Vol. 6, 673-680, june 1995 Cell Growth & Differentiation 673

Recombinant Human Retinoblastoma ProteinInhibits Cancer Cell Growth’

Lance C. Pagliaro, Douglas Antelman, Duane E. Johnson,Todd Machemer, Ernest A. McCulloch,2 Emil J. Freireich,Sanford A. Stass,3 H. Michael Shepard, Dan Maneval,and Jordan U. Gutterman4

Divisions of Medicine IL. C. P., E. J. F., J. U. G.J and Laboratory Medicine

IE. A. M., S. A. SI, The University of Texas M. D. Anderson CancerCenter, Houston, Texas 77030; and Canji, Inc., San Diego, California

92121 ID. A., D. E. J., T. M., H. M. S., D. M.I

Abstrad

Aberrant expression of the tumor suppressor gene RB1 isassociated with a variety of solid tumors andhematopoietic neoplasms. Certain cancer cell lines inwhich the protein encoded by RB1 (p1 1 0RB) is absenthave been reported to show decreased growth rate,clonogenicity, or tumorigenicity following insertion of atranscriptionally adive RB1 gene. We asked whetherthese RB-deficient cells could be growth inhibited bydired exposure to purified p1 ‘I 0R0#{149}We report adecrease in uptake of tritiated thymidine by 5637bladder carcinoma cells (RB-negative) when purifiedrecombinant p1 1 0RB is added to culture media.Internalization of the protein by cells and translocationto the nucleus are demonstrated byimmunohistochemistry, FACS, and detedion ofradiolabeled protein in subcellular fradions. Next, wechose a well-described leukemia cell culture model toinvestigate the potential effect of recombinant p1 1 0R8 inclinical disease. We observed dose-related decreases incell number or colony formation in vitro in 8 of 20acute myelogenous leukemia samples, 7 of which didshow endogenous p1 1 0RB detectable byimmunohistochemistry. Histological appearancefollowing exposure to p1 1 0RB shows cytoplasmicvacuolization and nuclear lobulation of degeneratingcells. We conclude that purified p1 1 0RB added toculture media is internalized by cells, translocated to thenucleus, and exerts a growth-inhibitory effed on certaincancer cell types.

IntrodudionTumor suppressor genes have been shown to play impor-tant roles in the pathogenesis of cancer in humans (1 , 2).Their attraction as potential therapeutic targets lies in thepossibility that restoration of their function in tumor cells

Received 2/1/95; revised 3/21/95; accepted 3/29/95.1 This research was supported in part by a grant from the Biomedical

Research Foundation.2 Present address: Ontario Cancer Institute/Princess Margaret Hospital,Toronto, Ontario, Canada M4X 1 K9.

3 Present address: Department of Pathology, The University of Maryland,Baltimore, MD 21201.4 To whom requests for reprints should be addressed, at The University of

Texas M. D. Anderson Cancer Center, Box 41 , 1 51 5 Holcombe Blvd.,Houston, TX 77030.

will interfere with cell proliferation. This idea is supportedby experimental evidence in the case ofthe RB!5 gene: thein vitro proliferation and in vivo tumorigenicity of RB-negative tumor cell lines is inhibited by insertion of a tran-scriptionally active copy of the gene (3-5). The RB! geneprotein, p1 1 0RB has been shown to block progressionthrough the G1 phase of the cell cycle when a truncatedmolecule containing the binding domain is microinjecteddirectly into cultured RB-negative tumor cells (6). The Mr

1 1 0,000 nuclear protein binds DNA and interacts withother nuclear proteins including the transcription factorE2F, the nuclear tyrosine kinase c-abl, and D cychins (7-10).Phosphorylation of multiple serine and threonine residuesinactivates p1 10RB and allows DNA synthesis and celldivision to occur (1).

The p1 10RB protein plays a crucial role in regulating theproliferation of cells (1). Patients who inherit an abnormalRB! allele develop retinal tumors, osteosarcomas, and softtissue sarcomas associated with loss of the remaining wild-type allele (1 1-1 3). Mice that are heterozygous for mutantRB! develop pituitary tumors, whereas homozygous mutantembryos die by the sixteenth day ofgestation with abnormaldevelopment of the hematopoietic and central nervous sys-tern tissues (1 4, 1 5). The mechanism through which p1 1 0RB

controls gene expression is not known, but the presence ofbinding to DNA and to transcription factors such as E2Findicates participation in a complex that binds to regulatoryelements in the genome. In experimental models and in thenatural history of diseases such as hereditary retinoblas-toma, loss of p1 10RB is associated with uncontrolled celldivision and progression toward malignancy.

We have produced recombinant human p1 1 0RB from E.

co/i at the laboratories of Canji, Inc., making it possible totest the protein at concentrations up to 1 �M for its effect oncells in culture. We asked whether direct exposure topllO�#{176}could regulate cell growth. There are several pre-cedents for extracellular proteins affecting gene expressionthrough cellular uptake and nuclear translocation. Theseinclude the tat protein of HIV, a 60-amino acid syntheticpeptide (pAntp) derived from the Drosophila antennapediahomeobox gene, bFGF, and insulin (16-20).

The bladder cancer cell line 5637 (ATCC HTB-9) ex-presses no functional pi 1 0RB and has been reported previ-ously to exhibit decreased growth in vitro, decreased col-ony formation in soft agar, and reduced tumorigenicity invivofollowing transfection with the wild-type RB! gene (1,5). First, we asked whether exposure of the 5637 parent cellline to recombinant p1 1 0RB would result in growth inhibi-tion. Second, to further investigate the effect of recombinantp1 10RB on self-renewal in a neoplastic cell population, welooked for an effect on clonogenicity of blast cells frompatients with AML. Previous reports have shown decreased

5 The abbreviations used are: RB, retinoblastoma; bFGF, basic fibroblastgrowth factor; AML, acute myelogenous leukemia; cdk, cyclin-dependentkinase; FBS, fetal bovine serum; CSF, colony-stimulating factor.

