Le(X) and related structures as adhesion molecules.

Summary and conclusion Le^x (13 fucosylated type 2 chain) functions as an adhesion molecule capable of Ca²⁺-mediated homotypic binding. Cells with high surface expression of Le^x therefore exhibit strong self-aggregation (based on Le^x-Le^x interaction) in the presence of Ca²⁺. In this review, I have summarized several lines of supporting data for this concept, and the role of Le^x-Le^x interaction in the process of embryo compaction and autoaggregation of F9 teratocarcinoma cells. In general, cell adhesion events based on Le^x-Le^x interaction may be followed and reinforced by integrin- or Ig receptor-based adhesion systems.SLe^x, the 23 sialosyl derivative of Le^x, and its positional isomer SLe^a, have been identified as the target molecules for selectin-dependent cell adhesion. Adhesion of leukocytes or tumour cells to ECs or platelets, which express E-selectin and P-selectin respectively, is initiated by this process. The target epitopes SLe^x and SLe^a are presented mainly on transmembrane glycoproteins having many clusters of O-linked carbohydrate chains. Therefore, inhibition of O-glycosylation may be effective for blocking selectin-mediated cell adhesion. The abundant presence of Le^x epitope in the central nervous system, and the physiological changes of Le^x expression as described in this monograph, reflect the adhesive properties of this molecule and its sialyosylated and/or fucosylated derivatives.     

Lymphocytes express Ia antigens of foreign haplotype following treatment with neuraminidase

Evidence is presented which indicates that neuraminidase (NA) treatment of spleen cells both destroys old Ia antigens and reveals new Ia specificities which are not normally expressed by splenocytes. It was found that NA treatment unmasked alien I-A^k-like specificities on A.TH (I ^s ) spleen cells, and I^s-like antigens on A.TL (I ^k ) spleen cells. These conclusions were based on direct testing of NA-treated targets with a range of alloantisera and on cell-absorption experiments. Furthermore, the cellular distribution of NA-exposed antigens resembled that of convential Ia antigens, the new antigens being expressed on more than 90 percent of splenic B cells and a subpopulation of splenic T cells. However, although some of the antigens exposed by NA on A.TH cells appeared to resemble the Ia. 3 and 15 specificities, additional antigens were involved which did not correlate with any previously described Ia antigens.Sugar inhibition experiments demonstrated the NA-exposed antigens to be carbohydrate in nature, D-galactose being an effective inhibitor in these studies. The proportion of- and-linked D-galactose residues associated with the new antigens depended upon the target cell used and the anti-Ia serum tested. Furthermore, glycolipid extracts from lymphoid cells were shown to contain the NA-exposed antigens.Collectively, these results support the existence of carbohydrate-defined Ia antigens. The simplest interpretation of the findings is that NA clips off terminal sialic acid residues from carbohydrate-defined Ia antigens on the cell surface and exposes subterminal sugars which resemble antigens expressed by otherI-region haplotypes.     

Traveling for the glycosphingolipid path

Our studies on glycosphingolipids (GSLs) were initiated through isolation and structural characterization of lacto-series type 1 and 2 GSLs, and globo-series GSLs. Lacto-series structures included histo-blood group ABH and I/i antigens. Our subsequent studies were focused on GSL changes associated with: (i) ontogenic development and differentiation; (ii) oncogenic transformation and tumor progression. Various novel types of GSLs such as extended globo-series, sialyl-Le^x (SLe^x), sialyl-dimeric-Le^x (SLe^x-Le^x), dimeric-Le^x (Le^x-Le^x), Le^y-on-Le^x, dimeric-Le^a (Le^a-Le^a), Le^b-on-Le^a, etc. were identified as tumor-associated antigens. These studies provide an essential basis for up- or down-regulation of key glycosyltransferase genes controlling development, differentiation, and oncogenesis. GSL structures established in our laboratory are summarized in Table 1, and structural changes of GSLs associated with ontogenesis and oncogenesis are summarized in Sections 2 and 3.Based on these results, we endeavored to find out the cell biological significance of GSL changes, focused on (i) cell adhesion, e.g., the compaction process of preimplantation embryo in which Le^x-to-Le^x, Gb4-to-GalGb4 or -nLc₄ play major roles; and (ii) modulation of signal transduction through interaction of growth factor receptor tyrosine kinase with ganglioside, e.g., EGF receptor tyrosine kinase with GM3. Recent trends of studies on i and ii lead to the concept that GSL clusters (microdomains) are organized with various signal transducer molecules to form glycosignaling domains (GSD). GSL-dependent adhesion occurs through clustered GSLs, and is coupled with activation of signal transducers (cSrc, Src family kinase, Rho A, etc.). Clustered GSLs involved in cell adhesion are recognized by GSLs on counterpart cells (carbohydrate-to-carbohydrate interaction), or by lectins (e.g., siglecs, selectins).Our major effort in utilization of GSLs in medical science has been for: (i) cancer diagnosis and treatment (vaccine development) based on tumor-associated GSLs and glycoepitopes; (ii) genetically defined phenotype for susceptibility to E. coli infection; (iii) clear identification of physiological E-selectin epitope (myeloglycan) expressed on neutrophils and myelocytes; (iv) characterization of sialyl poly-LacNAc epitopes recognized as male-specific antigens. Utilization of these GSLs or glycoepitopes in development of anti-adhesion approach to prevent tumor metastasis, infection, inflammation, or fertilization (i.e., contraceptive) is discussed. For each approach, development of mimetics of key GSLs or glycoepitopes is an important subject of future study.     

