blotting of RNA onto nylon membranes (Northernblotting) were done as previously described (22). DNA was prepared and elec-trophoresed by standard procedures (43).Oligodeoxynucleotides complementary to various regions of rat 28S rRNA

were 59 end labeled with 32P by using polynucleotide kinase (New EnglandBiolabs) and [g-32P]ATP (43) and used as probes for Northern blot hybridiza-tions. The oligodeoxynucleotides used and their sequences are given in Table 1.The RNA blots were prehybridized and hybridized at 608C in 0.5 M sodiumphosphate buffer (pH 7.2) containing 7% sodium dodecyl sulfate, 1 mM EDTA,and 0.1 mg of sheared and denatured herring sperm DNA per ml. The filterswere washed once in 23 SSC (13 SSC is 0.15 M NaCl plus 0.015 M sodiumcitrate) at room temperature (200 ml per filter) and then at high stringency in0.23 SSC at 508C for 30 min (200 ml per filter). Autoradiography was usuallydone without intensifying screens at room temperature.For semiquantitative evaluation of band intensities, negative photographs of

the ethidium-bromide stained gels (with Polaroid 665 positive/negative film)were scanned with an LKB UltroScan laser densitometer. For calculation of thedegree of 28S rRNA fragmentation, the amount of intact 28S rRNA relative tothat of the 18S rRNA was measured. The extent of DNA fragmentation wasmeasured as described previously (15), and the density plot for each lane wasintegrated with an MOP-AM03 integrator (Kontron Messgerate, Munich, Ger-many). For bovine endothelial cells the extent of DNA fragmentation was alsomeasured as the nonprecipitable fraction of labeled DNA, as previously de-scribed (42).Preparation of cell lysates and ribosomal fractions. IPC-81 cells (83 107 to 10

3 107) were spun down, washed twice in ice-cold phosphate-buffered saline, andresuspended in 1 ml of ice-cold lysis buffer (15 mM HEPES [N-2-hydroxyeth-ylpiperazine-N9-2-ethanesulfonic acid] [pH 7.6], 2.3 mM MgCl2, 120 mM KCl).The cell suspension was passed 20 times through an ice-cold cell homogenizer (ametal cylinder with an internal diameter of 8.020 mm containing a metal beadwith a diameter of 8.006 mm). The homogenate was spun at 6,600 3 g for 5 minat 48C, the supernatant was spun at 25,000 3 g for 10 min at 48C, and the finalsupernatant was aliquoted and either used immediately or frozen in liquid ni-trogen. This lysate supported in vitro protein synthesis with endogenous mRNAas the template. When the lysate was supplied with energy (0.5 mMATP, 0.1 mMGTP, 7 mM creatine phosphate, and 1 mg of creatine kinase [Boehringer] per

TABLE 1. Oligodeoxynucleotide probes used for the investigation of 28S rRNA fragmentation

Probe name Oligonucleotide sequencePosition inrat 28SrRNA

28 rRNA fragment recognitiona

59D2c 39D2b 39D2c 59D8 39D8

5928S 59-GACTAATATGCTTAAATTCAGCGGGTCGCCACGTC-39 16–51 1 159D2 59-TACCTCTTAACGGTTTCACGCCCTCTTGAACTCTC-39 397–432 1 1D2c-59 59-AGCGCGGCGACGGGTATCCGGCTCCCTCGG-39 880–910 1 1 1D2c-middle 59-ACGGGGGGTGGGAGAGCGGT-39 1001–1021 1 1D2-39int 59-ACATCGCCGCCGACCCCGTGCGCTCGGCTTC-39 1166–1196 1 139D2 59-CACGCGTTAGACTCCTTGGTCCGTGTTTCAAGACG-39 1213–1248 1 1 159D3 59-TCACCTTCATTGCGCCACGGCGGCTTTCGGAC-39 1264–1296 1 1 139D3 59-CTAACGCGTACGCTCGTGCTCCACCTCCCCG-39 1386–1417 1 1 159D6 59-GCGACGGCCGGGTATGGGCCCGACGCTCCA-39 1950–1980 1 1 139D6 59-AGGCTCACCGCAGCGGCCCTCCTACTCGTCG-39 2114–2145 1 1 159D8 59-TTAGAGCCAATCCTTATCCCGAAGTTACGGATCCG-39 2699–2734 1 1 1D8-59int 59-CAGCCCGACCGACCCAGC-39 2736–2754 1 1 1D8-middle 59-GCCGCAGCTGGGGCGATCCAC-39 3107–3128 1 1 1D8b-extr. 59-AACCCCCGACCCCGCGGAG-39 3206–3225 1 1 1D8-39int 59-TTCTAAGTCGGCTGCTAGGCGCCGGCCGAG-39 3306–3336 1 1 139D8 59-TAATTAAACAGTCGGATTCCCCTAGTCCGCACCAG-39 3336–3371 1 1 1D12-39int 59-CAGAAGCAGGTCGTCTACGAATGGTTTAGCGCCAA-39 4607–4642 1 1 13928S 59-ACAAACCCTTGTGTCGAGGGCTGA-39 4695–4719 1 1 1

a 1, fragment recognized in IPC-81 cells (also see Fig. 7).

2052 HOUGE ET AL. MOL. CELL. BIOL.


EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

ml), the protein synthesis proceeded at a linear rate for up to 60 min at 308C(data not shown).Polysomes were isolated as follows Cellular lysate (0.7 ml, corresponding to

;5 3 107 cells) was overlaid on a 12-ml linear 12 to 40% sucrose gradient in 20mM Tris Cl (pH 7.5)–120 mM KCl–7.5 mM MgSO4. After centrifugation in aBeckman SW40 rotor (39,000 rpm for 75 min at 48C), the gradients were har-vested from the bottom by using a peristaltic pump, and the A260 was monitoredwith a 10-mm flowthrough cuvette. Polysomes and monosomes were pooledseparately. Oligopolysomes containing fewer than three 80S subunits were notincluded in the polysomal fraction used for RNA preparation in order to avoidmonosomal contamination.Primer extension analysis of rRNA. 28S rRNA was sequenced directly by using

