1

mRNA LOCALIZATION AND CELL MOTILITY:ROLES OF HEPARIN-BINDING PROTEINS AMPHOTERIN

AND HB-GAM IN CELL MIGRATION

byCarole Fages

Laboratory of Molecular Neurobiology,Institute of Biotechnology

andDepartment of Biosciences, Division of Biochemistry,

Graduate School in Biotechnology and Molecular Biology,University of Helsinki

Finland

Academic DissertationTo be presented for public criticism, with the permission of the Faculty ofScience, University of Helsinki, in the auditorium 1041 at Viikki Biocenter,

Viikinkaari 5, Helsinki, on March 3rd, at 12 o’clock.

2

Supervised by:Professor Heikki Rauvala

Institute of Biotechnology andDepartment of Biosciences, Faculty of Science,

University of Helsinki, Finland

Reviewed by:Professor Jorma Keksi-OjaDepartment of Virology and

Dermatology and Venereology,University of Helsinki, Finland

and

Docent Pekka LappalainenInstitute of Biotechnology,

University of Helsinki, Finland

Opponent:Professor Antoine Triller

Laboratoire de Biologie Cellulaire de la SynapseNormale et Pathologique (Institut National de la Santé et

de la Recherche Médicale U-497),Ecole Normale Supérieure,

F-75005 Paris, France

ISBN 951-45-9126-7 (PDF version)Helsingin yliopiston verkkojulkisut

Helsinki 2000

3

á ma maman,pour ton amour et ton courage.

Tu aurais été si fière de moi.

5

CONTENTS

ORIGINAL PUBLICATIONS..............................................................................................................7

ABBREVIATIONS.................................................................................................................................8

ABSTRACT...............................................................................................................................................9

1. INTRODUCTION...............................................................................................................................10

2. REVIEW OF THE LITERATURE......................................................................................................11

2.1. MECHANISMS OF mRNA LOCALIZATION...........................................................................112.1.1. Localized RNAs.............................................................................................................................112.1.2. Localization tags..............................................................................................................................132.1.3. RNA recognition.........................................................................................................................142.1.4. The role of the cytoskeleton in mRNA localization...............................................................152.1.5. Transport particles......................................................................................................................152.1.6. The cell environment..................................................................................................................162.1.7. Translation of localized mRNAs...............................................................................................162.1.8. Functions of mRNA localization...............................................................................................16

Cell polarity...............................................................................................................................16Cell migration............................................................................................................................17Cell differentiation....................................................................................................................17Process outgrowth and synaptogenesis................................................................................18Synaptic plasticity.....................................................................................................................18

2.2. CELL MIGRATION.......................................................................................................................192.2.1. Extracellular signals.......................................................................................................................19

Laminin, HB-GAM and other ECM components............................................................19Growth factors and cytokines.............................................................................................21

2.2.2. Signal transduction.....................................................................................................................21Integrins.................................................................................................................................21Syndecan family.....................................................................................................................22Receptor tyrosine kinases...................................................................................................22Rho-family proteins..............................................................................................................22

2.3. AMPHOTERIN/HMG-1...............................................................................................................232.3.1. Characteristics of amphoterin...................................................................................................232.3.2. Localization and interactions of amphoterin in the extracellular space...............................242.3.3. Cell surface receptors of amphoterin........................................................................................24

Syndecans and other proteoglycans.................................................................................24Receptor for advanced glycation end products (RAGE)..............................................24

3. AIMS OF THE PRESENT STUDY..................................................................................................26

4. MATERIALS AND METHODS......................................................................................................27

6

5. RESULTS............................................................................................................................................28

5.1. ROLE OF HB-GAM/N-SYNDECAN INTERACTION IN CELL MOTILITY......................285.1.1. Enhancement of osteoblast recruitment and bone formation by HB-GAM......................285.1.2. Regulation of mRNA localization by HB-GAM/N-syndecan signalling pathway............285.2. SUBCELLULAR LOCALIZATION OF AMPHOTERIN mRNA AND PROTEIN............285.2.1. Localization of amphoterin mRNA............................................................................................285.2.2. Comparison of the localization of amphoterin mRNA and protein......................................295.3. COMPARISON OF AMPHOTERIN AND βββββ-ACTIN mRNA LOCALIZATION...............305.3.1. β-actin and amphoterin co-localize in the cell processes......................................................305.3.2. Amphoterin and β-actin mRNA localize to distinct granule-like structures in the samesubcellular compartments....................................................................................................................305.3.3. Amphoterin and β-actin are anchored and transported to the neurites by a microfilament-dependent mechanism...........................................................................................................................315.3.4. Locally applied ECM proteins induce amphoterin and β-actin mRNA localization in avery similar manner..................................................................................................................................325.4. ROLE OF AMPHOTERIN IN CELL MIGRATION.................................................................325.4.1. Export of amphoterin to the extracellular space from cells extending processes...............325.4.2. Inhibition of amphoterin decreases cell migration...................................................................335.4.3. RAGE mediates the neurite outgrowth induced by amphoterin..........................................34

6. DISCUSSION AND CONCLUSIONS...........................................................................................34

6.1. HB-GAM IN CELL MIGRATION................................................................................................346.2. SORTING OF AMPHOTERIN mRNA AND PROTEIN TO THE CELL PROCESSES...356.2.1. Amphoterin mRNA: a new example of localized mRNA........................................................356.2.2. Is amphoterin locally translated?..............................................................................................356.2.3. Is the local translation of amphoterin the key to its export to the extracellular space?....356.3. AMPHOTERIN IN CELL MIGRATION....................................................................................366.3.1. Amphoterin is required for cell motility...................................................................................366.3.2. Does RAGE act as the signalling receptor for amphoterin in cell migration?....................366.4. IS mRNA LOCALIZATION INVOLVED IN CELL MIGRATION?......................................376.4.1. Mechanisms that localize amphoterin and β-actin mRNAs..................................................376.4.2. Localization of amphoterin and β-actin mRNAs is directly coupled to transmembranesignalling involved in cell migration..................................................................................................386.5. HYPOTHETICAL MODEL FOR THE REGULATION OF CELL MIGRATIONBY AMPHOTERIN AND HB-GAM DURING NORMAL DEVELOPMENT AND INPATHOLOGY.......................................................................................................................................396.5.1. HB-GAM as a target-derived cue and amphoterin as an autocrine/paracrine regulator ofcell migration..........................................................................................................................................396.5.2. Implications of HB-GAM and amphoterin during development and in pathology............40

7. ACKNOWLEDGEMENTS...............................................................................................................43

8. REFERENCES....................................................................................................................................44

7

ORIGINAL PUBLICATIONS

This thesis is based on the following original articles, referred to by their Roman numerals inthe text.

I Imai, S., Kaksonen, M., Raulo, E., Kinnunen, T., Fages, C., Meng, X., Lakso, M., andRauvala H. (1998). Osteoblast recruitment and bone formation enhanced by cell ma-trix-associated heparin-binding growth-associated molecule (HB-GAM). Journal ofCell Biology. 143,1113-1128. http//www.jcb.org/

II Fages, C., Kaksonen, M., Kinnunen, T., Punnonen, E-L., and Rauvala, H. (1998).Regulation of mRNA localization by transmembrane signalling: local interaction ofHB-GAM (heparin-binding growth-associated molecule) with the cell surface local-izes β-actin mRNA. Journal of Cell Science. 111, 3073-3080.http//www.biologists.com/JCS/

III Punnonen, E-L., Fages, C., Wartiovaara, J., and Rauvala, H. (1999). Ultrastructurallocalization of β-actin and amphoterin mRNA in cultured cells: application of tyramidesignal amplification and comparison of detection methods. Journal of Histochemistryand Cytochemistry. 47, 99-112.http//www.jhc.org/

IV Fages, C., Nolo, R., Huttunen, H.J., Eskelinen, E-L., and Rauvala, H. (2000). Regulationof cell migration by amphoterin. Journal of Cell Science. In Press.http//www.biologists.com/JCS/

V Huttunen, H.J., Fages, C., and Rauvala, H. (1999). RAGE-mediated neurite-outgrowthand activation of NF-κB require the cytoplasmic domain of the receptor but differentdownstream signalling pathways. Journal of Biological Chemistry. 274, 19919-19924.http//www.jbc.org/

8

ABBREVIATIONS

Aβ amyloid-beta peptideAD Alzheimer´s diseaseAGE advanced glycation end productBDNF brain derived neurotrophic factorCaMK calmodulin kinaseCREB cyclic AMP response element bindingCSF colony stimulating factorECM extracellular matrixEN-RAGE extracellular newly identified RAGE binding proteinER endoplasmic reticulumGAP43 growth associated protein 43GAP GTPase-activating proteinsGEF guanine nucleotide exchange factorGFAP glial fibrillary acidic proteinGTP guanosine triphosphateHB-GAM heparin-binding growth-associated moleculeHGF/SF hepatocyte growth factor/scatter factorHMG high mobility group proteinHnRNP heterogenous nuclear ribonucleoproteinIL interleukinInsP inositol phosphateKH hnRNP K-homology domainLPA lysophosphatidic acidLTP long-term potentiationMAP microtubule associated proteinMBP myelin basic proteinMK midkineMOBP myelin-associated/oligodendrocytic basic proteinN-CAM neural cell adhesion moleculeNMDA N-methyl-D-aspartateNT neurotrophinODN oligonucleotideOMP orotidine monophosphateOSF oeteoblast specific factorPDGF platelet derived growth factorPI 3 kinase phosphoinositol 3 kinasePKC protein tyrosine kinasePTK protein kinase CRAGE receptor for advanced glycation end productsRLS RNA localization signalRRM RNA recognition motifRTS RNA transport signalTNF-β tumor necrosis factor-βt-PA tissue plasminogen activatortRNA transfer RNATSR thrombospondin repeatu-PA urokinase-type plasminogen activatorUTR untranslated regionZBP zipcode binding protein

9

ABSTRACT

To create and maintain its structure and organization, a cell has to target newly-synthetizedproteins to their correct destination. One way is to target the messenger RNA instead of theprotein. Although the mechanisms of mRNA targeting are unclear, mRNA localization andconsequently local translation have been shown to be essential in many cellular functions,such as cell polarity, cell migration, cell differentiation and synaptic plasticity. For example,localization of β-actin mRNA at the leading edges of fibroblasts augments cell translocation.Therefore, it seems reasonable to assume that the mechanisms regulating cell motility alsoaffect the localization of mRNAs. To address this question, extracellular matrix-associatedfactors that enhance migratory responses, were locally applied to cells via microbeads. In thepresent study, we show for the first time that extracellular stimuli through transmembrane sig-nalling are able to regulate the localization of mRNA within a cell.

Heparin-binding growth associated molecule (HB-GAM) is an extracellular matrix-associatedprotein implicated in the formation of neuron-target contacts during development and in synapticplasticity in adult brain. Interestingly, HB-GAM has been isolated from the murine osteosarcomacell line MC3T3. However, its function in bone tissue has not been previously addressed. Tothis end, expression of HB-GAM during bone formation was examined. HB-GAM is expressedin the cell matrices that act as target substrates for bone formation. Intriguingly, N-syndecan,the receptor of HB-GAM, is expressed by osteoblasts/osteoblast precursors, suggesting thatHB-GAM/N-syndecan interactions mediate osteoblast recruitment. Osteoblast cell lines, whichexpress abundantly N-syndecan, migrate rapidly to HB-GAM in a haptotactic transfilter assay.Furthermore, transgenic mice overexpressing HB-GAM are characterized by an increased bonethickness. Taken together, these results suggest that HB-GAM plays a role in bone formation,regeneration and maintenance.

Amphoterin, a major form of HMG (high mobility group)1 proteins, is strongly expressed inimmature and malignant cells. Although it lacks a classic secretion signal, extracellular roles inneurite outgrowth, cell differentiation and endotoxemia have been attributed to amphoterin.This work examines the sorting mechanism and the function of this protein in motile cells.Asymmetric distribution of amphoterin already occurs at the mRNA level, suggesting a localmode of translation for amphoterin. The targeting mechanism of amphoterin resembles that ofβ-actin mRNA since localization of both mRNAs depends on intact microfilaments and can betriggered by cell contact with extracellular matrix proteins, like HB-GAM or laminin. The distinctlocalization of the amphoterin mRNA and protein to the cell processes suggests a role in cellmotility. This premise is further supported by the finding that specific decrease of amphoterinmRNA and protein, using antisense oligonucleotides transfected into cells, significantly re-duces cell migration to laminin in a transfilter assay. Moreover, affinity-purified anti-amphoterinantibodies also inhibit cell migration to laminin, supporting an extracellular role for theendogenous amphoterin in cell motility. Finally, the fact that amphoterin promotes neurite ex-tension through its receptor RAGE (receptor of advanced glycation end products)/GTPasepathway, provides a mechanistic explanation for the role of amphoterin in cell motility.

In conclusion, we propose that HB-GAM and amphoterin play a crucial role in cell motility.However, their contributions to cell migration are distinct: HB-GAM acts as a target-derivedcue whereas amphoterin is an autocrine regulator, produced by the migrating cells themselves.

10

1. INTRODUCTION

One of the most important events in cell dif-ferentiation and development is the determi-nation of asymmetry. For more than a century,the idea, amongst embryologists, has emergedthat regional inhomogeneities in the distribu-tion of particular substances in the cytoplasmof the egg or an early embryo could lead todifferential inheritance of cell fate determinants(Von Baer, 1837 and Conklin, 1905). Apart fromthis concept, histologists have noticed thepolarized nature of differentiated cells andunderstood that it is likely to be a key to theirphysiological function (Van Beneden, 1883;Rabl, 1885). In almost all cells, asymmetry is required forsubsequent functions, as in the polarity ofneurons that transmit directional signals or inthe polarity of specific morphogens in oocytesthat drive different cell fates in the embryo.Morphological asymmetry requires the sort-ing of proteins unequally in the cell. Duringthe last decade, the problem of how proteinsare sorted to particular membrane-boundorganelles has been the focus of considerableefforts and now the secretory pathway of suchproteins via the ER/golgi route is well under-stood (Harter and Wieland, 1996). In contrast to the well described sorting ofmembrane proteins, very little is known abouthow cytosolic proteins are segregated withinthe cytoplasm. However, it has become in-creasingly clear that the transport of mRNAsconstitutes an important step in the localiza-tion of proteins. The first evidence for thesubcellular localization of mRNA came fromthe discovery that myelin basic protein (MBP)mRNA is localized to the cell processes ofoligodendrocytes (Colman et al., 1982).Shortly thereafter, several maternal mRNAs inDrosophila and Xenopus were shown to belocalized during oogenesis. More recently, lo-calized mRNAs have been notified in somaticcells. Thus, it is now reasonable to proposethat the subcellular localization of specificmRNAs serves as a general mechanism fortargeting proteins in polarized cell types.

Subcellular localization of mRNAs might haveseveral advantages as compared to localiza-tion at the protein level. Since mRNAs act astemplates for protein translation, their locali-zation allows specific proteins to besynthetized in subcellular compartments wherethey are required and prevents their expres-sion in regions where they are not required.For example, membrane-binding stickyproteins may be easier to localize, based onmRNA transport as compared to proteintransport. mRNA localization may also be anadvantage from the viewpoint of cell economysince one molecule of mRNA can be translatedmany times and thus generates a pool of aspecific protein in a particular region of thecell. Although the mechanisms of mRNA target-ing are not fully understood, mRNA localiza-tion, and consequently local proteintranslation, have been shown to be essential inmany cellular functions, such as cell polarity,cell migration, cell differentiation and synapticplasticity. These functional aspects implicatethat mRNA transport is spatially and tempo-rally regulated within the cell. In fibroblasts,the localization of β-actin mRNA at the lead-ing edge is required to promote cell motility(Kislauskis et al, 1997). Therefore, it seemsreasonable to assume that mechanisms regu-lating motility also affect the localization ofcertain mRNAs.

The migration of cells over substrata is an im-portant phenomenon that requires the co-or-dination of several cellular processes. It canbe divided in five steps: (1) extension of theleading edge; (2) adhesion to matrix contacts;(3) contraction of the cytoplasm; (4) releasefrom contact sites; and (5) recycling of mem-brane receptors from the rear to the front ofthe cell (Sheetz et al., 1997). Many componentsparticipating in cell migration have beenidentified. However, the manner in which thesecomponents work together to give rise to mi-gration is still unclear. The present study wasundertaken to elucidate, how the heparin-bind-ing proteins amphoterin and HB-GAM relateto the motility mechanisms. Special emphasis

11

was in mRNA localization since previousstudies, on β-actin, have suggested the im-portance of mRNA localization in the functionsof cell motility-associated genes.

2. REVIEW OF THE LITERA-TURE

2.1. MECHANISMS OF mRNA LOCALIZA-TION

Over the past decade, an increasing number ofmRNAs have been found to localize in manycell types in numerous animal species,highlighting the general importance of this tar-geting mechanism. During the emergence ofthis research field, it has become clear that themechanisms underlying the mRNA localizationare very similar in different cell types and

should be considered as a general model.However, many features of the mRNAtargeting are not well understood.

