Biorhimiur et Biophysku&a, @ 1992 Elwier

BBAMCR

1136W2)

253

E3-258

Sciew Publishers B.V. A!I rights re%?rved 0167-4889/92/$05.130

13225

CM-surfacechangesinducedby ectopicexpressionof the murine honzeoboxgene HLIX-3.3 Molecular Ethyohgv

Sebastian M. Shimeldand Paul T. Sharpe nfCelland S~nrc~urulBinlogy. Oni1rrsiry of Manc/res!er,Manclws~eriUKI

Lab~ru~ory.Depanmenr

(Received 16 March !‘)92)

Key words: Cell surface; Gene expression; Homeobox;

ffox-3.3:

(Mouse); Aqueous two-phase system; Transfection

Murk homcobox-cuntaining genes (HOI gencsj are postulated as playing key roles in the establishment of the antcrior-posterior embryonic body axis, possibly providing cells with positional cues. Little is known, however, concerning how cells might respond to home&ox gene expression to interpret these cues. Since changes in the ceil-surface are ccr.tral to manyprocesses in early development we reasoned that cells expressing different comp!ements of ~iox genesmight have different surface properties. In order to investigate this we have used the sensitive, no+disruptive technique of muttipte two-phase aqueous partition, which is able to detect small differences on the surface of intact cells. Using this technique we have found that ectopic expression of the murine ffax3.3 gene in cultured cells induces reproducible changes in the ceil surface. Changes only occurred above a threshold level of gene expression, but abow this level a correlation between surface change and gene expression was seen. The implications for the establishment of a ‘Hex’ code of homeobox genes acting to specifically change cell-surface properties arediscussed.

Homeobox genes are known to constitutea large and variedgroupof genes havingthr commonfeature of codingfor proteinscontaininga characteristichomeodomain capableof binding IMA 11,21. The presence

of this DNA-bindingdomainsuggeststhat borneoproteins act as transcriptionalregulatorsand suchproperties haveMJW been confirmedfor a numberof bomeoproteins(3-2). Much interest has fkussed on the role of borneoproteins in embryonic development, although it is probable that many Hex genes are also expressed in adult tissues [9,10].Comparison of the important characteristics of the murine Her; genes with the Drylscrphila homeoticgeneshas revealeda remarkable

homeoticloci [13]. Data are also emergingto suggest that similar cross-regulatoryinteractions exist in vertebrates [14,15].However, a majorgap in our current understanding of the role of homeoproteins in the control of development is the almost complete lack of information concerning the non-bomeobox-containing target genes of homeoproteins. The only current evi-

dence concerning the nature of non-transcriptional gene products that are regulated by homeoproteins is that the products of the Drosophila homeotic genes Ubx and Abdcl regulate the TGF-@-like gene decupentaplegc [16]. At a more general level,whateverthe functionof

evolu*;o?~= U 1can,mnwm4on -..ll.e-*.-__ of both gene structure and

homeoproteinsin development,it might be exwcted that they would exert an influence on cell surfaces, since changes in cell-surfacemoleculesare central to

expression [l&12]. This suggeststhat the proteinsencoded by these genes play similar roles in the control processes of embryonic development iu both mice and flies. In Droscrphila,studiesof the cross-regulatory inter-

probably all early developmental processes. It has been proposedthat positionaldeterminantsalong the pmbryonic anterior-posterior axis are determined by a ‘Hox code’, such that cells knowtheir position because of the complement of Hex genes they express[17].

actions between homeoproteinsand homeobox-contaming genes have established that homeotic gene productscan act as transcriptionalregulatorsof other

Comspwdenceto: P.T. Sh;upc, Mokular DepWnt versity

Embryology Laboratory,

of Ceil and Structural Biology. Stopford

of Manchester, Manchester AA139PT, UK.