avian myeloblastosis virus reverse transcriptase and oligodeoxynucleotide prim-ers on the basis of established procedures (1, 30). Briefly, 6 mg of RNA wasannealed with 2 pmol of 59-32P-labeled or, alternatively, 12 pmol of unlabeledprimer in 30 mM Tris Cl (pH 7.5)–50 mM KCl in a total volume of 10 ml.Annealing was done in a PCR machine (MiniCycler; MJ Research, Watertown,Mass.) with the following temperatures: 958C for 2 min, 568C for 1 min, and thena gradual decrease of 18C/45 s to 408C. At this point, 4.4 ml of 53 avianmyeloblastosis virus reverse transcriptase buffer (Promega), followed by 3.6 ml ofa-35S-dCTP (in the case of cold primers) and 1 ml (23 U) of avian myeloblastosisvirus reverse transcriptase (Promega), was added. The final volume was adjustedto 22 ml with H2O. After a brief centrifugation, 4 ml of this mixture was added tofive prewarmed (408C) deoxy/dideoxy sequencing mix tubes containing 4 ml of Amix, C mix, G mix, T mix, and N mix, respectively. These mixtures were made in13 reverse transcriptase buffer (62.5 mM Tris Cl [pH 8.4], 12.5 mM MgCl2, 2.5mM dithiothreitol) and contained a 10 mMddNTP–50 mMdNTP mixture (whereN is A, C, G, or T) and the other nucleotides at 200 mM each. The N mix waswithout dideoxynucleotides. When the 39 end of D2 was sequenced, the primer39D2 or 59D3 was used (Table 1), and when the 39 end of D8 was sequenced, theprimer 59TAGTCCGCACCAGTTCTAA39 was used. These primers correspondto the conserved 39 flanking regions of D2 and D8, respectively. The products ofthe sequencing reactions were separated in a 0.2-mm 6 or 10% polyacrylamide–urea gel by using Pharmacia’s Macrophor sequencing system.S1 nuclease protection assay. Total RNA (5 mg) from control IPC-81 cells or

apoptotic IPC-81 cells was annealed with 1 pmol of each 59-32P-labeled oligode-oxynucleotide (D2c-59 and D8-59int [Table 1]) in 13 S1 hybridization solution(166 mM HEPES [pH 7.5], 1M NaCl, 0.33 mM EDTA), with the total volumebeing 30 ml. The annealing was done at 658C overnight after the mixture wasbriefly warmed to 908C (2 min). Thereafter, 270 ml of S1 nuclease mixture (50

mM Na-acetate [pH 4.5], 280 mM NaCl, 4.5 mM ZnSO4, 17 ng of denaturedherring sperm DNA per ml, and 0.67 U of S1 nuclease per ml) (Boehringer) wasadded to the reaction mixtures. After incubation at 378C for 45 min, the pro-tected DNA-RNA hybrids were precipitated on dry ice by adding 3 ml of 0.5 MEDTA (pH 8), 1 ml of 10-mg/ml tRNA (Sigma), and 700 ml of ethanol to thereaction mixtures. After centrifugation (50,000 3 g for 15 min at 48C), theradioactive pellet was washed once with ice-cold 70% ethanol, dried, and dis-solved at 08C in 10 ml of 100 mM NaOH. A sample aliquot (3 ml) was mixed withan equal volume of loading buffer, denatured at 908C, and immediately loadedon a 10% polyacrylamide–urea sequencing gel.Induction of rRNA degradation by endogenous or exogenous RNase activity.

Degradation of IPC-81 cell RNA was induced by three different approaches.First, pellets of IPC-81 cells were frozen in liquid nitrogen, thawed, and incu-bated at 378C for up to 150 min. This was done to obtain membrane dysfunctionreminiscent of cellular necrosis. Second, an IPC-81 cell lysate, prepared asindicated above, was incubated with or without added RNase T1, Bacillus cereusRNase, or RNase U2 (all at 0.5 U/100 ml of lysate and obtained from Pharmacia,Uppsala, Sweden) for various periods of time. In some cases 1 mM ATP, 7 mMcreatine phosphate, and 1 mg of creatine kinase per ml were added to the lysatesas energy suppliers. Third, a cell lysate was obtained by quickly forcing the cellsfrom a high-pressure chamber equilibrated with nitrogen at 600 lb/in2 to atmo-spheric pressure. Preparation of a high-speed supernatant from this lysate wasdone as described above. The lysate prepared by this last procedure had a higherintrinsic RNase activity than the lysate prepared by the other procedures (prob-ably because of some lysosomal breakage) and was unable to support proteinsynthesis at a linear rate for more than 10 to 15 min at 308C.

RESULTS

Specific 28S rRNA cleavage was closely correlated with ap-optotic internucleosomal DNA fragmentation irrespective ofcell type and apoptosis-inducing agent. Total RNA was pre-pared from a human promyelocytic leukemia cell line (NB4)treated with the protein phosphatase inhibitor okadaic acid,which recently has been shown to induce apoptotic cell deathwith internucleosomal DNA fragmentation in this cell type(23). Upon denaturing RNA gel electrophoresis, the band of

FIG. 1. Apoptotic 28S rRNA fragmentation in human promyelocytic leukemia cells and rat thymocytes. Gels and autoradiographs of a Northern blot of total RNAisolated from NB4 cells treated with okadaic acid (left panel) and autoradiographs of a Northern blot of total RNA isolated from thymocytes treated with prednisolone(right panel) are shown. In the middle panel a blot of fragmented 28S rRNA from rat promyelocytic leukemia (IPC-81) cells is shown for comparison. The Northernblots were hybridized with various 28S rRNA oligonucleotides, as indicated below the lanes. The duration of the treatment is shown above each lane. The arrowsindicate the positions of the major 28S rRNA fragmentation bands.

VOL. 15, 1995 28S rRNA FRAGMENTATION IN APOPTOTIC CELL DEATH 2053


EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

intact 28S rRNA was diminished and novel distinct bandshybridizing with 28S rRNA probes appeared (Fig. 1, left pan-el). In contrast, 18S rRNA appeared to be intact, with no smallbands hybridizing with 18S rRNA probes (data not shown).Rat thymocytes represent another well-recognized model sys-tem for apoptosis, and internucleosomal DNA fragmentationwas first described as a feature of apoptosis in such cellstreated with the glucocorticoid prednisolone (54). Limited 28SrRNA fragmentation similar to that in NB4 cells was observedin rat thymocytes treated with either prednisolone or okadaicacid (Fig. 1). In both experimental systems the onset of 28SrRNA fragmentation coincided with the onset of internucleo-somal DNA fragmentation and morphological indices of ap-optosis (not shown). In order to compare more closely thefragments in rat thymocytes and human leukemia cells, theRNA filters were hybridized with a number of probes (Table 1)spanning most of the rat 28S rRNA sequence and (with theexception of the D2c-middle and D8b-extr. probes) represent-ing regions with a reasonably high level of homology betweenrats and humans. Selected hybridizations are shown for NB4cells (Fig. 1, left panel), thymocytes (Fig. 1, right panel), andpolysomal rRNA from a rat leukemia cell line (IPC-81) in-duced by cAMP to undergo apoptosis (Fig. 2, right panel). Thefragments in rat thymocytes and IPC-81 cells appeared to beidentical with respect to probe specificity and size and verysimilar to those in NB4 cells, with the only discernible differ-ences being that the major band from NB4 cells detected byprobes directed against the 59 part of the 28S molecule mi-grated slightly faster in denaturing gels than the corresponding

FIG. 2. Gels (left panel) and autoradiographs (right panel) showing apop-totic 28S rRNA cleavage in polysomal rRNA from IPC-81 cells. The times ofincubation with 8-CPT-cAMP (left) and the probes used for hybridization (right)are indicated above the lanes. The autoradiographs shown are from the ‘‘4-h withcAMP’’ polysomal RNA. On the right the positions of RNA molecular lengthmarkers are indicated.

