Normal Fibroblasts Responding to Anoxia Exhibit Features of the ...

5 downloads 0 Views 6MB Size Report
Jan 3, 1989 - Garth R. Anderson$, Daniel L. Stoler, and Lisa A. Scarcello. From the ...... Singh, K., Saragosti, S., and Botchan, M. (1985) Mol. Cell. Biol.
Vol. 264, No. 25, Issue of September 5, PP. 14885-14892,1989 Printed in U.S.A.

OF BIOLOGICAL CHEMISTRY THE JOURNAL 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc

Normal Fibroblasts Responding to Anoxia Exhibit Features of the Malignant Phenotype* (Received for publication, January 3, 1989)

Garth R. Anderson$, Daniel L. Stoler, and Lisa A. Scarcello From the Department of Molecular and CellularBiology, Roswell Park Memorial Institute, Buffalo, New York 14263

fore,reflectmedia depletion of glucose resultingfromincreased glycolysis (10, 11). VL30 elements are a multi-gene family with approximately 50-100 copies/cell in normal rat andmouse DNA (12). Structurally and biologically they are similar to retroviruses and other retrotransposable elements, although theylack an enu gene, and functional gag and pol genes have yet to be found (13-15).VL30 elementRNAis very strongly induced in normal rat fibroblasts responding to low oxygen tensions; induction can exceed 500-fold under complete anoxia, but a 2 % 0 2 (2,3). lesser response is seen under atmospheres up to It is not known if all or only a few members of the VL30 element family respond to anoxia. Anoxia-induced VL30 RNA is polyadenylated and polysome-associated (4), but no corresponding gene products have been heretofore identified. VL30 elements havebeen frequently associated with cancer, but their actual role in normal and cancer tissues has remained obscure. VL30 RNA hasbeen found expressed a t high levels in many rodent tumors and ainvariety of transformed cells (16,17). DNA rearrangements involving VL30 sequences have been observedin choline deficiency-induced livercancer in the rat (18). Major fractions of a rat VL30 element are present alongside ras oncogenes inserted in two acute transforming retrovirusgenomes, the Kirsten and Harvey sarcoma viruses (19,ZO). Cell-derived inserts in other acute transforming retrovirusgenomes have been shown in nearlyall cases to Normal rat fibroblasts subjected to anoxia exhibit a com- represent oncogenes (21). Although the precise role for VL30 plex response which begins with induction of retrotransposon- sequences in the Kirsten and Harvey viruses is as yet underelated VL30 element RNA (1-3). Several proteins are also fined, deletion of the VL30 sequences in the Harvey virus induced, which have been variously described as anoxic stress removes its ability to induce sarcomas (22). proteins or oxygen-regulated proteins (4-7). Anoxia has been We have investigated if anoxia and the inductionof VL30 generally regarded as a cellular stress akin to the two-well expression in normal fibroblasts by anoxia might be other studied systemsof heat shock and glucose deprivation. Indeed, than a simple stress response. Our central objectives were to heat shockgenes are induced uponthereturn of anoxic (i) evaluate if the induced VL30 element RNA is encoding fibroblasts to aerobic conditions.Quiteunlikeheat shock, gene products, (ii) togain insights into thephysiological role however, during the time frame of the response, anoxia is not of the fibroblast response to anoxia, and (iii) to ascertain if associated withanyreductionin cell viability (2). Many there might be some broad underlying relationship between authors have equated glucose deprivation and the glucose- this normal cell response and cancer. The cell line we have regulated protein response with anoxia, based on the induc- used as our model system is the FRE normal rat fibroblast tion by anoxia in moderately heavybut not sparse cell cultures line, with the NRK and REF 52 rat lines examined in parallel of two major glucose-regulated proteins of 94 and 78 kilodal- for most parameters. We describe here several aspects of the tons (4, 7-9). These two proteins are also readily induced by response of rat fibroblasts to anoxia. Theresponse occurs in glucose deprivation or 2-deoxyglucose treatment of aerobic stages, beginning with VL30 RNAinductionasthefirst cell cultures, and their induction in anoxic cells may, theredetectable event. TheVL30 RNA most likely encodes at least * This work was supported by grants from Abbott Laboratories, one 61-kilodalton protein, of whichsome is secreted. Our the United States Department of Agriculture, and the New York results indicate theresponse is not a simple survival response State Health Instruments and Devices Institute and a biomedical and instead appears to represent a physiological response to research support grant from the National Institutes of Health The a physiological signal.Additional aspects of the response costs of publication of this article were defrayed in part by the indicate it would be well suited to play an active role in payment of page charges. This article must therefore he hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 normal wound healing and, if expressed in an uncontrolled fashion, may contributecriticalfeaturestothemalignant solely to indicate this fact. $ To whom correspondence should be addressed. phenotype.