3H-Thy Uptake(% of Medium

Control)

120

i:

10 100

p11ORB (nM)

1000

Patient number

A

C80I

B

=

C0

(3

Fig. 2. A, dose-dependent reduction in AML blast clonogenic cell recovery.

Prior to plating, cells were incubated 48 h in liquid media with singleaddition of buffer (LI), p1 1 0” 1 00 nM (a), or p1 1 0R11 200 n�a (U). Significantinhibition is seen at 200 nM (P< 0.03 compared to buffer). The means of fourreplicate wells are shown; bars, SEM. Patients nos. 3 and 10 are not shown

(see text). B, a p1 10”-binding peptide, but not a mutant peptide, neutralizesgrowth inhibition by recombinant p1 10R��#{149}T peptide has the amino acidsequence NLFCSEEMPSSDDE, which is derived from SV4O T-antigen andbinds the p110’#{176}’A/B pocket. K peptide has a single amino acid substitution

(NLFCSKEMPSSDDE( and does not bind p1 � Prior to plating, blast cellsfrom patient no. 14 were incubated with mixtures as shown for 24 h

(concentrations: 200 ma pllO’#{176}#{176};250 nsa T peptide; and 250 nsi K peptide).Reduction in colony formation is significant with p1 l0�’ alone and with

p1 1 0R0 � K peptide (P = 0.03). The means offour replicate wells are shown;bars, SEM.

674 Recombinant RB Protein Inhibits Cancer Cell Growth

Fig. 1. Dose-response relationship of p1 1 �5�3 added to culture niedia andnwasured uptake of I Hithymidine by 5637 cells. Concentrations of p1 10R��

at the beginning of the experiment are shown. One-half of the cell medium

was removed and replaced with p1 10R���containing medium twice each dayfor 3 days. I HiThymidine incorporation assayed by liquid scintillation

counting; mean of tour replicate samples shown. Bars. SEM.

or absent levels of p1 1 0Rtt in AML blasts from 25-30% ofpatients studied, despite only rare occurrence of detectableRB! gene alterations (21-25). We chose the AML blast cellculture model to study the effect of pi i 0RB on clonogenic-ity because: (a) the genotype of freshly obtained blastsreflects that of cells in vivo, unlike cell lines that bearmutations accumulated during serial passage; (b) largenumbers of blast cells can be obtained from patients for cellculture; and (C) previous studies have demonstrated thateffects on clonogenic cell recovery in vitro are predictive ofevents in vivo (26-28).

In this report, we show that recombinant p1 1 0’� inhibitsthe growth rate of 5637 cells, and we describe 8 cases inwhich self-renewal of cultured AML blasts, measured asclonogenic cell recovery, is inhibited by pi 10RB Data arealso presented to show that pi 10RB is internalized by cellsin vitro and translocated to the nucleus.

Results

5637 Cells Exposed to p1 10R8 Are Growth Inhibited. Di-rect exposure to pi i 0RU is associated with a dose-relateddecrease in uptake of l#{176}Hlthymidine by 5637 cells, mdi-cating decreased DNA synthesis (Fig. i ). Continuous expo-sure to pi 10Rt3 concentrations greater than 100 nM weregrowth inhibitory. Cells incubated with medium aloneyielded counts of i 26,000 to i 32,000 cpm. At the highestconcentration of pi 1 0� (1 pM), incorporation was

approximately 50% of control (P < 0.Oi).Decreased Growth Rate and Clonogenicity of Blast Cells

from Patients with AML. Blasts were obtained from periph-eral blood or bone marrow specimens from 20 patients withAML. Inhibition of growth rate or clonogenic cell recoveryfollowing exposure to pi i 0RB was observed in eight casesand was dose dependent (Fig. 2A). Table i shows the resultsof immunostaining as a measurement of endogenouspi i 0R�� and the clinical characteristics of these eight cases.The sample obtained from patient no. 3 did not yield suf-ficient cells for the clonogenic assay but did show a markeddecrease in cell number relative to buffer control after 72 hin suspension culture with pi 1 0R0 added once at the be-ginning of the experiment (100 nM, 60% of buffer control;500 nra, 25#{176}hof buffer control). In the absence of pi i 0RB,

the number of blast cells in suspension culture increasedexponentially. Patient no. 10 did not show a significantdecrease in clonogenic cell recovery following exposure top1 i 0RB at 1 00 and 200 n�t concentrations but was sign ifi-cantly growth inhibited at 400 nM (P< 0.05) and completelyinhibited at 500 nM. Less pronounced growth inhibitionwith patients nos. 7 and i 0 may have been due to degra-dation of pi i 0RB during storage at 5#{176}C.There were notsufficient cells from patients nos. 7 and 10 to repeat thosedeterminations. Decreased progenitor cell recovery in theclonogenic assay after exposure to pi iORB was reproduc-ibly demonstrated over the course of numerous experimentswith samples from patients nos. 4, 5, i 2, and i 4, using cells

Cell Growth & Differentiation 675

Table 1 AML p atient sample s that are grow th inhihite d by p1 10R13

Patient rio. Source’ Age/Sex FAB” Karyotype RB stain100 nsi

Clono genic cell recovery’�___________________________ � - �

200 ns� 300 ma 400 � 501) no

:t PB 53/M M5 47#{247}8 Weak ND ND ND ND ND

4 PB 48/F BC t(9;22)” Moderate 62 (10) 48 (9.4) 34 (14) 62 Ii 1) 26 12

5 PB 59/F M4 diploid Moderate 83 (2.7( 47 (1.9( 66 (1.8) 57 (7.0) 56 (6.6)

7 BM 64/F M4 diploid Strong 90 (1.1) 81 1.2) 77 0.9) 65 (0.9) 59 1.0)

1() LP 74/M M2 47+8 Moderate 134 (5.4) 90 1.7) 79 (1 1) 30 5.3) 0 0)

12 BM 29/M Ml 47+21 Mod/strong 60)2.3) 2.6)0.2) ND ND ND

14 BM 31/M M4 46,1 lq- Mod/strong 83 (10) 37 12) ND 6.4 (0.7) ND

19 BM 81/M M2 diploid Strong 93 (7.5) 58 (8.6) 16 1.9) ND 27 (2.8)

., PB, peripheral blood; BM, bone marrow aspirate; LP, leukapheresis.I, Morphology according to French-American-British classification; BC, CML myeloid blast crisis.

