BRCA1 carries tumor suppressor activity distinct from that of ... - Nature

D 2001 Nature Publishing Group 0929-1903/01/$17.00/+0 www.nature.com/cgt

BRCA1 carries tumor suppressor activity distinct from that of p53 and p21 Voahangy Randrianarison,1,a Didier Marot,1,a Nicolas Foray,2 Jeannine Cabannes,1 Vincent Me´ret,2 Elisabeth Connault,1 Natacha Vitrat,3 Paule Opolon,1 Michel Perricaudet,1 and Jean Feunteun2 1

Laboratoire de Vectorologie et Transfert de ge`nes, CNRS UMR 1582, Institut Gustave Roussy, 94805 Villejuif Cedex, France; 2Laboratoire de Ge´ne´tique Oncologique, CNRS UMR 1599, Institut Gustave Roussy, 94805 Villejuif Cedex, France; and 3Laboratoire d’He´matologie et Cellules souches, INSERM U362, Institut Gustave Roussy, 94805 Villejuif Cedex, France.

The loss of BRCA1 function appears as an essential step in breast and ovarian epithelial cells oncogenesis and is consistent with the concept that BRCA1 acts as a tumor suppressor gene. However, the mechanism underlying this activity is not understood. In 1996, a retroviral vector was used for BRCA1 delivery to demonstrate that the transfer of BRCA1 inhibits breast and ovarian cancer cell growth. Since this early observation, the tumor growth inhibitory activity of BRCA1 in vivo has not been further documented. Here we re - address this issue and report experiments designed to evaluate the potential of adenovirus - mediated BRCA1 delivery to suppress the growth of cells with various status of endogenous BRCA1 in comparison with p53 and p21. Delivery of wild - type BRCA1 by an adenovirus vector in breast and ovarian tumor cells, decreases in vitro proliferation and tumorigenicity. Similarly, in vivo administration of BRCA1 provokes tumor growth retardation or regression comparable to that obtained with p53 or p21. The antitumor effect of BRCA1 is not observed upon transfer of a mutant lacking the 542 C - terminal residues. The p53 - or p21 - mediated antiproliferative activities are likely to bear on their capacity to induce apoptosis and / or interfere with cell cycle checkpoint. By contrast, the data presented here show that neither of these mechanisms can account for the BRCA1 - mediated antitumor activity and suggest the activation of an alternative route. Cancer Gene Therapy ( 2001 ) 8, 759 – 770 Key words: BRCA1; p53; p21; antitumor activity.

G

ermline mutation of BRCA1 is associated with a large fraction of hereditary breast and ovarian cancer syndromes.1 The BRCA1 gene, located on human chromosome 17q21, encodes a nuclear phosphoprotein of 1863 amino acid residues.2 The protein possesses distinct characteristic domains such as an N -terminal ring domain, a nuclear localization signal and two C - terminal BRCT repeat motifs also present in proteins involved in transcriptional regulation, in cell cycle checkpoint, or in DNA damage repair.3,4 Loss of heterozygosity affecting the wild -type allele is consistently observed in tumors from germline BRCA1 mutation carriers and is often coincident with high - grade stage in sporadic breast tumors.5 These observations strongly suggest that the loss of BRCA1 function is an essential step in breast and ovarian epithelial cells oncogenesis and support the role of BRCA1 as a tumor suppressor gene. The mechanism underlying this tumor suppressor activity is not understood.

Received June 19, 2001. Address correspondence and reprint requests to Dr Jean Feunteun, Laboratoire de Genetigue Oncolique, CNRS UMR 1599, Institut Gustave Roussy, 94805 Villejuif Cedex, France. E-mail address: feunteun@igr.fr a These two authors have contributed equally to this work.

Cancer Gene Therapy, Vol 8, No 10, 2001: pp 759 – 770

The role of BRCA1 in the surveillance of genome integrity is consensually recognized, although the precise biochemical activities involved in this function have not yet been identified. Several protein partners of BRCA1 are key players in DNA repair: Rad51,6 Rad50 /MRE11 / NBS,7 ATM,8 BLM.9 Furthermore BRCA1 is a substrate for several DNA damage– activated kinases.8,10 In addition, the association of BRCA1 with RNA polymerase II11 is consistent with the evidences for its contribution into transcription - coupled repair of oxidative lesions.12 – 14 Multiple studies have reported a role of BRCA1 as a transcription regulator. BRCA1 appears to co- regulate with p53 the transcription of reporter genes driven by p21 or mdm2 promoter.15 – 17 However, BRCA1 interacts with CBP /p300 and functions as a transcriptional coactivator.18 Finally it has been reported that a C -terminal BRCA1 domain alters the local chromatin structure and stimulates chromosomal DNA replication.19 The differential localization and phosphorylation of the protein throughout the cell cycle suggest that BRCA1 exerts some specific function at different stages of the cell cycle.6,20 Indeed, BRCA1 has been reported to play a role in the G0 / G1,21 G1 /S, or G2 /M checkpoints15,22,23 and induce apoptosis in various cell lines.24,25

759

760

RANDRIANARISON, MAROT, FORAY, ET AL: ANTITUMOR ACTIVITY OF BRCA1

The capacity of BRCA1 to exert an antiproliferative activity has been reported in several studies in vitro.21,26 – 29 Growth retardation and tumor inhibition in vivo were reported by Holt and colleagues28 after the transfer of the BRCA1 gene by retroviral vector. The aim of the present study was to evaluate the potential of adenovirus - mediated BRCA1 delivery as an inhibitor of tumor cell growth by comparison with the inhibitory activity of two well -known antiproliferative agents, namely p53 or p21. We demonstrate that BRCA1 exerts a growth inhibitory effect in vitro on three different tumor cells with various status for BRCA1. Such inhibitory effect is also observed in vivo on tumors established in nude mice with the same efficiency as that obtained with p53 or p21. Although the mechanism underlying this effect was not elucidated, it appeared not to involve cell cycle arrest or apoptosis as does p53 - or p21 -mediated effect. METHODS Construction of recombinant adenoviruses