Cancer hastwo fundamental features: neoplasia, representing the aberrant expression of a normal cell proliferation response, and malignancy, an ability to penetrate normal tissue boundaries. Although the role of oncogenes in neoplastic transformation is becoming clearer, malignancy remains far less wellunderstood. Normal rat fibroblasts exhibit a staged response to anoxia which, if expressed in anuncontrolled fashion, may contribute vital aspects to the malignant phenotype. The response beginswith induction of retrotransposon-like VL30 element transcription, progresses through induction of several intracellular proteins, and is followed by secretion of three major proteins including the protease cathepsin L. The induced VL30 element RNA encodes a 61-kDa secretory protein of unknown function. The response of fibroblasts to anoxiais evidently not a survival response. Instead, the response represents a close match to the role of fibroblasts during the early stages of wound healing where they are active under near-anoxic conditions. Malignant cancer cells are known to exhibit several of the characteristics we find induced in fibroblasts by anoxia. Conversion to the malignant phenotype may represent coordinate loss ofcontrol of thisnormal cellular response.

14885

14886

The Fibroblast Response to Anoxia

78, and asp34/LDH-k (7, 23). Lactate dehydrogenase-k enzymatic activity rose in parallel with the appearance of asp34 Cells and Tissue Culture and reached a level similar to that seen in cells transformed Cells were grown in the Dulbecco modification of Eagle's medium, by the Kirsten sarcoma virus (Table I). A fourth tentatively supplemented with 10% calf serum (GIBCO). The FRE' and NRK identified anoxia-inducible protein not in the pH range inrat fibroblast lines were originally obtained from Stuart Aaronson. cluded in Fig. 1 comigrated with the proliferating cell nuclear REF 52 was obtained from Michael Lambert at Cold Spring Harbor. antigen. After 16 h, an additional induced spot was seen which Kirsten sarcoma virus transformed cells were obtained from Stuart may represent post-translational modification of grp 94. As Aaronson and Michael Gottesman, National Institutes of Health. Cells were maintained either aerobically under a normal atmos- would be expected with the cessation of DNA synthesis and phere supplemented with 5% COZ or under totally anoxic conditions corresponding cell proliferation, reduction insynthesis of with 5% COz in a GasPak anaerobic chamber (Becton Dickinson) as several proteins during anoxiawas also evident. described (1, 23). An Induced 40-kDa Secretory Protein IsCathepsin L-['%] Methionine labeling at 4,8, 12, and 18 h into anoxia showed Nucleic Acid Analyses and Hybridization that anoxia induced three major secretory proteins when Cytoplasmic RNA assayed via slot blots was isolated by the method analyzed on one-dimensionalSDS-polyacrylamide gel electroof Godowski and Knipe (24). Blots were probed with oligonucleotide- phoresis. A 51-kDa secretory protein was seen beginning at 8 primed 32P-DNAsynthesized from electroeluted plasmid inserts.The plasmid used was pVL30 DS which was subcloned from the PuuII- h, whereas 61- and 40-kDa proteins were first evident at 12 PuuII sites of the Kirsten sarcoma virus. This represents rat VL30 h. Two-dimensional gel analysis revealed that two secretory sequences in the region from 3.0 to 4.5 kilobase pairs downstream proteins of approximately 40 kDa were actually present (Fig. from the Kirsten viral 5' long terminal repeat. 2). One, PI 4.8, was also produced by aerobic cells; the other, PI 5.9, was strongly induced by anoxia. Anoxia-inducible Proteins The V-8 peptide profile of the threemajor anoxia-inducible Intracellular Proteins-Cells were split 1 day before treatment with secretory proteins indicated they were not closely related to 2 X lo6 cells/dish. Cells were then media-changed and incubated or derived from one another (Fig. 3). With VL30 RNA, grp anoxically in complete media. At the time of labeling, they were 78 and asp34, all associated with cancer to a greater or lesser rinsed twice (via catheter) with methionine-free media containing 2% tested the relationship of dialyzed calf sera which had been anoxically equilibrated. Labeling extent (16, 19, 22, 23, 25-27), we was in such media containing 0.1 mCi/ml [35S]methionine for 4 h. the anoxia-inducible secretory proteins to known secretory The GasPak chamber was then opened, cells rinsed twice in phos- proteins associated with cancer. The molecular weights of