, Blasts incubated 48 h in liquid culture with P1 1 � at indicated concentrations prior to plating. Values are mean colony number expressed as a percentage ofbuffer control. SEM in parentheses. ND, not done.

(I Complete karyotype 47,XX,+8,t)9:22)(q34;ql 1 ),del)1 1 ))q23).

preserved in liquid nitrogen for up to i year. Buffer aloneshowed no growth-inhibitory effect on AML blasts whenadded to culture media.

Growth Inhibition Is Blocked by T Peptide. To demon-strate that decreased colony formation was due to func-tional pi i 0RtS a synthetic peptide was prepared with thesequence NLFCSEEMPSSDDE (T peptide) derived fromSV4O large-T antigen. A peptide with a lysine for glutamatesubstitution at position 6 (K peptide) was also produced.The peptides were 98 and 99% pure, respectively, byHPLC. T peptide has been reported previously to bind andblock the A/B pocket of pi i 0�, whereas K peptide does not(29). Blasts from patient no. i4 show decreased colonyformation following exposure to p1 1 0R13 alone or in com-bination with K peptide. Addition of T peptide to p1 i 0RB

abolished the growth inhibitory effect (Fig. 2B). Neither Tpeptide nor K peptide alone had an effect on colonyformation.

AML Blasts Show Degenerative Histological Changesfollowing Exposure to pllO�. Blasts from patient no. i9showed a striking change in morphology following expo-sure to p1 1 0Rt3 (Fig. 3, A-D). Increased adherence to theculture plate and elongated shape were visible under phasecontrast microscopy. Wright staining of nonadherent cellsdemonstrated cytoplasmic vacuolization and nuclear lob-

ulation. These ftndings are characteristic of degeneratingcultures and are further evidence of a growth-inhibitoryeffect.

Recombinant p1 1 0RB Is Internalized by Cells and Trans-located to the Nucleus. Immunostaining of 5637 cells fol-lowing exposure to recombinant pi i 0RU shows that theprotein is intracellular and localized to the nucleus,whereas unexposed cells show no detectable pi 10RB (Fig.3, E-F). Uptake of recombinant pi i 0RB by 5637 and AMLblasts was demonstrated by a signal shift on flow cytometry.5637 shows no increased fluorescence when labeled with3C8 anti-pi i 0RIt mAb versus irrelevant antibody. AMLblasts, on the other hand, express p1 i 0RB, which is detectedon flow cytometry. Following incubation with recombinantp1 i 0R0, both 5637 and AML blasts show fluorescenceabove that measured prior to exposure (Fig. 4). This was truefor patient samples that were inhibited by pi 1 � as wellas those which were not.

Accumulation of 251-Iabeled pi 10RB occurred mostrapidly in the nucleus, whereas accumulation of a radio-labeled control protein was not seen (Fig. 5A). Nuclear

Fig. 3. Microscopic appearance of cells exposed to P1 1 Q511#{149}Patient no. 1 9cells, phase-contrast appearance following buffer (A) or p1 J0R�� 400 no (B)for 72 h; Wright stain of nonadherent cells following buffer 0 or p1 i#{252}�”400 nsa (0) for 72 h. 5637 immunostained with 3C8 anti-pi J0RU antibody,

untreated (F) and following p1101#{176}’400 � (F) for 5 h.

fractions were isolated, and p1 10Rt3 was fractionated onSDS-acrylamide gels and subjected to autoradiography.A Mr ii0,000 radiolabeled band appeared in a time-dependent manner, indicating the presence of intact ex-ogenous pi i 0RB in the nucleus (Fig. 5B). At 500 ntvi,internalization of piiO’� by 5637 cells occurs over thecourse of 2-4 h. Approximately i% of total labeledp1 i 0RB was recovered from nuclear lysates after 2 h ofexposure.

A

A� L�

0E3

z3

0

. . �.l’02

B

Log Fluorescence Intensity

Fig. 4. FACS analysis of cellular pllORB content. All samples stained with

3C8 antibody. A, 5637 incubated with medium alone (unshaded) or with

p1 1 0R5 400 nsa (shaded) for 5 h. B, patient no. 1 1 incubated with mediumalone (unshaded) or with �110R8 200 nsa (shaded) for 24 h.

676 Recombinant RB Protein Inhibits Cancer Cell Growth

)_1

Discussion

Previous studies have demonstrated that reintroduction ofthe RB! gene into an RB-negative background reverses thetumorigenic phenotype in a variety of cell types as mea-sured by growth in soft agar and s.c. tumor formation innude mice (3-5). We have shown that a susceptible cancercell line and blast cells from patients with AML are growth-inhibited in vitro by direct exposure to recombinantpii0��. The possibility of growth inhibition from othercauses, such as bacterial toxin, is excluded by demonstra-tion that T peptide specifically blocks the antiprohiferativeeffect. Proliferation of 5637 cells, which do not expressRB!, was inhibited by p1 10RB in a dose-responsive fashion.Growth inhibition of this cell line resulting from RB! geneinsertion has also been reported (5). Transfected clones of5637 continue to proliferate, but at a slower rate, which isconsistent with our observation that DNA synthesis is nottotally inhibited by exogenous p110RB. AML blasts withimmunohistochemical evidence of endogenous p1 1 0RB

were exposed to recombinant pil0’� and showed de-creased clonogenicity. We cannot rule out the possibility ofaberrant RB! expression in AML that is not detected byother means, perhaps involving point mutations or differ-ences in phosphorylation of endogenous protein. Differ-ences in phosphorylation may also explain the range ofresponses to exogenous p1 i 0RB� At the concentrationstested, proliferation could be substantially suppressed inseveral cases (patients nos. 3, 10, 12, 14, and 19) but wasincompletely suppressed or unaffected in others.