Building,

Uni-

Such a code providedby nuclear proteinsis in effect nothing more tharr a ‘nuclear’code and must be translated into a ‘structural’ code with cells at different anterior-posterior positions having recognizable structuraldifferences.In theory,such a code could involve any non-nuclear protein but is likely,in viewof the fact

that cells must recognizeand nspund to their neighbouts, to invofve cell-surfacemolecules,The cell-

254 surface is a complexenoughstructureto allow for the wide range of moIeculardiversityrequired to specify different positions along the entire anterior-posterior axis. Obvious candidate molecules might be receptors for diffusible signals provided by growth factors or components of the extracellular matrix In this scenario, cells at different anterior-posterior positions interpret their complement of HOXgenes by synthesizing particular surface molecules that are different from cells at other anterior-posterior positions. The surface molecules providethe basis for cells ‘knowing their place’ with respect to their immediate neighbours and the anterior-posterior axis as a whde. Based on this presumption we have set uut to determine whetherectopicexpressionof a mouse homeobox gene WOX-3.3) can result in detectable changes in re!!-surface molecules [9]. Such predicted changes could

conceivablyinvolve any type of cell-surface component (protein, lipid or carbohydrate), although these might not necessarilybe majorchanges.In order to be able to identify what might be very subtle changes in any cell-surface molecule(s) we have used the very sensitive, non-disruptive technique of repeated aqueous

two-phase partition [18].The basis of this techniqueis the partitioning of cells between two immiscible aqueous pllases of high-molecular-weightpolymers, such as polyfethyleneglycoll(PEG) and dextran. In any given combination of the two polymers(phase system) the extent to which cells partitionbetween the interface and the upper (PEG-rich) phase is dependent on the gross surface characteristics of the cells. 4ny change in cell-surface molecules will result in a corresponding change in partition. Partition is extremely sensitive to these surface changes, being able, for example, to detect differencesin a singlecarbohydratespecies [19]. AdditionaIly,phase systemscan be adjustedto principally detect charged cell-surface molecules (chargesensitive phase systems) or non-charged cell-surface molecules (charge-insensitive phase systems). Partition differences between cell populations can be greatly enhanced by performing multiple successive partitions in a thin-layer counter-current distribution (TLCCD) apparatus, a process directly analogous to chromatography dependent on cell-surface properties [20]. In this communication we report the stable induction of ectopic expression of the murine homeobox gene Har-3.3 in the mouse breast cancer cell line Shionogi 115 iSll9. TLCCD analysis of Hex-3.3-expressing clones shows that this gene product can result in reproductble changes in cell-surface molecules. Matarials and M&ods RiVase protecth

assays

RNA was prepared from 5115 cells by the guanidinium thiocyanate method [21]. RNase protection assavs were oerformedas deseriid elsewhere 1221.

Fig. 1. ‘Ftre Had.3

arpressiorr construct. Conlml clones were lf2ns-

fectedwith he sameptasmidminus th HOX-3.3cDNA BGH. bovinegmwthf#LlrmlMle.

A cDNA encoding the entire Hex-3.? PRII open reading frame was isolated from an 85day-old mouse embryo cDNA library in AgtlO by standard DNA cloning techniques 1231.This was subcloned into the eukaryotic expression vector pRc/CMV (Invitrogenl, a plasmid bearing the gene neomycin phosphotransfcra (Fig. 1). Plasrtrid DNA WASp~rifkd through IJVO CsCl gradients prior to transfection [U]. Celltrati.$fctiuTl

S115 cells were stably transfectedwith 10 Fg of plasmid DNA using the calciumphosphateprecipitation methodkit suppliedby Stratagene)accordiig to the suppliersinstructions.Controlclones were generated by transfecting SllS cells with (!re expression construct lacking the HUX-3-3insert. The S115cell line was used as it was found to not express Hex-3.3 and to be easilydiirsed into a single-cellsuspensionby mild proteasetreatment(data not shown). S115 cells were routinelymaintained in DMEM supplemented with IO% foetal calf serum (FCS) and lo-’ M testosterone (Sigma).Colonies derived from single transfected cells were selected with the neomycin anaIogue Cl418 (Gibco) at 600 pg/mI and isolated using cloning rings. The resulting cell clones were maintained in G418 medium. Total cell RNA from each transfeeted cell line was prepared by the guanidiumthiocyanate method and analysed by Northernslot blot [21]. Blots were probed with 3zP-labelled Hex-33 or @-actin and autoradiographed at -70°C [U]. Blots were quantified on a flyingdot spot scanner (Shimadzu) and the relative Hox-S.3expression levelswerecalculatedby comparisonwith @actin.To calculate the growth rate of the clones a known number of cells was passaged to

severalflasks.These were then harvested’ at 24 h intervals and the growthcurveswere plotted.From