All the adenovirus vectors used throughout this study are nonreplicative E1 /E3 - defective recombinant adenoviruses. AdRSVBRCA1 carries a full cDNA of human BRCA1 driven by the RSV promoter.13 Ad1025 expresses a C terminal truncated BRCA1 gene producing a protein deleted for 542 C - terminal amino acid residues. Ad1025 was generated spontaneously during the process of homologous recombination. AdCO1 and AdRSV gal are

control viruses carrying, respectively no insert or the E. coli lacZ transgene ( gal ). AdCMVp21HA ( Adp21 ) was constructed by cloning an HA -tagged p21 fragment obtained by HindIII and XbaI digestion of pcDNA3- p21HA plasmid. This fragment was subcloned between the HindIII and XbaI sites of pAdCMVp35IL12m plasmid to generate pAdCMVp21HA adenoviral shuttle vector. AdPGKp53 (Adp53 ) contains a p53 cDNA driven by a mouse PGK promoter with a rat globin polyadenylation signal. Cell cultures and viral infection

Human 293 cell lines were cultured in MEM medium supplemented with 1 mM L -glutamine and 10% fetal calf serum ( FCS ). Human MCF- 7 and HCC1937 (ATCC, Manassas, VA ) are breast tumor cell lines. They are maintained in DMEM medium supplemented with 10% FCS. Human IGROV1 ovarian cell lines30 are cultured in RPMI containing 10% FCS. The status for BRCA1 and p53 of these cell lines are as follows: MCF7 is BRCA1 wild - type / wild - type, p53 wild - type / wild - type ; IGROV1 is BRCA1 wild - type / mutant,31 p53 wild - type / mutant 32 , and HCC1937 is BRCA1 mutant / mutant, p53 unizygote mutant 33 . The optimal multiplicity of recombinant virus infection ( MOI ) was determined by staining for lacZ after infection with AdRSV gal. For all cell lines, 200 plaque -forming units (pfu ) per cell was determined as an optimal compromise between a maximal efficiency

β

Figure 1. Adenovirus - mediated expression of wild - type BRCA1 in tumor cell lines. Fifty micrograms of total protein extracted from uninfected or infected cells at day 1 ( D1 ), day 2 ( D2 ), or day 5 ( D5 ) postinfection were fractionated by SDS - PAGE, transferred to nitrocellulose, and immunostained with an anti BRCA1 monoclonal antibody directed against an N - terminal epitope ( residues 1 – 304 ). Cells were mock - infected ( cells ) or infected with adenoviruses expressing either Gal ( Ad Gal ), wild type BRCA1 ( Ad BRCA1.1 or Ad BRCA1.9 are two independent stocks ) or truncated BRCA1 ( Ad 1025 ). A: IGROV1 cells; B: MCF - 7 cells; C: HCC1937 cells.

β

Cancer Gene Therapy, Vol 8, No 10, 2001

761

β


Figure 2. Adenovirus - mediated expression of p53 or HA - tagged p21 in tumor cell lines and p14ARF status. A, B, C: 50 g of total protein extracted from uninfected cells or cells infected with various combinations of recombinant adenovirus were fractionated by SDS - PAGE, transferred to nitrocellulose, and immunostained with both anti - p53 and anti - p21 monoclonal antibodies. The recombinant viruses used are: Ad gal, AdCO1, Adp21, Adp53, or AdBRCA1. A: IGROV1 cells; B: MCF - 7 cells; C: HCC1937 cells. D: p14ARF status: 50 g of total protein extracted from uninfected cells were fractionated by SDS - PAGE, transferred to nitrocellulose, and immunostained with anti - p14ARF polyclonal antibodies.

of infection and a minimal virus -induced cytopathic effect. Antibodies and Western blot analyses

Primary monoclonal antibodies against BRCA1 (OP92 ), p53 ( OP53 ), p21 /Waf1 ( OP64 ), and PARP1 (AM30 ) were purchased from Oncogene Research Products ( Cambridge, UK ). The secondary peroxidase - conjugated goat anti -mouse IgG (H +L ) was from Jackson Immunoresearch Laboratories ( West Grove, PA ). The rabbit polyclonal antibody against p14ARF ( PC409 ) was from Oncogene Research Products. The secondary donkey anti -rabbit Ig ( NA934 ) was from Amersham Life Sciences ( Buckinghamshire, UK ).


Subconfluent cell monolayers were infected with either AdRSVBRCA1, AdPGKp53, AdCMVp21, Ad CO1, or AdRSV gal. Whole cell lysates were prepared as described.6 Briefly, cells were lysed at 48C in EBC buffer ( 50 mM Tris pH 8, 100 mM NaCl, 0.5% NP40 ) containing protease inhibitors (in g /mL: 100 PMSF, 20 aprotinin, 10 leupeptin ) and phosphatase inhibitors (in mM: 50 sodium fluoride, 1 sodium orthovanadate ). Fifty micrograms of proteins were fractionated by sodium dodecyl sulfate – polyacrylamide gel electrophoresis ( SDS - PAGE ). Six percent acrylamide gels were used when the analysis was focused on BRCA1 alone and 4– 12% gels when multiple proteins were analyzed. After fractionation the proteins were transferred onto nitro-

762


B

A

10

Cell number (x105)

Cell number (x105)

15

10

5

0 0

24

48

5

0

72

0

Time (hours)

Cell number (x105)

72

C

10

Figure 3. BRCA1 inhibits the proliferation of cancer cells. Cells infected with AdCO1, AdBRCA1, Adp53 or Ad p21 or mock - infected cells were trypsinized and counted 24, 48, and 72 hours after infection. The graphs shown are growth curves representing the data set ( each value corresponding to the average of two cell counts ) from one representative of two independent experiments. A: IGROV1, B: MCF - 7, C: HCC1937.