Recombinant pii0RB may inhibit proliferation of RB-competent cells by achieving nuclear concentrations higherthan physiological levels. The importance of p1 1 0RB con-

centration in mammalian cells has been demonstrated intransgenic mice bearing multiple copies of the human RB!gene; p1 1 0RB levels were proportional to gene copy num-ber, and transgenic animals were smaller than their non-transgenic hittermates, with size inversely proportional toRB! gene copy number (30). Heterogeneity of response inAML does not, however, appear related to differences incellular uptake of p1 1 0RB (Fig. 4B). We observed one AMLsample that stained weakly for endogenous p1 1 0RB but wasnot growth inhibited, suggesting that the biological effect ofp1 1 0RB may be influenced by additional factors (discussedbelow).

Other growth-regulating macromolecules have beenshown to cross the plasma membrane through receptor-mediated or nonreceptor-mediated endocytosis. For exam-pIe, the HIV tat protein is found in the extracellular matrixand is internalized by cells; attachment to the cell surfaceinvolves interaction of basic domains on tat with aj35integrin (16, 31). Insulin enters cells by receptor-mediatedendocytosis at low concentrations and by fluid-phase en-docytosis at high concentrations (>1 00 ng/ml; Ref. 20). Bothtat and insulin reach the nucleus and affect gene expressiononce internalized. The Drosophila homeobox peptidepAntp is internalized by rat embryonic nerve cells, entersthe nucleus, and regulates morphological differentiation(1 7). Cultured endothelial cells exposed to bFGF are foundto have bFGF present in the cytoplasm and nucleus (18). Arequirement for receptor-mediated internalization and nu-clear translocation is seen for epidermal growth factor,platelet-derived growth factor, and interleukin 1 (32-34).

The mechanism for internalization of p1 10RB is not

known. One possibility is that recombinant p1 1 0RB reaches

the interior of cells through endocytosis when there is asufficient concentration gradient across the plasma mem-brane. Endocytosis of bFGF is mediated by binding to high-affinity receptors as well as lower affinity binding to hepa-ran sulfate proteoglycans (19). Internalization of pAntp byneurons also depends upon the presence of a specific car-bohydrate chain on the cell surface; in this case, the a-2,8-polysiahic acid chains of the neuronal cell adhesion mole-cule bear a structural similarity to double-stranded DNA,which is the usual binding partner ofthe antennapedia geneproduct (1 7). Although there is no evidence for a high-affinity p11oRB receptor, binding to proteoglycans on thecell surface might occur, and the lower affinity of suchbinding would explain the relatively high concentrations ofpli0ttu required for growth inhibition. An alternative, al-though less likely, mechanism for cellular uptake of p1 1#{216}RB

is by diffusion directly through the plasma membrane. Anintact plasma membrane does not normally permit diffusionof macromolecules as large as p1 1 0RB, but membranes maybe altered to increase permeability. Cellular uptake ofpll0’� has been reported to occur following permeabihi-zation with detergent (35). The small amount of glycerolpresent in our culture media (50 pM) may also have in-creased membrane permeability to p1 1 0RB, althoughapparently not to a control protein of similar size (Fig. 5A).

Localization to the nucleus is mediated by a putativenuclear localization sequence at the COOH-terminal por-tion of p1 1 0RB, and such signaling may also mediate pro-tein trafficking from endocytic vesicles (36). The transloca-tion of proteins from the cytoplasm to the nucleus is a highly

A1.2

I

0.8

0.6

0.4

0.2

0

0

B

50 100 150 200

Time (minutes)

250

10 30 60 120 240 24OSTD� r’ �

-� �- p1lO’�

Cell Growth & Differentiation 677

Fig. 5. Time-dependent local-

ization of ‘ 2�l-laheled P1 1 0’#{176}’tonucleus. A, � 2�t-labeled P1 1rt’�

recovered from nuclear (I) andnonnuclear )U) fractions, com-

pared with labeled murine lgGF(.ib’) recovered from nuclear 0)

and nonnuclear 0) fractions. Tn-chloroacetic acid-precipitable

2�I cpm was expressed as a per-

centage of the total cpm. B, auto-

radiographic detection of Al,

1 1 1)000 l)and on SDS gel electro-

phoresis of nuclear lysate. Lanes1-5 ) troni left) show time course;Lane 6 loaded with additional

protein shows that the labeled

protein is predominantly Al,1 l0,Ot)O; reombinant P1 10’#{176}’in

Lane .7 as internal standard (STO).

% of TotalCPM

regulated process. Interaction with transport proteins is me-diated by nuclear localization sequences (37). There isevidence that protein complexes shuttle along filamentoustracks leading to the nucleus, where a nuclear pore com-plex of proteins is formed and serves to regulate transloca-tion across the nuclear envelope (38). These mechanismsmay explain the selective localization of pii0� to thenuclear compartment. Internalization of smaller molecularweight fragments of partially degraded pi iORB may alsoaccount for some of the growth inhibition we observed.Some degradation of the protein does occur in medium at37#{176}C,although full-length pi i 0R8 was recovered from me-

dia at the end of experiments (data not shown). Althoughthe role of smaller molecular weight species remains unde-

termined, our data show that intact pi 1 0Rtt does reach thenucleus of 5637 cells (Fig. 58).