FIG. 3. Apoptotic 28S rRNA fragmentation in bovine endothelial cells. Gels (left panel) and autoradiographs of a Northern blot (right panel) of total RNAprepared from endothelial cells treated with TNF and cycloheximide for various times (indicated above each lane) are shown. The oligonucleotide probe used forhybridization is indicated below each lane (right panel), and the oligonucleotide pairs chosen were those corresponding to the conserved 28S rRNA regions flankingD2, D6, and D8. Major 28S rRNA fragments are marked.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

band from the apoptotic rat cells (Fig. 1). It is concluded thatthe apoptosis-associated rRNA cleavage was remarkably sim-ilar in all three apoptotic cell systems, suggesting that it was nota random occurrence.We have reported that the polysome chain termination in-

hibitor cycloheximide protected against cAMP-induced rRNAfragmentation in IPC-81 cells (21), raising the possibility thatrRNA in polysomes was protected against fragmentation orthat protein synthesis was required for the cleavage. One wayto test such a hypothesis was to compare the 28S rRNA frag-mentation in polysome-derived and monosome-derived RNAs(Fig. 2). Bovine endothelial cells offered another possibility totest the hypothesis, since the TNF-induced apoptosis isstrongly enhanced by cycloheximide in such cells (41, 42). Poly-some-associated 28S rRNA from cAMP-treated IPC-81 cellsdid show characteristic fragmentation, albeit to a lesser extentthan rRNA from monosomes (Fig. 2), suggesting that evenpolysome-associated rRNA was susceptible to apoptosis-asso-ciated cleavage. Furthermore, 28S rRNA became extensivelyfragmented in endothelial cells undergoing apoptosis in re-sponse to TNF and cycloheximide (Fig. 3). This indicates thatprotein synthesis is not required for rRNA cleavage. Also, inendothelial cells the cleavage was limited and discrete rRNAbands were formed. The fragments formed after cleavage inthe D8 domain corresponded to those found in the other cellsystems studied, but cleavage of the D6 domain was unique tothe bovine endothelial cells. Since bovine 28S rRNA has notbeen sequenced, an accurate probing of cleavage sites could

not be performed, but it is of interest that the D2, D6, and D8domains all have particularly variable nucleotide sequences(16, 35).Having demonstrated 28S-specific, limited rRNA fragmen-

tation in all apoptotic systems studied, we considered it impor-tant to know in detail the temporal correlation between theappearance of apoptotic morphology, internucleosomal DNAfragmentation, and 28S rRNA fragmentation. For this, cAMP-treated IPC-81 cells and TNF- and cycloheximide-treated en-dothelial cells were chosen because these cells undergo partic-ularly severe forms of apoptosis with easily scored cellfragmentation and extensive DNA fragmentation (27, 42). Theamount of intact 28S rRNA was estimated from laser scans ofethidium bromide-stained gels and related to the amount of18S rRNA, which remained stable. There was a close correla-

FIG. 4. Linear (A) and semilogarithmic (B) plots of the degrees of DNA(squares), 28S rRNA (circles), and cellular (triangles) fragmentation in 8-CPT-cAMP-treated IPC-81 cells versus time. The semilogarithmic plot was made tocalculate the starting points of the various fragmentations more exactly. The barsindicate standard errors of the mean calculated from results derived from threeto six different experiments (average, n 5 4).

FIG. 5. Correlation between DNA fragmentation (top panel) and rRNAfragmentation (middle panel) in bovine endothelial cell primary cultures treatedwith TNF and cycloheximide for various times as indicated. Total DNA wasseparated on a 1.5% agarose gel that was stained with ethidium bromide (toppanel), or total RNA was separated on a denaturing 1.5% formaldehyde-agarosegel, blotted, and hybridized with the 3928S probe (middle panel). The RNA wasprepared from the same cells as the DNA, but unfortunately the RNA yieldsfrom these cells tended to be low, resulting in somewhat unequal loading of thelanes. Note the lowermost band arising from a secondary cut 39 to the D8 region.The extents of DNA (squares), 28S rRNA (circles), and cellular (triangles)fragmentation were estimated in two parallel experiments, and the mean valuesare plotted versus time in the bottom panel.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

tion between fragmentations of the cells, 28S rRNA, and DNA(Fig. 4 and 5), all of which obeyed first-order kinetics, albeitwith different rate constants, in the rat leukemia system (Fig.4B). The first morphologically apoptotic leukemia cells startedto appear 15 to 20 min before the commencement of DNA and28S rRNA fragmentations (Fig. 4B). It is concluded thatpolynucleotide fragmentations were parts of the execution ofthe apoptotic process rather than activators of apoptosis.IPC-81 rat leukemia cells were found to undergo apoptosis

with widely different morphologies and variable degrees ofinternucleosomal DNA cleavage in response to okadaic acid,cycloheximide, 7-deaza-adenosine, and transient cooling. Asshown in Table 2, okadaic acid and 7-deaza-adenosine causeda rather inconspicious apoptotic morphology with little cellfragmentation but considerable DNA fragmentation, whereastransient cooling caused extensive cell fragmentation but lessDNA fragmentation. In all of these cases the extents of frag-mentation of 28S rRNA and of DNA were correlated. Suchcorrelation was also observed when various agents were usedto induce apoptosis in a human leukemia cell line and inbovine endothelial cells. On the other hand, there was lesscorrelation between the extent of apoptotic morphologicalchanges and polynucleotide fragmentation (Table 2). This con-clusion was further supported by experiments using the pro-tease inhibitor calpain inhibitor I, which caused only a moder-ate decrease in the number of morphologically apoptotic cellsbut efficiently inhibited both DNA and rRNA fragmentation inTNF- and cycloheximide-treated endothelial cells (Table 2).A final test of the genuine apoptotic nature of the observed