A Role for Lewis a antigens on salivary agglutinin in binding to Streptococcus mutans

Streptococcus mutans is a major etiological agent in dental caries. Salivary agglutinin is one of the main salivary components binding to S.mutans. To learn more about the interaction of salivary agglutinin with S.mutans, parotid, submandibular, sublingual and palatal saliva samples were incubated with S. mutans suspension. Both depleted saliva samples and bacterial extracts were analyzed by SDS-PAGE and immunoblotting. Salivary agglutinin was present in all types of glandular saliva and in all cases bound to S.mutans, also to PC337C, a P1^– mutant of S.mutans. Agglutinin was separated by SDS-PAGE under reducing and non-reducing conditions and then transferred to nitrocellulose. Non-reduced agglutinin bound S.mutans, but reduced agglutinin did not. Adhesion of S.mutans to agglutinin-coated microplates was inhibited by amine-containing components, 1 M NaCl or KCl and EDTA. Adhesion decreased with decreasing pH with no adhesion below pH 5.0. These data suggest that calcium-dependent electrostatic interactions play a role in binding. By immunoblotting was demonstrated that blood group antigens and Lewis antigens were present on agglutinin. Synthetic blood group antigens and Lewis antigens covalently coupled to polyacrylamide were tested for binding to S.mutans. Only Le^a(Gal1,3(Fuc1,4)GlcNAc) bound to S.mutans, whereas the blood group antigens Le^b, Le^x, Le^y, H1, H2, A, B and sialylated Le^a did not. Le^a without galactose (Fuc1,4GlcNAc) still bound to S. mutans, but Le^a without fucose (Gal1,3GlcNAc) did not. Binding of agglutinin to S. mutans was not inhibited by Le^a. In conclusion, S. mutans can bind to Le^a carbohydrate epitopes in which the fucose is an essential residue. Le^a carbohydrate epitopes are present on salivary agglutinin but play no major role in binding.     

Consequences of the expression of sialylated antigens in breast cancer

Changes in cell surface glycosylation are common modifications that occur during oncogenesis, leading to the over-expression of tumour-associated carbohydrate antigens (TACA). Most of these antigens are sialylated and the increase of sialylation is a well-known feature of transformed cells. In breast cancer, expression of TACA such as sialyl-Lewis^x or sialyl-Tn is usually associated with a poor prognosis and a decreased overall survival of patients. However, the specific role of these sialylated antigens in breast tumour development and aggressiveness is not clearly understood. These glycosylation changes result from the modification of the expression of genes encoding specific glycosyltransferases involved in glycan biosynthesis and the level of expression of sialyltransferase genes has been proposed to be a prognostic marker for the follow-up of breast cancer patients. Several human cellular models have been developed in order to explain the mechanisms by which carbohydrate antigens can reinforce breast cancer progression and aggressiveness. TACA expression is associated with changes in cell adhesion, migration, proliferation and tumour growth. In addition, recent data on glycolipid biosynthesis indicate an important role of G_D3 synthase expression in breast cancer progression. The aim of this review is to summarize our current knowledge of sialylation changes that occur in breast cancer and to describe the cellular models developed to analyze the consequences of these changes on disease progression and aggressiveness.     

The molecular nature of Fucus serratus sperm surface antigens recognised by monoclonal antibodies FS1 to FS12

Sperm of the brown alga Fucus serratus are highly differentiated, biflagellate, naked cells. Immunolocalisation studies, employing monoclonal antibodies (MAbs — designated FS1 to FS12) raised against antigens of these sperm cells, have revealed that some sperm surface components are distributed over the entire cell, whereas others are restricted to, or occur preferentially on, the surface of the anterior flagellum or cell body. This report describes the use of these MAbs in Western-blot procedures and antigen-modification binding assays to determine the nature of these sperm surface components. Monoclonal antibodies which bind to antigens found on the cell body and both flagella (FS3, FS4, FS6, FS8, FS10) recognise carbohydrate epitopes of a high-molecular-weight glycoprotein (M_r=205 kDa). These MAbs were initially chosen at random from a much larger number of antibodies which bound to sperm in a similar fashion, indicating that this glycoprotein is an immunodominant antigen. Though these MAbs compete under conditions of limited antigen availability, differences in the effects of periodate on antibody binding and differences in other binding data indicate that the MAbs recognise epitopes of this glycoprotein which are neighbouring or overlapping, rather than common. The MAb FS9, which has a similar binding pattern to the above antibodies, also seems to bind to carbohydrate epitopes, but the antigen recognised by this antibody could not be identified in Western-blotting procedures. The MAbs FS7 and FS12, which bind to the mastigonemes on the anterior flagellum and to the cell body and posterior flagellum, recognise a set of glycoproteins in the molecular-weight range 40–250 kDa. The evidence indicates that the antibodies are binding to N-linked carbohydrate side chains of these glycoproteins. Three MAbs that bind to the anterior flagellum (FS2, FS5 and FS11) recognise protein antigens in the molecular-weight range 90–250 kDa; it is not known whether these antigens are glycosylated. The MAb FS1, which binds primarily to the sperm cell body, could not be used in enzyme-linked immunosorbent assays or Western-blotting procedures and the antigen recognised by this antibody is so far uncharacterised.Abbreviations ELISA enzyme linked immunosorbent assay - HRP-RAMIG horseradish-peroxidase-labelled rabbit anti mouse immunoglobulin - Ig immunoglobulin - kDa kilodalton - MAb monoclonal antibody - M_r relative molecular mass - PBS phosphate-buffered saline - SDS-PAGE sodium dodecyl sulphate polyacrylamide gel electrophoresis We are grateful to AFRC for financial support under the cell signalling initiative.     

Time course study of H-2 and other antigen expression by hybrids of a myeloma cell line with inflammatory macrophages

The hybrids (the CANS lines) between inflammatory macrophages from C57BL/6N (B6) mice (H-2^b) and BALB/c mouse (H-2^d)-derived myeloma cell line NS1 in the early period after cell fusion showed no macrophage functions. However, most of the hybrids expressed these functions after prolonged cultivation accompanied with chromosome loss. In contrast, the hybrids initially displaying myeloma functions ( light chain production) lost this function when they exhibited macrophage functions. We studied the expression of cell-surface antigens in these hybrids and found that hybrids in the early period after cell fusion codominantly expressed both parental cell H-2 antigens (H-2K^b, H-2K^d, and H-2D^d) but not the H-2D^b antigen. On the other hand, aged hybrids strongly expressed the H-2 d antigen but lacked the H-2K^b antigen. Alternatively, these aged hybrids with macrophage functions expressed antigen(s) as detected with antiaged CANS-196 cell sera and asialo GM1 antigen, both of which were thought to be found exclusively on macrophages. Thus, the expression of cell-surface antigens in these hybrids was greatly altered after cell fusion.     