Study ofthe biological effect of pi i 0RB on blast cells frompatients with AML provided an in vitro model which isknown to be a valid surrogate for events in vivo, based onstudies relating biological properties determined in cultureto clinical outcome of the disease (28, 39, 40). AML is a

clonal hemopathy; in each patient, a single clone charac-terized by a large blast population becomes dominant,probably because of a proliferative advantage. The blastpopulation in the clone may be maintained independentlyby self-renewal occurring in stem cells committed to theblast lineage. Some drugs that have clinical value in thetreatment of AML, such as 1 -�-D-arabinofuranosylcytosine,may be effective because they have some selective toxicityfor blast stem cells undergoing self-renewal (4i, 42). Ourdata show that AML blast stem cell self-renewal is de-creased following exposure to exogenous pi 1 0RB, Recentlypublished results confirm the role of p1 i 0RB in AML blastcolony formation in that suppression of pi 1 0RB synthesis

using an antisense oligonucleotide results in increasedclonogenic cell recovery (25).

In both normal and malignant hematopoiesis, the prolif-eration and differentiation of progenitor cells is regulated bygrowth factors, cytokines (autocrine and paracrine), regu-lators of differentiation such as retinoic acid, cyclins, andcdks which converge to influence progression of the cellcycle (27, 42, 43). There is evidence that the effect of

678 Recombinant RB Protein Inhibits Cancer Cell Growth

(, D. Goodrich, personal communication.

certain cytokines on the cell cycle in hematopoietic cells ismediated by p1 1 0RB; these cytokines include interferons,transforming growth factor f3, and interleukin 6 (43-46).Transforming growth factor �3 down-regulates cdk2 andcdk4, which promote phosphorylation (inactivation) ofpi 1 0RB and are activated by D and E cyclins (45). Cell cyclepromotion and inhibition of differentiation induced byoverexpression of cyclins D2 and D3 in myeloid cells hasbeen shown to depend on cyclin binding to p1 1 0RB (47).Response to exogenous p1 1 0RB is most likely influenced bymultiple complex molecular interactions that vary depend-ing on the cell type. The observed heterogeneity of effect inAML is also consistent with previous observations regardingthe biological significance of loss or restoration of p1 10RBfunction in mammalian cells: (a) only certain cell typesform tumors in vivo when RB ! expression is lost; and (b) notall RB-negative tumor cell lines are growth inhibited bytransfection with the RB! gene (48).

In conclusion, we have described a novel example ofcancer cell growth inhibited by direct exposure to recom-binant human p1 1 0RB Abnormal expression of endogenousp1 1 0RB probably forms the basis for susceptibility to recom-binant p1 10RB although the nature of the biological re-sponse may also depend on other factors. Extracellularsignaling by soluble cytokines serves to inhibit cell prolif-eration and may be mediated by p1 1 0RB � that directexposure to recombinant p1 1 0RB exerts the same effect(46). The mechanism through which p1 1 0RB is internalizedby cells remains to be determined, and in vivo studies mayreveal more efficient methods for delivering recombinantp1 i 0RB to cells. The discovery that a tumor suppressorprotein gains access to nuclei from outside of cells suggests

possibilities for future research and medical applications.

Materials and MethodsReagents

Recombinant Human p1 10RB Recombinant full-lengthhuman pi 1 0RB was produced from Escherichia co/i andisolated to >90% purity. Stock solutions were kept at 1mg/mI concentration with 1 0% glycerol and EDTA 1 m�i toprevent aggregation (49). Individual lots of p1 1 0RB werecharacterized for binding to a peptide containing an 18-amino acid COOH-terminal sequence of the E2F transcrip-tion factor using methods reported previously to documentthe binding of pi 1 0RB to SV4O large-T antigen (50). E2F-binding function was intact for all protein samples used in

the experiments described, and recombinant p1 10RB produced and purified at Canji under the same conditions wasable to block progression from G1 to S phase in Saos2osteosarcoma cells following microinjection, such as wasreported previously for a Mr 56,000 RB fragment.6

Growth Factors. Granulocyte-CSF was obtained fromAmgen, Inc. (Thousand Oaks, CA), and granulocyte-mac-rophage/CSF was obtained from Genetics Institute(Cambridge, MA).

Antibodies and Immunochemical Reagents. 3C8 anti-p1 1 0R3 monoclonal antibody was obtained from Canji, Inc.Mouse lgG2a (control) was obtained from Fisher Scientific,Piano, TX; FITC-anti-mouse IgG was obtained from SigmaChemical Co. (St. Louis, MO); and Vectastain immunoper-oxidase staining kit was obtained from Vector Laboratories

(Burlingame, CA). Murine lgG F(ab’)2 was obtained fromPierce Scientific (Rockford, IL).

Cell Lines

Human bladder carcinoma cells (5637) were obtained fromAmerican Type Culture Collection (Rockville, MD) andpropagated in DMEM:F1 2 (1 :1 ) or a-MEM with 10% FBS.Cells were maintained at 37#{176}Cwith 5% CO2 and weretrypsinized and subcultured every 5-7 days.

Patient Samples

Peripheral blood specimens were collected in preservative-free heparin from patients with AML (bone marrow blastcounts >30% and peroxidase positivity �4%); bone mar-row aspirates were collected in EDTA. All samples wereobtained with informed consent from patients hospitalizedat the University of Texas M. D. Anderson Cancer Center.Mononuclear cells were isolated by density gradient cen-trifugation and depleted of T cells after rosette formationwith sheep RBC (51).

Blasts were maintained in liquid culture in a-MEM with

1 0% FBS, stem cell factor (conditioned medium producedfrom a transfected cell Ii ne), granulocyte-macrophage/CSF,and/or granulocyte-CSF. Cells were subcultured every 3-5days. Under these conditions, freshly obtained blastscontinue to proliferate for 2-10 weeks (52).