2.1.1. Localized mRNAs

Many model systems have been applied tostudy the localization of mRNAs, and each ofthem offers unique advantages. In Drosophilaand Saccharomyces cervicae, genetics andmolecular biology are conducted with ease.Xenopus possesses a very large oocyte facili-tating molecular, biochemical and spatial stud-ies. Neurons are highly polarized cells with verylong processes, which facilitates identificationof the subcellular differences. Cultured cellsoffer the advantage of being easily transfected,as well as allowing studies of environmentaleffects on the mechanisms of mRNAlocalization. Although a number steps of the

Table 1. Localized mRNAs in somatic cells

In non-neuronal somatic cells

mRNA Localization profile Class of protein or Protein function

α-actin Perinuclear (myoblasts) Actin isoform (Kislauskis et al., 1993)β-actin Peripheral (motile cells) Actin isoform (Kislauskis et al., 1993)γ-actin Perinuclear (myoblasts) Actin isoform (Hill and Gunning, 1993)Cardiac Perinuclear Myofibril components (Russell et al., 1992)α-myosin (active myocytes)heavy chainGFAP Peripheral and Intermediate filament component

perinuclear (glial) (Landry et al., 1994)Vimentin Costameres (muscle cells) Intermediate filament component

(Fulton and Alftine, 1997)MBP Cell processes Myelin constituent (Colman et al., 1982)

(oligodendrocytes)MOP (Myelin- Cell processes Myelin constituent (Holz et al., 1996)associated/ (oligodendrocytes)oligodendrocyticbasic protein)Carbonic Cell processes Myelin constituent (Ghandour and Skoff, 1991)anhydrase II (oligodendrocytes)mRNAsTitin Myofibril Myofibril component (Fulton and Alftine, 1997)

(skeletal muscles)Acetycholine Neuromuscular Acetycholine receptor (Merlie and Sanes, 1985;receptor subunits junction (muscle) Fontaine et al., 1988)Ash1 Presumptive daughter cells Cell-fate determinant (Takizawa et al., 1997)

(Saccharomyces cervicea)

12

Table 1. (continued)

In neurons

mRNA Localization profile Class of protein or Protein function

MAP2 Dendrites (cortex, dentate gyrus) Microtubule-associated protein (Garner et al., 1988)CAM II kinase Dendrites (cortex, dentate gyrus) Kinase, Ca2+ signalling (Burgin et al., 1990)Arc Dendrites (dentate gyrus first Cytoskeleton-associated protein

when stimulated, then cortex) (Lyford et al., 1995)Protein kinase C(γ) Dendrites(purkinje neurons) Kinase (Moriya and Tanaka, 1994)GAP43 Dendrites (cortex, hippocampus) PKC substrate (Chicurel et al., 1993)RC3/neurogranin Dendrites (cortex, hippocampus) PKC substrate (Chicurel et al., 1993)NMDA receptor I Dendrites Glutamate receptors

(Benson, 1997; Gazzaley et al., 1997)InsP3 receptor Dendrites (purkinje) Ca2+ signalling (Furuichi et al., 1993)Glycine receptor Dendrites (spinal cord) Glycine receptor (Racca et al., 1998)α-subunitBC1 Dendrites (hippocampus RNA polymerase III (Muslimov et al., 1998)

during development and in adultafter stimulation)

CREB Dendrites cAMP response element binding protein(cultured hippocampal neurons) (Crino et al., 1998)

BDNF Dendrites Neurotrophic factor (Tongiorgi et al., 1997)(cultured hippocampal neurons)

TrkB Dendrites BDNF receptor (Tongiorgi et al., 1997)(cultured hippocampal neurons)

β-actin Growth cone (cultured cortical Actin isoform (Bassell et al., 1998)neurons); dendrites (cerebellum) (Hannan et al., 1998)

Tropomyosin 5, 6 Axons Tropomyosin isoform (Hannan et al., 1998)Oxytocin Axons Neurotransmitter/Neuromodulator

(Hypothalamo-hypophyseal) (Jirikowski et al., 1990)Vasopressin Axons Neurotransmitter/Neuromodulator

(Hypothalamo-hypophyseal) (Lehmann et al., 1990)Prodynorphin Axons Neurotransmitter/Neuromodulator (

(Hypothalamo-hypophyseal) (Mohr et al., 1991)OMP and Axons (olfactory neurons) Odorant receptor (Wensley et al., 1995)Odorant receptorTAU Axons (cortex) Microtubule-associated proteins

(Litman et al., 1993)

m RNA localization mechanisms have been firstdiscovered in Drosophila or Xenopus, this re-view mainly focuses on mRNA localization insomatic cells. A list of localized mRNAs in so-matic cells is given in Table 1.

Many models have been proposed to explainhow the mRNAs are localized. Thus far, clearevidence exists only for mechanisms in whichthe mRNA itself is recognized by the machin-ery. In addition to the targeting via the cytosk-eleton (Fig. 1), as suggested for the majority oflocalized mRNAs, three other distinct mecha-

nisms have been identified (for review, see StJohnston, 1995). For example, RNA elementsmay act in local protection from RNA degrada-tion as in the case for Hsp83 mRNA duringDrosophila embryogenesis (Ding et al., 1993).Another mechanism is the anchoring to alreadylocalized components due to cytoplasmicstream. Finally, a vectorial nuclear exportmechanism has been proposed but the bio-logical relevance of this mechanism is still un-clear. For all of these mechanisms a localiza-tion signal in the RNA sequence is needed.

13

2.1.2. Localization tags

Localized RNAs contain cis-acting signals thattarget them to a specific domain within cells.Although a large number of localized mRNAshave been discovered, only in very few casesthe sequence of the localization signal has beenmapped. In most cases, with the exception ofnontranslated BC1 mRNA (Table 1) where thetag is in the 5´portion of the messenger, thesequence resides in the 3´UTR part of themessenger, in regions predicted to be rich insecondary structure elements. The recognitiontag is composed of multiple elements through-out a large region of the 3´UTR. For example,in oligodendrocytes, the cis-acting traffickingelements in the MBP mRNA have been iden-tified by deletion analysis (Ainger et al., 1993).The signal is composed of two distinct func-tional elements. One element, called ‘RNAlocalization signal‘(RLS), is required for lo-

calization of the mRNA to the myelin compart-ment. The postulated function of this elementis to anchor the MBP mRNA to the myelindomain. A second element, termed ´RNAtransport signal‘ (RTS), is required for thetransport along the processes. The RTS is alsopresent in MAP2A and MOBP (Gould et al.,1999) and therefore seems to represent ageneral localization signal used by differentmRNAs in different cell types. In chickenembryo fibroblasts, the cis-acting signal of theβ-actin mRNA has been mapped to two seg-ments of the 3´-untranslated part of the mes-senger. Indeed, a strong localization activityresides in a principal zipcode within the first54 nt of the 3´UTR part and a weaker one 43nt further downstream (Kislauskis et al., 1994).Interestingly, the zipcode sequences are highlyconserved in evolution, since the human β-actin mRNA shows 65 % homology in its first54 nt with that of the chicken (Kislauskis et al.,

Figure 1: mRNA trafficking in somatic cells. The cis-acting signals of the mRNA are recognized by RNA-bindingproteins in the nucleus (1), and the complex is exported from the nucleus. In the cytoplasm, this complex assemblesinto RNA granules, which may contain other RNA binding poteins and/or components of the translation machinery(2). These granules are then transported along cytoskeletal filaments (microtubules or actin) (3a) and the transportmay involve motor molecules (3b). Once at the final destination, the RNA may be translocated to actin filamentsprior translation (4). The various molecular components are not drawn to scale, and juxtaposition of particular com-ponents does not necessarily imply a direct interaction.

14

1994). Furthermore, the localization machineryof chicken fibroblasts can localize human β-actin mRNA, strengthening the idea of a highlyconserved mechanism (Kislauskis et al., 1994). It would be reasonable to think that RNAssharing the same destination would have a con-sensus sequence for localization. Thus far, noobvious homologies are apparent among dif-ferent RNAs. One possibility is that the pri-mary sequence itself is not the key for the lo-calization but the conformation that the tagtakes in a certain environment. No such com-parison has been possible so far and thereforethe code of mRNA localization remains locked.

2.1.3. RNA recognition

As mentioned above, in the majority ofexamples, the 3´UTR part of the mRNA allowstargeting of the messenger to its finaldestination. In several cases, the multiple stepsin the localization are coordinated by severalsequence elements. However, the recognitionstep is the most critical and the poorestunderstood. It has become apparent that the cis-actingsequence alone is not sufficient for RNA rec-ognition. Indeed, it seems that the RNA sec-ondary structures, such as stems, loops, bulges,tetraloops, double-stranded regions andpseudoknots, may be the critical part of therecognition. For example, in oligodendrocytes,various mutations affecting the predicted sec-ondary structure of the MBP mRNA tag blockthe localization of the mRNA (Ainger et al.,1997). It thus appears that understanding ofthe RNA localization requires a more detailedstructural mapping of the localization tags. In many cases, localized RNAs are trans-ported as large RNA/protein complexes (forreview, see Hazelrigg, 1998). The bestcharacterized RNA-binding protein, associatedwith the cytoskeleton, is the maternal proteinStaufen. Staufen has been implicated in thematernal mRNA localization, including bothanterior localization such as bicoid and poste-rior localization such as oskar mRNA in Dro-sophila (St Johnston et al., 1991). The proteinhas some characteristics of an RNA helicase,

although it is not clear what role Staufen playsin interaction with the cytoskeleton and withthe mRNA. However, it remains the only pro-tein to be found interacting between the cy-toskeleton and the mRNA. Interestingly, themammalian homologue of Staufen has beencloned recently and been found to localize tothe somatodendritic domain of cultured hip-pocampal neurons (Kiebler et al., 1999). It ispossible that in the neurons Staufen may havea similar role in the polarized transport andlocalization of mRNAs (Kiebler et al., 1999). An another interesting example of an RNA-binding protein is the zipcode-binding protein1 (ZPB-1, 60 kDa), which was isolated fromchicken fibroblasts (Ross et al., 1997). ZPB 1binds specifically to the 54 nt zipcode of β-actin mRNA. ZPB 1 contains five RNA-bind-ing domains: an RRM-type RNA-binding do-main and four KH domains. ZBP 1 colocalizeswith β-actin mRNA and its overexpressionincreases the localization of the β-actin mRNA(Ross et al., 1997). Interestingly, a Xenopushomolog of the chicken ZPB 1 has been puri-fied and shown to specifically bind to the Vg1 RNA cis-acting element (Deshler et al., 1998;Havin et al., 1998). These results suggest thatRNA localization in vertebrates may sharecommon features. Another ZBP protein, ZBP 2 (95 kDa), hasbeen shown to bind to the 54 nt zipcode of β-actin mRNA in neurons (Hazelrigg, 1998).ZBP 2 is also present in fibroblasts. In thesetwo cell types, β-actin mRNA associates withdifferent proteins, which in turn mediate its as-sociation with different cytoskeletal elements.It is noteworthy that ZPB 2 has also been foundin the nucleus. Recently, 6 proteins, which bind specificallyto the RTS tag of MBP mRNA have been pu-rified from rat brain extracts. The most abun-dant one has been identified as a heterogene-ous nuclear protein (hnRNP) A2 (de Vries etal., 1997). While this protein is primarily nu-clear, it has been shown to colocalize withMBP mRNA in the transport particles (de Vrieset al., 1997). This suggests that the first step inmRNA localization may occur in the nucleus,where the hnRNP A2 binds to MBP mRNA

15

and later on is transported to the cytoplasm. However, in addition to RNA-protein inter-actions in mRNA localization, RNA-RNA in-teractions may play a crucial role in the proc-ess. Indeed, it has been shown that the BC1RNA is present in the dendritic transport gran-ules (Muslimov et al., 1998). BC1 is a non-translated RNA polymerase III; therefore thefunction of its localization to these granulescan not be to target the protein. In Xenopus,the Xlsirts RNA localizes at the vegetal cor-tex and its targeting is necessary for the locali-zation of the Vg1 mRNA (Kloc and Etkin,1998). However, no direct evidence exists thatthis localized non-translated RNA participatesin the localization of other mRNAs.

2.1.4. The role of the cytoskeleton in mRNAlocalization

Active transport along the cytoskeleton is themost commonly suggested mechanism to ex-plain mRNA localization. All cytoskeletalcomponents including intermediate filaments,microtubules and microfilaments participate inmRNA localization. The cytoskeleton plays arole in both the transport and the anchoring ofthe mRNA at its final destination. Apart fromStaufen protein, nothing is known about howthe mRNA can be transported and anchoredon the cytoskeleton. Intermediate filaments may play a role in thelocalization of certain mRNAs in the center ofthe cell, as in the case of neurons andfibroblasts, or in the periphery as in the caseof the costamere mRNAs (Morris and Fulton,1994). Unfortunately, no direct evidence existsfor an interaction between these filaments andmRNAs. Microtubules have been associated with longdistance transport in the oocyte and in neu-rons (see review, Oleynikov and Singer, 1998).In these cells, disruption of the microtubules,using colcemid or taxol arrests the localiza-tion process. In neuronal cells, the localiza-tion of mRNAs to a distant area of the cell,like the growth cones, occurs mainly via themicrotubules and the actin filaments may there-after be used for anchoring prior to the trans-

lation. Transport of mRNAs along microtubulesrequires microtubule-dependent motors. Twomicrotubule-associated proteins, kinesin anddynein, have been implicated in the transportof organelles along the microtubules (forreview, see Hirokawa, 1998). Kinetic studies ofmRNA sorting for MBP are consistent withthe rate of kinesin transport (Barbarese et al.,1995), but no direct evidence exists for in-volvement of kinesin in the transport ofmRNAs. As mentioned above, microfilaments in theoocyte and in neuronal cells are involved inthe anchoring process, in the final destinationof the mRNA, rather than in the transport. How-ever, in non-neuronal somatic cells, microfila-ments are implicated both in transport and inanchoring. In situ hybridization using poly(A)mRNA, shows that the majority of the mRNAsare localized to actin-filament intersections,which contain ribosomes and the elongationfactor 1α (EF1α) in fibroblasts (Bassell et al.,1994a). More specifically, by using cytoskel-eton-disrupting drugs it has been shown thatthe β-actin mRNA is transported and anchoredby a microfilament-dependent mechanism(Sundell and Singer, 1991), although in neu-rons the transport of β-actin is dependent onmicrotubules (Bassell et al., 1998). The ten-dency of the β-actin mRNA to associate withmicrotubules in neurons may be due to thelong distance transport. In yeast, Ash1p mRNAlocalization also requires intact microfilaments(Takizawa et al., 1997). In contrast to the microtubule-mediated trans-port, microfilament-mediated transport is sug-gested to require actin-dependent motors (i.e.myosins). For example, Ash1p mRNA is incor-rectly localized in the absence of a myosin(Myo4p; Takizawa et al., 1997), but so far therehas been no direct demonstration that theAsh1p mRNA itself is transported by Myo4p.

2.1.5. Transport particles

RNA may be packaged into transport particles,that contain mRNA molecule(s) and transla-tional machinery. These particles may then be

16

transported along the cytoskeleton via inter-actions of the cis-acting signals, motor mol-ecules and/or other proteins. The first evidenceof such particles comes from the study of MBPmRNA in oligodendrocytes (Ainger et al.,1997). When MBP mRNA labelled with onefluorochrome was microinjected into livingcells, the mRNA accumulated in particles of ∼0.3 µm diameter closed to the microtubulebundles. These particles move at the rate of0.2 µm/s, which suggests that the transport maybe mediated by a kinesin-like motor protein. At present, these granules have only beenobserved at a light microscopic level and it istherefore impossible to affirm if these granulesare just an artefact. Only the analysis of theconstituents of the granules after biochemicalpurification would confirm the existence of spe-cific transport particles.

2.1.6. The cell environment

The regulation of mRNA localization can in-fluence cell morphology during developmentby providing a pool of protein in a specificregion. β-actin localization in fibroblasts is in-duced by serum and platelet-derived growthfactors (Latham et al., 1994; Hill et al., 1994),and this induction is required for maximal cellmotility rate (Kislauskis et al., 1997). Theseresults suggest a function for mRNA localiza-tion during motility of somatic cells. In neurons, a similar regulation could influ-ence morphology and neuronal activity. Re-cent evidence indicates that the neurotrophinNT-3 can promote the localization of dendriticmRNA granules (Knowles and Kosik, 1997;Zhan et al., 1999).

2.1.7. Translation of the localized mRNAs

The efficiency of protein synthesis benefitsfrom a close spatial association of all factorsinvolved in translation. As mentioned above,mRNAs are suggested to be transported alongthe cytoskeleton in granules. These transportparticles may contain not only the mRNA andRNA-binding proteins but also components ofthe translational machinery. In

oligodendrocytes, in situ hybridization com-bined with immunofluorescence revealed thatMBP mRNA, arginyl-tRNA synthetase, elon-gation factor 1α and ribosomal RNA arecolocalized in these granules (Barbarese et al.,1995). Using single dendrite transfection, it hasbeen shown that local protein synthesis occursin dendrites and also in growth cones of cul-tured hippocampal neurons (Crino andEberwine, 1996). This result points to localprotein synthesis in the dendrites. Is the translation occurring only when themRNA has reached is location? This seems tobe the case in Drosophila. Several mRNAslocalized in Drosophila oocyte are subject tolocalization-dependent translational control:only the mRNAs that have reached the finaldestination are translated (Wilson et al., 1996).Because translation, stability, association withthe cytoskeleton and localization, are all eventsmediated primarily through the mRNA´s3´UTR, it is reasonable to suggest that all theseprocesses may overlap in the mechanism anddepend on each other. In somatic cells, suchevidence is so far lacking.

2.1.8. Functions of mRNA localization

The function of mRNA localization is to gen-erate, in the majority of cases, a pool of a pro-tein to a specific region of the cell. LocalizedmRNAs encode all kinds of proteins, includingtranscription factors, cytoskeletal proteins andreceptor subunits (Table 1). How does the lo-cal synthesis of these proteins contribute to thecellular structure and function?

Cell polarity

Cell polarity is the ultimate consequence ofmechanisms that establish and maintain func-tionally specialized domains in the plasmamembrane and cytoplasm. In cells, polarity isinitiated by the formation of asymmetry at thecell surface by external and internal cues. In-deed, extracellular contacts between two cellsor between a cell and the extracellular matrixare sufficient to initiate the segregation ofmembrane and cytoskeletal proteins between

17

contacting and non-contacting surfaces. Asmentioned above, one way to create such poolof proteins within the cell is to localize mRNAs.The best evidence of the functional connec-tion between mRNA localization and the in-duction of cell polarity comes from the studyof β-actin mRNA in fibroblasts. When chickenfibroblasts are deprived of serum for 20 h, β-actin mRNA becomes delocalized and the cellshows a less polarized phenotype (Latham etal., 1994; Hill et al., 1994). If the cells are thenexposed to serum or chemotactic factors likePDGF or LPA, β-actin mRNA is relocalized tothe lamellipodia within a few minutes (Lathamet al., 1994). Moreover, treatment of the cellswith herbimycin, which is an inhibitor oftyrosine kinases, inhibits the localization of β-actin mRNA (Latham et al., 1994). These resultssuggest that extracellular factors operatingthrough a signal transduction pathway canregulate spatial sites of actin synthesis, whichmay in turn affect cell polarity. Theseexperiments do not, however, show whetherlocal transmembrane signalling would be ableto localize a mRNA or whether this could be anindirect result following e.g. cell spreading andprocess extension. The most convincing evidence demonstrat-ing that the β-actin mRNA plays a role in cellpolarity, comes from the artificialdelocalization of β-actin mRNA. As men-tioned above, the cis-acting signal of β-actinmRNA is a 54 nt sequence in the 3´UTR part.Using antisense oligonucleotides against thistag element, Kislauskis et al. showed that β-actin mRNA specifically becomes delocalized,without changing the amount of β-actin mRNAor the protein synthesis and actin integrity(Kislauskis et al., 1994). After 12 h in thepresence of the oligonucleotides, chickenfibroblasts no longer show a polarized pheno-type and the lamellipodia collapse (Kislauskiset al., 1994). Therefore, the localization of β-actin mRNA and the ensued local protein syn-thesis can provide a pool of newly synthe-sized actin monomers, which facilitate the as-sembly of actin filaments involved in cellpolarity. Thus, mRNA localization andpositional translation may act in concert with

protein transport to regulate microfilamentcomposition and consequently cell structure.