28S rRNA fragmentation was to see if it occurred in nonapop-totic cell death, which was induced in IPC-81 cells by 0.1%hydrogen peroxide, 2% sodium azide, or 50 nM calyculin A. Innone of these cases was the cleavage pattern typical of apop-tosis observed (Table 2).Fine mapping of the D8 and D2 cleavage sites of 28S rRNA

in cAMP-induced IPC-81 apoptosis. In order to start exploringthe mechanisms underlying cleavage of D2 and D8, the posi-

tions of the cleavage sites had to be determined more exactly.To this end, three approaches were used. First, careful rRNAfragments size determinations were done with IPC-81 rRNA(Fig. 2), and the fragments were probed with oligonucleotidespresumed to be in the vicinity of the cuts (Table 1). Second, the39 cleavage sites in D2 and D8 were determined by primerextension analysis with total RNA from apoptotic IPC-81 cells.Third, S1 nuclease protection was tried to determine the 59cleavage site in D2 more exactly.The 28S rRNA fragment sizes were determined in conven-

tional 1.5% denaturing agarose gels containing 6.7% formal-dehyde (Fig. 2) and in 2.6% polyacrylamide–7 M urea gels (notshown), using as standards rRNA (23S and 16S rRNAs fromEscherichia coli) and non-rRNA molecular weight markers.There was a general tendency of rRNA to migrate faster thanexpected from the molecular length, probably because of therRNA secondary structure being partially resistant to completedenaturation in 7 M urea or 6.7% formaldehyde. This wasparticularly evident when the migration of 23S rRNA (2.9 kb)was compared with an RNA ladder made from SP6/T7 poly-merase on linearized plasmids (the positions of these markersare indicated in Fig. 2).For a more precise estimation of fragments smaller than 18S

rRNA, vertical polyacrylamide-urea gels were run (data notshown). The sharp 28S rRNA-derived bands that could be seenon these gels indicated that the apoptosis-related cuts in 28SrRNA were not followed by random secondary exonucleasedegradation. It should be noted that the 1.5-kb 39D8 band (Fig.2) formed a finely separated doublet on the polyacrylamide-urea gel. It thus appeared that D8 could be cut at two closelyspaced locations separated by a maximum of 50 bases.For the exact determination of these cleavage sites, primer

extension analysis was performed with normal rRNA and frag-mented rRNA having a mixture of primary D8 cuts and pri-mary D2 cuts (Fig. 6). The most 39 cut in D8 was in a UUU-containing bulge loop in hairpin loop D8b (Fig. 7C). The slightsize heterogeneity of the 39 fragment (Fig. 6) indicated either

TABLE 2. Comparison of internucleosomal DNA fragmentation and 28S rRNA fragmentation during cell death

Cell type and treatment Death typeaFragmentationb

Cellular DNA 28S rRNA

IPC-81 cells200 mM 8-CPT-cAMP for 8 h ACD 1111 1111 111150 mM 7-deaza-adenosine for 8 h ACD (1) 11 1130 mg of cycloheximide per ml for 8 h ACD 11 11 1130 mg of cycloheximide per ml for 21 h ACD 11 111 1111,000 nM okadaic acid for 8 h ACD 1 111 1108C for 2 h, thereafter 378C for 1 h ACD 111 11 11c

50 nM calyculin A for 8 h pNCD 1 2 2458C for 6 h, thereafter 378C for 4 h NCD 2 2 20.1% (29 mM) H2O2 for 2 h NCD 2 2(smear) (1)c

2% (300 mM) NaN3 for 5 h NCD 2 2 (1)c

Endothelial cells250 nM okadaic acid for 8 h ACD 11 11 130 ng of TNF per ml for 6 h ACD (1) (1) (1)30 ng of TNF per ml 1 2 mg of cycloheximide per ml for 3 h ACD 11 11 1TNF 1 cycloheximide 1 25 mM calpain inhibitor I for 3 h ACD 1(1) (1) (1)

NB4 cells100 nM okadaic acid for 12 h ACD (1) 1 (1)316 nM okadaic acid for 12 h ACD 1 1 1100 nM okadaic acid for 24 h ACD 1 11 11

a ACD, apoptotic cell death; NCD, necrotic cell death; pNCD, pseudo-NCD (see reference 15 for a detailed classification of the different death phenotypes).b Fragmentation levels:2, none detectable; (1),,10% (rRNA fragmentation clearly seen only after hybridization);1, 10 to 30%;11, 30 to 50%;111, 50 to 70%;

1111, .70%.c 28S rRNA cleavage in the D8 region only.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

that this bulge loop was exposed to random endonucleaseactivity or that an inter-U cut was followed by exonucleolyticdegradation of the remaining bulge loop. A second cut in D8bwas more precisely located between two U’s in a bulge loop 27nucleotides upstream. The cuts in D8 were thus separated by31 nucleotides, accounting for the double 39D8 band describedabove. Hybridization data obtained with the probes D8b-extr.and D8-39int (Table 1) also limited the cuts in D8 to a maxi-mally 0.11-kb-long region in the descending strand of D8b. Theaccumulation of a 1.5-kb 39D8 fragment in situations withdominant primary D2 cleavage (21) indicated that D8 was also

cleaved secondarily to primary D2 cleavage. Whether the twocleavage sites in D8 represent primary or secondary cleavagesites is not known.Similar primer extension analyses were done to find the most

39 cut in D2 (Fig. 8). A precise cut was found between an A anda U that separates helix H1 from H2 (Fig. 7B) (for nomencla-ture for D2, see reference 32). The results obtained by hybrid-ization with the probes D2-39int and 39D2 (Table 1) were inagreement with this cut being a primary cut in the 39 end of D2.The most 59 primary cut in D2 proved more difficult to find

by primer extension. However, the size difference between thetwo 39 fragments arising from the double cut in D2 (39D2b and39D2c) (Fig. 7A) was estimated to be 0.3 kb, and the fragment59 to 39D2b was about 0.9 kb. This placed the 59 cut in D2 ator close to the bulge loop constituting the D2b-D2c junction,i.e., between subdomains D2b and D2c (Fig. 7B). To check thevalidity of this assumption, an oligonucleotide (D2c-59) com-plementary to this bulge loop region was used in an S1 nucle-ase protection assay and for probing Northern blots (Table 1and Fig. 8). A faint additional 19-nucleotide fragment ap-peared to be protected when S1 nuclease assays were donewith fragmented 28S rRNA annealed with 59-32P-labeledD2c-59 (data not shown), suggesting that the 59 cut in D2 wasin the middle of the bulge loop between D2b and D2c (Fig.7B). Similar bands were detected, although at lower levels,when the reaction was done with control RNA. However, hy-bridizations confirmed that the 59D2 cleavage site was in thesuspected bulge loop, because the D2c-59 oligonucleotide rec-ognized both the 59D2 and the 39D2b fragments (Fig. 8, leftpanel).Interestingly, a cleavage site was also detected inside D2c