the phate-buffered saline, and then lysed in boiling 0.3% SDS, 1%@anoxia-inducible secretory proteins particularlysuggested pomercaptoethanol, and 0.05 M Tris, pH8.0. If extracts were to be used tential identity with classical collagenase, urokinase, and cain immune precipitations, the immune lysis buffer containing deoxthepsin L. Accordingly, we reacted the anoxic secretcry proycholate and procedures we have described were used (2). Twoteins with antiserato these individual proteins and also dimensional gel analysis of intracellular proteins was carried out at compared reactivity with proteins secreted by aerobic FRE the Cold Spring Harbor Quest facility. Secretory Proteins-Cells were split and incubated anoxically as fibroblasts (Fig. 4). Although no reactivity was seen with for intracellularproteins. However, cells were rinsed twice with antisera to collagenase or urokinase, the 40-kDa anoxiamethionine-free, serum-free media, and labeling was for 30 min in inducible secretory protein reacted strongly with antisera to methionine-free serum-free media containing 0.1 mCi/ml [35S]methiL, and the immune precipitate comigrated with onine, with 2 ml/lO-cm dish. Following removal of labeling media via cathepsin catheter, dishes were rinsed twice with complete media (anoxically cathepsin L. Secretedproteins from aerobic cells showed equilibrated and serum-free). Three ml/dish of serum-free equili- substantially less of this antigen, confirming that immunobrated media were then added via catheter. After a secretion time of reactivity was not with a constitutively produced contami3.5 h, this media was collected and centrifuged 10 min at 2,000 rpm nant, and preimmune sera showed no reactivity. The extent to remove any contaminating cells. 10 kg/ml of bovine serum albumin of induction of cathepsin L in anoxic cells ranged from about were then added as carrier, and the media was made 10% in trichloroacetic acid. After precipitating 30 min at 4 'C, the precipitates were 3- to 10-fold at the high end similar to the level seen in cells collected by centrifugation for 30 min at 10,000 X g. The pellets were transformed by the Kirsten sarcoma virus. The identity of then rinsed in 1 ml of ether and centrifuged 10 min at 10,000 X g. p40 as cathepsin Lwas further verified by V-8 peptide analyFollowing drying under a stream of nitrogen, the precipitates were sis. was resuspended in 100 ~1 of H20/original culturedish.Material Anoxia-inducible V U 0 RNA Encodes a p61"Induction of stored at -70 "C. VL30 element RNA began within 4 hof anoxia and continued for more than 26 h.A small but reproducible plateau at RESULTS roughly 5-fold induction was observed around 6 h, suggesting Proteins Induced in Anoxic Fibroblasts-In FRE rat fibro- there might be multiple classes of VL30 RNA induced or that blasts, total RNA synthesis is minimally affected by anoxia individual elements may respond at different times or oxygen for at least 24 h. In contrast, DNA synthesis is severely tensions (Fig. 5). Induction of VL30 RNA did not occur in inhibited beginning between 4 and 8 h, and protein synthesis aerobic FRE fibroblasts treated with 10 mM 2-deoxyglucose, is reduced after 8 h. Beyond the net reduction in protein conditions which efficiently induced the two major glucosesynthesis, the actual profile of proteins synthesized under regulated proteins of 94 and 78 kDa. This result is consistent anoxia also undergoes changes (Fig. 1).High resolution two- withanoxia and glucose deprivation being two distinctly dimensional gel analysis was used to provide a clear picture separate responses which overlap to a very limited extent. of the anoxia inducible intracellular proteins and to define The level of expression of VL30 RNA in anoxic fibroblasts the kinetics of their expression. Although the majority of was somewhat less than that seen in cells transformed by the proteins were unaffected by anoxia, at least six major intra- Kirsten sarcoma virus or in Novikoff hepatoma cells (Fig. 6). cellular proteins were induced between 7 and 11h into anoxia; In thecase of Kirsten sarcoma virus-infected cells, such VL30 three were identified based on comigration with grp 94, grp expression reflects actual incorporation ofVL30 sequences the sarcoma virus genome (20). ' The abbreviations used are: FRE, Fischer rat embryo; NRK, into To ascertain if anoxia-induced VL30 RNA might encode normal rat kidney; REF, rat embryo fibroblast; SDS, sodium dodecyl any of the major intracellular or secretory proteins, we Presulfate. EXPERIMENTALPROCEDURES