Cell migration

To migrate, cells must acquire a spatialasymmetry enabling them to activateintracellularly generated forces into cellmovement. One feature of this asymmetry is tocreate a polarized morphology between cellfront and rear. An important consequence ofpolarization is that the extension of activemembrane processes, including bothlamellipodia and filopodia, takes place primarilyaround the front of the cell. Conversion ofprotrusion into cell translocation across asurface requires the coordination of the cy-toskeleton, adhesion, and transmembrane sig-nalling (for review, see Lauffenburger andHorwitz, 1996; Mitchison and Cramer, 1996). The ability to generate and maintain thisfunctional asymmetry involves the enrichmentof actin filaments at the lamellipodia. β-actinmRNA is required for cell polarity infibroblasts (Kislauskis et al., 1994). To extendthese observations to cell motility, inhibitionof β-actin mRNA localization byoligonucleotides has been shown to result insignificantly reduced cell migration, asevaluated by time-lapse video-microscopy(Kislauskis et al., 1997). It is likely that β-actinmRNA augments cell motility by providing newactin monomers for polymerization at theleading edges. It is noteworthy that localizationof β-actin mRNA is independent of proteinsynthesis (Sundell and Singer, 1990). It is rea-sonable to propose that translation of localizedβ-actin mRNA is important to achieve maximaltranslocation (Kislauskis et al., 1997). There-fore, mRNA localization represents a spatialcomponent of gene expression in the cyto-plasm determining the area where a proteinshould be synthesized. This control of geneexpression appears to be an essential step incell polarity and consequent in cell migration.

Cell differentiation

In Saccharomyces cerevisiae, recent studies

18

have shown that mRNA localization plays arole to determine cell fate. The protein Ash1pis a cell fate determinant in yeast and localizespreferentially to the daughter cell (Bobola etal., 1996; Sil and Herskowitz, 1996). Ash1pinhibits mating-type switching by repressingHO endonuclease expression in the daughtercell (Bobola et al., 1996). Indeed, the Ash1mRNA has been found to localize to the distaltip of the daughter buds, suggesting that thetransport and localization of the mRNA mayasymmetrically establish bud formation(Takizawa et al., 1997).

Process outgrowth and synaptogenesis

In immature neurons in culture, mRNAs thatlocalize to the peripheral area of the cell arepresent in neurites and are concentrated ingrowth cones (Kleiman et al., 1994). Growthcones are the motile tips that guide neurons totheir targets during development of the nerv-ous system. The growth cones contain bundlesand networks of actin filaments. Recently β-actin mRNA, as well as protein, were detectedin neurites and growth cones in corticalneurons in culture (Bassell et al., 1998). Ul-trastructural analysis revealed polyribosomeswithin growth cones that colocalize withcytoskeletal elements (Bassell et al., 1994b).Therefore the local synthesis of β-actin pro-tein may increase G-actin concentration andfacilitate the actin polymerization withingrowth cones. By using differential display to detectmRNAs in growth cones, it has been shownthat there is an increase in the complexity ofmRNA population in association with progres-sive dendritic arborization (Crino andEberwine, 1996). Moreover, protein kinase Cand calmodulin I mRNAs have been found inthe dendrites during the period of activesynaptogenesis (Steward, 1995; Moriya andTanaka, 1994). Both proteins have been im-plicated in synaptogenesis. Therefore, it istempting to hypothesize that local translationin growth cones may contribute to the func-tional transition, from pathfinding to special-ization in postsynaptic terminals.

Synaptic plasticity

As in synaptogenesis, dendritic mRNAs maybe used to synthesize some postsynapticcomponents when synapses are remodeled inadults during synaptic plasticity. Synapticplasticity, which is reflected in long-lastingchanges in synaptic strength, is dependenton gene transcription and translation.However, it is still an enigma howtranscriptional regulation, which takes placein the nucleus, can selectively modify synapticsites in the dendrites. Is the localization ofmRNA and the consequent local proteinsynthesis the key mechanism? Ribosomes andpolysomes as well as the reticular structuresare detected throughout the dendritic arbor(Chicurel et al., 1993 Steward, 1995; Torre andSteward, 1996). Recently, it has been shownthat all elements required for protein synthesisand insertion of folded and glycosylatedtransmembrane proteins are present in thesynapses (Gardiol et al., 1999). Therefore,localized mRNA in the dendrites could be alsotranslated. Local protein synthesis in dendrites can beregulated by synaptic activity as shown by theactivation of glutamatergic synapses (Weilerand Greenough, 1993). Most importantly, re-cent studies show that mRNA translation canbe stimulated by neurotrophic factors likeBDNF and NT3 (Kang and Schuman, 1996 ).These results provide the first evidence forlocal protein synthesis in neuronal plasticity.However, direct evidence showing that a spe-cific mRNA is translated in the dendrites, isstill lacking. Although the presence of mRNAs indendrites is clear, it is not known whetherevery dendritic mRNA is functionallyimportant and what role each localized mRNAplays. For some mRNAs, the localization indendrites has been suggested to have an im-portant functional implication. For example,Arc mRNA is strongly induced after seizureor NMDA receptor-dependent synaptic stimu-lation (Lyford et al., 1995). CaMKIIα mRNAis also upregulated following the induction ofLTP (Thomas et al., 1994). However, the evi-

19

dence for a role of dendritic mRNAs insynaptic plasticity is still largely indirect andweak.

2.2. CELL MIGRATION

Cell migration is involved in a wide variety ofbiological phenomena during normal physiol-ogy, such as embryogenesis, as well as inpathology, like metastasis. Cell migration is amulti-step process, which is spatially and tem-porally coordinated (for review, seeLauffenburger and Horwitz, 1996). The migra-tory process starts from the extension of thelamellipodia and filopodia at the leading edgedue to extracellular stimuli (cell-to-cell contactsor cell-ECM contacts), followed by theclustering of integrins and attachment to theunderlying substrate. Translocation is finallyachieved by the cytoskeletal reorganizationand backward movement of adhering integrinsand the retraction of the trailing edge due tointegrin detachment from the substrate (Fig.2). The cytoskeleton plays an essential role inthe coordination of the migratory processes(for review, see Mitchison and Cramer, 1996).For example, extension of filopodia andlamellipodia after migratory stimuli requireslocal actin polymerization. The increase ofsites for actin polymerization is partially pro-vided by an increase of actin monomers. Asmentioned earlier, the localization of β-actinmRNA is likely to be one of the mechanismsby which the concentration of monomeric ac-tin increases at the leading edge (Kislauskis etal., 1997). It is important to note that β-actinmRNA localization is a spatially and tempo-rally regulated process (Latham et al., 1994).It is thus reasonable to assume that localizedmRNAs and consequently local protein syn-thesis may participate in the mechanisms ofcell migration, for example by maintaining thepolarized morphology. Therefore, the mecha-nisms, which regulate cell migration may alsoregulate the localization of cell motility-asso-ciated mRNAs. To migrate, a cell first receives extracellularstimuli from the extracellular matrix, from other

cells or from soluble growth factors in theextracellular environment. These signals arethen transmitted to the cells via t celllmembrane receptors. Many extracellular mol-ecules, like growth factors, cytokines and ECMcomponents are able to promote cell migrationthrough specific receptors.

2.2.1. Extracellular signals

Laminin, HB-GAM and other ECM compo-nents

The attachment of cells to the ECM is essen-tial for cell migration. The ECM is composedof a variety of molecules such as collagens,fibronectin, laminin, proteoglycans, tenascin,thrombospondin and many others (for review,see Mosher et al., 1992). Although the amountof the ECM varies, it is most abundant duringthe period of cell migration and cell differen-tiation. Cells use a number of different adhe-sion receptors to bind to the ECM, including afamily of cell surface proteoglycans calledsyndecans (see below). The most prominentECM adhesion receptors are the integrins, alarge family of heterodimeric transmembranesproteins with different α and β subunits (dis-cussed below).

Laminins are a family of multifunctional mac-romolecules, ubiquitous in basementmembrane. They represent the most abundantstructural noncollagenous glycoproteins ofhighly specialized extracellular matrices.Laminins are composed of various isoformsarising from different A, B1 and B2 chains(Skubitz et al., 1991; Aumailley and Smyth,1998). The expression of laminin chains isspatio-temporally regulated, suggesting thatlaminin isoforms might have different functions(Ryan et al., 1996). In addition to the role inbasement membrane assembly, lamininsenhance cell migration through binding tointegrin family receptors. The ability of tumorcells to migrate and invade other tissues is acharacteristic of poor prognosis. In glioma,laminin is one of the permissive substrates, towhich glioma cells attach and migrate, although

20

in normal adult brain laminin is largely absent(Uhm et al., 1999).

HB-GAM (Heparin-Binding Growth-Associ-ated Molecule; p18; pleiotrophin) is an extra-cellular matrix-associated protein rich in lysineand cysteine residues (Merenmies andRauvala, 1990). The sequence of HB-GAM ishighly conserved across species; more than 95% homology is found in the human, rat, bo-vine and chicken. HB-GAM shares about 50% homology with midkine protein involved inretinoic acid-induced cell differentiation(Matsubara et al., 1990; Tomomura et al., 1990).HB-GAM was initially isolated from neonatalrat brain as a heparin-binding and neurite out-growth-promoting protein (Rauvala, 1989;

Rauvala and Peng, 1997), but it is also presentin many non-neuronal tissues during embry-onic development. It is abundant in develop-ing bone, and has been therefore also termedas osteoblast specific factor-1 (OSF-1; (Tezukaet al., 1990). HB-GAM induces neurite outgrowth fromcentral and peripheral neurons (Rauvala, 1989;Nolo et al., 1996). In addition to the neuriteoutgrowth-promoting activity, HB-GAM isexpressed in the fiber tracts at the stage of rapidaxonal growth, but its expression declines rap-idly after axonal development (Rauvala et al.,1994). Therefore, HB-GAM has been sug-gested to be a guidance cue during develop-ment in the central and peripheral nervous sys-tem (Nolo et al., 1995; Kinnunen et al., 1998a;

Figure 2: Cell migration in 5 steps. Extension of the membrane lamellipodia or filopodia is due to extracellularstimuli (e.g. growth factors and extracellular matrix proteins) (1); clustering of integrins and adhesion to matrix (2);contraction of the cytoplasm resulting in the cell body translocation (3); release from contact sites (4); recycling ofmembrane receptors from the rear to the front (5).

21

Kinnunen et al., 1999). Additional functionshave been proposed for HB-GAM, like a rolein neuronal injury, development of nerve/muscle contacts and synaptic plasticity (Szabatand Rauvala, 1996; Lauri et al., 1998; Yeh et al.,1998). HB-GAM binds with high affinity to N-syndecan/syndecan 3 (Raulo et al., 1994). Theexpression patterns of HB-GAM and N-syndecan are very similar in the neonatal fore-brain (Nolo et al., 1995). In addition to N-syndecan, phosphacan (receptor type tyrosinephosphatase ζ /β), a chondroitin sulfateproteoglycan, has been shown to bind HB-GAM with high affinity (Maeda et al., 1996),and this interaction might be involved in neu-ronal migration (Maeda and Noda, 1998).

Growth factors and cytokines

Growth factors and cytokines are involved inmany biological processes, such as cell prolif-eration, survival, process outgrowth andsynaptogenesis. In addition, a number ofgrowth factors and cytokines can promote cellmigration through autocrine or paracrine path-ways, mainly through a family of cell surfacereceptors called receptor tyrosine kinases.Hepatocyte growth factor/scatter factor (HGF/SF) is a well-known growth factor that con-trols cell motility (for review, see Birchmeierand Gerhardi, 1998). Disruption of the HGF/SF gene results in migration defect of myo-genic precursors (Bladt et al., 1995). Migrat-ing myogenic progenitors are generated fromthe dermomyotome by an epithelial-mesenchy-mal transition. Ectopic application of HGF/SFin the chick embryo induces ectopic epithe-lial-mesenchymal transition and migration ofthe myogenic precursors (hence the term scat-ter). The receptor for HGF/SF is the tyrosinekinase encoded by the c-met protooncogene(Bottaro et al., 1991). β-actin localization in fibroblasts is inducedby platelet-derived growth factor (Latham etal., 1994; Hill et al., 1994) and this inductionis required for maximal rate of cell motility(Kislauskis et al., 1997). In neurons, theneurotrophin NT-3 can promote the localiza-

tion of dendritic mRNA granules and β-actinmRNA in growth cones (Knowles and Kosik,1997; Zhan et al., 1999).

2.2.2. Signal transduction

There are two major classes of receptors thatregulate cell locomotion: the integrin receptorsthat bind specific extracellular adhesion mol-ecules and growth factor receptors that bindtheir respective ligands. Therefore, it is notsurprising that there is cross-talk betweenintegrins and growth factor receptors in cellmotility regulation. In this way, each receptortype can either amplify or attenuate the oth-er’s signal and downstream responses.

Integrins

Integrins regulate cell functions, for examplecell migration, proliferation, and behavior inalmost all cell types (for review, see Schwartzet al., 1995). Local regulations of cytoplasmicprocesses, including alterations of the cytosk-eleton, secretion, and the control of adhesion,are mainly mediated by integrins. Integrins area large family of cell surface receptors, whichare dimeric proteins, containing an α-subunitand a β-subunit (Schwartz et al., 1995). Thefunction of integrins as transmembrane recep-tors depends on their ability to bind theirligands, for example extracellular matrix pro-teins like laminin and fibronectin, and the cou-pling of integrin cytoplasmic domains to thecytoskeleton via the interaction of proteins,including talin, vinculin and α-actinin, lead-ing to the induction of intracellular signals(Schwartz et al., 1995). During cell motility, sig-nalling through integrins is also important fordirected migration (Leavesley et al., 1993). Depending on the extracellular ligand,integrin clustering triggers activation of a num-ber of signalling pathways, including MAPkinases, PKC (protein kinase C), PTK (pro-tein tyrosine kinases) such as pp125 focal ad-hesion kinase and src, changes in intracellularpH and calcium, membrane potential, and lipidmetabolism (Schwartz et al., 1995). Togetherwith cytoskeletal proteins that associate di-

22

rectly or indirectly with integrin cytoplasmicdomains, these signalling events initiate cellspreading due to cell adhesion mediated byintegrins, leading to cell migration. However, cell migration velocity also de-pends on the breakdown of adhesion contacts,due to the loss of integrins at the rear of thecell. Integrins are either left behind on the sub-stratum (Chen, 1981), dispersed by diffusiontoward the cell front, or internalized by en-docytosis (Palecek et al., 1996). This pointsout to the necessity of efficient delivery ofadhesion molecules to the leading edges.

Syndecan family

The syndecans are transmembrane heparansulfate proteoglycans and represent a familyof 4 members (syndecan-1, -2, -3 and -4) (forreview, see Carey, 1997). Most cells express atleast one member of the syndecan family, butthe expression of each syndecan is specifi-cally regulated during tissue development(Carey, 1997). For example, syndecan-3 or N-syndecan is strongly expressed in the devel-oping nervous system (Carey, 1997).Syndecans bind to their extracellular ligandsvia their heparan sulfate chains. Many extra-cellular proteins, which bind to syndecans, alsointeract with other receptors at a higher de-gree of affinity and communicate through thisbinding with intracellular signalling pathways.Thus, syndecans have been suggested to func-tion as co-receptors rather than signalling re-ceptors. For example, integrin-mediated re-sponses can be modified by syndecan 4(Couchman and Woods, 1999). Recently, it has been shown that the cytoso-lic domain of N-syndecan binds a protein com-plex containing src-family tyrosine kinases(pp60-src and fyn) and the src-substratecortactin (Kinnunen et al., 1998b). Moreover,binding of its ligand, HB-GAM, to N-syndecanleads to a phosphorylation and activation ofsrc followed by cortactin phosphorylation(Kinnunen et al., 1998b). Phosphorylation ofcortactin regulates actin crosslinking, whichis required for cell migration (Huang et al.,1997).

Receptor tyrosine kinases

A large number of growth factors and cytokinesact through transmembrane receptors with in-trinsic protein tyrosine kinase activity (see re-view, Schlessinger and Ullrich, 1992). Signal-ling by tyrosine kinase receptors is mediatedby selective interactions between individualSrc homology 2 (SH2) domains of cytoplas-mic effectors and a specific phosphotyrosinein the activated receptor (see review, Marshall,1995). For example, c-met, the receptor ofHGF/SF, has two tyrosine residues, which arephosphorylated upon ligand binding and mu-tation of these two residues is sufficient to sup-press the receptor activity (Ponzetto et al.,1994). Thus, it appears that the tyrosine resi-dues are essential for recruiting signal trans-ducers and adaptors, which act downstream ofc-met. In cell migration, c-met activates thephosphoinositide 3-kinase (PI 3-kinase) as wellas ras-rac/rho pathways (Birchmeier andGherardi, 1998).

Rho-family proteins

The driving force for cell movement is pro-vided by the dynamic reorganization of theactin cytoskeleton, directing protrusion at thefront and retraction at the rear of the cell. Mem-bers of the rho family of ras-related smallGTPases serve as molecular switches, whichregulate actin polymerization to distinct struc-tures. The activity of Rho GTPases is deter-mined by the ratio of their GTP-bound andGDP-bound states and is regulated by the op-posing effects of guanine exchange factors(GEFs), which enhance the exchange of boundGDP for bound GTP, and the GTPase-activat-ing proteins (GAPs), which increase the in-trinsic rate of hydrolysis of bound GTP. Inaddition, guanine nucleotide dissociation in-hibitors can inhibit both the exchange ofnucleotides and the hydrolysis of bound GTP(for review, see Hall, 1998). In fibroblasts,the three best characterized members of therho family control the reorganization of ac-tin-based structures induced by growth fac-tors: formation of stress fibers and focal adhe-

23

sion contacts are governed by RhoA, rufflingand lamellipodia extension is regulated byRac1 and filopodia formation is controlled byCdc42 (Ridley and Hall, 1992; Nobes and Hall,1995). RhoA activation is downstream of Rac1and both Rac1 and RhoA are downstream ofCdc42. Therefore, the interplay between acti-vation of these GTPases generates specificGTPase cascades, which influence the actincytoskeleton and cell behavior (Chant andStowers, 1995). Overexpression of RhoA, Rac1and Cdc42 stimulates cell motility and inva-sion, while inhibition of their activity inhibitscell motility and invasion (Hall, 1998).