24 48 Time (hours)

PBS Ad C01 5

Ad p21 Ad p53 Ad BRCA1.1

0

0

24

48

72

Time (hours)

cellulose membrane ( Schleicher and Schuell, 0.2 -m BA83 ). After overnight blocking at 48C in TBS buffer containing 0.05% Tween 20 ( TBS -T ) and 5% wt /vol of nonfat dry milk, the primary monoclonal antibody was added at 1.5 g /mL for 3 hours at room temperature. After extensive washing in TBS -T buffer, the horseradish peroxidase - conjugated goat anti - mouse IgG was added for 30 minutes. After three washes in TBS -T buffer, the proteins were revealed using the Amersham ECL kit. For analysis of expression of BRCA1 in tumors, the tissues collected 9 or 13 days postinfection were frozen, powdered in liquid nitrogen, and dissolved in lysis buffer ( PBS containing 10 mM N -ethyl maleimide, 1% Triton X100, 1 mM PMSF, 0.1 M NH4OH ). After centrifugation at 12 000 rpm for 10 minutes at 48C, 150 g of proteins were fractionated on SDS -PAGE and the membrane treated as described above. Expression of p14ARF was tested by Western blotting of SDS -PAGE fractionated total cell extracts using the rabbit polyclonal antibody PC409. Proliferation assays

Cells were plated in 100 -mm Petri dishes and infected the following day with recombinant adenovirus. Four hours

after infection, cells were collected in PBS buffer containing 1 mM EDTA, washed three time with PBS, counted, and plated onto 35- mm Petri dishes at identical concentrations. Cells were counted 24, 48, or 96 hours after infection. Growth curves were drawn from data set ( each value corresponding to the average of two cell counts ) from one representative of two independent experiments.

Table 1. Adenovirus - Mediated Expression of Wild - Type BRCA1 Does Not Induce Apoptosis in Cancer Cell Lines HCC1937

Cells Ad CO1 Ad p53 Ad BRCA1

IGROV - 1

MCF - 7

D1

D2

D1

D2

D1

D2

4.8 15.8 48.4 2.6

9.9 17.4 42.8 14.2

3.6 2.6 92.4 2.6

4.0 15.3 48.4 16.2

4.6 4.6 77.9 4.0

19.6 13.3 31 12.7

IGROV1, MCF - 7, and HCC1937 cells were infected with various viruses: AdCO1, AdBRCA1, Adp53, or mock - infected ( cells ). One or 2 days ( D1 and D2 ) after infection apoptosis was evaluated using the Apoptag kit ( Appligene ).


763


600 PBS Ad βGal

Figure 4. PARP1 cleavage in IGROV1 cells cultures. PARP1 cleavage was evaluated by Western blot of 50 g of proteins extracted from cells, 1 ( D1 ) or 2 days ( D2 ) after infection. Cells: mock - infected lane. AdCO1, AdBRCA1 or Adp53: virus - infected cells. The molecular masses are 112 kDa for intact PARP1 and 85 kDa for the cleaved PARP1 fragment.

tumor size mean (mm3)

500

Ad BRCA1.1 Ad BRCA1.9

400 300 200

100

Assays for tumorigenicity of HCC1937 and IGROV1 cells in nude mice

The capacity of BRCA1 to interfere with the growth of tumor cells in vivo was assayed in two different settings. Ex vivo infection. IGROV1 cells were infected in vitro with recombinant viruses for 24 hours. Viable cells (2106 ) were inoculated into the dorsa of 6- week -old female nude mice. The cell viability was monitored by Trypan blue exclusion. Intratumoral infection. Tumors were generated by subcuta-

neous injection of 2106 HCC1937 or IGROV1 cells into the dorsa of 6- week -old nude mice. Tumors of approximately 50 mm3 (IGROV1 ) or 80 mm3 ( HCC1937 ) were

Table 2. Adenovirus - Mediated Expression of Wild - Type BRCA1 Does Not Alter Cell Cycle Distribution of Cancer Cell Lines Day 1

Day 2

% G1

%S

% G2

% G1

%S

% G2

A: IGROV1 Cells Ad CO1 Ad BRCA1 Ad p53 Ad p21

51 52 56 72 69

37 31 31 6 18

12 17 13 22 13

68 51 54 72 76

23 18 17 12 7

9 31 29 16 17

B: MCF - 7 Cells Ad CO1 Ad BRCA1 Ad p53 Ad p21

55 53 53 75 63

34 36 38 13 29

11 11 9 12 8

62 56 57 79 76

25 30 28 10 13

13 14 15 11 11

C: HCC1937 Cells Ad CO1 Ad BRCA1 Ad p53 Ad p21

47 47 47 43 57

27 27 29 17 20

26 26 24 40 23

49 47 47 44 45

30 31 31 18 30

31 22 22 38 25

One day ( D1 ) or 2 days ( D2 ) after infection cells were fixed in 70% ethanol, stained by propidium iodide, and analyzed by flow cytometry. Viruses: AdCO1, AdBRCA1, Adp53, or Adp21. Cells: mock - infected.


0 0

7

14

21

28

35

42

49

56

Time (days) Figure 5. Wild - type BRCA1 inhibits cancer cell tumorigenesis in nude mice. IGROV1 cells were infected in vitro with recombinant adenoviruses. Cells harvested 24 hours after infection were injected subcutaneously into 6 - week - old female nude mice. Tumor volumes were measured at regular intervals. The data represent the mean value of five to six treated mice with the standard deviation.

injected three times at 4- day intervals with 109 pfu of recombinant adenovirus. In both settings, tumor sizes were monitored weekly. Histopathological analysis

Tumor tissues were fixed in alcohol formalin acetic acid and embedded in paraffin. Five - micrometer sections treated with toluene and rehydrated were microwaved three times for 5 minutes in 10 mM citrate buffer (pH 6.0 ) and quenched by 3% H2O2 for 5 minutes. Slices were stained with hematoxylin – eosin. Cell cycle and apoptosis assays

For cell cycle studies, cells were collected in PBS –1mM EDTA buffer 24 or 48 hours after infection. After centrifugation at 1200 rpm, the cells were washed twice in PBS and fixed with 70% methanol for at least 2 hours at 208C. After centrifugation at 4000 rpm and one wash in PBS, the cells were suspended in PBS containing 0.25% Triton X -100 for 10 minutes. The DNA was stained by incubation for 2 hours in a solution containing 100 g / mL of RNAse A and 50 g/ mL of propidium iodide. Samples were analyzed by flow cytometry (EPICS -profile II, Coulter, Fullerton, CA ). Apoptosis analyses were carried out as recommended by the manufacturer of Apoptag plus fluorescein kit ( Appligene Oncor, Gaithersburg, MD ). Total cellular DNA was stained in red with propidium iodide and the 30OH ends marked in green by a fluorescein -labeled antidigoxigenin antibody. Samples were analyzed by flow cytometry.