255 these curvesthe doubling-timeduringthe logarithmic growth phase was calculated. The doubling-timeof untransfeeted5115 cells is 24 h. TLCCDamlysk of tramfected

cells

Cellswere isolatedat the same point in the growth curve by incubation in 0.25% Dispase@CL),2 mM EDTA in cakium- and magnesium-freeHanks’balanced salt solution(HBSSCMFIGibcol at 37°C for 30 min. The resultingsingle-cellsuspensionwas centrifugedat 150xg for 3 min to pellet the cellsand the enzymemediumwas removedby aspiration.Cell suspensions were then incubated in 2 mM EDTA in HBSS-CMFat 37°Cfor 2 h, a period found adequate for full recoveryof surfacepropertiesas detected by 7LCCD (dati not shown). After this recoveryperiod cells were pelleted at IsOXg for 5 min, resuspended in top phase and loadedimmediatelyinto the TLCCD rotor.TLCCDruns of 60 transferswere conductedat 4°Cwith a settlingtime of 9 min on a Bioshef MarkV TLCCD apparatus.Charge-insensitivephase systems consisted of 35% dextran T500 (Pharmacia),4%

polyfethyleneglycol), a.05 M NaCl, 8.01 M phosphate bufferand 4.8%sucrose.Upperand bwer p&es were vigorouslymixedand allowedto separateat 4°C until both phaseswere clear. The upper and ‘!.ver phases were separatedand stored until run OT.ihe TLCCD. The TLCCDrotorswerefilled to 0.85%of the bottom rotorvolumewith lowerphaseand an equal volumeof upper phase was added. After the TLCCDrun, fractions were diluted with phosphate-bufferedsaline (pIi 7.6)and lsoton Kcndter Electronics)and countedon a model ZB Coulter counter Koulter Electronics).Results were plotted as graphs of cell counts against fractionnumber. Itmiltsand-

Multipleaqueoustwo-phasepartition usingTLCCD is a well establishedtechniquefor cell separationand for detecting changes in cell-surfacemolecules1181. The techniqueis highlyreproducibleand is extremely sensitive, since partition is exponentiallyrelated to changes in eekuface molecules.Thus small changes on the ceII+mfacecan be reflectedby largechangesin

partition.TLCCDhas, for example,been used to show that red blood cells from individualsof an inbrti rat-strain partition identically,whereasthose from individualsof an outbred strain have differentpartition distributions,implyingthat TLCCDcan detect small genetic differenceson the cell surface[24].Siiilarly, closelyrelated cultured leukemiccell sublines can be distinguishedby TKCD [25k‘TLCCDhas also proved usefulfor studyingsurfacechangesin development,for examplethe identificationof differentiatingcell types in Dicryosrelium G%SC&MI and chick-limb-budmes-

enchyme[26,27].We opted to use this techniqueas a sensitive assay for surface changes that might IX induced by homeoboxgene expression,as it is the only

methodthat can detectchangesin anysurfacemolecule (protein,carbohydrateor lipid)on intactcells.Multiple partition producespeaks of cells based on a Poisson distribution.Deviationfrom this distributionindicates surface heterogeneitywithin the populationand the

relative position of the peaks is a ‘measure’of the grosssurfacecharacteristics. Thus a changein the peak fractionnumber indicatesa changein surfaceproperties. Two HOX-3.3 transcripts are found in mouse embryos(Shimeldet al., manuscriptin preparation).These are derivedfromtwo independentpromoters(PRI and PRII) and produce two proteins sharing the same homeodomainbut differingat the amino terminus.A mouse cell line, Sli5, was chosenthat couldbe trans-