(Fig. 8; cut A in Fig. 7B). Because there is no intermediateband between the two primary 39D2 bands (39D2b and 39D2c;see above), this cut must be secondary to primary cleavage inD8. Such a cut should also give rise to secondary 59D2 bands,and such bands have indeed been found (Fig. 1 and 2). Takentogether, these results indicate that the 59D8 fragment iscleaved in D2 secondarily to primary cleavage in D8 but thatthis secondary D2 cleavage site is inside D2c, while the primaryD2 cleavage sites are flanking the D2c subdomain.rRNA cleavage patterns obtained by RNase activity in per-

meabilized cells or cellular lysates. It was important to deter-mine if these regions in D2 and D8 were targets for rRNAcleavage not only in apoptosis but also during 28S rRNA deg-radation caused by normal rRNA turnover or artificially in-creased RNase activity. Accordingly, degradation of rRNA wasinduced in IPC-81 cells and cell lysates (see Materials andMethods for a detailed description of the various approaches).In both cases 28S and 18S rRNAs were degraded to similarextents (Fig. 9). Degradation of 28S rRNA tended to start athot spots in and just 39 to the D8 region. Cuts here gave twomajor 5928S bands and three major 3928S bands. Size compar-isons showed that the distal cut(s) in D8 had to be in thevicinity of the D8 cuts described above (see bottom right panelof Fig. 9 for a direct comparison of apoptosis- and degrada-tion-induced 59D8 fragments). Notably, no primary or earlysecondary cuts in D2 or D6 were induced by endogenousRNase activation.Cell lysates were also exposed to RNases having different

nucleotide specificities in order to test whether apoptotic frag-mentation patterns could be obtained. However, when degra-dation of 28S rRNA was seen, it appeared as extensive smear-ing of both 28S and 18S rRNAs. Cutting limited to only certainsites or regions was not obtained (data not shown).Finally, it should be noted that the bands created by normal

28S rRNA turnover (Fig. 1 and 2 and data not shown) were not

FIG. 6. Primer extension analysis of the 39 end of the D8 domain. Primerextensions done with RNA isolated from normal and apoptotic IPC-81 cells werecompared. Note the many unspecified terminations, which are also found in thelanes showing reactions done without dideoxynucleotides. However, a clear dif-ference between control rRNA and fragmented rRNA was found in two loca-tions (arrowheads). For comparison, the published genomic rat D8 sequence isindicated on the right (3).



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

the same as the apoptotic 28S rRNA fragmentation bands.Interestingly, cleavage in the D8 region was not prominentduring normal ribosomal turnover.

DISCUSSION

The present study has demonstrated limited cleavage re-stricted to a few of the D domains of 28S rRNA during apop-totic death in several cell types and in response to a number ofdifferent apoptotic induces. Similar rRNA fragmentation couldnot be induced during necrosis-like processes (Table 2), byexposing intact ribosomes to RNase activity in the presence orabsence of added energy (Fig. 9 and Table 2), or during normalribosomal turnover (Fig. 1 and 2 and data not shown). Thediscrete nature of the observed cleavages argues against gen-erally increased RNase activity or random polynucleotide deg-radation as the explanation for apoptotic rRNA fragmentation.Another argument is that the regions cleaved in the D2 sub-domain are known to be inaccessible to chemical modificationin the intact ribosome (18, 20), suggesting that the ribosomeconformation is changed during apoptosis in such a way as toexpose these D2c regions to RNase. We believe that 28S rRNAfragmentation is an example of the limited, selective, and con-trolled macromolecular degradation characteristic of apoptoticcells before secondary necrosis occurs (23, 24, 36).The order of the two apoptosis-associated cuts in D2 was not

random. The fact that a 1.2-kb 59D2 fragment was never de-tected (only a 0.9-kb fragment was detected [Fig. 1, 2, and 8])indicates that the 39 cut in D2 probably was secondary to the 59cut and maybe was dependent on conformational changes ofrRNA induced by the 59D2 cut. The double cutting in D2might lead to excision of the hypervariable 0.3-kb D2c subdo-main from the ribosome. Probing for the 0.3-kb D2c fragmentin rRNA or total RNA from cells with extensive D2 cleavagerevealed about 10% of the amount expected if all of the 0.3-kbD2c fragment had remained intact. This means that the frag-ment became short-lived after excision. Interestingly, cleavagein D8 caused secondary cleavage in the middle of the D2csubdomain (Fig. 8), thus hindering such excision of a completeD2c fragment. The fact that D2 can be differentially cleaveddepending upon (primary) cleavage in D8 makes a direct struc-tural interaction between D2 and D8 more likely. This sup-ports the hypothesis that the coevolution of the D2 and D8domains with respect to both size and GC content reflects afunctional link between these two major ribosomal variableregions (35). In all cases the fragmentation patterns obtainedwere reproduced when the same cells were exposed to okadaicacid (data not shown), an inducer of apoptosis in several ex-perimental systems (2). This indicates that it is the cell type and

FIG. 7. (A) Schematic drawing showing a linearized rat 28S rRNA moleculewith constant (white) and variable (crosshatched) regions. The positions of theoligonucleotides used for hybridization are marked at the top (see Table 1). Themajor 28S rRNA fragmentation bands in IPC-81 cells are indicated at thebottom. (B) Schematic drawing of the D2c subdomain showing the cleavage sites.The primary cleavage sites are numbered (1 and 2), and a secondary cleavage siteis marked A. The small arrow between the CC and GG in cleavage site 1 marksthe cleavage position suggested by the S1 nuclease protection analysis. Theproposed secondary structure of the complete D2 domain is shown in the box(35). Arrows orient the sequence in the 59-to-39 direction. (C) Schematic drawingof the D8b subdomain showing the apoptotic cleavage sites (numbered 3 and 4).The proposed secondary structure of the complete D8 domain is in the box (35).Arrows orient the sequence in the 59-to-39 direction. The circled U was notincluded in the published genomic rat 28S rRNA sequence (3), but a U in thisposition was found in human 28S rRNA (30, 35). The two bases in lowercaseletters indicate nucleotides in the genomic rat sequence that were not found inIPC-81 rRNA.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

not the agent causing apoptosis that determines the 28S rRNAfragmentation pattern.Cuts in or close to the D8 region were also seen during

rRNA degradation caused by necrosis-like processes or bycellular disintegration and in cell lysates (Fig. 9 and Table 2).In studies of ribosomal degradation in cell extracts supple-mented with ppp(A29p)3A, a ligand activating RNase L, cleav-age of 28S rRNA gave rise to at least some bands with sizesreminiscent of 59D8 and 39D8 fragments (47, 52). Thus, the D8region seems to be exposed to attacks by intracellular RNaseactivity. Recently it was found that several D domains in Dro-sophila melanogaster (D1 to D3, D6, D7b, D8 to D10, and D12)were partly accessible to chemical modification when intactribosomes were exposed to EDTA-Fe(II), while the conservedparts of the 28S rRNA remained largely unaffected (18). Sim-ilar observations were made with mouse 28S rRNA, in whichthe U’s corresponding to the U’s involved in apoptotic D8cleavage (Fig. 7C) were highly accessible to chemical modifi-cation (20). In addition, the 39 bulge loop that is cleaved in D8has been found to be prone to RNase T2 digestion in studiesdone with in vitro-transcribed human D8b regions (30). Ap-parently, in the D8b tertiary structure this bulge loop is acces-sible to RNases (30), and the D8 region itself is exposed onthe ribosomal surface (18, 20). When the apoptotic D8cleavage sites are compared, the consensus sequence is