764


β

Figure 6. Wild - type BRCA1 inhibits the growth of HCC1937 and IGROV1 tumors preestablished in nude mice. Tumors were obtained by injecting subcutaneously 2106 HCC1937 or IGROV1 cells in 200 L of PBS into 6 - week - old female nude mice. Tumors of approximately 50 mm3 ( IGROV1 ) or 80 mm3 ( HCC1937 ) received three successive intratumoral injections ( with 4 - to 6 - day intervals ) of PBS or various combinations of recombinant adenovirus. The vertical arrows indicate the time of injection. One experiment was performed with HCC1937 tumors ( A ) and three independent experiments ( B, C, D ) were performed with IGROV1 tumors. Each rectangle illustrates the mean value of tumor volumes of five to six mice with the standard deviation.


765


Construction of recombinant adenoviruses

The aim of the present study was to evaluate the potential of BRCA1 as an inhibitor of tumor cell growth by reference to p53 and p21. Recombinant adenoviruses expressing either wild -type or mutant alleles of BRCA1 were constructed for gene -expression studies both in vivo and in vitro. Adenovirus vectors expressing either wild - type p53 or wild - type p21 were also used. The construction of viral vectors expressing a gene putatively carrying a growth suppressor activity requires special attention with regard to the integrity of the cloned gene.34 The potential selective advantage of clones producing an allele with no antiproliferative activity should not be overlooked. Although such bias is expected to be less prevalent in the production of recombinant adenovirus, which can infect and express cloned genes in resting cells, homologous recombination carried out in 293 cells with a plasmid bearing a wild -type BRCA1 allele yielded predominantly recombinants producing truncated forms of BRCA1. However, several clones of recombinant AdRSVBRCA1 expressing full length and full sequence proved wild - type BRCA1 cDNA were obtained. The mutant Ad1025, generated in 293 cells, encodes a polypeptide comprising the 1399 N - terminal residues. Ad gal carries the E. coli galactosidase gene and AdCO1 has no expression cassette. The three coding sequences are driven by the RSV promoter. Western blot analysis ( Fig 1 ) performed on extracts prepared from recombinant virus – infected cells at day 1, 2, or 5 postinfection, shows overexpression of a 220 -kDa BRCA1 protein in IGROV1 ( A ) and MCF -7 (B ) cells as compared with mock -infected cells (Ad gal ). Ad1025infected IGROV1 cells express a truncated 150 -kDa BRCA1 protein. In HCC1937 cells ( C ), the wild - type protein is overexpressed by the recombinant virus although in this experiment the mobility shift due to the size difference with the truncated endogenous species (1863 vs. 1853 residues ) is not obvious. Recombinant Adp53, driven by the PGK promoter, expresses a 53 -kDa polypeptide in all cell types studied ( Fig 2A – C ). Infections with Adp21 -HA harboring HA -tagged p21 gene under CMV promoter induce the production of p21 protein, easily distinguishable from the endogenous p21 because of its higher electrophoretic mobility due to the three HA repeats tag (p21 HA ) ( Fig 2A –C ).

Expression of BRCA1 does not induce apoptosis

We have explored the possibility that the growth -inhibitory effect of BRCA1 expression observed markedly on IGROV1 and to a lesser extent on MCF7 and HCC1937 cells, may be due to apoptotic cell death. Apoptosis was measured at day 1 or day 2 after infection using two different assays: the Apoptag kit, which estimates the fragmentation of DNA and the PARP1 cleavage, which illustrates the activation of Caspase - 3 and Caspase -6. The results of the Apoptag assays presented on Table 1 demonstrates that adenovirus - mediated BRCA1 delivery does not induce apoptosis in neither cell type. In the same conditions, p53 delivered by the same vector induces massive apoptosis. These conclusions are reinforced by the results of the PARP1 cleavage experiment carried out on IGROV1 cells (Fig 4). The expression of BRCA1 does not lead to PARP1 cleavage as does the expression of p53. Cycle distribution of cells infected with BRCA1 - expressing vector

The cycle distribution of cells infected with BRCA1expressing vector was studied by cytofluorometric analysis over a period of 2 days following infection (Table 2 ). One day after infection of IGROV1 cells, there was no effect on cell cycle distribution attributable to BRCA1 expression whereas the delivery of p53 or p21 led to a marked G1 specific arrest (Table 2A ). Two days after infection there was still no change in cell cycle distribution specifically due to the expression of BRCA1, because the accumulation of G2 arrested cells observed upon infection with AdBRCA1 was also induced by the control adenovirus. The other cell types

β

RESULTS

nonspecific reduction of growth rate was observed upon infection with the empty adenovirus (AdCO1 ). This effect was enhanced in cells infected with a BRCA1 -expressing vector leading to an almost complete growth arrest 72 hours postinfection. A more pronounced effect is observed upon the delivery of either p21 or p53. Two other cell types derived from mammary tumors, namely MCF7 and HCC1937, appear to be less sensitive to BRCA1 expression (Fig 3B and C ).

β

Evaluation of apoptosis by PARP1 cleavage assay was performed on total cell extracts by Western blot using the PARP1 - specific AM30 antibody.