2.3. AMPHOTERIN/HMG-1

2.3.1. Characteristics of amphoterin

Amphoterin (p30) was isolated from perinatalrat brain as a heparin-binding protein, whichpromotes neurite outgrowth in brain neuronsin vitro by binding to the cell surface (Rauvalaand Pihlaskari, 1987). This protein has a di-polar structure (designated therefore asamphoterin), consisting of a 184-amino acidbasic region followed by a cluster of 30 acidicresidues in the carboxy terminal end(Merenmies et al., 1991). The dipolar struc-ture is the apparent reason for the formationof dimers and oligomers under physiologicalconditions (Rauvala and Pihlaskari, 1987). Theamphoterin sequence proved to be identical tothe sequence cloned for the high mobilitygroup 1 protein (HMG-1) (Bianchi et al.,1989). The designation “HMG” refers to non-histone components of chromatin (for review,see Bustin and Reeves, 1996). Since the cur-rent review mainly deals with extracellular in-teractions of the protein, the designation“amphoterin” will be used instead of “chro-mosomal HMG-1”. Amphoterin/HMG-1 is a ubiquitous, highlyconserved molecule. All eukaryotes possessa similar protein. However, the physiologicalfunction of amphoterin/HMG-1 and relatedevolutionary conserved proteins still remainsunclear (Baxevanis and Landsman, 1995).Many roles have been suggested, such as in

DNA replication, chromatin assembly and dis-assembly (Travers et al., 1994) or transcription(Ge and Roeder, 1994), but none of thesepotential functions have been confirmed un-equivocally. The protein contains two domainsthat are homologous with the DNA-bindingHMG box. A weak binding has been shown todouble-stranded linear DNA (Stros et al., 1994)and a high affinity binding to DNA containingsharp bends (Bianchi et al., 1989; Bianchi etal., 1992). However, it is currently thought thatHMG-1 binds only weakly to DNA, withoutsequence specificity, but that this interactionwould enhance the function of steroid hormonereceptors (Boonyaratanak-ornkit et al., 1998;Calogero et al., 1999). Generation of knockout mice shows nochanges in the chromatin organization in thenucleus (Calogero et al., 1999). Hmg1-/- miceare born alive, but the majority die within 24 hdue to hypoglycaemia (Calgero et al., 1999).Moreover, a small proportion of mutant micesurvive for 4 weeks after birth. These miceshow severe developmental problems, includ-ing size reduction and a generalized atrophyof internal organs (Calogero et al., 1999).However, the reasons for this generalized at-rophy remain unexplored. Amphoterin expression is developmentallyregulated. In rat brain the expression ofamphoterin is high during the embryonic pe-riod and decreases after birth (Rauvala andPihlaskari, 1987; Merenmies et al., 1991).Furthermore, amphoterin is abundant inoligodendrocytes during late embryonic devel-opment (Daston and Ratner, 1994). Therefore,the abundance of amphoterin seems to corre-late with an undifferentiated cell stage andearly maturation. Although the designation HMG-1 means anuclear component, the protein in cells hasbeen found to localize mainly to the cytoplasm(Bustin and Neihart, 1979). Amphoterin was,in many recent studies, detected in the cyto-plasm of the cell somas as well as in the proc-esses of many cells in vitro and in vivo, butlittle or no localization was observed in thenucleus (Rauvala et al., 1988; Merenmies et al.,1991; Daston and Ratner, 1991). These results

24

suggest that either nuclear and cytoplasmicamphoterin have a different structure or that amechanism bypassing the nuclear transporttakes place. Proteins that are closely related intheir sequence to amphoterin exist in cells(Parkkinen et al., 1993; Bianchi, 1998), buttheir subcellular distribution is currently un-clear.

2.3.2. Localization and interactions ofamphoterin in the extracellular space

Despite the lack of a secretion signal,amphoterin also shows extracellular localiza-tion. In rat neurons and in a neuroblastoma cellline, amphoterin is localized at the surface ofcell bodies and neurites in vitro (Rauvala etal., 1988) and is quickly secreted to the cul-ture medium from induced erythroleukemia(Melloni et al., 1995b). In vivo, amphoterin isfound to be associated with neurons and glialcells (Daston and Ratner, 1991; Nair et al.,1998) Different functions have been suggested foramphoterin in cells. Since antibodies againstamphoterin inhibit neurite initiation, it has beenproposed that amphoterin promotes neuriteoutgrowth in vivo (Rauvala et al., 1988;Merenmies et al., 1991). These results suggestan autocrine function for amphoterin in medi-ating neurite outgrowth. Amphoterin mightalso mediate neuron-Schwann cell interactionsthat are important in myelination (Daston andRatner, 1991). Furthermore, it has been shownthat amphoterin plays an essential role in pro-moting differentiation of murineerythroleukemia (Melloni et al., 1995a). It isnoteworthy that amphoterin binds to plasmino-gen and its activators (tissue plasminogen ac-tivator (t-PA) and urokinase-type plasminogenactivator (u-PA)), resulting in plasmin genera-tion and degradation of amphoterin (Parkkinenand Rauvala, 1991; Parkkinen et al., 1993).These results suggest that amphoterin-plas-minogen activator complex might contributeto penetration of the processes in tissue dur-ing development and regeneration after tissueinjury. The secretion mechanism of amphoterin is

not known. Amphoterin is released to the ex-tracellular space after cell stimulation, and thesecretion is not due to cell damage(Passalacqua et al., 1997; Passalacqua et al.,1998; Wang et al., 1999). Moreover, the ER/golgi pathway is not involved in the secretionof amphoterin, but intracellular increase ofCa2+ and protein kinase C may be required(Passalacqua et al., 1997). Recently,amphoterin has been shown to be secretedfrom macrophages after stimulation withcytokines, including TNF-α and IL-1 (Wanget al., 1999). Accumulation of high levels ofamphoterin in plasma has been suggested toact as a late mediator of the endotoxin shock(Wang et al., 1999).

2.3.3. Cell surface receptors of amphoterin

Syndecans and other proteoglycans

Several studies on the biochemical propertiesof amphoterin suggest that it can interact withthe cell surface and extracellular matrix viamultiple mechanisms. Since amphoterin canbe detached from cell surfaces by heparin,amphoterin might bind to heparan sulfateproteoglycan (Rauvala et al., 1988). Indeed,amphoterin binds syndecan 1 in a specificmanner and this binding requires the heparansulfate chains of syndecan 1 (Salmivirta et al.,1992). Recently, it has been shown thatamphoterin also binds to the chondroitinsulfate proteoglycan neurocan andphosphacan/receptor-type tyrosine phos-phatase ζ /β (Milev et al., 1998) and tosulfoglycolipids (Nair et al., 1998).

Receptor for advanced glycation end prod-ucts (RAGE)

RAGE is a member of the immunoglobulinsuperfamily and bears closest homology toneural cell adhesion molecule (NCAM)(Schmidt et al., 1992; Neeper et al., 1992).RAGE was originally isolated as the receptorfor advanced glycation end products (AGEs),which are nonenzymatic glycation productsof proteins and lipids that accumulate during

25

senescence and in diabetes (Brownlee et al.,1988). This interaction leads to vascular disor-ders by generating cellular oxidant stress(Wautier et al., 1996). RAGE has been alsoshown to interact with amyloid-β peptide (Aβ)(Yan et al., 1996). RAGE-Aβ interaction re-sults in oxidant stress in vitro and thereforeRAGE has been proposed to mediate the neu-rotoxic effects of Aβ in Alzheimer’s disease(Yan et al., 1996). In a recent study, RAGE hasbeen shown to bind EN-RAGE (extracellularnewly identified RAGE-binding protein), amember of the S100/calgranulin proteinsuperfamily, and this interaction is able to trig-ger cellular activation, with the generation ofproinflammatory mediators (Hofmann et al.,1999).Search of endogeneous ligands of RAGE fromtissue extracts has recently resulted in the iso-

lation of amphoterin (Hori et al., 1995). Fur-thermore, the idea of a specific interaction be-tween amphoterin and RAGE is supported bythe findings that anti-RAGE antibodies andsoluble RAGE inhibit neurite outgrowth fromcortical neurons on amphoterin-coatedsubstrates, and that amphoterin and RAGEcolocalize in developing rat brain (Hori et al.,1995). Since the other ligands are not presentduring development, it has been proposed thatamphoterin might act as the physiological lig-and for RAGE and that their interaction mightbe relevant for the development and the matu-ration of the central nervous system (Hori etal., 1996). Interestingly, amphoterin has alsobeen reported to compete with Aβ (Yan et al.,1996) and with EN-RAGE (Hoffman et al., 1999)for binding to RAGE.

26

3. AIMS OF THE PRESENT STUDY

Cell migration is a multi-step process, which requires the coordination between extracellu-lar and intracellular components. Considerable evidence indicates the involvement of twoheparin-binding proteins, HB-GAM and amphoterin in cell motility. However, many aspectsof their roles in cell migration remain unclear. For this purpose, the aim of this study wasspecifically to:

1) understand the role of HB-GAM outside of the nervous system, especially in the mesen-chymal tissue.

2) clarify the unexpected peripheral localization of amphoterin in motile cells.

3) investigate the role of amphoterin in the migration of immature and transformed cells.

4) analyze how the localization of motility-associated mRNAs is regulated and particularly,whether transmembrane signalling is involved in the regulation of the mRNA localization.

27

4. MATERIALS AND METHODS

Only materials and methods that have not beenpublished are described below. Other experi-mental procedures are listed in the table 2, andmore detailed descriptions can be found in theoriginal publications.

Double-in situ hybridization

For double-in situ hybridization, the cells wereincubated in the presence of the digoxigenin-labeled amphoterin and biotin-labeled β-actinprobes at 60oC overnight. After washing un-der stringent conditions, the cells were blockedfor 1 h at room temperature in 0.25% Tween-20 (Sigma, St Louis, MO)/5% goat serum(PAA Labor-und Forschungesellschaft mbH,Austria) in PBS. Both mRNAs were detectedindependently of each other. First, thedigoxigenin-labeled amphoterin probe wasvisualized using sheep anti-digoxigenin anti-bodies (Boehringer Mannheim, Mannheim,Germany) followed by affinity-purifiedfluorescein isothiocyanate (FITC)-conjugatedrabbit anti-sheep IgG (Janssen Biochemica,Belgium). Before the detection of β-actinmRNA, the cells were washed with 0.25%Tween-20 in PBS. The β-actin mRNA wasrevealed using mouse anti-biotin antibodies(Boehringer Mannheim) followed by affinity-purified tetramethyl rhodamine(TRITC)-conjugated goat anti-mouse IgG (Jackson

Table 2: List of methods used in the study (* experimental procedures not accomplished by myself )

Experimental procedure Described in

Cloning of β-actin cDNA from rat hippocampal library IIIn Situ hybridization and probes I, II, III and IVImmunocytochemistry and immuno-electron microscopy I, II, III, IV and VNorthern blotting I, IV and VWestern blotting I and IVIsolation of HB-GAM binding fraction* IDrug treatments I, III, IV and VBead assays and inhibition assays II and IVTransient transfections IV and VAntisense oligonucleotides IVNeurite outgrowth assays I, IV and VTransfilter migration assays I and IVNF-κB-luciferase assay* VA model of postarthritic periosteal ossification* IProduction of transgenic mice overexpressing HB-GAM* I

ImmunoResearch Laboratories Inc., WestGrove, PA). All antibody dilutions were pre-pared in 0.25% Tween-20/5% goat serum inPBS. The primary antibodies were incubatedat 4oC overnight and the secondary antibod-ies at room temperature for 1 h. A confocallaser fluorescence microscope (Biorad, Rich-mond, CA) was used to analyze the results.

Drug treatments

In order to analyze mRNA transport, N18 cellswere plated on laminin (10 µg/ml) in the pres-ence of cytochalasin D (1 µg/ml, Sigma, ) orcolcemid (5 µg/ml, Sigma) for different timeperiods. To study mRNA anchoring, the cellswere kept on laminin for 3 h to induce processgrowth, after which they were further culturedin the presence of cytochalasin D (0.5 µg/ml)or colcemid (5 µg/ml) for various time peri-ods. Control cells were cultured in parallelwithout drugs. The cells were then fixed with4% paraformaldehyde/0.05% glutaraldehyde inPBS and in situ hybridization was carried outeither with only one probe (eiher digoxigenin-labeled amphoterin or β-actin probe) or withtwo probes (digoxigenin-labeled amphoterinand biotin-labeled β-actin probe) as describedabove. After the color detection, the resultswere analyzed under a light microscope (AX-70, Olympus Optical Co Ltd., Japan) and thecells were counted using a 20x objective.

28

5. RESULTS

5.1. ROLE OF HB-GAM/N-SYNDECANINTERACTION IN CELL MOTILITY

5.1.1. Enhancement of osteoblast recruit-ment and bone formation by HB-GAM (I)

Although it appears clear that HB-GAM,through its receptor N-syndecan, enhancesmigratory response in neurons, the possiblerole of HB-GAM/N-syndecan interaction out-side the brain has not been addressed. For thispurpose, HB-GAM expression was studiedduring bone formation. HB-GAM is highlyexpressed in the cell matrices that act as targetsubstrates for osteoblast migration and boneformation (I: Fig. 1 and Fig. 2). Moreover, N-syndecan is expressed by osteoblasts/osteob-last precursors, which show a motile pheno-type (I: Fig. 1). These results suggest that os-teoblast recruitment may be mediated throughHB-GAM/N-syndecan interaction. Osteoblast-type cells that express N-syndecan were there-fore studied in a haptotactic cell migration as-say, using HB-GAM as a cue. All osteoblastcell types used in this study migrated rapidlyto HB-GAM, whereas N18 and 3T3 cells,which do not express N-syndecan, did not re-spond to HB-GAM (I: Fig. 3). In addition, in-hibitors of tyrosine kinases, as well as solu-ble HB-GAM and soluble N-syndecan addedinto the medium, significantly inhibited themigratory response to HB-GAM. These re-sults suggest that HB-GAM/N-syndecan in-teraction enhances osteoblast migration in asimilar manner previously described for HB-GAM-induced migratory response of neurons(Kinnunen et al., 1998b). To further study the role of HB-GAM/N-syndecan interaction in bone formation, twodifferent models were used (I). In the adjuvant-induced injury model, HB-GAM and N-syndecan expression was upregulated in theperiosteum during the bone regeneration andthen disappeared after bone mineralization (I:Fig. 5). In the HB-GAM transgenic model, themaintenance of HB-GAM expression in peri-osteum, resulted in an increased bone thick

ness (I: Fig. 8).

5.1.2. Regulation of mRNA localization byHB-GAM/N-syndecan signalling pathway(II and unpublished data)

β-actin mRNA localization and the ensuinglocal protein synthesis in the leading edges infibroblasts augment cell motility (Kislauskiset al., 1997). Although the mechanism ofmRNA localization is intensively examined,studies demonstrating the possible role of ex-tracellular cues in the mRNA targeting has notbeen analyzed. It is reasonable to assume thattransmembrane signalling participating in cellmotility may also regulate the localization ofmRNAs such as β-actin mRNA. For this pur-pose, HB-GAM/N-syndecan signalling path-way, which enhances migratory responses incells such as C6 glioma cells and osteoblastscell lines, was studied. When locally appliedto C6 (II: Fig. 2) or osteoblast cell line (un-published data), HB-GAM is able to localizeβ-actin mRNA by a microtubule andmicrofilament-dependent mechanism (II: Fig.2). In contrast, HB-GAM-coated beads haveno effect on the localization of N-syndecanmRNA (II: Fig. 2C). Inhibitors of src tyrosinekinase, as well as soluble HB-GAM and solu-ble N-syndecan, significantly decreased β-ac-tin mRNA localization induced by HB-GAM-coated beads (II: Fig. 4 and 5). Moreover, anti-N-syndecan antibodies coated on microbeadswere able to induce specifically the localiza-tion of β-actin mRNA (II: Fig. 6).

5.2. SUBCELLULAR LOCALIZATIONOF AMPHOTERIN mRNA AND PRO-TEIN

5.2.1. Localization of amphoterin mRNA(II, III; IV; unpublished data)

Amphoterin has been shown to localize periph-erally when cells start to grow processes(Parkinnen et al., 1993). In order to study themechanisms of the peripheral localization ofamphoterin, several transformed cell linesgrown on laminin were subjected to in situ

29

hybridization. Amphoterin mRNA is localizedto the cell processes in rat glioma C6 (III: Fig.2; IV: Fig. 1A), rat osteoblasts UMR-106 (notshown), mouse melanoma B16 (not shown),mouse neuroblastoma N18 and human fibro-sarcoma HT1080 cells (not shown).Ultrastructural in situ hybridization in C6 cellsconfirmed the peripheral localization ofamphoterin mRNA (III: Fig. 4C). Moreover, localpresentation of laminin via microbeads wasefficient in localizing amphoterin mRNA as themajority of the mRNA signal was found aroundthe beads, irrespectively of the position of the

beads on the surface of the cell soma or of thegrowing processes (IV: Fig. 1D; unpublisheddata: Fig. 6).

5.2.2. Comparison of the localization ofamphoterin mRNA and protein (IV; unpub-lished data)

To compare the protein and the mRNA distri-butions, in situ hybridization and immunocy-tochemistry were performed on C6 cells grownon laminin. The localization of the mRNA andprotein resemble each other (IV: fig. 1A).

Figure 3: Amphoterin and β-actin mRNAs colocalize at the tips of membrane processes in N18 cells and REF cells (B).N18 cells were grown on laminin (10 µg/ml) for 3 hours. REF cells were obtained from sixteen-day-old rat embryos andcultured for two days in DMEM medium supplemented with 10% fetal calf serum. Double-in situ hybridization wascarried out using a digoxigenin-labeled amphoterin and a biotin-labeled β-actin probe. The cells were stained with asheep anti-digoxigenin antibody, followed by a FITC-conjugated rabbit anti-sheep IgG to reveal the amphoterin mRNAand a mouse anti-biotin antibody, followed by TRITC-conjugated goat anti-mouse IgG to detect the β-actin mRNA. Thefluorescence micrographs were obtained using a BioRad MRC-1024 confocal microscope. (A) shows the amphoterin(green) and β-actin (red) mRNA in control N18 cells and (B) in REF cells. (C) shows the effect of cytochalasin D (0.5 µg/ml, 1 hours) and (D) the effect of colcemid (0.5 µg/ml, 1 hours) on the anchoring of the messengers in N18 cells. Bars: 50µm (A-D).

30

Figure 4: Amphoterin and β-actin mRNAs localize as separate spots in N18 (A) and C6 (B) cells. The mRNAs werevisualized in cells grown on laminin (10 µg/ml) for 3h using three-dimensional digital imaging. Amphoterin (red)and β-actin (green) mRNA spots that localize at the tip of the processes are distinct from each other (arrowheads).Bar: 50 µm.

Moreover, the same proportion of laminin-coated beads induced the localization ofamphoterin mRNA and protein, suggesting thesame regulatory mechanism (IV: Fig. 1C andD). Numerous polyribosomes were detectedin the processes of C6 cells (IV: Fig. 1B).Moreover, puromycin or cycloheximide, in-hibitors of translation, did not have any effecton the localization of the mRNA (IV). Thissuggests strongly that translation and locali-zation of amphoterin mRNA occur independ-ently. Therefore, it appears reasonable that thelocal mRNA is translated.

5.3. COMPARISON OF AMPHOTERINAND βββββ-ACTIN mRNA LOCALIZATION

5.3.1. βββββ-actin and amphoterin mRNA co-localize in the cell processes (III; unpub-lished data).