Tritiated Thymidine Uptake Assay

For dose-response curves, 5637 cells were seeded in 96-well plates with 5 x i0� cells/well, allowed to adhere

overnight, then amounts of p1 1 0RB in cell medium wereadded to achieve the concentrations shown, and 1 pCi[3H]thymidine (specific activity, 83 Ci/mmol) was added toeach well. One-half of the cell medium was removed andreplaced with p1 10RB�containing medium twice each dayfor 3 days. After exposure to p1 1 0RB and [3H]thymidine for72 h, cells were harvested and assayed for [3H]thymidineincorporation by liquid scintillation counting. As a control,cells were incubated with medium and [3H]Thy alone andassayed for [3H]thymidine incorporation by liquid scintil-lation counting. Four replicate wells were analyzed for eachconcentration of p1 10RB#{149}

Clonogenic Assay for Leukemic Blast Cell Progenitors

To measure the effect on colony formation, blasts wereincubated 48 h in liquid culture medium with buffer aloneor p110RB, 100 to 500 nM, with an initial cell density of1 06/mI. p1 1 0RB was added once at the beginning of theexperiment. To control for the presence of glycerol andEDTA, which are present in the p1 1 0RB stock solution,concentrations were adjusted to 0.5% and 50 pM, respec-tively, in all samples. Buffer controls consisted of glyceroland EDTA alone. At 48 h, blasts were washed and resus-pended in medium with growth factors, 15% FBS, and 0.8%methylcellulose, then plated in quadruplicate in an un-treated 96-well microtiter plate. Colonies consisting ofmore than 20 cells were scored after 4-7 days at 37#{176}Cwith5% CO2. Differences in clonogenic cell recovery wereconsidered to reflect leukemic stem cell self-renewal andthe effect of recombinant p1 1 0RB on self-renewal (26).

Blast cells from patient no. 1 4 were incubated for 24 hwith no addition versus T peptide or K peptide 250 nM,

Cell Growth & Differentiation 679

p1 1 0RB 200 nM alone, p1 1 0RB 200 nivi plus T peptide 250nM, or p1 1 0RB 200 nisi plus K peptide 250 nM. Cells werethen washed and plated in methylcellulose-containing me-

dium as described above. The experiment was performedtwice.

Histochemical Assay for Endogenous p1 10RB

To detect the presence and relative amounts of endogenousp1 1 0RB cells were pelleted onto glass slides via cytocen-trifugation and stained using 3C8 monoclonal anti-pi 1 0RB

antibody and Vectastain ABC immunoperoxidase kit. 3C8 isa mouse lgG2a that binds near the COOH-terminal end ofp1 1 0RB (amino acids 886-905), which is outside of theregion involved in interactions with other proteins (50).

Slides were scored blind by one of us (S. A. S.). The scoringsystem grades the intensity of nuclear staining as weak,moderate, moderate-strong, or strong.

FACS Analysis for Measurement of Cellular p1 10RB

AML blasts were incubated with p1 1 0RB, 200 to 400 nM, ormedium alone at 37#{176}Cwith 5% CO2 for 24 h; 5637 cellswere grown to confluence as an adherent monolayer in35-mm Petri dishes, incubated with p1 1 0RB 400 nM for 5 h,then trypsinized. Cells were then washed three times withPBS, fixed in 1 % paraformaldehyde with 0.1 % Triton-X,and stained with 3C8 monoclonal antibody or irrelevantmurine lgG2a, followed by FITC-anti-mouse IgG. Sampleswere then analyzed with an Epics Profile flow cytometer.

Cellular Uptake of ‘25l-labeled p1 10R8 by 5637

E. co/i produced p1 1 0RB and a murine IgG F(ab’)2 wereradiolabeled with � 251 using the chloramine-T method. Spe-cific activities ofthe two proteins were 3.6 and 19.2 pCi/pg,respectively, and the labeled protein was added to unla-

beled protein to achieve the desired final concentrations.

Plates (10-cm) were seeded with approximately 3 X 1065637 cells and allowed to grow to confluence over 48 h.Cells were incubated with 500 n�i p110RB or murine lgGF(ab’)2 (approximately 1 0 pCi), and plates were harvested at1 0, 30, 60, 1 20, and 240 mm. At harvest, cells were washedtwice with cold HBSS and removed from the plates viatrypsin EDTA. After additional washing with cold HBSS,cells were centrifuged at 500 x g for 5 mm. A lysis buffer(iO m�i Tris-HCI, 10 ms� NaCI, 3 ms�t MgCI2, 0.5% NP4O,and 10 pg/mI aprotinin) was used to prepare nuclear frac-tions. Phase contrast microscopy after the addition of lysisbuffer revealed intact nuclei with <10% intact cells.

Nuclei were separated from nonnuclear (cytoplasm plusmembrane) fractions (500 X gfor 5 mm), and fractions wereassayed for total and trichloroacetic acid-precipitable 1251

radioactivity. Aliquots of nuclear fractions were dilutedwith double-strength SDS buffer and loaded onto 8-16%Tris-glycine electrophoresis gels. Dried gels were exposedto storage phosphor screens to produce autoradiograms. Toassess the integrity of the fractionation procedure, samplesfrom both nuclear and nonnuclear fragments were evalu-ated for NADH oxidase and 5’-nucleotidase activity.

References1 . Goodrich, D. W., and Lee, W-H. Molecular characterization of the

retinoblastoma gene. Biochim. Biophys. Acta, 1 155: 43-61 , 1993.

2. Klein, G. Genes that suppress cancer. FASEB )., 7: 821-825, 1993.

3. Bookstein, R., Shew, J. Y., Chen, P. K., Scully, P., and Lee, W-H. Sup-

pression of tumorigenicity of human prostate carcinoma cells by replacing amutated RB gene. Science (Washington DC), 247: 71 2-71 5, 1990.

4. Huang, H-J. S., Yee, i-K., Shew, J-Y., Chen, P-L., Bookstein, R., Fried-mann, T., Lee, E. Y., and Lee, W-H. Suppression of the neoplastic phenotypeby replacement of the retinoblastoma gene product in human cancer cells.

Science (Washington DC), 242: 1 563-1 566, 1988.

5. Takahashi, R., Hashimoto, T., Xu, H. J., Hu, S. X., Matsui, T., Miki, T.,

Bigo-Marshall, H., Aaronson, S. A., and Benedict, W. F. The retinoblastoma

gene functions as a growth and tumor suppressor in human bladder carci-noma cells. Proc. NatI. Acad. Sci. USA, 88: 5257-5261 , 1 991.