BRCA1 effects on cell proliferation

The consequences of adenovirus - mediated BRCA1 delivery on in vitro growth properties of three cell lines with various status for BRCA1 was evaluated by comparison with p53 or p21 delivery by the same vector. Figure 3A shows growth curves of IGROV1 cells derived from an ovarian tumor, infected with various recombinant viruses. A significant


Figure 7. Adenovirus - mediated expression of wild - type BRCA1 in IGROV1 tumors established in nude mice. PBS or 109 pfu of either control Ad Gal or AdBRCA1 were injected in situ in tumors at days 0 and 4. Tumors were collected from sacrificed animals at day 9 ( A ) or day 13 ( B ) after the last injection. One hundred fifty micrograms of total protein extracts were analyzed by Western blotting.

766


Figure 8. Evidence for tumor regression in Ad - BRCA1 – treated IGROV1 tumors. Tumor sections were obtained at day 56 after injection of recombinant viruses and stained by hematoxylin – eosin. Each panel presents two or three slices of the same tumor. A: PBS - treated tumor, B: control AdCO1 - treated tumor, C and D: AdBRCA1 - treated tumors, E: Adp21 - treated tumor, F: Adp53 - treated tumor.

tested were not sensitive to this nonspecific effect, which has been previously described.35 In MCF7 cells, overexpression of BRCA1 did not change the cell cycle distribution in contrast with the strong G1 arrest effect induced by p21 and p53 expression ( Table 2B ). Similarly, restoration of wild - type BRCA1 protein in the BRCA1- deficient HCC1937 cells, had essentially no effect

on cell cycle distribution (Table 2C ). These cells carry mutant p53 alleles and express very low level of p21. The G1 arrest mediated by p21 delivery is limited and temporary. Upon p53 delivery, shown to induce endogenous p21 expression ( see Figure 2C ), the G2 arrest is overrepresented. Together these observations suggest that the G1 checkpoint may be loose in HCC1937 cells.



Transactivation properties of p53 and BRCA1

Transduction of cells with Adp53 concurrently activates the endogenous p21 expression in the three cell lines ( Fig 2A – C ). The p21 activation is more pronounced in IGROV1 cells ( A ) and HCC1937 cells ( C ) in which the level of endogenous protein is low, probably because of the mutant status of endogenous p53. However, it is noteworthy that in all cell types, wild -type BRCA1 by itself does not increase the level of endogenous p21. These observations rule out an additive effect of BRCA1 and p53 together to activate p21 expression. Somasundaram et al36 have reported an increase of p53 level mediated by BRCA1 only in cells expressing functional p14ARF. This prompted us to analyze the status of p14 ARF in the three cell lines used throughout this study. This was done by Western blotting and illustrated on Figure 2D. p14ARF is undetectable in IGROV1 and MCF7 cells whereas its presence is clearly demonstrable in HCC1937 cells. Mutations in the p14 ARF coding sequence of IGROV1 and MCF7 have been demonstrated ( Bressac de Paillerets et al, unpublished observations ). They may lead to protein instability and account for the difference in protein level. These data support the hypothesis that the BRCA1- mediated antiproliferative activity is independent of the p53 pathway. Expression of BRCA1 inhibits breast and ovarian cancer cell growth in vivo

In vivo assays for antitumor activity were carried out in nude mice bearing tumors obtained by subcutaneous injection of HCC1937 or IGROV1 cells. The other cell lines ( MCF7 ) used throughout the in vitro experiments grew poorly in the animals. The capacity of wild - type BRCA1 to interfere with the tumorigenicity of these cells was assayed either by grafting tumor cells infected in vitro (IGROV1 ) or by intratumoral infection of preestablished tumors (HCC1937 and IGROV1 ). IGROV1 cells were infected in vitro with recombinant virus for 24 hours before injection into the dorsa of nude mice. No difference was observed in the number of viable cells monitored by Trypan blue exclusion, between recombinant virus and mock -infected cells. A representative of three concordant independent experiments is shown in Figure 5. Transplanted mock -infected or Ad Gal - infected cells grow efficiently in vivo, whereas the growth of cells infected with two independent wild - type AdBRCA1 virus stocks ( AdBRCA1- 1 and AdBRCA1 - 9) is totally prevented. This strong effect prompted us to determine whether the delivery of BRCA1 gene by recombinant adenovirus would induce an inhibitory effect on established tumors. HCC1937 and IGROV1 calibrated tumors established in female nude mice were injected by three successive doses of 109 pfu of recombinant adenovirus every 4 days. Three independent experiments involving five or six mice per group gave consistent results. Figure 6A illustrates the monitoring of the HCC1937 tumor size over a period of approximately 1 month. The growth of the tumors injected with either AdBRCA1 or Adp53 was essentially arrested compared to


767

that of controls tumors injected with PBS or the empty vector ( AdCO1 ). Figure 6B shows a similar effect on the growth rate of IGROV1 tumors injected with AdBRCA1 ( two independent stocks). At day 58 postinfection, the volume of the tumors injected with AdBRCA1 reached in average one third of the volume of Ad Gal -injected control tumors ( P= .05). The mutant BRCA1 producing a truncated protein lacking 542 C terminal amino acid residues ( Ad1025 ) was defective in this assay. Western blot analysis of tumor extracts demonstrated an efficient in situ expression of full -length BRCA1 protein ( Fig 7 ). It is noteworthy that the steady -state level of the BRCA1 protein in the tumor tissue was stable over a period of at least 13 days postinfection. The sensitivity of IGROV1 growth to BRCA1 expression was comparable to that obtained with recombinant adenovirus expressing two different and well - characterized antiproliferative agents namely p53 or p21 (Fig 6C). Furthermore, no additive effect was observed in infections combining BRCA1 and p53 or BRCA1 and p21 (Fig 6D ). Representative histopathological slides of IGROV1 tumor samples stained by hematoxylin – eosin are presented on Figure 8. The tumors display the classical morphology of ovarian adenocarcinoma. Neither PBS treatment nor infection with empty adenovirus vector ( AdCO1 ) prevented tumor growth (Fig 8A and B ). By contrast, among the five AdBRCA1 - treated tumors one complete ( Fig 8C ) and four partial regression leading to very small tumor residues (Fig 8D ) were observed. There was no evidence for an increased necrosis to account for the tumor reduction. A strong antitumor growth effect was also obtained after infection with adenovirus vectors expressing either p21 (Fig 8E ) or p53 (Fig 8F ). Similar observations were made on HCC1937 tumors ( data not shown ). DISCUSSION