GCGGUêU(U)CCC. The presence of U residues in both D8cleavage sites makes activated RNase L a candidate for causingcleavage, even though RNase L prefers to cleave after UpNand not GpU doublets, at least in vitro (14).The cuts in D2 were not in bulge loops but rather in single-

stranded regions forming junctions of at least three large hair-pin loops (Fig. 7B). Unlike the case with the cuts in D8, noconsensus sequence for cleavage could be found. It is temptingto compare these junctional regions with the structure of ham-merhead ribozymes, in which the catalytic core also is at thebranching point of three hairpin loops (38). Interestingly, abreak between the stems H2 and H1 in D2 (Fig. 7B), twonucleotides 39 to the break found in apoptosis, was a constantfinding in the large subunit rRNA molecule of the dinoflagel-late Procrocentrum micans (31). Autocatalytic or RNA-cata-lyzed cleavage, triggered by a change in RNA tertiary struc-ture, is certainly a possibility that must be kept in mind in thesecases, espesially since the apoptotic D2c cleavage sites appearto be protected within the ribosome (18, 20).The 28S rRNA divergent domains have neither a known

evolutionary origin nor any known functions (13), and theyhave evolved within strict frames for secondary, but not pri-mary, structure (19). The D2, D6, and D8 domains, which wereimplicated in apoptosis (Fig. 1 to 3), typically show large ex-pansions only in the 28S rRNAs of higher eukaryotes (19).

FIG. 8. (Left panel) Hybridizations of Northern blots and primer extension analyses done to find the cleavage sites in D2. Hybridization with the oligonucleotideD2c-59 indicated that this oligonucleotide bridged the 59 cut in D2. (Middle and right panels) Results of primer extension reactions done with fragmented IPC-81 28SrRNA. The primary 39D2 cut was found by using a-35S-dCTP and cold 59D3 oligonucleotide primers in the sequencing reaction. No bands were found when a reactionwith no ddNTPs was performed with RNA from normal IPC-81 cells (not shown), and these results were reproduced by using a 59-32P-labeled 59D3 oligonucleotideas a primer (not shown). A cut was detected inside D2c upstream of the primary 39 cut when the avian myeloblastosis virus reverse transcriptase was primed with a59-32P-labeled 39D2 oligonucleotide (right panel). In this case, a high background level in the primer’s most proximal sequence blurred the primary 39D2 cut (notshown). Both cleavage sites are marked with arrowheads.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

FIG. 9. Experiments demonstrating that artificially induced 28S rRNA degradation gives bands distinct from apoptotic 28S rRNA fragmentation bands. Gels (leftpanels) and autoradiographs of a Northern blot (middle and right panels) from RNA prepared from IPC-81 cells after freeze-thawing (top) or from an IPC-81 derivedcellular lysate (bottom) are shown. Incubation times are indicated above the lanes. It should be noted that the lysates used in the left bottom panels were preparedin a way that gave a rather high level of endogenous RNase activity, while the lysates used in the right bottom panel were prepared to work optimally for in vitro proteinsynthesis (see Materials and Methods for details). The blots were hybridized with oligonucleotide probes for a crude classification of the major 28S rRNA degradationbands (marked). The region around D8 appears to be prone to RNase digestion as indicated by the multitude of small bands detected by the D8-39int probe (left bottompanels). Note that the 28S rRNA fragmentation bands (marked with arrows; see right bottom panel, which also shows RNA prepared from apoptotic IPC-81 cells) aredistinct from the degradation bands.

2060


EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

Their great intraspecies variation makes them not only variablebut hypervariable, and it would not be surprising if the ribo-somal divergent domains have adopted functions of increasingimportance in species of highly evolved complexity that are notdirectly related to the task of synthesizing proteins.The observed rRNA fragmentation either could be signifi-

cant for the apoptotic process itself or could serve some otherpurpose in the organism harboring the dying cell. An earlyricin-like effect on rRNA leading to strand breaks later wasunlikely because aniline treatment of rRNA (12) from latepreapoptotic cells did not give rise to any 28S rRNA fragments(data not shown). In all systems studied the specific rRNAfragmentation coincided with internucleosomal DNA frag-mentation, with both processes occuring simultaneously withor slightly after morphological signs of apoptosis (Fig. 4 and 5;Table 2). This suggests that concerted polynucleotide cleavagewas a feature of the apoptotic process but not part of themechanism activating apoptosis. There was no clear correla-tion between morphological type of apoptosis and polynucle-otide cleavage, but there was a close covariation in extents of28S rRNA fragmentation and internucleosomal DNA frag-mentation (Fig. 4 and 5; Table 2). The functional significanceof internucleosomal DNA fragmentation is still not clear; di-gestion of DNA to facilitate disposal of apoptotic cells as wellas protection against illegitimate, potentially harmful DNA hasbeen proposed (6, 50). One possibility is that DNA fragmen-tation is coincidental, since it easily can be triggered in isolatednuclei (49). However, the observed correlation between inter-nucleosomal DNA fragmentation and highly specific and non-random cleavage of 28S rRNA (Fig. 4 to 8) supports a role forDNA fragmentation in apoptosis, since the cuts observed in28S rRNA could not be reproduced in isolated lysates, evenwhen an energy source was supplied (Fig. 9). We propose thatconcerted polynucleotide cleavage is a feature which is prom-inent in some forms of apoptosis and that its function can be toprotect neighboring cells from illegitimate polynucleotides oth-erwise released from the dying cell. Obvious targets for de-struction are RNA or DNA of viral origin, it being reasonablethat dying virus-infected cells have evolved a system designedto inactivate viral polynucleotides (7, 8, 33, 50, 51). Since re-sistance (or tolerance) to a given virus can be selected for (28),it would be expected that a system must exist which is subjectto (random) variation, analogous to variable parts of the anti-bodies. The ribosomal divergent domains, showing extensiveintra- and interspecies variation (16, 30, 35), could constitutesuch a flexible intracellular viral defense system. It is intriguingthat the cytokine TNF is both an antiviral molecule (51) and anefficient inducer of 28S rRNA fragmentation (Fig. 3 and 5).Furthermore, the double-stranded-RNA-activated protein ki-nase (p68 kinase), a ribosome-associated kinase related to theinterferon-stimulated viral defense system (44), can stimulateapoptosis (29). The finding that 2-aminopurine, an crude in-hibitor of double-stranded RNA kinase and interferon-stimu-lated gene expression, efficiently inhibits cAMP-induced apop-tosis of IPC-81 cells when present at above 5 mM in the growthmedium (data not shown) suggests that p68 kinase activationcould be a necessary component in a cAMP-triggered apop-totic pathway as well. Taken together, these findings suggestthe idea that apoptosis developed as an antiviral defense be-fore adaptive immunity arose in evolution (7, 8, 33).In conclusion, 28S rRNA fragmentation should be added to

the list of parameters to be considered when studying apopto-sis, and a role for ribosomal D domains in apoptosis is sug-gested.