Studies using chicken embryo fibroblasts haveshown that the β-actin mRNA localizes closeto the leading edges in migrating cells(Lawrence and Singer, 1986; Kislauskis et al.1993). Since the distribution of amphoterinmRNA in the cell types studied resembles theleading edge distribution found for β-actinmRNA, double in situ hybridization was car-ried out in different cell types. The β-actin and

amphoterin mRNAs co-localize in the cellprocesses of N18 (Fig. 3 A), rat embryonicfibroblasts (Fig. 3 B) and C6 cells (III, Fig. 1,2 and 4). Co-localization of the β-actin andamphoterin mRNAs in the cell processes wasfound in the majority of cells (Fig. 3 A-E).

5.3.2. ΑΑΑΑΑmphoterin and βββββ-actin mRNA local-ize to distinct granule-like structures in thesame subcellular compartments (unpublisheddata)

Particles functioning as mRNA transport as-semblies have been described (Ainger et al.,1993). Such particles may contain all the com-ponents needed for the transport, anchoringand translation of the messenger. However, itis unclear whether one transport particle couldcontain more than one mRNA specie.Amphoterin and β-actin mRNA were detectedin C6 and N18 cells by enhancing the sensi-tivity of the detection with a tyramide com-plex and the signals were analyzed by confocalmicroscopy. Colocalization of the signals (in-dicated by yellow spots) was only observed invery few particles, suggesting that β-actin andamphoterin mRNAs are sorted in differentgranule-like structures in the same subcellularcompartments (Fig. 4, arrowheads).Ribosomes were also found in these granules,

31

by immunostaining with the antibodies againstthe 60S ribosomal subunit (not shown).

5.3.3. Amphoterin and β-actin mRNA areanchored and transported to the neuritesby a microfilament-dependent mechanism(III; unpublished data)

Previous studies have shown that the β-actinmRNA is localized to the leading edges by amicrofilament-dependent mechanism (Sundelland Singer, 1991). In order to compare themechanism localizing amphoterin mRNA to

that localizing β-actin mRNA, the effects ofcytoskeleton-disrupting drugs on the distri-bution of these mRNAs were studied. To study the mechanism of mRNA anchor-ing, N18 cells grown on laminin were exposedfor different time periods to colcemid (5 µg/ml), which depolymerizes microtubules andinduces collapse of the intermediate filaments.Fig. 5 E shows that approximately the sameproportion (60%) of the cells contain periph-erally localized β-actin and amphoterin mRNAand that this proportion is not essentiallychanged, even after 120 min incubation in

Figure 5: Time courses and quantification of the effects of colcemid and cytochalasin D on anchoring (A and B) andtransport (C and D) of amphoterin and β-actin mRNAs to cell periphery. (A) shows the effect of colcemid (5 µg/ml)on cells that were grown on laminin (10 µg/ml) for 4 hours and then treated with the drug for various time periods.The mRNAs were localized using digoxigenin-labeled probes. (B) shows the corresponding analysis of the effect ofcytochalasin D (0.5 µg/ml). To study the transport of the β-actin (C) and amphoterin (D) mRNAs, cells were platedon laminin (10 µg/ml) and grown in the presence of colcemid (5 µg/ml), cytochalasin D (0.1 µg/ml) or without anydrugs. Every time point is an average of two independent experiments in which 2000 cells were counted (the cellcounts at each time point were within the range of 15% (A), 7% (B), 12% (C) and 5% (D)).

32

colcemid for either mRNA species. Thecolcemid treatment, however, changes the pat-tern of peripheral localization since mostneurites retract during the drug treatment. In-terestingly, both mRNAs still co-localize at thecell periphery and are found in patches closeto the plasma membrane, as shown by double-in situ hybridization in Fig. 3 D. In contrast to colcemid, cytochalasin D (0.5µg/ml), that disrupts microfilaments, induceda rapid displacement of both amphoterin andβ-actin mRNA from the neurites to the cellsoma. Fig. 5 F shows that both mRNAs weredisplaced from the neurites at a similar rate.Double-in situ hybridization (Fig. 3 C) showedthat in cytochalasin D-treated cells, β-actin andamphoterin mRNA co-localize in a diffusepattern in the soma of cells that have retainedmost of their neurites during the drug treat-ment (Fig. 3 C; arrows). The collapse of thedifferent cytoskeletal components was con-firmed by immunostaining (not shown). To examine the effect of the cytoskeleton-disrupting drugs on mRNA transport, the N18cells were plated on laminin in the absence andpresence of the drugs and grown for differenttime periods. The rate of mRNA transport tothe cell periphery in the absence of the drugswas found to be approximately the same forβ-actin (Fig. 5 G) and for amphoterin (Fig. 5H). In the presence of colcemid, the cells wereable to spread and to transport both β-actinmRNA (Fig. 5 G) and amphoterin mRNA (Fig.5 H) to the cell periphery. Furthermore, thekinetics of mRNA transport was similar to thatobserved for untreated cells (Fig. 5 G-H). Incontrast, low doses of cytochalasin D (0.1 µg/ml) inhibited laminin-induced morphologicalchanges and transport of the RNAs to the cellperiphery (Fig. 5 G-H). The same results wereobtained in C6 cells, when grown on laminin(not shown). Association of the β-actin mRNA to the ac-tin filaments was confirmed by electron mi-croscopy (III: Fig.6). The detection ofamphoterin mRNA at the electron microscopiclevel was difficult without a long tyramideamplification, which gives poor ultrastructuralpreservation (III).

5.3.4. Locally applied ECM proteins induceamphoterin and β-actin mRNA localization ina very similar manner (II and unpublisheddata)

To compare the regulation of amphoterin andβ-actin mRNA localization, laminin-coatedbeads were applied to C6 cells for 15 min. Asshown in Fig. 6, local presentation of lamininwas very efficient in localizing amphoterin andβ-actin mRNA as the majority of the mRNAsignal was concentrated around the beads,irrespectively of the position of the beads onthe surface of the cell soma or of the growingprocesses. Moreover, the same proportion ofbeads induced the localization of both mRNAs,suggesting a similar regulatory mechanism (Fig.6 B). Both amphoterin and β-actin proteins arealso localized around the beads (not shown),supporting the local mode of translation.Furthermore, cytoskeleton-disrupting drugswere incubated with the cells when beads wereapplied. The disruption of actin filaments usingcytochalasin D (1 µg/ml) inhibited significantlythe laminin-induced localization of amphoterinand β-actin mRNA at a very similar rate,whereas no significant effect was observedwith colcemid (Fig. 6 B). HB-GAM, through its receptor N-syndecan,can regulate the localization of β-actin mRNAas well as the localization of amphoterin mRNA(not shown). Interestingly, in the case of HB-GAM, the localization of mRNA is dependenton intact microtubules and microfilaments,suggesting that depending on thetransmembrane signalling, mRNA might belocalized through a different cytoskeletal com-ponent (II). Recently, tubulin was found in theprotein complex associating specifically withN-syndecan (Kinnunen et al., 1998a), suggest-ing that tubulin might stabilize the interactionof N-syndecan with HB-GAM.

5.4. ROLE OF AMPHOTERIN IN CELLMOTILITY

5.4.1. Export of amphoterin to the extracellu-lar space from cells extending processes (IV)

33

To investigate the distribution of amphoterinin the medium and in cells, N18 cells were usedbecause they display a motile phenotype onlaminin-coated substrate but do not extendprocesses on some other matrix proteins, likeHB-GAM (for review, see Rauvala and Peng,1997), despite of their strong adhesion. Inter-estingly, amphoterin is exported to the extra-cellular space from process-extending cells,whereas only cell binding to the matrix doesnot significantly enhance extracellular locali-zation (IV: Fig. 2A). No changes in the cellularlevel of amphoterin were observed in the ex-periments (IV: Fig. 2A). Although amphoterin and β-actin revealed asimilar staining at the cell periphery whenstudied at the light microscopic level, a cleardifference was found in the localization at theplasma membrane; both in living (IV: Fig. 3A)and in permeabilized cells (IV: Fig. 3B).Amphoterin was stained in most cells asextracellular patches, which do not contain β-actin, at the plasma membrane, suggesting thatthe release of amphoterin is not due tononspecific cell damage (IV: Fig. 3A and B).Occurrence of amphoterin at the cell surfacewas also clearly visualized by immunoscanningelectron microscopy (Fig. 3D). Doubleimmunostaining of amphoterin and PDI(protein disulfide isomerase), which is a solubleER-lumen enzyme, showed no colocalization,suggesting that amphoterin does not enter theclassical secretory pathway (IV).

5.4.2. Inhibition of amphoterin decreasescell migration (IV)

Amphoterin mRNA is strongly expressed inmigrating cells but is downregulated duringcell-to-cell contact formation (IV: Fig. 6).Alsothe localization of amphoterin to the processesof migrating cells suggests a role in cell motil-ity. To investigate the possible role ofamphoterin in cell migration, specific decreaseof amphoterin mRNA and protein wasachieved by using C-5 propynylpyrimidine-2´-deoxyphosphorothioate-modified amphoterinantisense ODNs transfected into cells (IV: Fig.4A and B).

A morphological effect was observed in C6glioma cells, into which the inhibitoryoligonucleotides (see above) had beentransfected. Most ODN-transfected cellsshowed a round morphology, whereas nearlyall cells (90-100 %) extended long cytoplasmicprocesses when grown for 6 h on laminin (V).A decrease to 40% of the controls (non-transfected cells, or cells transfected with senseor nonsense ODNs) was observed in theproportion of process-extending cells, whenamphoterin anti-sense ODNs were transfectedinto C6 cells. Since these results suggested that a decreaseof amphoterin mRNA affects the migratoryresponse of cells, a haptotactic transfilter as-say was used. Decrease of amphoterin expres-sion using antisense ODNs inhibited C6 cellmigration to laminin to about 40% of control

Figure 6: Amphoterin (arrowheads in A) and β-actinmRNA localize around laminin-coated beads in a mi-crofilament-dependent manner. The cells were incubatedwith the beads for 15 minutes. Almost 40% of the laminin-coated beads are able to localize amphoterin as well asβ-actin mRNA (B). B shows the effects of cytochalsin D(1 µg/ml) and colcemid (5 µg/ml). Only the disruptionof actin filaments has a significant effect on the mRNA-localizing effect of laminin-coated beads. In B, each valuerepresents the mean ± S.D. calculated from 12 randomlyselected microscopy fields in 3 independent experiments.***P<0.001 (Anova test). Bars: 50µm (A).

34

values (IV: Fig 4C and D). The inhibitory effecton cell migration correlates to the reduction ofamphoterin mRNA and protein; for example,no effect on cell migration was observed forthe ODNs used in the culture medium (IV: Fig.4D). Moreover, anti-amphoterin antibodiesinhibited cell migration to laminin, stronglysuggesting that amphoterin acts extracellularlyduring cell migration (IV: Fig. 5). A similarmigration inhibition was observed in other celltypes tested (N18 and HT1080) and by usingHB-GAM as a cue (unpublished data: notshown).

5.4.3. Amphoterin mediates neurite outgrowththrough RAGE (V)

RAGE, a member of the immunoglobulinsuperfamily, has been suggested to mediate theneurite outgrowth effect of amphoterin in vitro(Hori et al., 1995). Moreover, RAGE has beenshown to interact with AGEs and amyloid β−peptide in cell injury mechanisms involvingcellular stress and activation of the transcrip-tion factor NF-κB. However, it was unclearwhether RAGE acts as a signalling receptor ina manner that requires the cytoplasmic domainof the receptor. To investigate this, transfectionof full-length RAGE or a deletion mutant lack-ing the cytoplasmic domain was performed inN18 cells grown on amphoterin. Transfectionof the full-length RAGE was sufficient to spe-cifically induce filopodia and neurites onamphoterin substrate and to enhance NF-κB-dependent transcription, whereas cells ex-pressing the deletion mutant or the mockcontrol did not respond to amphoterin (V: Fig.1 and 4A). In addition, RAGE-mediated neuriteoutgrowth on amphoterin was shown todepend on rac and cdc42 but not rho or ras-MAP kinase pathway (V: Fig. 2 and 3). Incontrast, the activation of NF-κB by RAGErequired ras but not the rho family GTPase (V:Fig. 4B), suggesting RAGE uses differentsignalling pathways to induce neuriteoutgrowth and to regulate gene expressionthrough NF-κB.

6. DISCUSSION AND CONCLU-SIONS

6.1. HB-GAM IN CELL MIGRATION

In the nervous system, the roles for HB-GAMin development of neuron-target contacts andin regulation of LTP have been previously at-tributed. However, the function of HB-GAMoutside of the brain has been poorly under-stood. Because HB-GAM (or OSF-1) was iso-lated from murine osteosarcoma cell line(Tezuda et al., 1990), its possible function inbone tissue was examined. During bone de-velopment, HB-GAM is strongly expressed inthe cartilage matrix, onto which osteoblast pre-cursors are recruited. Interestingly, N-syndecanis found in cells, which show a motile pheno-type and were identified as the motile precur-sors of osteoblasts (I). In addition, osteoblastcell lines, expressing abundantly N-syndecan,migrated to HB-GAM and this migration wasinhibited by soluble N-syndecan, soluble HB-GAM and by inhibitors of the src-family tyro-sine kinases (I). According to these results,HB-GAM/N-syndecan interaction appears tomediate osteoblast migration through the samesignalling mechanism as in the migratory re-sponse of neurons (Kinnunen et al., 1998b).Furthermore, the upregulation of HB-GAMand N-syndecan after bone injury and the en-hancement of bone formation in HB-GAMtransgenic mice strongly support the idea thatosteoblast recruitment would be mediated byHB-GAM (I). In conclusion, we propose a hypotheticalmodel in which the deposition of HB-GAM,for example on the bone surface, induces therecruitment of the osteoblast precursors to thesites of bone formation. Furthermore, theosteoblasts differentiate into fully functional,osteoid-producing osteoblasts due to stimula-tion stimulation by diffusible growth factors,and finally new bone is formed. HB-GAM, likein the case of growth cone migration along theneuronal pathway (Kinnunen et al., 1998a),might be a cue to target the osteoblast to thesite of bone formation.

35

6.2. SORTING OF AMPHOTERIN mRNAAND PROTEIN TO CELL PROCESSES

6.2.1. Amphoterin mRNA: a new exampleof localized mRNA

Amphoterin is a highly adhesive protein thatis enriched in the processes of immature andmalignant cells (Rauvala and Pihlaskari, 1987;Parkkinen et al., 1993). The structural charac-teristics and the peripheral localization ofamphoterin are, however, controversial. During the last few years the idea hasemerged that mRNA transport to specific lo-cations in the cell may be an important mecha-nism in molecular sorting. The localization ofa mRNA would then allow the protein in ques-tion to be synthesized close to the presumedsite of function. One advantage of sorting atthe mRNA level may be to avoid sorting of acytosolic protein that binds to various biologi-cal membranes and might therefore disturbnormal cellular functions. This may be the ra-tionale for sorting of myelin basic protein atthe mRNA level to the processes ofoligodendrocytes (Colman et al. 1982; Aingeret al. 1993). About one third of the amino acid residuesin the amphoterin sequence are cationic (lysineor arginine), and the protein also contains acharacteristic polyanionic tail (Merenmies etal., 1991). These structural features appar-ently explain, at least in part, the highly adhe-sive nature of amphoterin (Rauvala andPihlaskari, 1987). These characteristics raisethe question as to how this abundant cytosolicprotein can be sorted to cellular processes, aphenomenon observed both in vitro (Rauvalaet al., 1988; Parkkinen et al., 1993) and invivo (Daston and Ratner, 1991 and 1994; Horiet al., 1995; Milev et al., 1998). The present results (II, III and IV) demon-strate that amphoterin mRNA provides a strik-ing example of messenger localization toneurites and other forms of cytoplasmic pro-cesses. It therefore appears that, as in the caseof myelin basic protein, the sorting problemof amphoterin to cellular processes has beenresolved by sorting its mRNA.

6.2.2. Is amphoterin locally translated?

The localization of amphoterin protein andmRNA correspond to each other (see IV: Fig.1A). Furthermore, the localization of theamphoterin protein is tightly connected to thelocalization of the mRNA, as shown by ex-periments with the laminin-coated beads (IV:Fig. 1C and D). These results suggest that themRNA is locally translated. The occurrenceof ribosomes in the processes of the cells iscompatible with a local mode of translation(IV). Since amphoterin is the major component inthe HMG1 of tissues and cells (Merenmies etal., 1991; Parkkinen et al., 1993) and containspolycationic HMG boxes suggested to medi-ate DNA binding (Landsman and Bustin,1993), it may have a functional role in the cellnucleus. Since injection of labelled HMG1 intothe cell soma results in rapid nuclear uptake,HMG1 and related proteins are suggested tobe effectively transported from the cytosol intothe nucleus (Rechsteiner and Kuehl, 1979; Wuet al., 1981; Landsman and Bustin, 1993). Pre-vious studies on the distribution of the endog-enous protein in vitro (Rauvala et al., 1988;Merenmies et al., 1991) and in vivo (Dastonand Ratner, 1991; Hori et al., 1995; Milev etal., 1998; Nair et al., 1998) have, however,indicated a peripheral, non-nuclear distribu-tion. This suggests that the endogenous sort-ing system is somehow able to bypass the trans-port mechanism that conveys polycationic pro-teins into the nuclei. The present results there-fore suggest that this bypass occurs at themRNA level, and that the protein remains closeto its site of synthesis in the cell periphery.

6.2.3. Is the local translation of amphoterinthe key of its export to the extracellularspace?

Amphoterin has been implicated in severalphenomena like neurite outgrowth, neuron-gliainteractions, in early phases of cell differen-tiation and in endotoxemia (Rauvala andPihlaskari, 1987; Rauvala et al., 1988; Dastonand Ratner, 1991; Hori et al., 1995; Melloni

36

et al., 1995a; Wang et al., 1999). Sinceamphoterin lacks a classic-type secretion sig-nal, the mechanism of its extracellular localiza-tion is not well understood. In the presentstudy, amphoterin was detected as patches atthe cell surface, that do not contain β-actin(IV). Furthermore, accumulation of amphoterinto the extracellular space accompanies processextension due to laminin but is much lower incells that only bind to the cell matrix withoutextending processes. Interestingly, the extra-cellular export during process extension corre-lates with the laminin-induced localization ofthe mRNA. Unfortunately, extracellularamphoterin has not been observed around thebeads due to the bead invagination into thecells after 30 min incubation. However, westernblotting showed that the level of amphoterinin the cell lysate does not change in cells grownon laminin or HB-GAM; only the extracellularlevel of the protein is dependent on matrixinteractions that enhance process outgrowth.Taken together, we speculate that the laminin-induced localization of the mRNA and theconsequent local translation of amphoterin fa-cilitate the extracellular export. Monoclonal anti-amphoterin antibodies havebeen used to show an effective transport to thecell surface under conditions that do not in-volve cell damage, and appear to depend onprotein kinase C and influx of calcium ionsinto the cell (Sparatore et al. 1996; Passalacquaet al. 1997; Passalacqua et al. 1998). Very re-cently, cytokines such as TNF and IL-1 wereshown to induce extracellular export ofamphoterin both in vitro and in vivo (Wang etal., 1999). As shown in the present study (IV),cell-matrix contact is also able to enhance ex-tracellular export of amphoterin. It is conceiv-able that local synthesis of amphoterin closeto the plasma membrane would facilitate ex-tracellular export. It would be interesting tolearn, whether mRNAs of other proteins lack-ing a secretion signal but having apparent ex-tracellular roles would also be targeted closeto the plasma membrane.