Multiple independent evidences converge to support the role of BRCA1 as a tumor suppressor gene in breast and ovarian cancers.1 So far, the most decisive argument remains the loss of wild -type BRCA1 allele observed in tumors from BRCA1 mutant carriers. Furthermore, although there is no demonstration for BRCA1 somatic mutations in sporadic cancers, the level of BRCA1 protein was reported to be consistently lower in advanced grade tumors.5 Finally, the role of brca1 as a caretaker in murine mammary tissue has been documented by gene -inactivation experiments.37 These observations have led to the hypothesis that restoration of wild -type BRCA1 expression may suppress the tumorigenic potential of cells that lack functional BRCA1 and therefore represents a possible therapeutic approach for mammary or ovarian cancers. In 1996, Holt et al28 used a retroviral vector for BRCA1 delivery to demonstrate that the transfer of BRCA1 inhibits breast and ovarian cancer cell growth. Intraperitoneal treatment of established MCF- 7 tumors with this retroviral vector were reported to trigger tumor growth retardation, and to increase survival. Finally, when used in a phase I trial in ovarian cancer, this vector mediated tumor growth stability

768


or reduction.38,39 Subsequently, the retroviral vector was shown to express a variant of BRCA1 lacking 71 N -terminal amino acid residues.40 Since this early observation, the tumor growth inhibitory activity of BRCA1 in vivo has not been further documented. Here we have re -addressed this issue and report experiments designed to evaluate the potential of adenovirus -mediated BRCA1 delivery to suppress the growth of cells with various status of endogenous BRCA1. These assays were carried out in tissue culture conditions for three cell lines MCF7, IGROV1, and HCC1937, and in nude mice for IGROV1 and HCC1937 cells. Growth -inhibitory effect were observed in cell culture. Blockade within the cell cycle or apoptotic cell death are the most common mechanisms underlying growth arrest. Although BRCA1- induced growth arrest has been reported in various cell systems, so far there is no consensus as to a common mechanism underlying this biological effect. Differential response such as G0 / G1 arrest,21 G1 arrest,29 G1 or G2 arrest,22 or apoptosis25 have been described. These apparent discrepancies may indeed reflect differences in the status of specific cell types for major cell cycle regulators such as Rb29 or p21.22 In addition, murine brca1 might regulate the G2 -M transition by controlling the assembly of mitotic spindles.23 In our hands, adenovirus -mediated BRCA1 expression induces neither significant changes in cell cycle profile nor apoptosis. We hypothesize that the reduced growth rate of the cells infected with AdBRCA1 reflects a generalized slowing down of the cell cycle without any specific checkpoint blockage. A similar observation has been reported in prostate tumor cells stably expressing BRCA1.41 Here, we report the absence of BRCA1- induced apoptosis demonstrated by three independent approaches: lack of apoptag staining, lack of sub -G1 fraction, and lack of PARP1 cleavage. Increased apoptosis of serum - deprived NIH3T3 cells and calcium ionophore – treated MCF7 cells have been described upon transfection of BRCA1.24 Furthermore, BRCA1 expression was reported to induce apoptosis via JNK /SAPK activation apoptosis.25 In contrast, others have reported the lack of apoptosis response in different situations.22,41 The possibility that these discrepancies might be related to the p53 status seems to be excluded by the data presented here, because the lack of apoptosis is observed in cells that are either homozygote wild -type (MCF7 ), heterozygote (IGROV1 ), or unizygote mutant for p53 ( HCC1937 ). In our hands, the expression of BRCA1 does not lead to the accumulation of endogenous p53 in consistence with the lack of stimulation of p21 expression. Moreover, we have no evidence for a direct effect of BRCA1 on the steady -state level of p21. These observations are in agreement with the absence of significant G1 or G2 cell cycle arrest. The potential activation of p53 and p21 by BRCA1 remains controversial. Harkin et al25 have observed neither activation nor repression of endogenous p21 and no activation of p53 following the inducible expression of BRCA1. Similarly, Fan et al41 did not observe significant modification of p21 or p53 expression in clones of prostate tumor cells DU145 stably expressing wild -type BRCA1. By contrast MacLachlan et al22 have reported the onset of a pathway leading to