The roles of cell adhesion molecules in tumor suppression and cell migration: A new paradox

In addition to mediating cell adhesion, many cell adhesion molecules act as tumor suppressors. These proteins are capable of restricting cell growth mainly through contact inhibition. Alterations of these cell adhesion molecules are a common event in cancer. The resulting loss of cell-cell and/or cell-extracellular matrix adhesion promotes cell growth as well as tumor dissemination. Therefore, it is conventionally accepted that cell adhesion molecules that function as tumor suppressors are also involved in limiting tumor cell migration. Paradoxically, in 2005, we identified an immunoglobulin superfamily cell adhesion molecule hepaCAM that is able to suppress cancer cell growth and yet induce migration. Almost concurrently, CEACAM1 was verified to co-function as a tumor suppressor and invasion promoter. To date, the reason and mechanism responsible for this exceptional phenomenon remain unclear. Nevertheless, the emergence of these intriguing cell adhesion molecules with conflicting roles may open a new chapter to the biological significance of cell adhesion molecules.Key words: hepaCAM, cell adhesion molecules, tumor suppressor, migration, E-cadherin, CADM1, integrin α7, CEACAM1It is well known that many cell adhesion molecules function as tumor suppressors (reviewed in ref. ¹). These molecules exert their tumor suppressive effect mainly through cell-adhesion-mediated contact inhibition. Cell adhesion molecules allow cells to communicate with one another or to the extracellular environment by mediating cell-cell or cell-extracellular matrix (ECM) interactions (reviewed in refs. ² and ³). Broadly, these proteins can be classified into five families including immunoglobulin superfamily, integrins, cadherins, selectins and CD44. Apart from participating in the development and maintenance of tissue architecture, cell adhesion molecules serve as cell surface receptors critical for capturing, integrating and transmitting signals from the extracellular milieu to the cell interior (reviewed in refs. ² and ³). These signaling events are vital for the regulation of a wide variety of cellular functions including embryogenesis, immune and inflammatory responses, tissue repair, cell migration, differentiation, proliferation and apoptosis. Alterations of these cell adhesion molecules are a common event in cancer (reviewed in refs. ^{1, 2, 4} and ⁵). The disrupted cell-cell or cell-ECM adhesion significantly contributes to uncontrolled cell proliferation and progressive distortion of normal tissue architecture. More importantly, changes in cell adhesion molecules play a causal role in tumor dissemination. Loss of cell adhesion contacts allows malignant cells to detach and to escape from the primary mass. Gaining a more motile and invasive phenotype, these cells break down the ECM and eventually invade and metastasize to distal organs.Based on the above understanding, it is conventionally accepted that cell adhesion molecules with tumor suppressor activity, when expressed in cancer cells, are able to exert inhibitory effect on cell motility. The ability of cells in migration/motility is a prerequisite for cancer invasion and metastasis (reviewed in refs. ¹ and ⁵). Indeed, a number of cell adhesion molecule-tumor suppressors have been reported to be capable of reducing cell migration. The most classical example is E-cadherin, a calcium-dependent cell adhesion molecule. E-cadherin is expressed exclusively in epithelial cells and its expression is commonly suppressed in tumors of epithelial origins. The cytoplasmic domain of E-cadherin interacts with catenins to establish an intracellular linkage with the actin cytoskeleton (reviewed in ref. ⁶). The assembly of E-cadherin with the cytoskeleton via catenins at the sites of adherens junctions is important for the stabilization of cell-cell adhesions. Disruption of E-cadherin-mediated cell-cell adhesion, due to loss of expression or function of E-cadherin and/or catenins, is assocated with tumor development and progression (reviewed in ref. ⁷). Forced expression of E-cadherin in several cancer cell lines not only slows down cell growth^8,9 but also significantly reduces the invasiveness of the cells.^10,11 On the other hand, inhibition of E-cadherin by function-blocking antibodies and antisense RNA restores the invasiveness in non-invasive transformed cells.¹¹ Furthermore, using a transgenic mouse model of pancreatic beta-cell carcinogenesis, it has been demonstrated that E-cadherin-mediated cell adhesion is important in preventing the transition from well differentiated adenoma to invasive carcinoma.¹²Cell adhesion molecule 1 (CADM1), another example, has also been implicated in cancer progression. CADM1 is a member of the immunoglobulin superfamily and mediates cell-cell adhesion.¹³ The molecule associates with the actin cytoskeleton via the differentially expressed in adenocarcinoma of the lung (DAL1) protein; and the formation of CADM1-DAL1 complex is dependent on the integrity of actin cytoskeleton.¹⁴ Inactivation of the CADM1 and/or DAL1 gene usually through methylation has been reported in diverse human cancers.^15,16 A paper by Ito et al. showed that restoration of CADM1 expression in esophageal squamous cell carcinoma cells not only suppresses cell growth, but also retards cell motility and invasion.¹⁶In contrast to E-cadherin and CADM1, integrin α7 is a cell-ECM adhesion molecule which also possesses tumor suppressor activity. Ren et al. showed that integrin α7 gene is mutated in several human malignances; and the mutations are associated with an increase in cancer recurrence.¹⁷ Forced expression of integrin α7 in integrin α7-deficient leiomyosarcoma cells results in decreased colony formation and slower cell motility. Conversely, knockdown of integrin α7 in lung cancer cells expressing wild-type integrin α7 increases the colony number and cell motility rate. In addition, the researchers revealed that mice bearing xenograft tumors overexpressing integrin α7 have reduced tumor size with no obvious metastasis.In 2005, we first reported the identification of a cell adhesion molecule belonging to the immunoglobulin superfamily, designated as hepaCAM.¹⁸ To date, we have shown that the gene is frequently downregulated in a variety of human cancers.^18,19 Re-expression of hepaCAM in the hepatocellular carcinoma HepG2 cells¹⁸ and breast cancer MCF7 cells¹⁹ inhibits colony formation and retards cell proliferation. In addition, expression of hepaCAM in MCF7 cells results in cell cycle arrest at the G₂/M phase and cellular senescence. Concurrently, the expression of several senescence-associated proteins including p53, p21 and p27 is enhanced. Moreover, downregulation of p53 by p53-specific small interfering RNA in cells expressing hepaCAM clearly reduces p21 without changing p27 and alleviates senescence, indicating that hepaCAM induces senescence through a p53/p21-dependent pathway.¹⁹ Together, the data suggest that hepaCAM is a tumor suppressor. Interestingly, the expression of hepaCAM in both HepG2 and MCF7 cells stimulates both cell-ECM adhesion and cell migration.^18,20,21 The function of hepaCAM as a tumor suppressor in cell migration is contradictory to other cell adhesion molecule-tumor suppressors. Noteworthily, hepaCAM-mediated cell motility is evidenced by its direct interaction with the actin cytoskeleton.²¹Evidences are currently emerging to support the contradictory roles of cell adhesion molecules that both inhibit cell growth and promote cell motility when restored in cancer cells. In addition to hepaCAM, the immunoglobulin superfamily carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) is implicated to function as a tumor suppressor and a metastasis promoter. The characteristics and functions of CEACAM1 have been demonstrated in individual reports. CEACAM1 is frequently downregulated or dysregulated in multiple human tumors,^22–25 and is capable of suppressing cell growth and inducing apoptosis.^26–28 Ebrahimnejad et al. demonstrated that exogenous expression of CEACAM1 enhances melanoma cell invasion and migration; and this enhanced motility can be reverted by anti-CEACAM antibodies.²⁹ The ability of CEACAM to co-stimulate tumor suppression and invasion was finally established by Liu et al. in restricting thyroid cancer growth but promoting invasiveness.³⁰ Introduction of CEACAM1 into CEACAM1-deficient thyroid cancer cells results in G₁/S phase cell cycle arrest accompanied by elevated p21 expression and diminished Rb phosphorylation. Overexpression of CEACAM1 also increases cell-ECM adhesion and promotes cell migration and tumor invasiveness. In xenografted mice, CEACAM1 expression results in reduced tumor growth but increased tumor invasiveness. Conversely, silencing of endogenous CEACAM1 accelerates tumor growth and suppresses invasiveness.³⁰It is an exciting issue to address why a cell adhesion molecule is able to suppress tumor growth yet promote tumor progression. Could there be a molecular switch that controls the functions of the gene between a tumor suppressor and a migration promoter in cancer or are the functions executed simultaneously? The expression level, the extracellular cues as well as the interacting partners of the cell adhesion molecules may likely play a critical role in regulating its functions. The question is under what circumstances these factors come into play. To answer all these questions, and maybe more, on the intriguing findings of these proteins, more extensive and intensive experimentation is required. Nevertheless, it is obvious that the emergence of these cell adhesion molecules that function in a contradictory manner opens a new chapter to the biological significance of cell adhesion molecules.     