6.3. AMPHOTERIN IN CELL MIGRA-TION

6.3.1. Amphoterin is required for cell mo-tility

Amphoterin is a ubiquitous protein, but its ex-pression levels are downregulated during celland tissue differentiation (Rauvala andPihlaskari, 1987; Parkkinen et al., 1993). Theexpression level of amphoterin is very high intransformed cell types as compared to theirnon-transformed counterparts (Parkkinen et al.,1993). Therefore, the high expression ofamphoterin in malignant cells may be an im-portant factor in the regulation of cell motilityduring tumor invasion. For this purpose, theeffects of a specific decrease in the level ofamphoterin expression in cells by using a re-cently described modification ofoligonucleotides (Wagner et al., 1993) wastested in cell migration assay. When transfectedinto several cell types, the antisenseoligonucleotides reduce the amount ofamphoterin mRNA to a similar level that isobserved during cell density-dependentdownregulation of amphoterin expression. Theantisense treatment indeed inhibits migrationof various cell types, as is demonstrated in ahaptotactic transfilter assay, whereas all thecontrols used in this study did not have a sig-nificant effect on migration (IV). In addition,anti-amphoterin antibodies added in the me-dium significantly reduced cell migration(IV), suggesting a role for extracellularamphoterin in cell motility. Taken together,the present results strongly argue for a regu-latory role of amphoterin in cell migration.Because of its ubiquitous expression,amphoterin might act as an autocrine regula-tor of migration in many cell types.

6.3.2. Does RAGE act as the signallingreceptor of amphoterin in cell migration?

RAGE (receptor of advanced glycation endproducts), an immunoglobulin superfamilymember with closest homology to N-CAM(neural cell adhesion molecule), has been re-cently shown to bind amphoterin (Hori et al.,1995). RAGE is the receptor for the advancedglycation end products (AGEs) that accumu-

37

late in diabetes and during senescence, and itis also suggested to act as a receptor of the β-amyloid (Aβ) peptide in the neurodegenerativemechanism of Alzheimer´s disease (Yan et al.,1996), as well as that of EN-RAGE, a mem-ber of the S100/calgranulin family, in chronicinflammation (Hofmann et al., 1999). SinceRAGE is already expressed during embryonicdevelopment, physiological RAGE ligandshave been searched from tissue extracts, whichhas resulted in the isolation of amphoterin as aligand that binds to the ectodomain of RAGE(Hori et al., 1995). Intriguingly, amphoterincompetes with AGEs, Aβ and EN-RAGE forbinding to the RAGE ectodomain (Hofmannet al., 1999). Interestingly, S100 proteins alsolack a classic secretion signal. It would be ofinterest to know whether their mRNAs local-ize close to the plasma membrane. The soluble ectodomain of RAGE has beenshown to inhibit neurite outgrowth onamphoterin matrix (Hori et al., 1995). Further,the present study (V) shows that transfectionof cells with full-length RAGE, but not thecytosolic domain-deleted RAGE, strongly en-hances extension of filopodia and neurites incells grown on matrix coated with recombinantamphoterin. The amphoterin/RAGE-mediatedprocess extension depends on the Rho-familyGTPases Cdc42 and Rac1, but not on RhoAor Ras. The amphoterin/RAGE/GTPase path-way thus provides one mechanistic explana-tion for the role of amphoterin in cell migra-tion. It is also noteworthy that amphoterinbinds to cell surface proteoglycans (Salmivirtaet al. 1992; Milev et al., 1998) andsulfoglycolipids (Nair et al., 1998) that mayhave a role in amphoterin-induced cell migra-tion. The ligation of RAGE with Aβ and AGEs issuggested to generate the cellular oxidant stressthat activates the ras-MAP kinase pathwayleading to the activation of NF-κB, but it hasnot been directly demonstrated whether RAGEis the signalling receptor. The present study(V) shows that activation of NF-κB requiresthe cytoplasmic domain of RAGE and the raspathways but not the rho family GTPases. Inconclusion, RAGE may activate two different

signalling pathways that in turn result in NF-κB activation or cytoskeletal reorganizationduring cell migration.

6.4. IS mRNA LOCALIZATION IN-VOLVED IN CELL MIGRATION?

6.4.1. Mechanisms that localize amphoterinand β-actin mRNAs

The localization of β-actin that is implicatedin cell motility, becomes more peripheral dur-ing cell spreading and migration and less soduring cell confluency and differentiation, bothat the mRNA and protein levels (Singer et al.,1989; Kislauskis et al., 1993, Hill et al.,1993). The distribution of amphoterin mRNAat the leading edges of motile cells resem-bles the localization of β-actin mRNA. Dou-ble-in situ hybridization confirmed the co-localization of the β-actin and amphoterinmRNAs in the cell processes (II, III and IV). The similarities found in the distribution ofthe β-actin and amphoterin mRNA, suggest acommon mechanism for the localization of thetwo messengers. It is conceivable that proteinsinvolved in process extension might use thesame cytoskeleton-dependent transport mecha-nism to the cytoplasmic processes. Experi-ments using cytoskeleton-disrupting drugsclearly suggest that this indeed is the case.Localization of the β-actin mRNA has beenpreviously shown to depend on microfilaments(Sundell and Singer, 1991). As was shown inthe present study, both the anchoring and thetransport of amphoterin mRNA also dependon microfilaments. Furthermore, the rates oftransport into and dissociation from the cellprocesses are similar for the β-actin andamphoterin mRNAs (Fig. 5). In addition, themRNAs still co-localize in drug-treated cells.These findings clearly argue for a commonpathway of mRNA transport and anchoring inthe cases of both β-actin and amphoterin. The interaction of amphoterin and β-actinmRNA with microfilaments is not a uniqueexample: vimentin and tubulin mRNA alsointeract with microfilaments (Singer et al., 1989).A major proportion of polyadenylated mRNA

38

is associated with microfilaments (Taneja etal., 1992) in fibroblasts while in neurons,microtubules provide the major framework forpolyadenylated mRNA transport and/oranchoring (Bassell et al., 1994b). However, inthe case where the localization pathway for aparticular mRNA, like amphoterin and β-actin,overlaps, we might expect to find overlap inthe localization signals as well. β-actinlocalization signal was shown to be in the3´UTR part of the messenger (Kislauskis et al.,1994). This signal is composed of two regionscalled zipcodes, the first 54 nt of the 3´UTRpart and a homologous but less active 43 ntsegment within the 3´UTR. Comparison withthe β-actin zipcode sequence revealed 60%homology within the 3´UTR sequence ofamphoterin but not in its 5´region as is thecase of β-actin (Fig. 7) (Kislauskis et al., 1993).However, further studies to determine theamphoterin zipcodes are required to confirm acommon localization signal. Perhaps otherparameters, like secondary or even tertiarystructures, of the 3´UTR should be considered.It is also noteworthy that amphoterin/HMG1contains a long and variable 3´UTR(Merenmies et al., 1991, Ferrari et al., 1994), thefunctions of which should be furthercharacterized. Particles functioning as transport assemblieshave been recently described (Ainger et al.,1993). Such particles may contain all the com-ponents needed for the transport, anchoringand translation of the messenger. Until now, ithas been unclear whether these granules con-

tain one mRNA species or several differentmRNAs. The present study demonstrates that,in the case of amphoterin and β-actin, bothmRNAs are mainly transported in differentgranules, despite the similarities in the trans-port and anchoring mechanisms (Fig. 4).

6.4.2. Localization of amphoterin and βββββ-actin mRNAs is directly coupled to trans-membrane signalling involved in cell migra-tion

The mechanism of migration involves a com-bination of different phenomena, which arecontrolled by extracellular signals. Previousstudies have shown that extracellular stimuli,like growth factors, are required for the locali-zation of mRNAs (see review of the litera-ture, p. 14). Until now, no direct links betweencell environment and mRNA localization havebeen demonstrated. However, it is reasonableto assume that extracellular stimuli might regu-late both spatially and temporally the locali-zation of mRNAs. In the case of amphoterinand β-actin mRNAs, stimuli influencing cellmigration should specifically induce the locali-zation of the mRNAs, independently of proc-ess extension or the growth stage of the cell.To test the requirement of extracellular fac-tors on the distribution of amphoterin and β-actin mRNA irrespectively of the growingprocesses, HB-GAM was applied locally viamicrobeads on the C6 or osteoblastic cells.Accumulation of the amphoterin and β-actinmessengers around the beads both in the cell

Figure 7: Comparison of the chicken β-actin zipcode sequence with amphoterin, rat and human β-actin sequencesusing the GCG (Genetic Computer Group software) program bestfit in order to find the optimal alignment. Theunderlined regions show the 2 motifs AATGC and GGACT of the chicken actin defined as fine structure of thezipcode. The AATGC motif has been shown to be more active than the GGACT motif. In the rat and human actins,AATGC, described above, is represented as AACTTGC. Comparison between rat ß-actin and rat amphoterin shows ahigh homology in the 3´region of the zipcode where the ‘AATGC´is present.

39

soma and the processes indicates that HB-GAM is able to regulate the mRNA localiza-tion independently on the process growth (II).This result shows that amphoterin and β-ac-tin mRNA localizations are tightly connectedto the mechanisms involved in cell migration. In addition to HB-GAM, other extracellu-lar matrix proteins, like laminin andfibronectin, are able to regulate the localiza-tion of both mRNAs in C6 cells. However solu-ble heparin or heparitinase have no significanteffects on the mRNA-localizing activity oflaminin, although they strongly inhibit the ef-fect of HB-GAM. It is therefore likely thatthese extracellular components act throughdifferent transmembrane components. Lamininand fibronectin might act through integrinreceptors. Very recently, Chicurel and col-leagues have shown that integrins regulate thelocalization of mRNAs, by using a similarmethod (Chicurel et al., 1998). In contrast, HB-GAM appears to act through N-syndecan. Thefinding that soluble N-syndecan andheparitinase treatment inhibit the mRNA-lo-calizing activity of HB-GAM supports thisinference. In addition, the mRNA-localizingactivity of N-syndecan is also suggested by thefinding that affinity-purified anti-N-syndecanantibodies applied via microbeads are able tolocalize β-actin mRNA (II). The binding of HB-GAM to N-syndecan in-creases the phosphorylation of c-src andcortactin, suggesting that N-syndecan acts asan HB-GAM receptor by modulating the cy-toskeleton via the src kinase/cortactin pathway(Kinnunen et al., 1998b). Like in the neuriteoutgrowth-promoting effect of HB-GAM(Kinnunen et al., 1998b), the mRNA-localiz-ing activity was also significantly inhibited byPP1, a specific inhibitor of src-family kinases.The induction of src-family kinase activity byHB-GAM via N-syndecan might lead to phos-phorylation of proteins required for the locali-zation of β-actin. Ross et al., have recently iso-lated a complex of proteins responsible for thelocalization of β-actin mRNA (Ross et al.,1997). Some of them possess putative phos-phorylation sites. It would be of great interestto elucidate whether the same proteins are

also involved in amphoterin mRNA localiza-tion. In conclusion, this study shows, for the firsttime, that a local cue of a cell matrix-associ-ated protein is able to control the localizationof motility-associated mRNAs, like β-actinand amphoterin. Therefore, local proteintranslation in response to a local extracellu-lar cue may be an important mechanism inenhancing and guiding the migratory re-sponses of cells.

6.5. HYPOTHETICAL MODEL FOR THEREGULATION OF CELL MIGRATIONBY AMPHOTERIN AND HB-GAM DUR-ING NORMAL DEVELOPMENT AND INPATHOLOGY

6.5.1. HB-GAM as a target-derived cue andamphoterin as an autocrine/paracrine regu-lator of cell migration

According to the present study, when stimu-lated by an extracellular migratory signal, likeHB-GAM, cells start to target amphoterin tothe processes at the mRNA level. This resultsin the local translation of amphoterin closeto the plasma membrane. Amphoterin is thenreleased to the extracellular space where itinteracts with the RAGE protein and sulfatedglycan epitopes in proteoglycans andglycolipids, which may enhance invasiveprocess extension in cells. During motilephenomena amphoterin may be released bythe migrating cell itself (“autocrine mecha-nism”) or by the neighboring cell (“paracrinemechanism”). It is also noteworthy thatamphoterin binds both plasminogen and t-PA(tissue-type plasminogen activator), whichresults in degradation of amphoterin andstrong enhancement of proteolytic activity(Parkkinen and Rauvala, 1991; Parkkinen etal., 1993), which may facilitate cell invasionduring normal development and metastasis.The RAGE-mediated effect on process ex-tension would be thereby terminated by deg-radation of amphoterin by plasmin. Theamphoterin/RAGE-mediated effect on proc-ess extension would thus be transient, as ex-

40

pected during cell motility, and would be cou-pled to proteolytic activation enhancing pen-etration of a process into a tissue (Fig. 8).

6.5.2. Implications of HB-GAM andamphoterin during development and inpathology

HB-GAM has been implicated in many devel-opmental processes in the nervous system, in-cluding neuronal migration, neurite outgrowth,axonal guidance and synaptogenesis (Rauvalaand Peng, 1997; Lauri et al., 1998) In addi-tion, HB-GAM has been shown to participatein the mechanism of LTP (long termpotentiation; Lauri et al., 1998). Mechanismsof synaptic plasticity, like LTP, are thought toinvolve reorganization of neuronal connec-tions, mimicking the neurite outgrowth proc-ess (Constantine-Paton and Cline, 1998).Whether the role of HB-GAM in LTP and alsoin synaptogenesis is related to its mRNA-lo-calizing activity is an interesting notion thatwarrants further investigations. It is reason-able that the local synthesis of cytoskeletalstructures in synaptic areas might be an effi-cient way to reorganize neural connectionsduring stabilization of LTP. The present findings attribute to HB-GAMa role in bone formation, maintenance and re-pair. However, the knockout mice show nomajor development defects, indicating thatHB-GAM is not essential for bone develop-ment (Hienola, Imai and Rauvala, unpublisheddata). HB-GAM shares 50 % homology withmidkine; these proteins thus constitute a fam-ily of 2 members. However, HB-GAM andmidkine sequences contain 2 domains, whichshow significant homology with thethrombospondin type 1 (TSR) repeats in F-spondin (Kilpeläinen et al., 2000). F-spondinhas been implicated in axonal guidance in thefloor plate (Klar et al., 1992). TSR repeats arefound in several other extracellular matrix andcell surface proteins (Kilpeläinen et al, 2000).This suggests that the lack of HB-GAM mightbe compensated by other proteins. However,it would be of interest to test whether in bonemaintenance and repair, the lack of HB-GAM

is compensated. Recently, Watanabe and colleagues showedfor the first time a correlation between anmRNA sorting abnormality with a mamma-lian disease. They observed that osteoblastsfrom the skeletal mutation toothless, an os-teopetrotic mutation in rat, fail to sort β-ac-tin mRNA to the cell processes and γ-actinto the perinuclear regions, whereas in nor-mal osteoblasts, these mRNA´s localize(Watanabe et al., 1998). The mutantosteoblasts show a lack of long stress fibersseen in the normal osteoblasts. However,treatment with CSF-1 (colony stimulating fac-tor 1), a growth factor whose receptor, c-fms,is a tyrosine kinase, restores the stress fibersin mutant osteoblats. Therefore, theyhypothetize that the primary genetic lesionin toothless rats results in interruption of asignalling pathway that is critical for theproper sorting of actin mRNAs. Because HB-GAM is involved in the recruitment ofosteoblasts and bone formation, it would beof interest to see whether HB-GAM and itsmRNA-localizing activity may rescue the mu-tation.

Amphoterin is highly expressed during devel-opment and in transformed cells in compari-son to their normal counterparts. Roles in neu-rite outgrowth, cell differentiation, neural-glialinteraction and endotoxemia have been attrib-uted for amphoterin. The present findings showthat amphoterin regulates cell migration. Inaddition, amphoterin interacts directly withplasminogen and tissue type plasminogen ac-tivator (tPA), which enhances plasmin genera-tion at the cell surface (Parkkinen et al., 1991).Because of its high expression during embry-onic development, amphoterin may regulatedevelopmental phenomena, such as the inva-sive phenotype of migrating growth conesduring the development of neuronal connec-tions. In addition, the regulation of cell migrationby amphoterin should be of interest in the in-vasive behavior of tumor cells. Indeed, theability of malignant cells to infiltrate struc-tures that are adjacent to or distant from the

41

Figure 8: (A) A model forthe HB-GAM/N-syndecaninteraction in cell migration.HB-GAM/N-syndecan in-teraction activates thecortactin/src-kinase signal-ling pathway that is respon-sible for cytoskeletal reorga-nization required during cellmigration (left panel). ThemRNA-localizing activity ofthe HB-GAM/N-syndecaninteraction is one angle in thecell motility control. (B) Ahypothesis explaining thefunctional role of amphoterinin a motile cell. AmphoterinmRNA is transported alongthe microfilaments to the cellperiphery, where the proteinis translated and exported tothe extracellular space. Dueto its highly adhesive naturethat may be caused by thepoly-cationic andpolyanionic regions,amphoterin binds to cell ma-trices and surfaces, facilitat-ing transient binding of thecell process to the surround-ing tissue during migration.Amphoterin mediates pro-cess extension via RAGE/GTPase pathway during nor-mal development and/or re-generation after injury. The figure depicts an autocrine-type interaction, but amphoterin may also be derived from neighboring cells. The amphoterin-mediated cell surface effect is coupled toproteolytic activation since amphoterin binds both plasminogen and a tissue-type plasminogen activator enhancing the generation of active enzyme. The generation of local proteolyticactivity, due to the ternary complex of amphoterin/t-PA/plasminogen, hydrolyzes amphoterin and facilitates further extension of a process into a tissue.

42

primary tumor site is an important determi-nant of the prognosis of tumors. Infiltrativebehavior requires migration and invasion,which in addition to locomotion involves pro-teolytic activities to degrade the extracellu-

lar matrix. Therefore, amphoterin is an idealcandidate to enhance tumor invasion. How-ever, the role of amphoterin in metastasisshould be carefully examined in an in vivomodel.