fectedas a monolayerbut also grownin static suspension. This was essentialto ensure that minimumdamage to cell-surfacemoleculeswouldresultfromdissociation. An RNaseprotectionassay,usinga probewhich could identifyboth transcripts,detectedonlylowlevels of the PRI transcript(Fig. 2). in embryosPRII is the major transcriptand this was not found in S115cells. Sixclones transfectedwith the contro1plasmid(prefixed l-14 and il transfectedwith the &x-3-3 expression construct+refmed l-2. and 5-2-Jwere isolated. Clones prefiied l- and 5- were isolatedfrom chrono logicallyseparatedtransfections.All cloneswereanalysedfor HOX-3.3 expressionand the levelstandardized by comparisonto @-actinexpression.All the control clones had approximatelythe same level of standardizedhybridization(not shown)and the averageof these was used to calculate the relative level of Hex-3.3 expressionin the experimentalclones (Table I). Since TLCCDis Knin to detectdifferencesbetweencellsin difterent pzrts of the growthcurve,the growthcurves for all clonesto be assayedbyTLCCDwerecalculated [28-301.From these the doubling-timesduring the logarithmicgrowthphase were deduced and used to ensure that all cells harvestedfor TLCCDwereat the same point in the growthcurve (Table I). Single-cell suspensionswere isolatedfrom monolayersof cells by enzymaticdigestion and allowed to recover surface propertiesin suspensionprior to TLCCDanalysis. TLCCD runs were conducted with two separate batches of a charge-insensitivephase systemand the profileswereplottedgraphically(Fig.3).Controlctones and very-low-expressing clones consistentlyproduced distributions which peaked in fraction 39 or 40. Higher-expressing clones,up to efeventimesabovethe controllevel,also produceddistributionswithpeaksin fractions 39/40. Clones expressing Iilox-3-3above twelvetimesthat of the controlsproduceddistributions whichpeakedin fractionslowerthan39/40 andshowed

2% a Corre~dtktIltaween the level of f-h-3.3 expreshn

andthe d&tree of the peak from the eontrot peaks. No

such

co&a&m

was seen between

peak fractian

rate. Ah these distributions were found to be high& repeatabk with different batches of cehs in tk smw batch of phase went. Although repeat TLCCD runs with different batches of phase system occasim& produced peaks differing by one fktion, these were eon&tent for all clones, so the relative differences between the chutes remained the same. No differences in partitkn cotrid be detected using chargesensitive phase systems (data not shown). It thus appears tkt the level of ectopic HOX-3.3 expressioncan be equated with the extent of cellsurface change as revealed by multiple partition in a charge-insensitive phase system. Ahhough we obtained a progressive shift in the cell distribution correlated with high levels of Har-3.3 expression, this does not n-y mean that the progressive changes are produced by the same surface molecule. Different levels of rotein cot&Lfor example, be regut genes. Also, the surface changes deteeted need not be due to direct regulation by HQXaad

growth

425bp 361 bp 310bp

i,sL Fig.2RNasepalHecm . asay of S115 tmal RNA using a probe spanning the Har-3-3PRI/PRIl spike site. Lane 1. probe onk lane 2 SO~g Slt5 RNA; law 3. !N pg EKhprirltio cafi IRNA One m protectedbandat 3!0 bp is seenfrom hybridization IO the FRI trawxript. Mouse ll.May+mbryu RNA typically show a setoad Moe intense band at 361 bp from hybridization IOthe PRlI tip? Wiki et aiI, mmmipt in preparation). No protected band of this size is mesa in the 5115 track. Full-length probe, inc!uding 61 bp of trmcrii polylinker. is marked at 425 bp.

e

1-I-2 l-1-3 I-9-4

X.0 315 27.0

LO0 1.14 0.86

39 40 39

-

3.3 of &-get genes ettco-fingc&surface mokcuks, but could be an indirect function -t-ring via, for exampie, other transkption factors. All the distriitions obtained were narrow peaks of similar shape, indicating thai the cells were homogeneous with respect to the surface mokuks being detected This suggests that all the cells in a single clone had undergone the same relative surface changes and that all observations were based on living healthy eeLIs,since dead cells typically partition in the fust few fractions. This was supported by the fact that cells from aLIpeak fractions were equahy viable and could bc recuhured (data not shown). Failure to detect signiftcant surface changes in cells with lower levels of Hex-3.3 expression may be due to a number of factors. There may be a threshoLdlimit of expression below which the downstream target genes are not activated, as the eoneentration of some other transcription factors has been found to be critical in determiniig target-gene activation [31]. The fact that S115cells do express detectable levels of HOX-3.3PRI transcript may also be significant, since it has been found that the PRI and PRII proteins may act antagonistically [321.Thus, low levels of PRI protein may block low levels of PRII protein so ceU-surfacechange is only seenwith high levels. A third possible explanation is that fk-3_3 RNA Levelsdo not correlate with HOX-3.3protein level:. Despite repeated attempts, anti-Hruc-33antibodies were unable to detect protein in transfected clones or in L-cells, which express Hox3.3 in very large amounts 1331.Until a more suitatle antibody is available,this remains unresohed. Charge-insensitive phase systems principally detect changes in non-charge-associated surface mokeuks, often referred to as cell-surface hydrophubicity.A shift of the cell distribution to lower fraction numbers in this phase system refkcts a decrease in surface hydrophobic&yor an increase in hydrophilicity. Ahhough