Cdk5: A regulator of epithelial cell adhesion and migration

Cell adhesion is a fundamental property of epithelial cells required for anchoring, migration and survival. During cell migration, the formation and disruption of adhesion sites is stringently regulated by integration of multiple, sequential signals acting in distinct regions of the cell. Recent findings implicate cyclin dependent kinase 5 (Cdk5) in the signaling pathways that regulate cell adhesion and migration of a variety of cell types. Experiments with epithelial cell lines indicate that Cdk5 activity exerts its effects by limiting Src activity in regions where Rho activity is required for stress fiber contraction and by phosphorylating the talin head to stabilize nascent focal adhesions. Both pathways regulate cell migration by increasing adhesive strength.Key words: Cdk5, Src, Rho, stress fibers, epithelial cells, cell adhesion, cell migrationAnchoring of epithelial cells to their basement membrane is essential to maintain their morphology, normal physiological function and survival. Cells attach to extracellular matrix components by means of membrane-spanning integrins, which cluster and link to the actin cytoskeleton via components of focal adhesions. At focal adhesions, actin is bundled into stress fibers, multi-protein cellular contractile machines that strengthen attachment and provide traction during migration.¹ Stress fiber contraction is generated by myosin II, a hexamer containing one pair of each non-muscle heavy chains (NMHCs), essential light chains, and myosin regulatory light chains (MRLC). Myosin motor activity is regulated by phosphorylation of MRLC at Thr18/Ser19 and is required to generate tension on actin filaments and to maintain stress fibers.¹ Although a number of kinases have been identified which phosphorylate MRLC at Thr18/Ser19, the principal kinases in most cells are myosin light chain kinase (MLCK)² and Rho-kinase (ROCK),³ a downstream effector of the small GTPas, RhoA.Rho family small GTPases play a central role in regulating many aspects of cytoskeletal organization and contraction.⁴ These GTPases are subject to both positive regulation by guanine nucleotide exchange factors (GEFs), such as GEF-H1,^5,6 and negative regulation by GTPase-activating proteins (GAPs), such as p190RhoGAP.⁷ As cells spread, the Rho-family GTPase, Cdc42, is activated at the cell periphery, leading to the formation of numerous filapodia. Focal adhesion formation is first seen at the tips of these filapodia as focal adhesion proteins such as talin and focal adhesion kinase (FAK) bind to the intracellular domains of localized integrins.⁸ Src is recruited to activated FAK at the nascent focal adhesion and generates binding sites for additional focal adhesion proteins by phosphorylating FAK and paxillin.⁹ Src activity is essential for the further maturation of the focal adhesion and for activating the Rho-family GTPase, Rac, leading to Arp2/3-dependent actin polymerization, formation of a lamellipodium and extension of the cell boundary. Simultaneously, Src inhibits RhoA by phosphorylating and activating its upstream inhibitor, p190RhoGAP. As the focal adhesion matures, Src is deactivated, allowing the Rho activation necessary for mDia-dependent actin polymerization,¹⁰ myosin-dependent cytoskeletal contraction⁵ and tight adhesion to the extracellular matrix. Since new focal adhesions continually form at the distal boundary of the spreading cell, the most mature and highly contracted stress fibers are localized at the center of the cell.Cell adhesion is an essential component of cell migration: if adhesion is too weak, cells can not generate the traction necessary for migration; if it is too strong, they are unable to overcome the forces that anchor them in place. Thus, the relationship between adhesion force and migration rate is a bell-shaped curve.¹¹ Migration rate increases as adhesive strength increases until an optimum value is reached. Thereafter, increases in adhesive strength decrease migration rate. Since the strength of adhesion depends on extracellular matrix composition as well as the types of integrin expressed in the cell, a decrease in adhesive strength may result in either faster or slower cell migration.Several lines of evidence indicate that the proline-directed serine/threonine kinase cyclin dependent kinase 5 (Cdk5) plays an integral role in regulating cell adhesion and/or migration in epithelial cells.^12–17 Cdk5 is an atypical member of cyclin dependent kinase family, which is activated by the non-cyclin proteins, p35 or p39.¹⁸ Cdk5 is most abundant in neuronal cells where it also regulates migration and cytoskeletal dynamics.¹⁹ In neurons, Cdk5 exerts its effects on migration at least in part by phosphorylating FAK,¹⁹ and the LIS1 associated protein, NDEL1.²⁰ In contrast, recent findings have revealed two novel pathways involved in Cdk5-dependent regulation of migration in epithelial cells.^16,17One of these newly discovered mechanisms links Cdk5 activation to control of stress fiber contraction.¹⁶ We have found that Cdk5 and its activator, p35, co-localize with phosphorylated myosin regulatory light chain (MRLC) on centrally located stress fibers in spreading cells.¹⁶ Moreover, Cdk5 is strongly activated in spreading cells as central stress fiber contraction becomes pronounced.²¹ Since contraction of these central stress fibers is primarily responsible for tight attachment between the cell and the extracellular matrix,⁵ the above findings suggested that Cdk5 might regulate cell adhesion by regulating MRLC phosphorylation. To test this possibility we inhibited Cdk5 activity by several independent means and found that MRLC phosphorylation was likewise inhibited. In addition, we found that inhibiting Cdk5 either prevented the formation of central stress fibers or led to their dissolution. The concave cell boundaries characteristic of contracting cells were also lost.¹⁶ Since MRLC lacks a favorable site for phosphorylation by Cdk5, we asked whether Cdk5 might affect the upstream signaling pathways that regulate MRLC phosphorylation. Experiments with specific pathway inhibitors indicated that the MRLC phosphorylation involved in stress fiber contraction in lens epithelial cells was regulated largely by Rho-kinase (ROCK). Inhibiting Cdk5 activity not only significantly reduced ROCK activity, but also blocked activation of its upstream regulator, Rho. To explore the mechanism behind the Cdk5-dependent regulation of Rho, we turned our attention to p190RhoGAP, which appears to play a major role in regulating Rho-dependent stress fiber contraction.⁷ This RhoGAP must be phosphorylated by Src to be active; as a result, Rho activity is low in the early stages of cell spreading, when Src activity is high. At later times, Src activity falls, p190RhoGAP activity is lost, and Rho-GTP is formed, enabling Rho-dependent myosin phosphorylation and stress fiber contraction.^9,10 We have found that inhibiting Cdk5 activity during this later stage of cell spreading increases Src activity and Src-dependent phosphorylation of its substrate, p190RhoGAP. This in turn leads to decreased Rho activity accompanied by loss of Rho-dependent myosin phosphorylation, dissolution of central stress fibers, and loss of cell contraction (Fig. 1). Moreover, inhibiting Src protects cells from the loss of Rho activation and dissolution of central stress fibers produced by inhibiting Cdk5.