6. Goodrich, D. W., Wang, N. P., Qian, Y-W., Lee, E. Y-H. P., and Lee,

W-H. The retinoblastoma gene product regulates progression through the G,phase of the cell cycle. Cell, 67: 293-302, 1991.

7. Lee, W-H., Shew, J-Y., Hong, F. D., Sery, T. W., Donoso, L. A., Young,

L-J., Bookstein, R., and Lee, E. Y-H. The retinoblastoma susceptibility geneproduct is a nuclear phosphoprotein associated with DNA binding activity.

Nature (Lond.), 329: 642-645, 1987.

8. Dowdy, S. F., Hinds, P. W., Louie, K., Reed, S. I., Arnold, A., andWeinberg, R. A. Physical interaction of the retinoblastoma protein with

human D cyclins. Cell, 73:499-511, 1993.

9. Nevins, J. R. E2F: a link between the Rb tumor suppressor protein and viral

oncoproteins. Science (Washington DC), 258: 424-429, 1992.

1 0. Welch, P. 1., and Wang, Y. J. A C-terminal protein-binding domain in theretinoblastoma protein regulates nuclear c-able tyrosine kinase in the cellcycle. Cell, 75: 779-790, 1993.

11. Godbout, R., Dryja, T., Squire, J., Gallie, B. L., and Phillips, R. A.

Somatic inactivation of genes on chromosome 1 3 is a common event in bothhereditary and nonhereditary retinoblastoma tumors. Nature (Lond.), 304:

451-453, 1983.

1 2. Abramson, D. H., Ellsworth, R. M., Kitchin, D., and Tung, G. Secondnonocular tumors in retinoblastoma survivors. Ophthalmology (Phila.), 91:1351-1355, 1984.

13. Hansen, M. F., Koufos, A., Gallie, B. L., Phillips, R. A., Fodstad, 0.,

Brogger, A., Gedde-DahI, T., and Cavenee, W. K. Osteosarcoma and reti-noblastoma: a shared chromosomal mechanism revealing recessive predis-

position. Proc. NatI. Acad. Sci. USA, 82:6216-6220, 1985.

14. Lee, E. V-H. P., Chang, C-V., Hu, N., Wang, V-C. J., Lai, C-C., Herrup,

K, Lee W-H., and Bradley, A. Mice deficient for Rb are nonviable and show

defects in neurogenesis and hematopoiesis. Nature (Lond.), 359: 288-294,1992.

1 5. Jacks, T., Fazeli, A., Schmitt, E. M., Bronson, R. T., Goodell, M. A., and

Weinberg, R. A. Effects of an Rb mutation in the mouse. Nature (Lond.), 359:

295-330, 1992.

16. Frankel, A. D., and Pabo, C. 0. Cellular uptake ofthe tat protein fromhuman immunodeficiency virus. Cell, 55: 1189-1193, 1988.

1 7. Joliot, A. H., Triller, A., Volovitch, M., Pernelle, C., and Prochiantz, A.a-2,8-Polysialic acid is the neuronal surface receptor of antennapedia ho-meobox peptide. New Biol., 1 1: 1 1 21-1 1 34, 1991.

18. Baldin, V., Roman, A-M., Bosc-Bierne, I., Amalric, F., and Bouche, G.Translocation of bFGF to the nucleus is G, phase cell cycle specific in bovine

aortic endothelial cells. EMBO I., 9: 151 1-1517, 1990.

1 9. Roghani, M., and Moscatelli, D. Basic fibroblast growth factor is inter-

nalized through both receptor-mediated and heparan sulfate-mediatedmechanisms. J. Biol. Chem., 267: 22156-22162, 1992.

20. Harada, S., Loten, E. G., Smith, R. M., and Jarett, L. Nonreceptormediated nuclear accumulation of insulin in H35 rat hepatoma cells. J. Cell.

Physiol., 153:607-613, 1992.

21 . Furukawa, V., DeCaprio, J. A., Belvin, M., and Griffin, J. D. Heteroge-neous expression of the product of the retinoblastoma susceptibility gene inprimary human leukemia cells. Oncogene, 6: 1343-1346, 1991.

22. Ahuja, H. G., Jat, P. S., Foti, A., Bar-Eli, M., and Cline, M. J. Abnormal-ties of the retinoblastoma gene in the pathogenesis of acute leukemia.

Blood, 78:3259-3268, 1991.

23. Kornblau, S. M., Xu, H-J., del Giglio, A., et al. Clinical implications ofdecreased retinoblastoma protein expression in acute myelogenous leuke-mia. Cancer Res., 52:4587-4590, 1992.

24. Chen, V-C., Chen, P-J., Veh, S-i., Tien, H-F., Wang, C-H., Tang, J-L., andHong, R-L. Deletion ofthe human retinoblastoma gene in primary leukemia.

Blood, 76:2060-2064, 1990.

25. Zhu, V-M., Bradbury, D., and Russell, N. Decreased retinoblastomaprotein expression in acute myeloblastic Ieukaemia is associated with the

autonomous proliferation of clonogenic blasts. Br. J. Haematol., 86:533-539, 1994.

680 Recombinant RB Protein Inhibits Cancer Cell Growth

26. Buick, R. N., Till, J. E., and McCulloch, E. A. Colony assay for prolifer-

ative blast cells circulating in myeloblastic leukemia. Lancet, 1: 862-863,

1977.

27. Miyauchi, J., Kelleher, C. A., Wang, C., Minkin, S., and McCulIoch, E.A. Growth factors influence the sensitivity of leukemic stem cells to cytosine

arabinoside in culture. Blood, 73: 1272-1278, 1989.

28. Rayappa, C., and McCulloch, E. A. A cell culture model for the treatmentof acute myeloblastic leukemia with fludarabine and cytosine arabinoside.

Leukemia, 7:992-999, 1993.