activation of p53 and p21 upon expression of BRCA1. The use of cell lines with different status for pRb, p53, p21, p16, p14 ARF, and others may account for these discrepancies. We show no activation of p53 and p21 upon expression of BRCA1 in the three cell lines used in this study. This result is not unexpected for IGROV1 and HCC1937, which carry mutant p53 alleles, but is surprising for MCF7, which bears wild -type p53. It may be explained by differences in p14 ARF status as suggested in the model of Somasundaram et al.36 Indeed, the lack of activation observed in our MCF7 could be related to their status of p14ARF deficiency. However, this remains in discrepancy with MacLachlan et al22 who reported activation in their MCF7 cells. This difference may reflect clonal divergence in MCF7 cells from one laboratory to another. Concerning p21, it is noteworthy that IGROV1 cells display elevated endogenous level of cdk2 and PCNA, which sequester p21.42 Therefore IGROV1 cells can be considered as functionally null for p21 making it unlikely that p21 plays a role as an intermediate in the antitumor activity of BRCA1. We show in this report that this sequestering effect can be bypassed upon infection with a recombinant adenovirus expressing p21, which results in cell growth inhibition both in vivo and in vitro. The data presented here clearly show that adenovirus mediated delivery of BRCA1 inhibits ovarian and breast cancer cell growth in vivo. We observe that: (i) in vitro infection of an ovarian tumor cell (IGROV1 ) by the AdBRCA1 vector prevents its capacity to grow in nude mice. (ii ) In situ injections of the AdBRCA1 vector in IGROV1 and HCC1937 established tumors lead to growth arrest, and in some cases to regression. This inhibitory effect is comparable to that obtained upon delivery of either p53 or p21, although the expression of each of the three genes is driven by a different promoter. No additive effect was observed when BRCA1 was coexpressed with either p53 or p21. The activity of p53 on ovarian tumors in such setting has already been documented,43 as well as the activity of p21 in murine mammary tumors.44 Finally, our observation that both in vivo and in vitro of a growth - inhibitory effect of BRCA1 on HCC1937 cells that lack functional p53 rules out the possibility that p53 mediates this BRCA1 activity. Aprelikova et al29 reached a similar conclusion and emphasized the role of Rb in this BRCA1 -associated growth arrest. AdBRCA1 injection in IGROV1 and HCC1937 tumors established in nude mice induces a marked growth inhibition and even a complete total regression in some cases. Considering the limited diffusion of the virus in these solid tumors ( data not shown), such response cannot be attributed solely to BRCA1 -infected cells and must involve a paracrine bystander effect on neighboring noninfected cells. Inhibition of angiogenesis may have contributed for this effect as previously reported for p53.45 This hypothesis should be explored in the future. In conclusion, we have developed a tractable genetic system owing to adenoviral vector for wild -type BRCA1 delivery. Restoration of high expression of BRCA1 in breast and ovarian tumor cells decreases in vitro proliferation and tumorigenicity. Similarly, in vivo administration of BRCA1



provokes tumor- growth retardation or regression comparable to that obtained with p53 or p21. It is likely that these p53 - or p21 -mediated effects bear on their capacity to induce apoptosis and/ or interfere with cell cycle checkpoint. The data presented here suggest that neither of these mechanisms can account for the BRCA1- mediated antitumor activity that is clearly distinct from that of p53 and p21. ACKNOWLEDGMENTS

We thank Arlette Vervisch ( Cytometry laboratory — CNRS ) for her excellent technical assistance in cell cycle analysis, Patrice Ardouin for his help in animal care, Vladimir Lazar for a major contribution in sequencing, and Nazanine Modtjahedi for stimulating discussions. This work was supported by CRC contract 97 -07 from Institut Gustave -Roussy and a Contrat Libre (No. 9023 ) from ARC. N.F was supported by postdoctoral fellowships from the Fondation pour la Recherche Me´dicale ( FRM ) and from Electricite´ de France (De´partement de Radioprotection ). REFERENCES 1. Feunteun J. Breast cancer and genetic instability: the molecules behind the scenes. Mol Med Today. 1998;4:263 – 267. 2. Miki Y, Swensen J, Shattuck - Eidens D, et al. A strong candidate for the breast and ovarian cancer susceptibility gene BRCA1. Science. 1994;266:66 – 71. 3. Callebaut I, Mornon JP. From BRCA1 to RAP1: a widespread BRCT module closely associated with DNA repair. FEBS Lett. 1997;400:25 – 30. 4. Koonin VF, Altschul SF, Bork P. BRCA1 protein products: functional motifs. Nat Genet. 1998;13:266 – 267. 5. Wilson CA, Ramos L, Villasen˜or MR, et al. Localization of human BRCA1 and its loss in high - grade, non - inherited breast carcinomas. Nat Genet. 1999;21:236 – 240. 6. Scully R, Chen JJ, Plug A, et al. Association of BRCA1 with Rad51 in mitotic and meiotic cells. Cell. 1997;88:265 – 275. 7. Zhong Q, Chen CF, Li S, et al. Association of BRCA1 with the hRad50 – hMre11 – p95 complex and the DNA damage response. Science. 1999;285:747 – 750. 8. Cortez D, Wang Y, Qin J, et al. Requirement of ATM dependent phosphorylation of BRCA1 in the DNA damage response to double - strand breaks. Science. 1999;286:1162 – 1166. 9. Wang Y, Cortez D, Yazdi P, et al. BASC, a super complex of BRCA1 - associated proteins involved in the recognition and repair of aberrant DNA structures. Genes Dev. 2000;14:927 – 939. 10. Lee JS, Collins KM, Brown AL, et al. hCds1 - mediated phosphorylation of BRCA1 regulates the DNA damage response. Nature. 2000;404:201 – 204. 11. Anderson SE, Schlegel BP, Nakajima T, et al. BRCA1 protein is linked to the RNA polymerase II holoenzyme complex via RNA helicase a. Nat Genet. 1998;19:254 – 256. 12. Abbott DW, Thompson ME, Robinson - Benion C, et al. BRCA1 expression restores radiation resistance in BRCA1 defective cancer cells through enhancement of transcription coupled DNA repair. J Biol Chem. 1999;274:18808 – 18812.