43

8. ACKNOWLEDGEMENTS

This study was carried out in the Department of Biosciences and at the Institute of Biotechnol-ogy in the University of Helsinki during the years 1996-2000. I thank professor Carl G. Gamhberg,head of the Division of Biochemistry, and Professor Mart Saarma, director of the Institute ofBiotechnology, not only for providing excellent facilities during my work but as well as for theirprecious help in various occasions. Merci beaucoup.

I have had the great privilege to work under the supervision of Professor Heikki Rauvala. I amsincerely grateful to you, Heikki, for providing me with the opportunity of working in yourgroup and to learn from your experience in science. Your door was always open for my problemsconcerning the scientific and not scientific issues. Merci beaucoup, je vous en suisreconnaissante à jamais.

I wish to express my gratitude to Professor Jorma Keski-Oja and Docent Pekka Lappalainen fortheir constructive criticism and their help in improving this manuscript. I want to thank DavidRice for correcting my ̀ frenchy´ mistakes.

To all the former and present collegues of the Rauvala´s lab: Henri, Anu, Sami, Sari, Tarja, Ewa,Erkki, Marko, Anni, Maria, Marko, Seija, Eveliina and.....: a BIG THANKS. Without your helpand your friendship, my work in the lab and my life in the ̀ pub´ would have never been so great.I want to thank specially Henri for the great discussion we had about the ̀ Janus´project...Mercià vous tous.

I wish to thank all my friends in the great BI Knockout team. I promise I will stop with my ̀ KungFu´ tricks....I want to thanks all my friends in Finland for their supports outside the lab. Merci devotre amitié.

Papa, je te suis reconnaissante pour tout ce que m´appris (Miracle, tu n´iras pas à Lourdes àpied...). Papa, Eric, Sylvie, Mélanie, Marine, Mado, Jacky et Philippe, sans vous cette thèsen´aurait jamais abouti. Je vous exprime toute ma profonde gratitude. Cette thèse est aussi pourvous. Merci de tout mon coeur.

I want to thank someone who is very important in my life. Frank, je veux te remercier pour tonencouragement, ta patience et ton amour. Ich liebe dich so sehr.

This study has been supported by the TEKES Foundation, Sigrid Jusélius Foundation and theAcademy of Finland.

Kiitos paljon,

Helsinki, January 20th, 2000

Carole

44

8. REFERENCES

Ainger K, Avossa D, Morgan F, Hill SJ, Barry C, Barbarese E and Carson JH (1993) Transport and localization ofexogenous myelin basic protein mRNA microinjected into oligodendrocytes. Journal of Cell Biology 123:431-441.

Ainger K, Avossa D, Diana AS, Barry C, Barbarese E and Carson JH (1997) Transport and localization elements inmyelin basic protein mRNA. Journal of Cell Biology 138:1077-1087.

Aumailley M, and Smyth N (1998) The role of laminins in basement membrane function. Journal ofAnatomy 193:1-21

Barbarese E, Koppel DE, Deutscher MP, Smith CL, Ainger, Morgan F and Carson JH (1995) Protein translationcomponents are colocalized in granules in oligodendrocytes. Journal of Cell Science 108:2781-2790.

Bassell GJ, Powers CM, Taneja KL and Singer RH (1994a) Single mRNAs visualized by ultrastructural in situhybridization are principally localized at actin filament intersections in fibroblasts. Journal of Cell Biology126:863-876.

Bassell GJ, Singer RH and Kosik KS (1994b) Association of poly(A) mRNA with microtubules in culturedneurons. Neuron 12:571-582.

Bassell GJ, Zhang H, Byrd AL, Femino AM, Singer RH, Taneja KL, Lifshitz LM, Herman IM and Kosik KS(1998) Sorting of beta-actin mRNA and protein to neurites and growth cones in culture. Journal of Neuroscience18:251-265.

Baxevanis AD and Landsman D (1995) The HMG-1 box protein family: classification and functional relation-ships. Nucleic Acids Res 23:1604-1613.

Benson DL (1997) Dendritic compartmentation of NMDA receptor mRNA in cultured hippocampal neurons.Neuroreport 8:823-828.

Bianchi ME, Beltrame M and Paonessa G (1989) Specific recognition of cruciform DNA by nuclear proteinHMG1. Science 243:1056-1059.

Bianchi ME, Falciola L, Ferrari S and Lilley DM (1992) The DNA binding site of HMG1 protein is composed oftwo similar segments (HMG boxes), both of which have counterparts in other eukaryotic regulatory proteins.EMBO Journal 11:1055-1063.

Bianchi ME and Beltrame M (1998) Flexing DNA: HMG-Box Proteins and Their Partners. American journal ofhuman genetic 63:1573-1577.

Birchmeier C and Gherardi E (1998) Developmental roles of HGF/SF and its receptor, the c-met tyrosine kinase.Trends in Cell Biology 8:404-410.

Bladt F, Riethmacher D, Isenmann S, Aguzzi A and Birchmeier (1995) Essential role for the c-met receptor in themigration of myogenic precursor cells into the limb bud. Nature 376:768-771.

Bobola N, Jansen RP, Shin TH and Nasmyth K (1996) Asymmetric accumulation of Ash1p in postanaphase nucleidepends on a myosin and restricts yeast mating-type switching to mother cells. Cell 84:699-709.

Boonyaratanakornkit V, Melvin V, Prendergast P, Altmann M, Ronfani L, Bianchi ME, Taraseviciene L, NordeenSK, Allegretto EA and Edwards DP (1998) High-mobility group chromatin proteins 1 and 2 functionally interactwith steroid hormone receptors to enhance their DNA binding in vitro and transcriptional activity in mammaliancells. Molecular and Cellular Biology 18:4471-4487.

45

Bottaro DP, Rubin JS, Faletto DL, Chan AM, Kmiecik TE, Vande WG and Aaronson SA (1991)Identification of the hepatocyte growth factor receptor as the c-met proto-oncogene product. Science251:802-804.

Brownlee M, Cerami A and Vlassara H (1988) Advanced glycosylation end products in tissue and the biochemicalbasis of diabetic complications. New England Journal of Medicine 318:1315-1321.

Burgin KE, Waxham MN, Rickling S, Westgate SA, Mobley, WC and Kelly PT (1990) In situ hybridizationhistochemistry of Ca2+/calmodulin-dependent protein kinase in developing rat brain. Journal of Neuroscience10:1788-1798.

Bustin M and Neihart NK (1979) Antibodies against chromosomal HMG proteins stain the cytoplasm ofmammalian cells. Cell 16:181-189.

Bustin M and Reeves R (1996) High-mobility-group chromosomal proteins: architectural components thatfacilitate chromatin function. Prog Nucleic Acid Res Mol Biol 54:35-100.

Calogero S, Grassi F, Aguzzi A, Voigtlander T, Ferrier, Ferrari S and Bianchi ME (1999) The lack of chromosomalprotein Hmg1 does not disrupt cell growth but causes lethal hypoglycaemia in newborn mice. Nature Genetics22:276-280.

Carey DJ (1997) Syndecans: multifunctional cell-surface co-receptors. Biochemical Journal 327:1-16.

Chant J and Stowers L (1995) GTPase cascades choreographing cellular behavior: movement, morphogenesis, andmore. Cell 81:1-4.

Chen WT (1981) Mechanism of retraction of the trailing edge during fibroblast movement. Journal of CellBiology 90:187-200.

Chicurel ME, Singer RH, Meyer CJ and Ingber DE (1998) Integrin binding and mechanical tension inducemovement of mRNA and ribosomes to focal adhesions. Nature 392:730-733.

Chicurel ME, Terrian DM and Potter H (1993) mRNA at the synapse: analysis of a synaptosomal preparationenriched in hippocampal dendritic spines. Journal of Neuroscience 13:4054-4063.

Colman DR, Kreibich G, Frey AB and Sabatini DD (1982) Synthesis and incorporation of myelin polypeptidesinto CNS myelin. Journal of Cell Biology 95:598-608.

Conklin EG (1905) The organization and cell-lineage of the ascadian egg. J.Acad.Natural Sci.Philadelphia 13:1-119.

Constantine-Paton, M and Cline, H. T (1998) LTP and activities dependent synaptogenesis: the morealike they are, the more different they become. Current Opinion in Neurobiology 8, 139-148.

Couchman, J. R and Woods, A (1999) Syndecan-4 and integrins: combinatorial signaling in cell adhesion.Journal of Cell Science 112, 3415-3420.

Crino PB and Eberwine J (1996) Molecular characterization of the dendritic growth cone: regulated mRNAtransport and local protein synthesis. Neuron 17:1173-1187.

Crino P, Khodakhah K, Becker K, Ginsberg S, Hemby S and Eberwine J (1998) Presence and phosphorylation oftranscription factors in developing dendrites. Proceedings of the National Academy of Sciences of the UnitedStates of America 95:2313-2318.

Daston MM and Ratner N (1991) Expression of P30, a protein with adhesive properties, in Schwann cells andneurons of the developing and regenerating peripheral nerve. Journal of Cell Biology 112:1229-1239.

46

de Vries H, de Jonge JC, Schrage C, van der Haar ME and Hoekstra D (1997) Differential and cell develop-ment-dependent localization of myelin mRNAs in oligodendrocytes. Journal of Neuroscience Research47:479-488.

Deshler JO, Highett MI, Abramson T and Schnapp BJ (1998) A highly conserved RNA-binding protein forcytoplasmic mRNA localization in vertebrates. Current Biology 8:489-496.

Ding D, Parkhurst SM, Halsell SR and Lipshitz HD (1993) Dynamic Hsp83 RNA localization during Drosophilaoogenesis and embryogenesis. Molecular and Cellular Biology 13:3773-3781.

Ferrari S, Ronfani L, Calogero S and Bianchi ME (1994) The mouse gene coding for high mobility group 1protein (HMG1). Journal of Biological Chemistry 269:28803-28808.

Fontaine B, Sassoon D, Buckingham M and Changeux JP (1988) Detection of the nicotinic acetylcholine receptoralpha-subunit mRNA by in situ hybridization at neuromuscular junctions of 15-day-old chick striated muscles.EMBO Journal 7:603-609.

Fulton AB and Alftine C (1997) Organization of protein and mRNA for titin and other myofibril componentsduring myofibrillogenesis in cultured chicken skeletal muscle. Cell Structure and Function 22:51-58.

Furuichi T, Simon-Chazottes D, Fujino I, Yamada N, Hasegawa M, Miyawaki A, Yoshikawa S, Guenet JL andMikoshiba K (1993) Widespread expression of inositol 1,4,5-trisphosphate receptor type 1 gene (Insp3r1) in themouse central nervous system. Receptors and Channels 1:11-24.

Gardiol A, Racca C and Triller A (1999) Dendritic and postsynaptic protein synthetic machinery. Journal ofNeuroscience 19:168-179.

Garner CC, Tucker RP and Matus A (1988) Selective localization of messenger RNA for cytoskeletal proteinMAP2 in dendrites. Nature 336:674-677.

Gazzaley AH, Benson DL, Huntley GW and Morrison JH (1997) Differential subcellular regulation of NMDAR1protein and mRNA in dendrites of dentate gyrus granule cells after perforant path transection. Journal ofNeuroscience 17:2006-2017.

Ge H and Roeder RG (1994) The high mobility group protein HMG1 can reversibly inhibit class II genetranscription by interaction with the TATA-binding protein. Journal of Biological Chemistry 269:17136-17140.

Ghandour MS and Skoff RP (1991) Double-labeling in situ hybridization analysis of mRNAs for carbonicanhydrase II and myelin basic protein: expression in developing cultured glial cells. Glia 4:1-10.

Gould RM, Freund CM and Barbarese E (1999) Myelin-associated oligodendrocytic basic protein mRNAs resideat different subcellular locations. J Neurochem 73:1913-1924.

Hall A (1998) Rho GTPases and the actin cytoskeleton. Science 279:509-514.

Hannan AJ, Gunning P, Jeffrey PL and Weinberger RP (1998) Structural compartments within neurons:developmentally regulated organization of microfilament isoform mRNA and protein. Molecular and CellularNeurosciences 11:289-304.

Harter C and Wieland F (1996) The secretory pathway: mechanisms of protein sorting and transport. Biochimicaet Biophysica Acta 1286:75-93.

Havin L, Git A, Elisha Z, Oberman F, Yaniv K, Schwartz, SP, Standart N and Yisraeli JK (1998) RNA-bindingprotein conserved in both microtubule- and microfilament-based RNA localization. Genes and Development

Daston MM and Ratner N (1994) Amphoterin (P30, HMG-1) and RIP are early markers ofoligodendrocytes in the developing rat spinal cord. Journal of Neurocytology 23:323-332.

47

12:1593-1598.

Hazelrigg T (1998) The destinies and destinations of RNAs. Cell 95:451-460.

Hill MA and Gunning P (1993) Beta and gamma actin mRNAs are differentially located within myoblasts.Journal of Cell Biology 122:825-832.

Hill MA, Schedlich L and Gunning P (1994) Serum-induced signal transduction determines the peripherallocation of beta-actin mRNA within the cell. Journal of Cell Biology 126:1221-1229.

Hirokawa N (1998) Kinesin and dynein superfamily proteins and the mechanism of organelle transport.Science 279:519526.

Hofmann MA, Drury S, fu C, Qu W, Taguchi A, Lu Y, Avila C, Kambham N, Bierhaus A, Nawroth P,Neurath MF, Beach D, McClary J, Nagashima M, Morser J, Stern D and Schmidt AM (1999) RAGEmediates a novel proinflammatory axis: a central cell surface receptor for S100/calgranulin polypeptides.Cell 97:889-901.

Holz A, Schaeren-Wiemers N, Schaefer C, Pott U, Colello, RJ and Schwab ME (1996) Molecular and develop-mental characterization of novel cDNAs of the myelin-associated/oligodendrocytic basic protein. Journal ofNeuroscience 16:467-477.

Hori O, Brett J, Slattery T, Cao R, Zhang J, Chen JX, Nagashima M, Lundh ER, Vijay S and Nitecki D (1995) Thereceptor for advanced glycation end products (RAGE) is a cellular binding site for amphoterin. Mediation ofneurite outgrowth and co-expression of rage and amphoterin in the developing nervous system. Journal ofBiological Chemistry 270:25752-25761.

Hori O, Yan SD, Ogawa S, Kuwabara K, Matsumoto M, Stern, Schmidt AM (1996) The receptor for advancedglycation end-products has a central role in mediating the effects of advanced glycation end-products on thedevelopment of vascular disease in diabetes mellitus. Nephrology, Dialysis, Transplantation 11 Suppl 5:13-16.

Huang C, Ni Y, Wang T, Gao Y, Haudenschild CC v Zhan X (1997) Down-regulation of the filamentous actincross-linking activity of cortactin by Src-mediated tyrosine phosphorylation. Journal of Biological Chemistry272:13911-13915.

irikowski GF, Sanna PP and Bloom FE (1990) mRNA coding for oxytocin is present in axons of thehypothalamo-neurohypophysial tract. Proceedings of the National Academy of Sciences of the United States ofAmerica 87:7400-7404.

Kang H and Schuman EM (1996) A requirement for local protein synthesis in neurotrophin-induced hippocampalsynaptic plasticity. Science 273:1402-1406.

Kiebler MA, Hemraj I, Verkade P, Kohrmann M, Fortes P, Marion RM, Ortin J and Dotti CG (1999) Themammalian staufen protein localizes to the somatodendritic domain of cultured hippocampal neurons: implica-tions for its involvement in mRNA transport. Journal of Neuroscience 19:288-297.

Kilpeläinen I, Kaksonen M, Kinnunen T, Avikainen H, Fath M, Linhardt RJ, Raulo E and Rauvala H (2000) HB-GAM (heparin-binding grwoth associated molecule) contains two heparin-binding beta-sheet domains that arehomologous to the thrombospondin type I repeat. Journal of Biological Chemistry. InPress.

Kinnunen T, Kaksonen M, Saarinen J, Kalkkinen N, Peng HB and Rauvala H (1998a) Cortactin-Src kinasesignaling pathway is involved in N-syndecan-dependent neurite outgrowth. Journal of Biological Chemistry273:10702-10708.

Kinnunen A, Kinnunen T, Kaksonen M, Nolo R, Panula P and Rauvala H (1998b) N-syndecan and HB-GAM (heparin-binding growth-associated molecule) associate with early axonal tracts in the rat brain.European Journal of Neuroscience 10:635-648.

48

Kislauskis EH, Li Z, Singer RH and Taneja KL (1993) Isoform-specific 3'-untranslated sequences sortalpha-cardiac and beta-cytoplasmic actin messenger RNAs to different cytoplasmic compartments. Journalof Cell Biology 123:165-172.

Kislauskis EH, Zhu X and Singer RH (1994) Sequences responsible for intracellular localization of beta-actinmessenger RNA also affect cell phenotype. Journal of Cell Biology 127:441-451.

Kislauskis EH, Zhu X and Singer RH (1997) beta-Actin messenger RNA localization and protein synthesisaugment cell motility. Journal of Cell Biology 136:1263-1270.

Klar A, Baldassare M and Jessel TM (1992) F-spondin: a gene expressed at high levels in the floor plate encodes asecreted protein that promotes neural cell adhesion and neurite extension. Cell 69:95-110.

Kleiman R, Banker G and Steward O (1994) Development of subcellular mRNA compartmentation in hippocam-pal neurons in culture. Journal of Neuroscience 14:1130-1140.

Kloc M and Etkin LD (1998) Apparent continuity between the messenger transport organizer and late RNAlocalization pathways during oogenesis in Xenopus. Mechanisms of Development 73:95-106.

Knowles RB and Kosik KS (1997) Neurotrophin-3 signals redistribute RNA in neurons. Proceedings of theNational Academy of Sciences of the United States of America 94:14804-14808.

Landry CF, Watson JB, Kashima T and Campagnoni AT (1994) Cellular influences on RNA sorting in neurons andglia: an in situ hybridization histochemical study. Brain Research Molecular Brain Research. 27:1-11.

Landsman D and Bustin M (1993) A signature for the HMG-1 box DNA-binding proteins. Bioessays 15:539-546.

Latham VMJ, Kislauskis EH, Singer RH and Ross AF (1994) Beta-actin mRNA localization is regulated by signaltransduction mechanisms. Journal of Cell Biology 126:1211-1219.

Lauffenburger DA and Horwitz AF (1996) Cell migration: a physically integrated molecular process. Cell 84:359-369.

Lauri SE, Rauvala H, Kaila K and Taira T (1998) Effect of heparin-binding growth-associated molecule (HB-GAM) on synaptic transmission and early LTP in rat hippocampal slices. European Journal of Neuroscience10:188-194.

Lawrence JB and Singer RH (1986) Intracellular localization of messenger RNAs for cytoskeletal proteins. Cell45:407-415.