¹⁶ Since the effects of Cdk5 on Rho-dependent cytoskeletal contraction appear to be mediated almost entirely through Cdk5-dependent regulation of Src, it will be particularly important to determine how Cdk5 limits Src activity.Open in a separate windowFigure 1Cdk5 inhibition reduces contraction of preformed stress fibers. (A) Cells were spread on fibronectin for 60 min to adhere, allowing them to form focal adhesion and stress fibers (pre-incubation) and then further incubated for 2 h in absence (control) or presence of Cdk5 inhibitor (olomoucine) and stained with phalloidin. The cells without olomoucine (control) had concave boundaries and well-formed stress fibers. Olomoucine treated cells showed loss of central stress fibers and failure to contract. Scale bar = 20 µ. (B) experimental conditions were same as shown in (A). Cdk5 inhibitor, olomoucine, was added after 1 h of spreading (indicated as t = 0) and cells were incubated for an additional 2h in absence or presence of olomoucine. Cell lysates were immunoblotted with antibodies for pMRLC (upper) and MRLC (middle). Tubulin was used as a loading control (lower). Lane 1: untreated (at 0 h); Lane 2: untreated (at 2 h); Lane 3: Cdk5 inhibitor (olomoucine) treated. (C) results of three independent experiments of the type shown in (B) were quantified by densitometry and normalized to determine the relative levels of pMRLC at each time. Statistical analysis demonstrated a significant (p < 0.05) decrease in pMRLC level in olomoucine treated cells compared to untreated cells.The central stress fibers regulated by Cdk5 play a central role in anchoring cells to the substratum, and their loss when Cdk5 is inhibited will reduce adhesion. As discussed above, reduction in cell adhesion may either increase or decrease the rate of cell migration, depending on the cell type and extracellular matrix composition. In lens and corneal epithelial cells, the reduction in adhesion produced by Cdk5 inhibition promotes cell migration.^13,15,16,22 Moreover, regulation of Rho/Rho-kinase signaling by Cdk5 seems to be a major factor in determining the migration rate, since inhibitors of Cdk5 and Rho-kinase increased lens epithelial cell migration rate equivalently and inhibiting both produced no additional effect.¹⁶Interestingly, an independent line of investigation has shown that this is not the only mechanism underlying Cdk5-dependent regulation of cell adhesion and migration. Cdk5 also localizes at focal contacts at the cell periphery and phosphorylates the focal adhesion protein talin.¹⁷ The talin phosphorylation site has been identified as S425, near the FERM domain in the talin head region. Upon focal adhesion disassembly, this region is separated from the talin rod domain by calpain-dependent cleavage.²³ Phosphorylation at S425 by Cdk5 blocks ubiquitylation and degradation of the talin head by inhibiting interaction with the E3 ligase, Smurf1. This leads, ultimately, to greater stability of lamellipodia and newly formed focal adhesions, thus strengthening adhesion to the substrate.¹⁷ Although the exact molecular events involved in this stabilization are not yet clear, it has been suggested that the talin head may “prime” integrins to bind full length talin.²⁴ One possible scenario describing how this might occur is shown in Figure 2. By permitting the isolated head region to escape degradation following calpain cleavage, Cdk5-dependent phosphorylation may stablize a pool of talin head domains to bind focal contacts within the lamellipodium. It is known that the isolated talin head region can bind and activate integrins during cell protrusion.²⁵ The resulting integrin activation would be expected to stabilize the lamellipodium by strengthening integrin-dependent adhesion. Since the head domain lacks sites for actin binding, which are located in the talin rod domain,²⁶ the bound head domain would have to be replaced by full length talin to enable focal adhesion attachment to the cytoskeleton.²⁵ The head domain might promote this replacement by recruiting the PIP-kinase needed to generate PI(4,5) P₂,²³ which facilitates binding of full length talin to integrin by exposing the auto-inhibited integrin binding sites.²⁷ The binding of full length talin and the resulting link between the integrins and the actin cytoskeleton would then further strengthen adhesion.²⁵ This model predicts that full length talin would bind poorly in the absence of Cdk5 activity, due to degradation of the talin head and the resulting limited availability of PI(4,5)P₂, and thus provides a possible explanation for the observed rapid turnover of peripheral focal adhesions.¹⁷ Clearly, other models may be proposed to explain the increase in adhesion produced by talin head phosphorylation, and deciding among them will be an active area for future investigation. Nonetheless, it is now certain that talin is a key substrate for Cdk5 at focal adhesions.Open in a separate windowFigure 2Mechanism of Cdk5-dependent regulation of cell adhesion and migration. Binding of p35 to Cdk5 forms the active Cdk5/p35 kinase, which regulates cell adhesion and migration in two distinct ways. Cdk5-dependent phosphorylation of the talin head domain at Ser425 prevents its ubiquitylation and degradation, allowing it to persist following calpain cleavage. The phosphorylated talin head may then bind to integrin at peripheral sites and recruit PIP-K, which converts PI(4)P to PI(4,5)P₂. PI(4,5)P₂ may promote replacement of the talin head by full length talin. Full length talin recruits other focal adhesion proteins to form the mature focal adhesion. The talin tail provides the site for the actin binding and polymerization. Polymerized actin is subsequently bundled into stress fibers. Cdk5/p35 also regulates the Rho-dependent myosin phosphorylation necessary for stress fiber stability and cytoskeletal contraction by limiting Src activity. This in turn decreases Src-dependent phosphorylation of p190RhoGAP, favoring Rho-GTP formation, Rho-dependent stress fiber polymerization, stabilization and contraction. Both pathways modulate cell migration by increasing adhesive strength.In summary, the presently available data indicate that Cdk5 has at least two distinct functions in cell adhesion (Fig. 2). On the one hand, it stabilizes peripheral focal adhesions and promotes their attachment to the cytoskeleton by phosphorylating the talin head. On the other hand, once the actin cytoskeleton has been organized into stress fibers, Cdk5 enhances the Rho activation essential for stability and contraction of central stress fibers by limiting Src activity. The discovery that Cdk5 is involved in two separate events required for efficient migration, suggests that it may coordinate multiple signaling pathways. The known involvement of Cdk5 and its activator, p35, in regulating microtubule stability suggests yet another mechanism by which Cdk5 activity may regulate cytoskeletal function. Microtubules are closely associated with stress fibers²⁸ and their depolymerization has been shown to release the Rho activating protein, GEF-H1, leading to Rho activation and Rho-dependent myosin contraction.⁶ Since cell adhesion and migration play an important role in the progression of many pathological conditions, Cdk5, its substrates and its downstream effectors involved in cell adhesion may provide novel targets for therapeutic intervention.^15,29     