769

13. Le Page F, Kwoh EE, Avrutskaya A, et al. Transcription coupled repair of 8 - oxoguanine: requirement for XPG, TFIIH, and CSB and implications for Cockayne syndrome. Cell. 2000;101:159 – 171. 14. Le Page F, Randrianarison V, Marot D, et al. BRCA1 and BRCA2 are necessary for the transcription - coupled repair of the oxidative 8 - oxoguanine lesion in human cells. Cancer Res. 2000;60:5548 – 5552. 15. Somasundaram K, Zhang HB, Zeng YX, et al. Arrest of the cell cycle by the tumour - suppressor BRCA1 requires the CDK inhibitor p21WAF1 / CiP1. Nature 1997;389:187 – 190. 16. Zhang HB, Somasundaram K, Peng Y, et al. BRCA1 physically associates with p53 and stimulates its transcriptional activity. Oncogene. 1998;16:1713 – 1721. 17. Ouchi T, Monteiro ANA, August A, et al. BRCA1 regulates p53 - dependent gene expression. Proc Natl Acad Sci USA. 1998;95:2302 – 2306. 18. Pao GM, Janknecht R, Ruffner H, et al. CBP / p300 interact with and function as transcriptional coactivators of BRCA1. Proc Natl Acad Sci USA. 2000;97:1020 – 1025. 19. Hu YF, Hao ZL, Li R. Chromatin remodeling and activation of chromosomal DNA replication by an acidic transcriptional activation domain from BRCA1. Genes Dev. 1999;13:637 – 642. 20. Chen YM, Farmer AA, Chen CF, et al. BRCA1 is a 220 - kDa nuclear phosphoprotein that is expressed and phosphorylated in a cell cycle - dependent manner. Cancer Res. 1996;56:3168 – 3172. 21. Campbell M, Aprelikova ON, Van der Meer R, et al. Construction and characterization of recombinant adenoviruses expressing human BRCA1 or murine Brca1 genes. Cancer Gene Ther. 2001;8:231 – 239. 22. MacLachlan TK, Somasundaram K, Sgagias M, et al. BRCA1 effects on the cell cycle and the DNA damage response are linked to altered gene expression. J Biol Chem. 2000;275:2777 – 2785. 23. Xu X, Weaver Z, Linke SP, et al. Centrosome amplification and a defective G2 - M cell cycle checkpoint induce genetic instability in BRCA1 exon 11 isoform – deficient cells. Mol Cell. 1999;3:389 – 395. 24. Shao NS, Chai YL, Shyam E, et al. Induction of apoptosis by the tumor suppressor protein BRCA1. Oncogene. 1996;13:1 – 7. 25. Harkin DP, Bean JM, Miklos D, et al. Induction of GADD45 and JNK / SAPK - dependent apoptosis following inducible expression of BRCA1. Cell. 1999;97:575 – 586. 26. Thompson ME, Jensen RA, Obermiller PS, et al. Decreased expression of BRCA1 accelerates growth and is often present during sporadic breast cancer progression. Nat Genet. 1995;9:444 – 450. 27. Rao VN, Shao NS, Ahmad M, et al. Antisense RNA to the putative tumor suppressor gene BRCA1 transforms mouse fibroblasts. Oncogene. 1996;12:523 – 528. 28. Holt JT, Thompson ME, Szabo C, et al. Growth retardation and tumour inhibition by BRCA1. Nat Genet. 1996;12:298 – 302. 29. Aprelikova ON, Fang BS, Meissner EG, et al. BRCA1 associated growth arrest is RB - dependent. Proc Natl Acad Sci USA. 1999;96:11866 – 11871. 30. Benard J, Da Silva J, De Blois MC, et al. Characterization of a human ovarian adenocarcinoma line, IGROV1, in tissue culture and in nude mice. Cancer Res. 1985;45:4970 – 4979. 31. Foray N, Randrianarison V, Marot D, et al. Gamma - rays induced death of human cells carrying mutations of BRCA1 or BRCA2. Oncogene. 1999;18:7334 – 7342. 32. Fajac A, Da Silva J, Ahomadegbe JC, et al. Cisplatin - induced apoptosis and p53 gene status in a cisplatin - resistant human ovarian carcinoma cell line. Int J Cancer. 1996;68:67 – 74.

770


33. Tomlinson GE, Chen TTL, Stastny VA, et al. Characterization of a breast cancer cell line derived from a germ - line BRCA1 mutation carrier. Cancer Res. 1998;58:3237 – 3242. 34. Estreicher A, Iggo R. Retrovirus - mediated p53 gene therapy. Nat Med. 1996;2:1163. 35. Wersto RP, Rosenthal ER, Seth PK, et al. Recombinant, replication - defective adenovirus gene transfer vectors induce cell cycle dysregulation and inappropriate expression of cyclin proteins. J Virol. 1998;72:9491 – 9502. 36. Somasundaram K, MacLachlan TK, Burns TF, et al. BRCA1 signals ARF - dependent stabilization and coactivation of p53. Oncogene. 1999;18:6605 – 6614. 37. Xu XL, Wagner KU, Larson D, et al. Conditional mutation of Brca1 in mammary epithelial cells results in blunted ductal morphogenesis and tumour formation. Nat Gen. 1999; 22:37 – 43. 38. Tait DL, Obermiller PS, Hatmaker AR, et al. Ovarian cancer BRCA1 gene therapy: phase I and II trial differences in immune response and vector stability. Clin Cancer Res. 1999; 5:1708 – 1714. 39. Tait DL, Obermiller PS, Redlin - Frazier S, et al. A phase I trial of retroviral BRCA1sv gene therapy in ovarian cancer. Clin Cancer Res. 1997;3:1959 – 1968.

40. Holt JT, Thompson ME, Szabo C, et al. Growth retardation and tumor inhibition by BRCA1 ( correction ). Nat Genet. 1998;19:102. 41. Fan SJ, Wang JA, Yuan RQ, et al. BRCA1 as a potential human prostate tumor suppressor: modulation of proliferation, damage responses and expression of cell regulatory proteins. Oncogene. 1998;16:3069 – 3082. 42. Barboule N, Baldin V, Jozan S, et al. Increased level of p21 in human ovarian tumors is associated with increased expression of cdk2, cyclin A and PCNA. Int J Cancer. 1998;76:891 – 896. 43. Santoso JT, Tang DC, Lane SB, et al. Adenovirus - based p53 gene therapy in ovarian cancer [ see comments ]. Gynecol Oncol. 1995;59:171 – 178. 44. Schreiber M, Muller WJ, Singh G, et al. Comparison of the effectiveness of adenovirus vectors expressing cyclin kinase inhibitors p16INK4A, p18INK4C, p19INK4D, p21( WAF1 / CIP1 ) and p27KIP1 in inducing cell cycle arrest, apoptosis and inhibition of tumorigenicity. Oncogene. 1999;18:1663 – 1676. 45. Dameron KM, Volpert OV, Tainsky MA, et al. The p53 tumor suppressor gene inhibits angiogenesis by stimulating the production of thrombospondin. Cold Spring Harbor Symp Quant Biol 1994;59:483 – 489.