ACKNOWLEDGMENTS

We are grateful to Erna Finsås and Aud K. H. Eliassen for experttechnical assistance, to Anders Molven for helpful discussion of themanuscript, and to Jean Pierre Bachellerie for helpful technical advice.This work was supported by the EC Biomedical and HCM Pro-

grams, the Association Claude Bernard, the Norwegian ResearchCouncil, the Blix Foundation, and the Nansen Foundation.

REFERENCES1. Bachellerie, J.-P., and L.-H. Qu. 1993. Ribosomal RNA probes for detectionand identification of species, p. 249–263. In J. E. Hyde (ed.), Methods inmolecular biology, vol. 21. Protocols in molecular parasitology. HumanaPress Inc., Totowa, N.J.

2. Bøe, R., B. T. Gjertsen, O. K. Vintermyr, G. Houge, M. Lanotte, and S. O.Døskeland. 1991. The protein phosphatase inhibitor okadaic acid inducesmorphological changes typical of apoptosis in mammalian cells. Exp. CellRes. 195:237–246.

3. Chan, Y.-L., J. Olvera, and I. G. Wool. 1983. The structure of rat 28Sribosomal ribonucleic acid inferred from the sequence of nucleotides in agene. Nucleic Acids Res. 11:7819–7830.

4. Chomczynsky, P., and N. Sacchi. 1987. Single-step method of RNA isolationby acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Bio-chem. 162:156–159.

5. Clarke, P. G. H. 1990. Developmental cell death: morphological diversityand multiple mechanisms. Anat. Embryol. 181:195–213.

6. Clem, R. J., and L. K. Miller. 1993. Apoptosis reduces both the in vitroreplication and the in vivo infectivity of a baculovirus. J. Virol. 67:3730–3738.

7. Clem, R. J., and L. K. Miller. 1994. Induction and inhibition of apoptosis byinsect viruses, p. 89–110. In L. D. Tomei and F. O. Cope (ed.), Apoptosis II:the molecular basis of apoptosis in disease. Cold Spring Harbor LaboratoryPress, Cold Spring Harbor, N.Y.

8. Clouston, W. M., and J. F. R. Kerr. 1985. Apoptosis, lymphocytotoxicity andthe containment of viral infections. Med. Hypoth. 18:399.

9. Cohen, J. J. 1993. Apoptosis. Immunol. Today 14:126–130.10. Delic, J., M. Coppeymoisan, and H. Magdelenat. 1993. Gamma-ray-induced

transcription and apoptosis-associated loss of 28S rRNA in interphase hu-man lymphocytes. Int. J. Rad. Biol. 64:39–46.

11. Driscoll, M. 1992. Molecular genetics of cell death in the nematode Cae-norhabditis elegans. J. Neurobiol. 23:1327–1351.

12. Endo, Y., and K. Tsurugi. 1987. RNA N-glycosidase activity of ricin A-chain.J. Biol. Chem. 262:8128–8130.

13. Engberg, J., H. Nielsen, G. Lenaers, O. Murayama, H. Fujitani, and T.Higashinakagawa. 1990. Comparison of primary and secondary 26S rRNAstructure in two Tetrahymena species: evidence for a strong evolutionary andstructural constraint in expansion segments. J. Mol. Evol. 30:514–521.

14. Floyd-Smith, G., E. Slattery, and P. Lengyel. 1981. Interferon action: RNAcleavage pattern of a (29-59)oligoadenylate-dependent endonuclease. Sci-ence 212:1030–1031.

15. Gjertsen, B. T., L. I. Cressey, S. Ruchaud, G. Houge, M. Lanotte, and S. O.Døskeland. 1994. Multiple apoptotic death types triggered through activa-tion of separate pathways by cAMP and inhibitors of protein phosphatases inone (IPC leukemia) cell line. J. Cell. Sci. 107:3363–3377.

16. Gonzales, I. L., J. L. Gorski, T. J. Campen, D. J. Dorney, J. M. Erickson,J. E. Sylvester, and R. D. Schmickel. 1985. Variation among human 28Sribosomal RNA genes. Proc. Natl. Acad. Sci. USA 82:7666–7670.

17. Gutell, R. R., and G. E. Fox. 1988. A compilation of large subunit RNAsequences in a structural form. Nucleic Acids Res. 16(Suppl.):r175–r216.

18. Han, H., A. Schepartz, M. Pellegrini, and P. B. Dervan. 1994. Mapping RNAregions in eukaryotic ribosomes that are accessible to methidiumpropyl-EDTA z Fe(II) and EDTA z Fe(II). Biochemistry 33:9831–9844.

19. Hancock, J. M., and G. A. Dover. 1988. Molecular coevolution among cryp-tically simple expansion segments of eukaryotic 26S/28S rRNAs. Mol. Biol.Evol. 5:377–391.

20. Holmberg, L., Y. Melander, and O. Nygard. 1994. Probing the structure ofmouse Ehrlich ascites cell 5.8S, 18S and 28S ribosomal RNA in situ. NucleicAcids Res. 22:1374–1382.

21. Houge, G., S. O. Døskeland, R. Bøe, and M. Lanotte. 1993. Selective cleavageof 28S rRNA variable regions V3 and V13 in myeloid leukemia cell apop-tosis. FEBS Lett. 315:16–20.

22. Houge, G., O. K. Vintermyr, and S. O. Døskeland. 1990. The expression ofcAMP-dependent protein kinase in primary rat hepatocyte cultures. Cy-clic AMP down-regulates its own effector system by decreasing theamount of catalytic subunit and increasing the mRNAs for the inhibitory(R) subunits of cAMP-dependent protein kinase. Mol. Endocrinol.4:481–488.

23. Jensen, P. H., L. I. Cressey, B. T. Gjertsen, P. Madsen, G. Mellgren, P.Hokland, J. Gliemann, S. O. Døskeland, M. Lanotte, and O. K. Vintermyr.1994. Cleaved intracellular plasminogen activator inhibitor 2 in human my-eloleukemia cells is a marker of apoptosis. Br. J. Cancer 70:834–840.

24. Kaufmann, S. H., S. Desnoyers, Y. Ottaviano, N. E. Davidson, and G. G.Poirier. 1993. Specific proteolytic cleavage of poly(ADP-ribose) poly-



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

merase—an early marker of chemotheraphy-induced apoptosis. Cancer Res.53:3976–3985.