Significance of microtubule catastrophes at focal adhesion sites

Directional cell migration requires cell polarization and asymmetric distribution of cell signaling. Focal adhesions and microtubules are two systems which are essential for these. It was shown that these two systems closely interact with each other. It is known that microtubule targeting stimulates focal adhesion dissociation. Our recent study shows that focal adhesions, in turn, specifically induce microtubule catastrophe via a biochemical mechanism. We were able to track down one of the focal adhesion proteins paxillin which is involved in this process. Paxillin phosphorylation was previously shown to be the key component in the regulation of focal adhesion assembly or disassembly. Since microtubule catastrophe dynamic differs at the leading edge and cell rear, similar to paxillin phosphorylation levels, we suggest a model connecting asymmetric distribution of focal adhesions and asymmetric distribution of microtubule catastrophes at adhesion sites as a feedback loop.Key words: microtubule catastrophe, focal adhesion, microtubules, paxillin, cell motilityCell migration is important for many biological processes. It requires organized asymmetric dynamics of focal adhesions (FAs), sites where cells interact with extra cellular matrix. FAs appear at the leading edge as small transient dot-like structures termed focal complexes (FXs).^1,2 FX assembly and disassembly is regulated by phosphorilation status of paxillin a major FX protein.^3,4 Most of FXs form and disassemble rapidly. However, some adhesions mature in a force-dependent manner, into larger late adhesions. This process, involves both an increase in size and change in molecular composition^3,5 and is accompanied by a reduction in local paxillin phosphorylation.⁴ Late adhesions are more stable, immobile and undergo forced disassembly by multiple microtubule targeting events⁶ only underneath the approaching cell body or transform into fibrillar adhesions by a Src-dependent mechanism.⁷Similarly to the leading edge, proper adhesion patterns at the cell rear are also essential. Most trailing adhesions are initiated in protrusions at the rear and flanks of the cell as FX rapidly mature in response to tension and transform into sliding trailing adhesions.⁸ The process of sliding is complex. While adhesion proteins coupled with the actin cytoskeleton can be translocated relative to substratum, those that are associated with the membrane are thought to undergo treadmilling within the adhesion site.^9,10 Treadmilling, which includes disassembly of adhesion proteins at the distal end and reassembly at the proximal end,¹⁰ is accompanied by fusion with new adhesions formed in front of the sliding one.⁶ Thus, despite a protein composition similar to late adhesions, sliding adhesions are more dynamic. Not surprisingly, sliding adhesions have high paxillin phosphorylation at the distal end of the adhesion site, indicating very dynamic assembly/disassembly rates.⁴Several mechanisms have been proposed for the regulation of adhesion turnover (reviewed in ref. ¹¹). However, these have not accounted for the observed asymmetry of adhesion turnover. Understanding this requires examining the connection with another asymmetric intracellular system, the microtubule network. This dynamic network closely interacts with FAs. Microtubules play an essential role in cell migration and polarized distribution of signals within the cell. Multiple microtubule targeting to FA leads to their disassembly both at the leading edge and at the cell rear.⁶Unlike microtubule growth in other cell regions, growth at its leading edge is persistent, characterized by short periods of shrinkage.⁸ Simultaneous observation of microtubules and FAs show that microtubules specifically target adhesion sites.¹² More detailed analysis of microtubule dynamics reveals that FAs are preferable sites for microtubule catastrophes.¹³ Although FAs cover only about 5% of cell area more than 40% of catastrophes occur at these sites. The likelihood of microtubule catastrophe is seven times higher when a microtubule grows through a FA rather than through an adhesion-free area¹³ and about 90% of microtubules approaching adhesion sites undergo catastrophe. Although most of the catastrophes occur at late adhesions, due to their increased stability and lifespan, there is no difference in efficiency of catastrophe induction between small focal complexes and large rigid late adhesions.¹³ As FX do not have dense adhesion or actin plaque, it is likely that microtubule catastrophe is triggered by a biochemical mechanism rather than mechanical rigidity. This is also supported by the fact that mechanical obstacles in a cell do not necessarily cause microtubule catastrophe.¹³At the cell rear, microtubule dynamics differ from those at the leading edge. Microtubules spend less time in a growing phase and more time in pauses and shrinkage.⁸ Polymerization and depolymerization occur within a very limited area close to the cell edge.⁸ Live-cell imaging of cells expressing both microtubule and focal adhesion markers show that this complex dynamic sequence often happens within a single sliding adhesion. Microtubules that are captured at the proximal end of adhesion undergo multiple repetitive catastrophes at the distal end (Fig. 1) accompanied by rescue at the capture site. Thus, the capture mechanism significantly increases the lifetime of a microtubule and ensures that repetitive catastrophes occur at the single adhesion. This scenario leads to high catastrophe frequency at the cell rear, resulting in intensive catastrophe-dependent regulation in this cell region.Open in a separate windowFigure 1Multiple microtubule catastrophes at the sliding adhesion. (A) Frame from TIRF video sequence of a fish fibroblast cell (CAR) co-transfected with GFP-tubulin (green) to visualize microtubules and Cherry-Zyxin (red) to mark focal adhesions. The boxed region is presented in the kymograph in (B). Bar, 10 µm. (B) Kymograph of microtubule dynamics at a trailing end focal adhesion. Top panel shows microtubule (MT) only. Bottom panel shows life history plot of MT (green line shows movement of MT end) in relation to focal adhesions (red). Arrows show catastrophes at the distal end of adhesion, arrowheads show capture at the proximal end of adhesion.Detailed analysis of microtubule catastrophe localization shows that they occur at the areas of FAs where paxillin is enriched and highly phosphorylated.^4,13 Paxillin was shown to interact with microtubules through its Lim2/Lim3 domain.¹⁴ Purified GST-Lim2/Lim3 fragment injected into the cell localizes to FAs, displacing endogenous paxillin.¹³ This leads to a 40% decrease in the number of microtubule catastrophe events at adhesion sites,¹³ indicating that paxillin is needed for catastrophe initiation.In summary, we conclude that microtubule catastrophes at focal adhesions are specific events that are triggered by a biochemical mechanism. This process involves the focal adhesion protein paxillin, which may serve as a docking site for microtubules and/or microtubule catastrophe factors. The nature of catastrophe factors remains to be clarified. Possible mechanisms include molecules which induce microtubule catastrophe directly, such as stathmin,¹⁵ or molecules which regulate catastrophe-inducing factors activity. Alternatively, catastrophe factors at adhesion sites could act by removing stabilizing factors from microtubule tips. Thus, allowing already active catastrophe-inducing molecules such as kinesin-13 family member MCAK^16,17 to complete their function. Furthermore, microtubule catastrophe at paxillin-enriched areas, followed by release of microtubule-associated factors, may be involved in paxillin phosphorylation. This local regulation of adhesion disassembly would close the feed-back loop to microtubule regulation of FA turnover.In this model, asymmetric distribution of microtubule catastrophes is tightly linked to asymmetric regulation of FA. Since asymmetric FA dynamics in a cell are critical for organization of the actin cytoskeleton, tensile force distribution and directional cell migration, we conclude that microtubule catastrophes serve as important regulatory events for asymmetric signaling and dynamics of the whole cell (Fig. 2).Open in a separate windowFigure 2Model for asymmetric focal adhesion and microtubule dynamics. Focal complexes at the leading edge either disassemble or mature in response to tension. Microtubules undergo catastrophe both at focal complexes and late adhesions. Late adhesions disassemble in response to multiple microtubule targeting. At the cell rear a microtubule is captured at the proximal end of sliding adhesion and undergoes multiple catastrophes at its distal end, supporting disassembly of this region.     