257 aqueous two-phase partition allows detection of

changes in c&surface properties, it cannot predict or identify the mokcuIes involved,although the failure to detect any surface changes in charge-sensitive phase systems implies that the surface molecules involvedare not appreciably charged A~WCMIS two-phase partition does, however, provide a rapid and sensitive method of evakating the effects of iwmeobox gene expression on ceellsurfaces. In the case of HOX-3.3,expressiondependent cell-surfa& changes can be detected. providing

Tmnsiected doe&

evidence that home&ox gene expression in embryos is providing cells with positional cues to which they can respond by specific cell-surface changes. Such chmges could involveany surface componeng but likely andidates might be proteins

acting as receptors

for &if-

fusible ligands or extracellular matrix components. Interesting!y, it has already beer! suggested that the X4nopus homologue of H053.3, XLHbox 1, may be able to effect specific changes in the cell-surface 1341. This suggestion was based on the observation that

tow to high expresokn. 5000 4000 3000 2000 1000 a FrHtion

numbw

Frwtion

nurbw

SO00

4000 3000 2000 1000 0 FrrtU-ii

-numkr

aooo 6000 4600 2000 0 Frreclan number Frwlkn number Fig. 3. TLCCD distr&~tk~~~ of transfectedcell &rm amlysed in onephasebatch.The peak fractionof each distributionis indicated.Clones a&sed ia a smmd phasebatchshowedpeakfractionsia identicalrelativepositions.