Leavesley DI, Schwartz MA, Rosenfeld M and Cheresh DA (1993) Integrin beta 1- and beta 3-mediatedendothelial cell migration is triggered through distinct signaling mechanisms. Journal of Cell Biology 121:163-170.

Lehmann E, Hanze J, Pauschinger M, Ganten D and Lang RE (1990) Vasopressin mRNA in the neurolobe of therat pituitary. Neuroscience Letters 111:170-175.

Litman P, Barg J, Rindzoonski L and Ginzburg I (1993) Subcellular localization of tau mRNA in differentiatingneuronal cell culture: implications for neuronal polarity. Neuron 10:627-638.

Lyford GL, Yamagata K, Kaufmann WE, Barnes CA, Sanders, LK, Copeland NG, Gilbert DJ, Jenkins NA,Lanahan AA and Worley PF (1995) Arc, a growth factor and activity-regulated gene, encodes a novel cytoskel-eton-associated protein that is enriched in neuronal dendrites. Neuron 14:433-445.

Kinnunen A, Niemi M, Kinnunen T, Kaksonen M, Nolo R and Rauvala H (1999) Heparan sulphate andHB-GAM (heparin-binding growth-associated molecule) in the development of the thalamocorticalpathway of rat brain. European Journal of Neuroscience 11:491-502.

49

Maeda N, Nishiwaki T, Shintani T, Hamanaka H and Noda M (1996) 6B4 proteoglycan/phosphacan, anextracellular variant of receptor-like protein-tyrosine phosphatase zeta/RPTPbeta, binds pleiotrophin/heparin- binding growth-associated molecule (HB-GAM). Journal of Biological Chemistry 271:21446-21452.

Maeda N and Noda M (1998) Involvement of receptor-like protein tyrosine phosphatase zeta/RPTPbetaand its ligand pleiotrophin/heparin-binding growth-associated molecule (HB-GAM) in neuronal migration.Journal of Biological Chemistry 142:203-216.

Marshall CJ (1995) Specificity of receptor tyrosine kinase signaling: transient versus sustained extracellularsignal-regulated kinase activation. Cell 80:179-185.

Matsubura S, Tomomura M, Kadomatsu K and Muramatsu T (1990) Structure of a retinoic acid-responsivegene, MK, which is transiently activated during the differentiation of embryonal carcinoma cells and themid-gestation period of mouse embryogenesis. Journal of Biological Chemistry 265:9441-9443.

Melloni E, Sparatore B, Patrone M, Pessino A, Passalacqua M and Pontremoli (1995a) Identity inmolecular structure between “differentiation enhancing factor” of murine erythroleukemia cells and the 30kD heparin-binding protein of developing rat brain. Biochemical and Biophysical Research Communica-tions 210:82-89.

Melloni E, Sparatore B, Patrone M, Pessino A, Passalacqua M and Pontremoli (1995b) Extracellular release of the‘differentiation enhancing factor’, a HMG1 protein type, is an early step in murine erythroleukemia cell differen-tiation. FEBS Letters 368:466-470.

Merenmies J and Rauvala H (1990) Molecular cloning of the 18-kDa growth-associated protein of developingbrain. Journal of Biological Chemistry 265:16721-16724.

Merenmies J, Pihlaskari R, Laitinen J, Wartiovaara J and Rauvala H (1991) 30-kDa heparin-binding protein ofbrain (amphoterin) involved in neurite outgrowth. Amino acid sequence and localization in the filopodia of theadvancing plasma membrane. Journal of Biological Chemistry 266:16722-16729.

Merlie JP and Sanes JR (1985) Concentration of acetylcholine receptor mRNA in synaptic regions of adult musclefibres. Nature 317:66-68.

Milev P, Chiba A, Haring M, Rauvala H, Schachner M, Ranscht B, Margolis RK and Margolis RU (1998) Highaffinity binding and overlapping localization of neurocan and phosphacan/protein-tyrosine phosphatase-zeta/betawith tenascin-R, amphoterin, and the heparin-binding growth-associated molecule. Journal of BiologicalChemistry 273:6998-7005.

Mitchison TJ and Cramer LP (1996) Actin-based cell motility and cell locomotion. Cell 84:371-379.

Mohr E, Fehr S and Richter D (1991) Axonal transport of neuropeptide encoding mRNAs within thehypothalamo-hypophyseal tract of rats. EMBO Journal 10:2419-2424.

Morris EJ and Fulton AB (1994) Rearrangement of mRNAs for costamere proteins during costameredevelopment in cultured skeletal muscle from chicken. Journal of Cell Science 107:377-386.

Moriya M and Tanaka S (1994) Prominent expression of protein kinase C (gamma) mRNA in thedendrite-rich neuropil of mice cerebellum at the critical period for synaptogenesis. Neuroreport 5:929-932.

Mosher DF, Sottile J, Wu C and McDonald JA (1992) Assembly of extracellular matrix. Current Opinion in CellBiology 4:810-818.

Muslimov IA, Banker G, Brosius J and Tiedge H (1998) Activity-dependent regulation of dendritic BC1 RNA inhippocampal neurons in culture. Journal of Cell Biology 141:1601-1611.

50

Nolo R, Kaksonen M and Rauvala H (1996) Developmentally regulated neurite outgrowth response fromdorsal root ganglion neurons to heparin-binding growth-associated molecule (HB-GAM) and the expressionof HB-GAM in the targets of the developing dorsal root ganglion neurites. European Journal ofNeuroscience 8:1658-1665.

Oleynikov Y and Singer RH (1998) RNA localization: different zipcodes, same postman? Trends in Cell Biology8:381-383.

Palecek SP, Schmidt CE, Lauffenburger DA and Horwitz AF (1996) Integrin dynamics on the tail region ofmigrating fibroblasts. Journal of Cell Science 109:941-952.

Parkkinen J and Rauvala H (1991) Interactions of plasminogen and tissue plasminogen activator (t-PA) withamphoterin. Enhancement of t-PA-catalyzed plasminogen activation by amphoterin. Journal of BiologicalChemistry 266:16730-16735.

Parkkinen J, Raulo E, Merenmies J, Nolo R, Kajander EO, Baumann M and Rauvala H (1993) Amphoterin, the30-kDa protein in a family of HMG1-type polypeptides. Enhanced expression in transformed cells, leading edgelocalization, and interactions with plasminogen activation. Journal of Biological Chemistry 268:19726-19738.

Passalacqua M, Zicca A, Sparatore B, Patrone M, Melloni E and Pontremoli S (1997) Secretion and binding ofHMG1 protein to the external surface of the membrane are required for murine erythroleukemia cell differentia-tion. FEBS Letters 400:275-279.

Passalacqua M, Patrone M, Picotti GB, Del Rio M, Sparatore B, Melloni E, Pontremoli S (1998) Stimulatedastrocytes release high-mobility group 1 protein, an inducer of LAN-5 neuroblastoma cell differentiation.Neuroscience 82:1021-1028.

Ponzetto C, Bardelli A, Zhen Z, Maina F, dalla ZP, Giordano S, Graziani A, Panayotou G and Comoglio PM(1994) A multifunctional docking site mediates signaling and transformation by the hepatocyte growth factor/scatter factor receptor family. Cell 77:261-271.

Rabl C (1885) Uber Zellteilung. Ebenda Bd 10:214-330.

Racca C, Gardiol A and Triller A (1998) Cell-specific dendritic localization of glycine receptor alpha subunitmessenger RNAs. Neuroscience 84:997-1012.

Raulo E, Chernousov MA, Carey DJ, Nolo R and Rauvala H (1994) Isolation of a neuronal cell surface receptor ofheparin binding growth-associated molecule (HB-GAM). Identification as N-syndecan (syndecan-3). Journal ofBiological Chemistry 269:12999-13004.

Rauvala H and Pihlaskari R (1987) Isolation and some characteristics of an adhesive factor of brain thatenhances neurite outgrowth in central neurons. Journal of Biological Chemistry 262:16625-16635.

Nair SM, Zhao Z, Chou DK, Tobet SA and Jungalwala FB (1998) Expression of HNK-1 carbohydrate andits binding protein, SBP-1, in apposing cell surfaces in cerebral cortex and cerebellum. Neuroscience85:759-771.

Neeper M, Schmidt AM, Brett J, Yan SD, Wang F, Pan YC, Elliston K, Stern D and Shaw A (1992) Cloningand expression of a cell surface receptor for advanced glycosylation end products of proteins. Journal ofBiological Chemistry 267:14998-15004.

Nobes CD and Hall A (1995) Rho, rac, and cdc42 GTPases regulate the assembly of multimolecular focalcomplexes associated with actin stress fibers, lamellipodia, and filopodia. Cell 81:53-62.

Nolo R, Kaksonen M, Raulo E and Rauvala H (1995) Co-expression of heparin-binding growth-associatedmolecule (HB-GAM) and N-syndecan (syndecan-3) in developing rat brain. Neuroscience Letters 191:39-42.

51

Rauvala H, Merenmies J, Pihlaskari R, Korkolainen M, Huhtala ML and Panula P (1988) The adhesive andneurite-promoting molecule p30: analysis of the amino-terminal sequence and production of antipeptideantibodies that detect p30 at the surface of neuroblastoma cells and of brain neurons. Journal of CellBiology 107:2293-2305.

Rauvala H (1989) An 18-kd heparin-binding protein of developing brain that is distinct from fibroblastgrowth factors. EMBO Journal 8:2933-2941.

Rauvala H, Vanhala A, Castren E, Nolo R, Raulo E, Merenmies J and Panula P (1994) Expression of HB-GAM (heparin-binding growth-associated molecules) in the pathways of developing axonal processes invivo and neurite outgrowth in vitro induced by HB-GAM. Brain Research Developmental Brain Research.79:157-176.

Rauvala H and Peng HB (1997) HB-GAM (heparin-binding growth-associated molecule) and heparin-typeglycans in the development and plasticity of neuron-target contacts. Progress in Neurobiology 52:127-144.

Rechsteiner M and Kuehl L (1979) Microinjection of the nonhistone chromosomal protein HMG1 intobovine fibroblasts and HeLa cells. Cell 16:901-908.

Ridley AJ and Hall A (1992) The small GTP-binding protein rho regulates the assembly of focal adhesionsand actin stress fibers in response to growth factors. Cell 70:389-399.

Ross AF, Oleynikov Y, Kislauskis EH, Taneja KL and Singer, RH (1997) Characterization of a beta-actinmRNA zipcode-binding protein. Molecular and Cellular Biology 17:2158-2165.

Russell B, Wenderoth MP and Goldspink PH (1992) Remodeling of myofibrils: subcellular distribution ofmyosin heavy chain mRNA and protein. American Journal of Physiology 262:339-345.

Ryan MC, Christiano AM, Engvall E, Wewer UM, Miner JH, Sanes JR and Burgeson RE (1996) Thefunctions of laminins: lessons from in vivo studies. Matrix Biology 15:369-381.

Salmivirta M, Rauvala H, Elenius K and Jalkanen M (1992) Neurite growth-promoting protein(amphoterin, p30) binds syndecan. Experimental Cell Research 200:444-451.

Schlessinger J and Ullrich A (1992) Growth factor signaling by receptor tyrosine kinases. Neuron 9:383-391.

Schmidt AM, Vianna M, Gerlach M, Brett J, Ryan J, Kao, Esposito C, Hegarty H, Hurley W and Clauss M(1992) Isolation and characterization of two binding proteins for advanced glycosylation end productsfrom bovine lung which are present on the endothelial cell surface. Journal of Biological Chemistry267:14987-14997.

Schwartz MA, Schaller MD and Ginsberg MH (1995) Integrins: emerging paradigms of signal transduction.Annual Review of Cell and Developmental Biology 11:549-599.

Sheetz MP, Felsenfeld D, Galbraith CG and Choquet D (1997) Cell migration as a five-step cycle. BiochemSoc Symp 65:233-243.

Sil A and Herskowitz I (1996) Identification of asymmetrically localized determinant, Ash1p, required for lineage-specific transcription of the yeast HO gene. Cell 84:711-722.

Singer RH, Langevin GL and Lawrence JB (1989) Ultrastructural visualization of cytoskeletal mRNAs and theirassociated proteins using double-label in situ hybridization. Journal of Cell Biology 108:2343-2353.

Skubitz AP, Letourneau PC, Wayner E and Furcht LT (1991) Synthetic peptides from the carboxy-terminalglobular domain of the A chain of laminin: their ability to promote cell adhesion and neurite outgrowth, andinteract with heparin and the beta 1 integrin subunit. Journal of Cell Biology 115:1137-1148.

52

modulation of DNA binding by the acidic C-terminal domain. Nucleic Acids Research 22:1044-1051.

Sundell CL and Singer RH (1990) Actin mRNA localizes in the absence of protein synthesis. Journal of CellBiology 111:2397-2403.

Sundell CL and Singer RH (1991) Requirement of microfilaments in sorting of actin messenger RNA. Science253:1275-1277.

Szabat E and Rauvala H (1996) Role of HB-GAM (heparin-binding growth-associated molecule) inproliferation arrest in cells of the developing rat limb and its expression in the differentiatingneuromuscular system. Developmental Biology (Orlando) 178:77-89.

Takizawa PA, Sil A, Swedlow JR, Herskowitz I and Vale RD (1997) Actin-dependent localization of an RNAencoding a cell-fate determinant in yeast. Nature 389:90-93.

Taneja KL, Lifshitz LM, Fay FS and Singer RH (1992) Poly(A) RNA codistribution with microfilaments:evaluation by in situ hybridization and quantitative digital imaging microscopy. Journal of Cell Biology119:1245-1260.

Tezuka K, Takeshita S, Hakeda Y, Kumegawa M, Kikuno R and Hashimoto-Gotoh T (1990) Isolation ofmouse and human cDNA clones encoding a protein expressed specifically in osteoblasts and brain tissues.Biochemical and Biophysical Research Communications 173:246-251.

Thomas KL, Laroche S, Errington ML, Bliss TV and Hunt SP (1994) Spatial and temporal changes insignal transduction pathways during LTP. Neuron 13:737-745.

Tomomura M, Kadomatsu K, Nakamoto M, Muramatsu H, Kondoh, Imagawa K and Muramatsu T (1990)A retinoic acid responsive gene, MK, produces a secreted protein with heparin binding activity.Biochemical and Biophysical Research Communications 171:603-609.

Tongiorgi E, Righi M v Cattaneo A (1997) Activity-dependent dendritic targeting of BDNF and TrkBmRNAs in hippocampal neurons. Journal of Neuroscience 17:9492-9505.

Torre ER and Steward O (1996) Protein synthesis within dendrites: glycosylation of newly synthesizedproteins in dendrites of hippocampal neurons in culture. Journal of Neuroscience 16:5967-5978.

Travers AA, Ner SS and Churchill ME (1994) DNA chaperones: a solution to a persistence problem?. Cell77:167-169.

Uhm JH, Gladson CL and Rao JS (1999) The role of integrins in the malignant phenotype of gliomas.Frontiers in Bioscience 4:188-199.

Von Baer KE (1837) Uber Entwicklungsgesichte der Tiere: Beobachtung und Reflexion. In: Konigsberg ed.

Sparatore B, Passalacqua M, Patrone M, Melloni E and Pontremoli S (1996) Extracellular high-mobilitygroup 1 protein is essential for murine erythroleukaemia cell differentiation. Biochemical Journal320:253-256.

St Johnston D, Beuchle D and Nusslein-Volhard C (1991) staufen, a gene required to localize maternalRNAs in the Drosophila egg. Cell 66:51-63.

St Johnston D (1995) The intracellular localization of messenger RNAs. Cell 81:161-170.

Steward O (1995) Targeting of mRNAs to subsynaptic microdomains in dendrites. Current Opinion in Neurobiol-ogy 5:55-61.

Stros M, Stokrova J and Thomas JO (1994) DNA looping by the HMG-box domains of HMG1 and

53

Van Beneden E (1883) Recherches sur la maturation de l´oeuf et la fécondation et la division cellulaire.Arch.Biol. 4:265-640.

Wagner RW, Matteucci MD, Lewis JG, Gutierrez AJ, Moulds and Froehler BC (1993) Antisense gene inhibitionby oligonucleotides containing C-5 propyne pyrimidines. Science 260:1510-1513.

Wang H, Bloom O, Zhang M, Vishnubhakat JM, Ombrellino M, Che J, Frazier, Yang H, Ivanova S,Borovikova L, Manogue KR, Faist E, Abraham E, Andersson J, Andersson U, Molina PE, Abumrad NN,Sama A and Tracey KJ (1999) HMG-1 as a late mediator of endotoxin lethality in mice. Science 285:248-251.

Watanabe H, Kislauskis EH, Mackay CA, Mason-Savas A and Marks SC, Jr. (1998) Actin mRNA isoformsare differentially sorted in normal osteoblasts and sorting is altered in osteoblasts from a skeletal mutationin the rat. Journal of Cell Science 111:1287-1292.

Wautier JL, Zoukourian C, Chappey O, Wautier MP, Guillausseau PJ, Cao R, Hori O, Stern D and SchmidtAM (1996) Receptor-mediated endothelial cell dysfunction in diabetic vasculopathy. Soluble receptor foradvanced glycation end products blocks hyperpermeability in diabetic rats. Journal of ClinicalInvestigation 97:238-243.

Weiler IJ and Greenough WT (1993) Metabotropic glutamate receptors trigger postsynaptic proteinsynthesis. Proceedings of the National Academy of Sciences of the United States of America 90:7168-7171.

Wensley CH, Stone DM, Baker H, Kauer JS, Margolis FL and Chikaraishi DM (1995) Olfactory markerprotein mRNA is found in axons of olfactory receptor neurons. Journal of Neuroscience 15:4827-4837.

Wilson JE, Connell JE, Schlenker JD and Macdonald PM (1996) Novel genetic screen for genes involved inposterior body patterning in Drosophila. Developmental Genetics 19:199-209.

Wu, L., Rechsteiner, M., and Kuehl, L. (1981). Comparative studies on microinjected high-mobility-groupchromosomal proteins, HMG1 and HMG2. J. Cell Biol. 91, 488-496.

Yan SD, Chen X, Fu J, Chen M, Zhu H, Roher A, Slattery, Zhao L, Nagashima M, Morser J, Migheli A,Nawroth P, Stern D and Schmidt AM (1996) RAGE and amyloid-beta peptide neurotoxicity in Alzheimer’sdisease [see comments]. Nature 382:685-691.

Yeh HJ, He YY, Xu J, Hsu CY and Deuel TF (1998) Upregulation of pleiotrophin gene expression indeveloping microvasculature, macrophages, and astrocytes after acute ischemic brain injury. Journal ofNeuroscience 18:3699-3707.

Zhan, H. L, Singer, R. H., and Bassell, G. J. (1999). Neurotrophin regulation of beta-actin mRNA and proteinlocalization within growth cones. Journal of Cell Biology 147, 59-77.