Gas vesicle formation and buoyancy regulation in Pelodictyon phaeoclathratiforme (Green sulfur bacteria)

Gas vesicle formation and buoyancy regulation in Pelodictyon phaeoclathratiforme strain BU1 (Green sulfur bacteria) was investigated under various laboratory conditions. Cells formed gas vesicles exclusively at light intensities below 5 mol · m^-2 · s^-1 in the stationary phase. No effect of incubation temperature or nutrient limitation was observed. Gas space of gas vesicles occupied always less than 1.2% of the total cell volume. A maximum cell turgor pressure of 330 kPa was determined which is comparable to values determined for cyanobacterial species. Since a pressure of at least 485 kPa was required to collapse the weakest gas vesicles in Pelodictyon phaeoclathratiforme, short-term regulation of cell density by the turgor pressure mechanism can be excluded.Instead, regulation of the cell density is accomplished by the cease of gas vacuole production and accumulation of carbohydrate at high light intensity. The carbohydrate content of exponentially growing cells increased with light intensity, reaching a maximum of 35% of dry cell mass above 10 mol · m^-2 · s^-1. Density of the cells increased concomitantly. At maximum density, protein and carbohydrate together accounted for 62% of the total cell ballast. Cells harvested in the stationary phase had a significantly lower carbohydrate content (8–12% of the dry cell mass) and cell density (1010–1014 kg · m^-3 with gas vesicles collapsed) which in this case was independent of light intensity. Due to the presence of gas vesicles in these cultures, the density of cells reached a minimum value of 998.5 kg · m^-3 at 0.5 mol · m^-2 · s^-1.The cell volume during the stationary phase was three times higher than during exponential growth, leading to considerable changes in the buoyancy of Pelodictyon phaeoclathratiforme. Microscopic observations indicate that extracellular slime layers may contribute to these variations of cell volume.     

Tissue-specific expression of LeY antigen in high endothelial venules of human lymphoid tissues

In this study, we demonstrated that the anti-Le^Yantibody (BM-1) especially reacted with high endothelial venules (HEVs) in peripheral lymph nodes of blood group O individuals. The Le^Yexpression on HEVs showed a unique tissue-specific pattern, i.e., a large amount of the Le^Yexpression in peripheral lymph nodes and no or small amounts in mesenteric lymph node. Statistical analysis showed that there was the significant difference between the percentage of Le^Y-positive HEVs in peripheral lymph nodes and mesenteric lymph nodes. No expression of Le^Ywas observed in vessels of Payer's patch, thymus, spleen and other non-lymphoid organs. In blood group A or B individuals, the reactivity between HEVs and anti-Le^Yantibody increased after enzyme digestion with -N-acetylgalactosaminidase or -galactosidase. These findings show that the expression of difucosylated blood group ABH antigens are especially expressed on HEVs in peripheral lymph nodes. Furthermore, the tissue-specific pattern suggests that these antigens may be related to intercellular adhesion between lymphocytes and HEVs.     