3 Otting. G, Qiah Y., Mulkr. M.. Affolter, M., Gehrifig, W. and duringthe position-specific induction of aeurcctoderm wum K t1%7l EMBO J. 7.4305-43Q9. by mesoderm in the Xenopus embryo, XLHbox I ex4 CR. Td S. sod Parker, CS. 11989) Nature 341, prexsion is found to be exactly aligned in these two germ layers. Ia mouse embm aligned expression of S Driver, W. and N%kin-Volhard C (19891Nature 337,136-141. some Hm genes also occurs early in development 6 Lcke. MS. and Manley,J-L (1989)Cell 56.573-583. 7 GM_ Hapshi, S, Krasnw. M., Hognus,DS. and abhough e_xpre&n later shifts out of a@nmen~ possiP. (1989)cell S7, lG;7-103% biy due to increased growth of the qurine central 8 Rep!ski, M, Desk, S, McGinnis, N. aad McGinnis.W. (1991) netvotts 5lstetn [3335].IIris led to the intriguing idea Gees lb, 5.278-286. that XL&w 2 expressed in the mesderrn may be R Eus, EP, Bu~enshaw, M.D. and indtring XLHbar 2 expression in the overiying ectotlll0.z 397-4%. . (1991)J Biol. Ckm. zh6,3246derm. Central to this assumption would be the ability 3%. of Xi!,J to direct specific changes in the cellP. (l!B9J EMEG J. 8, I497-1505. 11 DubkD. surface, an ability we have now demonstated for Hex- 12Gmlta~G u, N. awI Krwulaarf.R (1989) Cell 57. 3.3. x7-37%. Specific changes in cel&nface mdectdes induced 13 ~~EZ..lbmkldKaodH~DS. u9@1 Cell S7,1031-1033. by Hm gene expressim can provide a way of explain14 cha. KW.Y, Gee@ J, Wrigk CVE, Fria k Hardwickc J. ing how differences in an axial Hex a& could be aw! De ktberk, EM (1988)EMBO J. 7,2139-2149. interpreted by cells as position. Central to this idea 15 Odemwld, W_F_Garben, J, Amkiter. I-L Toumier-Laaerve. E would be the concept that different Hax genes can Genes De%3,1S8-172” i&ce different CeiLsurfacemdecula. TLCCII can .EF., Hc%nan, F.M. and SaxI, M.P. provide a simpbz,rapid method for in-vitro analysisof 17 Kesel Bt d Grws P (1991)Cell 67.89-104. surface changes due to ectopic expression of different 18 Walter, H, m 0. and Fii, D. (1985) Partitioning in Hm genes. In the embryunichindbrain a Hax code is Aqw=sTw.wl=Sssttaa.Awkmic~Landon. postulated as providing cells with positional spec%ca19 Sharp, PI. and Witrm, GS. WX34~Biodh. BiQphys.Acta ticm and appears to be linked to segmentation in the 72.z 1X-182. 20 sharpe, PX (l%41 Trends Ekcbn. Sci. 9.374-377. form of rhombotnercs and neural crest migration 21 CbwymlkPandSaccKN.~l%7)Anat0ii.162,1%136,371. A possible baa for the development and main159. tenance of lineage rest&ion in the rhombomeres may I.4s.!% tzTta&uri, L wliilter, u, Clark% 22 wna, PL be specific changes in c&surface molecules induced J.P.ti m P-T-(19911Mech. Deva35,129-142 by Hm genes and other genes such as Kr~x-20.Fur23 !kwk.k. 1. Frirsh, E.F. and biatis, T. (1989) Molecular m-L--C-. k.biiga~lbt?al+CO!dS~gHkUbJU Lavur_,“., thermore, tk subsequent migration of neural crest and - preaG4d~~~induction of Hex expression in branchial arch ecto24 Waker. H. and Krab. EJ. (1985) Eksbk Biophp. Acta 8%. derrn may be a conscqttenceof Ho-x-relatedcell-surface 8-15. changes. 25 Walter, H, kob, El., Al-Romaihi, FA Jdhmon, D. and Lanio. It shotdd now be possiie to identify the surface CR wl%3)Cell BiQpbB. t3,173-187. 26 Sharpe, P.Tq Trcffrey,T-E and Waits DJ. (1982)J. Eanb.Erg. molecules that are different in the Ho.r-33-expressing Morph. &7*181-193. celIs and also to use the phase partition technique to 27 Cobll. CPy !%arpe, P-T. and Wolpert, L (1986) I. Emb. l%p. investigatethe effects of expression of other home&ox Ma@_ 94,267-275. genes, either akrne or in combinations,on cell-surfaces, 28 Gemein. DM and Boanzon, KB. (19741Exp. Cell Res. 87, and thus provide a biochemical assay for Hex codes. 73-78. 29 Gelstein, D.wL aBd Baslmann,H.3. (1974) Exp. Cell Res. 88, Also since TJAXD is possible with very small numbers * 225-m. of celJs it shodI even be possible to investigate the 33 Sharpe, P-T. ad Watts. DJ.(198%J. Cell Sci X339-346. surface properties of nettrectoderm cell poputations in - 31 Zuo, P, saaadjnrir, D, colgaa J, wan, K., tine, M. ad the developing hindbrain. Manley,Lt.” (1991)Gene5 DeK 5,254-264-

We thank Dr. Jim Glover for the gift of the Sli5 cell line and Dr. P. lnui Coletta for critical reading of the manuscript. This work was supported by a Morrison Watson fellowship in anatomy to S.M.S. Rekrences 1 Getrtig$ WJ. w87) scklxe 236.1245-1251. 2 Kewl, M. and Gmss, P. (19901Sciincx 249.374-379.

32 Wright. CVE, Clw, K.W.Y, lbdwkk, J, Cobs, RI-i. ati Lk Robenk EM. wxw Cell 59.61-93. 33 Oliver. G, Wright, C.V.E, lkbkke, J. and De Robertis. EM. wlB) ElrlBo 1.7,3199-3209. 34 De Robertis E..M, Oliver. E aad Wrighl CVE (1989)Cell 57, l&9-191. 35 Frolun, % BoyIs, M. ap$ Marfin, G. (l!B@ Development 110,S89-607. 36 Hut P, Guilsaw M., Cuok, M, Sham, M.-K l%ella, A., Wilkkq Q., Bcnxkelli. E and Ktumlaut, R (19911Name 35;. 861-w. 37 Hunb P, Wibwn, D. and Krwlauf, R. It9911 Debzlopment 112 43-50,