The Fibroblast Response to Anoxia

FIG. 1. Intracellular polypeptides induced by anoxia in FRE rat cells. Subconfluent cell cultures were labeled with 0.1 mCi/ml [35S]methioninefor 4 h, and polypeptides were resolved on nonequilibrium two-dimensional gel electrophoresis and then autoradiographed as described (23). I and 2 indicate the glucose-regulated proteins of 94 and 78 kilodaltons; 3 indicates asp34/LDH-k. The other anoxia-induced polypeptides marked with u are unidentified. A , control aerobic culture; B-F, anoxic cultures. l?, 3-7 h; C, 7-11 h; D, 16-20 h; E, 24-28 h; F, labeled from 20 to 24 h and chased with cold methionine from 24 to 28 h. The anoxia-inducible intracellular proteins aremarked for that time period a t which they first appear.

C

14887



pared antisera to a synthetic peptide corresponding to VL30 sequences. The sequences we chose were from that fraction of a rat VL30 element which was incorporated into theHarvey murine sarcoma virus genome. These VL30 sequences have a number of small open reading frames of around 100 amino acids, separated inseveral cases by a single termination codon. Although the Harvey VL30 sequences themselves appear to represent a defective individual member of the VL30 element family, they presumably would share substantial homology with nondefective VL30 elements if such exist (14, 28). The synthetic peptide we prepared corresponded to bases 22452295 in the VL30 insert of the Harvey virus and had the sequence RKCPTNLAKIRKRYKEK. This peptide was within an open reading frame beginning with an AUG codon at position 2147 meeting Kozak’s criteria and ending with a consensus splice donor signal at position 2306 (29,30). Antisera to thissynthetic VL30 peptide detected an anoxiainducible protein of 61 kilodaltons. This antigen was detected both intracellularly and extracellularly (Fig. 7) and comigrated with the secretory p61 obtained by direct methionine

A.

labeling. As would beexpected if p61 were indeed encoded by a VL30 RNA, the p61 antigenwas strongly induced by anoxia but was not induced by 2-deoxyglucose. It must be noted that the secretory p61 still might be encoded by an RNA other than a true VL30 element, but one which shares with VL30 the property of anoxia inducibility and which shares close homology with VL30 in theregion we selected for preparation of the anti-peptide antisera.This p61 was not phosphorylated and not reactive with antisera to the malignancy-associated secretory phosphoprotein p61 described by Senger (31). The major anoxic stress protein asp34lLDH-kis often seen in human cancer and isfrequently found at elevated levels in patient sera (23, 25). Although antiserato asp34 readily detected this anoxia-inducible antigen in total cell extracts, no reactivity was seen with any secretory proteins (Fig. 7). As with VL30 RNA, 2-deoxyglucose had no effect on the intracellular expression of this protein, further differentiating the fibroblast anoxia response from induction of glucose-regulated proteins. Not aCell Survival Response-A nearly identical pattern of

The Fibroblast Response to Anoxia

14888

TABLE I Induction of lactate dehydrogenase-k activity by anoxia Five pg of protein in aTriton X-100 lysate were assayed for lactate dehydrogenase-k activity by activity staining of nondenaturing imidazole/borate-buffered polyacrylamide gels, as described (23,27). G. U. refers to gel units measured on an integrating densitometer. Cells were the normal rat fibroblast line FRE or the same cells transformed by Kirsten sarcoma virus (KiMSV-FRE).