25. Lanotte, M., S. Hermouet, G. Gombaud-Saintonge, and I. Dobo. 1986. Onthe growth regulation of the rat promyelocytic leukemia (BNML): growthinhibition and eradication of clonogenic cells by cholera toxin. LeukemiaRes. 10:1319–1326.

26. Lanotte, M., V. Martin-Thouvenin, S. Najman, P. Ballerini, F. Valensi, andR. Berger. 1991. NB4, a maturation inducible cell line with t(15;17) markerisolated from a human acute promyelocytic leukemia (M3). Blood 77:1080–1086.

27. Lanotte, M., J. B. Riviere, S. Hermouet, G. Houge, O. K. Vintermyr, B. T.Gjertsen, and S. O. Døskeland. 1991. Programmed cell death (apoptosis) isinduced rapidly and with positive cooperativity by activation of cyclic aden-osine monophosphate-kinase I in a myeloid leukemia cell line. J. Cell.Physiol. 146:73–80.

28. Lee, J. C., H. H. Chen, H. L. Wei, and Y. C. Chao. 1993. Superinfection-induced apoptosis and its correlation with the reduction of viral progeny incells persistently infected with Hz-1 baculovirus. J. Virol. 67:6989–6994.

29. Lee, S. B., and M. Eesteban. 1994. The interferon-inducible double-strandedRNA-activated protein kinase induces apoptosis. Virology 199:491–496.

30. Leffers, H., and A. H. Andersen. 1993. The sequence of 28S ribosomal RNAvaries within and between human cell lines. Nucleic Acids Res. 21:1449–1455.

31. Lenaers, G., L. Maroteaux, B. Michot, and M. Herzog. 1989. Dinoflagellatesin evolution. A molecular phylogenetic analysis of large subunit ribosomalRNA. J. Mol. Evol. 29:40–51.

32. Linares, A. R., J. M. Hancock, and G. A. Dover. 1991. Secondary structureconstraints on the evolution of Drosophila 28S ribosomal RNA expansionsegments. J. Mol. Biol. 219:381–390.

33. Martz, E., and D. M. Howell. 1989. CTL: virus control cells first and cytolyticcells second? Immunol. Today 10:79–86.

34. McConkey, D., P. Hartzell, M. Jondal, and S. Orreniu. 1989. Inhibition ofDNA fragmentation in thymocytes and isolated thymocyte nuclei by agentsthat stimulate protein kinase C. J. Biol. Chem. 264:13399–13402.

35. Michot, B., and J.-P. Bachellerie. 1987. Comparisons of large subunit rRNAsreveal some eukaryote-specific elements of secondary structure. Biochimie69:11–23.

36. Oberhammer, F., J. W. Wilson, C. Dive, I. D. Morris, J. A. Hickman, A. E.Wakeling, P. R. Walker, and M. Sikorska. 1993. Apoptotic death in epithe-lial cells—cleavage of DNA to 300 and/or 50 kb fragments prior to or in theabsence of internucleosomal fragmentation. EMBO J. 12:3679–3684.

37. Oren, M. 1992. The involvement of oncogenes and tumor supressor genes inthe control of apoptosis. Cancer Metastasis Rev. 11:141–148.

38. Pan, T., D. M. Long, and O. C. Uhlenbeck. 1993. Divalent metal ions in RNAfolding and catalysis, p. 271–302. In R. F. Gesteland, and J. F. Atkins (ed.), TheRNA world. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.

39. Raff, M. C. 1992. Social controls on cell survival and cell death. Nature(London) 356:397–399.

40. Raue, H. A., J. Klootwijk, and W. Musters. 1988. Evolutionary conservationof structure and function of high molecular weight ribosomal RNA. Prog.Biophys. Mol. Biol. 51:77–129.

41. Robaye, B., A. P. Døskeland, N. Suarez-Huerta, S. O. Døskeland, and J. E.Dumont. 1994. Apoptotic cell death analyzed at the molecular level bytwo-dimensional gel electrophoresis. Electrophoresis 15:503–510.

42. Robaye, B., R. Mosselmans, W. Fiers, J. E. Dumont, and P. Galand. 1991.Tumor necrosis factor induces apoptosis (programmed cell death) in normalendothelial cells in vitro. Am. J. Pathol. 138:447–453.

43. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: alaboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, ColdSpring Harbor, N.Y.

44. Samuel, C. E. 1979. Mechanisms of interferon action. Kinetics of interferonaction in mouse L929 cells: phosphorylation of protein synthesis initiationfactor eIF-2 and ribosome-associated protein P1. Virology 93:281–285.

45. Schwartz, L. M., and B. A. Osborne. 1993. Programmed cell death, apoptosisand killer genes. Immunol. Today 14:582–590.

46. Schwartzman, R. A., and A. Cidlowski. 1991. Internucleosomal deoxyribo-nucleic acid cleavage activity in apoptotic thymocytes: detection and endo-crine regulation. Endocrinology 128:1190–1197.

47. Silverman, R. H., P. J. Cayley, M. Knight, C. S. Gilbert, and I. M. Kerr. 1982.Control of the ppp(A29p)nA system in HeLa cells. Eur. J. Biochem. 124:131–138.

48. Skalka, M., J. Matyasova, and M. Cejkova. 1976. DNA in chromatin ofirradiated lymphoid tissue degrades in vivo into regular fragments. FEBSLett. 72:271–274.

49. Sun, D. Y., J. S., L.-M. Zheng, D. M. Ojcius, and J. D.-E. Young. 1994.Separate metabolic pathways leading to DNA fragmentation and apoptoticchromatin condensation. J. Exp. Med. 179:559–568.

50. Vaux, L. D., G. Haecker, and A. Strasser. 1994. An evolutionary perspectiveon apoptosis. Cell 76:777–779.

51. White, E., and L. R. Gooding. 1994. Regulation of apoptosis by humanadenoviruses, p. 111–141. In L. D. Tomei and F. O. Cope (ed.), Apoptosis II:the molecular basis of apoptosis in disease. Cold Spring Harbor LaboratoryPress, Cold Spring Harbor, N.Y.

52. Wreschner, D. H., T. C. James, R. H. Silverman, and I. M. Kerr. 1981.Ribosomal RNA cleavage, nuclease activation and 2-5A (ppp(A29p)nA) ininterferon-treated cells. Nucleic Acids Res. 9:1571–1581.

53. Wyllie, A. H. 1980. Glucocorticoid-induced thymocyte apoptosis is associatedwith endogenous nuclease activation. Nature (London) 284:555.

54. Wyllie, A. H., J. F. R. Kerr, and A. R. Currie. 1980. Cell death: the signifi-cance of apoptosis. Int. Rev. Cytol. 68:251–305.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

cleaved in cells undergoing apoptosis. fine mapping of 28s rrna

Documents