In vitro differentiation and antigenic changes in human melanoma cell lines

Summary Malignant transformation of melanocytes may be associated with changes in the expression of HLA antigens and melanoma-associated antigens (MAA). To determine whether these changes reflect the differential expression of HLA antigens and MAA by melanocytes at different stages of differentiation, we have studied the effect of the reversible induction of differentiation by fibroblast interferon (interferon ) and/or 12-O-tetradecanoyl-phorbol 13-acetate (TPA) on the expression of HLA antigens and MAA by the melanoma cell lines DU-2, FO-1 and HO-1. The three melanoma cell lines differed in their sensitivity to the differentiating and antiproliferative activity of these two compounds and displayed an increased growth suppression and induction of differentiation, when incubated with the combination of TPA and interferon . Incubation of the three melanoma cell lines with interferon , TPA or their combination resulted in a differential modulation of the expression of membrane-bound high-molecular-mass melanoma-associated antigen, 115-kDa MAA, 100-kDa MAA, intercellular adhesion molecule 1, HLA class I antigens and gene products of the HLA-D region. Each melanoma cell line displayed a unique pattern of antigenic modulation when exposed to the two differentiating agents alone or in combination. No direct relationship was found between the effects of interferon and/or TPA on the growth and differentiation of the three melanoma cell lines and the expression of HLA antigens or the MAA evaluated in the present study. These findings argue against a direct role of any of the antigens tested in the reversible induction of human melanoma cell differentiation in the in vitro system.     

Cellular interaction through LewisX cluster: theoretical studies

It is well known that cell surface carbohydrates play a role in cell–cell adhesion and communication. LewisX glycosphingolipids form microdomains on cell surfaces. Homotypic and calcium-mediated LewisX–LewisX (LeX-LeX) interactions were proposed to be responsible for the initial steps of cell adhesion, and to mediate embryogenesis and metastasis. Various techniques have been used to investigate such interactions, but little information is available on the geometry and the mechanism of dimerisation. To better understand these interactions, a new molecular model was developed to simulate homotypic interactions in explicit solvent with and without calcium ions. Accurate analysis of both trajectories yielded valuable information about the energetics of LeX-LeX dimerisation. Detailed interpretation of the hydrogen bond network and the presence of calcium ions along the trajectory provide valuable insights into the role of calcium ions in this carbohydrate–carbohydrate interaction. Figure Calcium population density around the LewisX carbohydrate (after the trajectory has been fitted to the primary unit cell). All central dimer coordinates are fitted along the time axis, whereas calcium ion positions are recorded and represented as points. The clouds of points indicate that the ions are not randomly placed around the dimer but take up preferred positions     

Inhibition of human HT-29 colon carcinoma cell adhesion by a 4-fluoro-glucosamine analogue

Cell surface glycoconjugates play an important role in cellular recognition and adhesion. Modification of these structures in tumour cells could affect tumour cell growth and behaviour, including metastasis. 2-Acetamido-1,3,6-tri-O-acetyl-4-deoxy-4-fluoro--D-glycopyranose (4-F-GlcNAc) was synthesized as a potential inhibitor and/or modifier of tumour cell glycoconjugates. The effect of this sugar analogue on the adhesive properties of human colon carcinoma HT-29 cells was evaluated. Treatment of HT-29 cells with 4-F-GlcNAc led to reduced cell surface expression of terminal lactosamine, sialyl-Le^x and sialyl-Le^a, as determined by Western blotting and flow cytometry. The aberrant expression of these oligosaccharide structures on the HT-29 cell surface resulted in: (1) decreased E-selectin mediated adhesion of human colon cells to human umbilical cord endothelial cells (HUVEC); (2) impaired adhesion of HT-29 cells to -galactoside binding lectin, galectin-1; and (3) reduced ability to form homotypic aggregates. After exposure to 4-F-GlcNAc, lysosomal associated membrane proteins (lamp) 1 and 2, and carcinoembryonic antigen (CEA) detected in HT-29 cells were of lower molecular weight, probably due to impaired glycosylation. These results strongly suggest that modification of tumour cell surface molecules can alter tumour cell adhesion and that tumour cell surface oligosaccharides may be suitable targets for therapeutic exploitation.Abbreviations 4-F-GlcNAc 2-acetamido-1,3,6-tri-O-acetyl-4-deoxy-4-fluoro--glucopyranose - GlcNAc N-acetylglucosamine - s-Le^x sialyl-Lewis^x - s-Le^a sialyl-Lewis^a - lamp-1 and lamp-2 Lysosomal Associated Membrane Protein 1 and 2 - CEA carcinoembryonic antigen - DMEM Dulbecco's Modified Eagle Medium - PBS Phosphate Buffered Saline (2.7 mM KCl, 1.5 mM KH₂PO₄, 137 mM NaCl, 6.5 mM Na₂HPO₄, pH 7.3) - BSA Bovine Serum Albumin - PMSF Phenylmethylsulfonylfluoride - TBS Tris Buffered Saline (10 mM Tris, 20 mM NaCl, pH 7.3) - TCA Trichloroacetic Acid - DSA Datura stramonium agglutinin     

Immunogenic (tum−) variants obtained by mutagenesis of mouse mastocytoma P815

Mutagen treatment of mouse mastocytoma P815 produces highly inununogenic tum^– variants. Most of these variants express potent transplantation antigens which are not present on the original P815 tumor cells. These tum^– antigens, which appear to be specific for each variant, elicit a strong cytolytic T lymphocyte (CTL) response, but do not seem to induce a specific antibody response. As a first step in the isolation of the gene of a tum^– antigen, we attempted DNA-mediated gene transfer. As a DNA recipient cell we used P1.HTR, a highly transfectable P815 cell line, whose selection has been previously described. For the detection of antigen-expressing cells in transfected populations we developed a procedure that relies on the ability of these cells to stimulate the proliferation of the relevant CTL. Using DNA from tum^– variant P91 mixed with a plasmid carrying an antibiotic resistance gene, we obtained several independent transfectants expressing a tum^– antigen, at a frequency of approximately 1 in 13 000 antibiotic-resistant transfectants. These transfectants express only one of the two tum^– antigens that were identified on P91, suggesting that these tum^– antigens correspond to different genes. We expect that the detection procedure described here will be-suitable for the identification of transfectants for any gene that determines the expression of an antigen recognized by CTL.