- 43,000

- 25,700

Lactate dehydrogenase-k Cells

Culture

G.u.,ccg

Relative activity”

- 18,400

Exp. 1 FRE FRE

Aerobic 0.01 0.01 Anoxic, 4 h 0.05 0.04 0.05 0.04 Anoxic, 8 h Anoxic,1.03 16 h 1.27 1.23 Anoxic,1.00 24 h Exp. 2FRE Aerobic 0.09 0.05 0.54 1.00 FRE Anoxic, 24 h KiMSV-FRE Aerobic0.65 0.35 ’Activity is relative to thatseen in extracts of FRE cells cultured anoxically for 24 h.

I

FIG. 3. V-8 peptide analysis of the major anoxia inducible fibroblast secretory proteins. [35S]Methionine-labeledsecretory proteins produced at 16 hinto anoxia were separated on SDSpolyacrylamide gel electrophoresis, and then Cleveland gel analysis with digestion by V-8 protease was carried out as described (23, 51). Lane 1, secretory p61; lane 2, secretory p51; lane 3, secretory p40.

ANOXIC

- s7.000

-

64000

-

43,000

-

28,000

-

;flm

200

FIG. 4. The IO-kilodalton anoxia inducible secretory protein is the protease cathepsinL. A, secreted proteins from aerobic

-

64000

-

4a000

-

s7,OOo

24000

FIG. 2. Two-dimensional analysis of fibroblast secretory proteins induced by anoxia. Secretory proteins were labeled with [”Slmethionine after 18 h of anoxia, as described under the “Experimental Procedures”. Equilibrium isoelectric focusing and SDS-polyacrylamide gel electrophoresis on 10% gels for the second dimension were as per O’Farrell (50). 1, 2, and 3 represent anoxia-inducible secretory proteins of 61, 51, and 40 kDa, respectively.

(odd numbers) or anoxic (euen numbers) FRE rat fibroblasts were reacted with preimmune sera (lanes 1 and 2); antisera to cathepsin L (lanes 3 and 4 ) ; antisera to classical collagenase (lanes 5 and 6 ) ; antisera to urokinase (lanes 7 and 8). Total input proteins are shown in lanes 9 and 10, and “C molecular weight standards are inlane 11. 15,000 cpm of [35S]methionine-labeledmaterial were used in each immune precipitation. Secretory proteins were labeled for 30 min, 16 h into anoxia. B, the extent of induction of cathepsin L in Kirsten sarcoma virus transformed cells. Antisera to cathepsin L was reacted with secretory proteins from: lane I, Kirsten sarcoma virus-infected FRE, cultured aerobically; lane 2, aerobic FRE; lane 3, anoxic FRE.

induction of VL30 RNA and intracellular and secretory proteins was seen in both FRE and NRK normal rat fibroblast cells. In contrast, REF52 rat fibroblasts cells showed none of the responses. The presence or absence of the response showed no correlation with the ability of the cells to tolerate anoxia as shown by growth curve analysis (Fig. 8). Additionally, all three rat fibroblast lines maintain complete viability for more than 3 days of anoxia as determined by trypan blue exclusion and more than 2 days as determined by colonyforming efficiency. These results indicate that the response pattern seen in FRE and NRK cells does not represent a simple cellular survival response to anoxic stress or else that

The Fibroblast Response to Anoxia

14889

C HOURS ,.....-

..