29. DeCaprio, J. A., Ludlow, J. W., Lynch, D., Furukawa, V., Griffin, J.,Piwnica-Worms, H., Huang, C-M., and Livingston, D. M. The product of theretinoblastoma susceptibility gene has properties of a cell cycle regulatoryelement. Cell, 58: 1085-1095, 1989.

30. Bignon, V-i., Chen, V., Chang, C-V., Riley, D. J., Windle, J. J., Mellon, P.L., and Lee, W-H. Expression of a retinoblastoma transgene results in dwarfmice. Genes Dev., 7: 1654-1662, 1993.

31 . Vogel, B. E., Lee, S-J., Hildebrand, A., Craig, W., Pierschbacher, M. D.,Wong-Staal, F., and Ruoslahti, E. A novel integrin specificity exemplified by

binding of the a�f3s integrin to the basic domain of the HIV tat protein andvitronectin. I. Cell. Biol., /21:461-468, 1993.

32. Burwen, S. J., and Jones, A. L. The association of polypeptide hormonesand growth factors with the nuclei of target cells. Trends Biochem. Sci., 12:159-162, 1987.

33. Grenfell, S., Smithers, N., Miller, K., and Solari, R. Receptor-mediated

endocytosis and nuclear transport of human interleukin 1 a. Biochem. J., 264:813-822, 1989.

34. Maier, J. A., Statuto, M., and Ragnotti, G. Endogenous interleukin 1 amust be transported to the nucleus to exert its activity in human endothelial

cells. Mol. Cell. Biol., 14: 1845-1851, 1994.

35. Mittnacht, S., Hinds, P. W., Dowdy, S. F., and Weinberg, R. A. Modu-lation of retinoblastoma protein activity during the cell cycle. Cold Spring

Harbor Symp. Quant. Biol., 56: 197-209, 1991.

36. Zacksenhaus, E., Bremner, R., Phillips, R. A., and Gallie, B. L. A bipartitenuclear localization signal in the retinoblastoma gene product and its im-

portance for biological activity. Mol. Cell. Biol., 13: 4588-4599, 1993.

37. Silver, P. A. How proteins enter the nucleus. Cell, 64: 489-497, 1991.

38. Meier, U. T., and Blobel, G. Noppl4O shuttles on tracks betweennucleolus and cytoplasm. Cell, 70: 127-138, 1992.

39. Koistinen, P., Wang, C., Curtis, J. E., and McCulloch, E. A. Granulocyte-

macrophage colony-stimulating factor and interleukin-3 protect leukemicblast cells from Ara-C toxicity. Leukemia, 5: 789-795, 1991.

40. Estey, E., Thall, P. F., Kantarjian, H., O’Brien, S., Koller, C. A., Beran, M.,Gutterman, J., Deisseroth, A., and Keating, M. Treatment of newly-diagnosedacute myelogenous leukemia with GM-CSF prior to and during continuous-infusion high dose ara-c (CHDAC) + daunorubicin: comparison to patients

treated without GM-CSF. Blood, 79: 2246-2255, 1992.

41 . Nara, N., Curtis, J. E., Senn, J. S., Tritchler, D. L., and McCulloch, E. A.The sensitivity to cytosine arabinoside of the blast progenitors of acutemyeloblastic leukemia. Blood, 67: 762-769, 1986.

42. McCulloch, E. A. Stem cell renewal and determination during clonal

expansion in normal and leukaemic haemopoiesis. Cell Prolif., 26:399-425, 1993.

43. Hatzfeld, J., Li, M-L., Brown, E. L., Sookdeo, H., Levesque, i-P., O’Toole,T., Gurney, C., Clark, S. C., and Hatzfeld, A. Release of early hematopoieticprogenitors from quiescence by antisense transforming growth factor 131 or

Rb oligonucleotides. J. Exp. Med., 174: 925-929, 1991.44. Melamed, D., Tiefenbrun, N., Varden, A., and Kimchi, A. Interferons andinterleukin-6 suppress the DNA-binding activity of E2F in growth-sensitivehematopoietic cells. Mol. Cell. Biol., 13: 5255-5265, 1993.

45. Ewen, M. E., Sluss, H. K., Whitehouse, L. L., and Livingston, D. M. TGF�inhibition of cdk4 synthesis is linked to cell cycle arrest. Cell, 74:

1009-1020, 1993.

46. Gutterman, J. U. Cytokine therapeutics: lessons from interferon a. Proc.NatI. Acad. Sci. USA, 91: 1 198-1205, 1994.

47. Kato, i-V. and Sherr, C. J. Inhibition of granulocyte differentiation by G1cyclins D2 and D3 but not Dl . Proc. NatI. Acad. Sci. USA, 90:

11513-11517, 1993.

48. Zacksenhaus, E., Bremner, R., Jiang, Z., Gill, R. M., Muncaster, M.,

Sopta, M., Phillips, R. A., and Gallie, B. L. Unraveling the function of theretinoblastoma gene. Adv. Cancer Res., 6 1 : 1 1 5-1 41 , 1993.

49. Antelman, D., Machemer, T., Huyghe, B. G., Shepard, H. M., Maneval,D., and Johnson, D. E. Inhibition of tumor cell proliferation in vitro and invivo by exogenous p1 10RB, the retinoblastoma tumor suppressor protein.

Oncogene, 10: 697-704, 1995.

50. Wen, S-F., Nodelman, M., Nared-Hood, K., Duncan, J., Geradts, J., andShepard, H. M. Retinoblastoma protein monoclonal antibodies with novel

characteristics. J. Immunol. Methods, 169: 231-240, 1994.

51 . Minden, M. D., Buick, R. N., and McCulloch, E. A. Separation of blast

cell and T-Iymphocyte progenitors in the blood of patients with acutemyeloblastic leukemia. Blood, 54: 186-195, 1979.

52. Nara, N., and McCuIIoch, E. A. The proliferation in suspension of theprogenitors of the blast cells in acute myeloblastic leukemia. Blood, 65:

1484-1493, 1985.