-

"

"

~

"

1 CONTROL

ANOXIC

D

(B)

FIG. 5. VL30 RNA induction by anoxia. Cytoplasmic RNA was assayed on slot blots probed for VL30 with the plasmid pVL30 DS, as described (3). A , the kinetics of VL30 RNA induction in FRE cells subjected to anoxia. VL30 RNA is expressed relative to that level in aerobic cells a t time 0, asquantitated by densitometric scanning of multiple slot blots. 0, aerobic cells; 0, anoxic cells. B, VL30 RNA induction occurs independently of induction of the glucose-regulated protein response. RNAsfrom aerobic, anoxic, and aerobic cells cultured in the presence of 10 mM 2-deoxyglucose (2d G ) were assayed by slot blot analysis. Anoxia or deoxyglucose treatment was for 16 h.

FIG.6. VL30 RNA levels in anoxic rat fibroblasts compared to that in transformed cells. Slot blot analyses for VL30 RNA were carried outasin Fig. 5. RNAs were from: A , aerobic FRE fibroblasts; B, anoxic FRE; C, FRE transformed by Kirsten sarcoma virus; D, Novikoff hepatoma cells.

fibroblasts migrate into thewound and secrete collagen; much later they participate in wound contraction. For the first few days fibroblasts do not divide and function in an essentially anoxic environment until neovascularization is completed (36). The responses to anoxia exhibited by fibroblasts in tissue culture, summarized in Fig. 9, appear likely to relate to this in uiuo role. A metabolic shift to glycolysis with concurrent REF 52 cells have atotallyindependentanoxia survival lactate production is required of active anoxic cells and is response system which does not include synthesis of any major consistent with the elevation of hexose transport and asp341 new proteins. Interestingly, REF 52 cells have also been found lactate dehydrogenase-k levels seen. Secretion of the protease to be anomalously resistant to neoplastic transformation by cathepsin L suggests that fibroblasts areparticipatingin transfected oncogenes (32). wound debridement, an activity generally ascribed only to The response does appear to be associated with a continued macrophages and monocytes (35,37). The fact that cathepsin high level of metabolic activity. Hexose transport rateswhen L is activated only upon exposure to acidic conditions and measured by 2-deoxyglucose uptake of FRE fibroblasts has optimal activity below pH 5 would be consistent with a showed that a 2-fold increase occurred between 5 and 24 h role in glycolytic regions where lactate is being abundantly into anoxia. The final values were 10.6 X 10"' mol/cell for secreted (38, 39). anoxic cells, compared with 5.0 x 10"' mol/cell for aerobic A Possible Basis of Malignancy-Malignant cancer cells controls. FRE cells transformed by the Kirsten sarcoma virus exhibit two features not shared with benign neoplasias: invamol/ siveness, by definition, and in most cases a metabolic pattern showed a value equal to the anoxic cells of10.4 X cell. Similarinduction has been observed when cells are of aerobic glycolysis (40,41). The lethal difference between a treated with avariety of respiratory poisons and is associated malignant tumor and a benign tumor is the ability of the with an increased reliance on glycolysis (33, 34). proliferating cells of the former to break down intercellular matrices in surrounding tissues, permitting tumor cell invaDISCUSSION sion across tissue boundaries. Such invasion does not reflect Although often regarded as generic cells, fibroblasts are a the simple effects of tumor cell proliferation and can occur in discrete cell type whose primary role in adult animals is in the absence of proliferation (42). The destruction of intercelwound healing (35). Beginning around 1day after wounding, lular matrices most likely also contributes to the formation

The FibroblastAnoxia Response to

14890

(A)

""-

1 2 3 4 5 6 7 8 9

10 l " .

A

10

-

I

B

-200

- e7 -68

-43 I 0

I

I

2

4

6

0

2

4

6

days

-26

FIG. 8. The staged response to anoxia exhibited by FRE rat fibroblasts does not appear to be linked to cellular survival. REF 52 cells which do not exhibit the anoxic stress response survive anoxia equally well, if not better. Growth curves are shown for A, FRE cells, and B, REF 52 cells. Culture conditions were: continuous aerobic culture (O), anoxia for 1day, followed by aerobic culture (A), and anoxia for 2 days, followed by aerobic culture (0).

-18

(6)

1 2 3 4 5 6

7 8 9

14

loll -200

.e7

!akLw&