If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Co-expression of p16INK4A and Laminin 5 by Keratinocytes: A Wound-Healing Response Coupling Hypermotility with Growth Arrest that Goes Awry During Epithelial Neoplastic Progression
Department of Dermatology and Harvard Skin Disease Research Center, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
Department of Dermatology and Harvard Skin Disease Research Center, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
Department of Dermatology and Harvard Skin Disease Research Center, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
The replicative lifespan of human keratinocytes in culture is restricted by a telomere-unrelated induction of p16INK4A (p16) and p14ARF. We have found that, in vivo, p16 is expressed by epidermal and oral keratinocytes at the migrating fronts of healing wounds and at the stromal interface of severely dysplastic and early invasive lesions and that such cells also invariably display increased expression of Laminin 5 (Lam5). In culture, p16 and Lam5 are coexpressed in keratinocytes at senescence, at the edges of wounds made in confluent cultures, and when cells are plated on dishes coated with the γ2 precursor form of Lam5 (Lam5γ2pre). Lam5/p16 coexpression in all three in vitro settings is associated with directional hypermotility and growth arrest. Hypermotility and growth arrest are uncoupled in p16- and p14ARF/p53-deficient keratinocytes and squamous cell carcinoma (SCC) cells; such cells become hypermotile is response to Lam5γ2pre but do not growth arrest. Thus, the Lam5/p16 response is activated in normal wound healing, causing growth arrest of migratory keratinocytes that lead wound reepithelialization. This response also becomes activated at a critical stage of neoplastic progression, acting as a tumor suppressor mechanism. Rare premalignant cells that lose p16 remain motile and proliferative, thereby resulting in invasive growth as SCC.
Squamous cell carcinoma (SCC) is a malignancy of the oral mucosal epithelium, the epidermis, and other stratified squamous epithelia. SCC arise within areas of abnormal, preinvasive cell growth (dysplasia), which may take many years to progress to invasive cancer (
). A consistent feature of invasive SCC and detectable in many premalignant dysplasias is loss, by mutation, deletion, or promoter hypermethylation, of the ability to express functional p16INK4A (p16) (
) (see Figure 1). Normally, cells do not express p16 and whether cells cycle or not is controlled by cyclin D1, the expression of which is regulated by mitogen availability and cell–substratum anchorage. Cyclin D1 forms complexes with and activates cdk4 and cdk6, which phosphorylate and inactivate the Rb protein, permitting E2F-dependent transcription of genes whose products are necessary to permit the G1/S transition and initiate chromosome replication (reviewed by
Western blot analysis of normal primary human keratinocytes in culture has identified an increase in p16 levels with serial passage, as the cells approach the end of their replicative lifespan (
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
Concurrence of replicative senescence and elevated expression of p16(INK4A) with subculture-induced but not calcium-induced differentiation in normal human oral keratinocytes.
). Immunocytochemical analysis of such cultures has revealed that p16 expression occurs heterogeneously and abruptly, followed by growth arrest, and that the probability that a keratinocyte will express p16 increases steadily with each passage until all cells are p16-positive and senescent (
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
). Our studies have provided strong evidence that, if p16 expression fails to arrest growth, induction of p14ARF expression follows soon thereafter to effect a p53-dependent, p21cip1-enforced arrest in either G1 or G2, as described in
and summarized in the model shown in Figure 2. Keratinocytes engineered to express the catalytic subunit of telomerase, TERT, extend and stabilize their telomeres; yet, still they undergo p16-enforced senescence (
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
). This fate can be mofified by mutational, epigenetic, or regulatory loss/reduction of p16 expression in TERT-transfected cells, yielding immortalized lines (
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
). p16-enforced keratinocyte senescence is distinct from the telomere-sensitive, p53/p21cip1-dependent replicative senescence mechanism that enforces the replicative lifespan limit of human skin fibroblasts and several other cell types (
) are predisposed to a variety of spontaneous and carcinogen-induced cancers, but they undergo normal development and form structurally and functionally normal stratified squamous epithelia. These results confirm the tumor suppressor function of p16 and are also consistent with the finding that p16 protein is not expressed as a feature of normal stratified squamous epithelial renewal or differentiation (
). The stage of neoplastic progression toward SCC at which the p16 protein becomes expressed and functions as a tumor suppressor was revealed by our recent study, summarized below, in which we characterized p16 expression immunohistochemically in a set of normal, benign hyperplastic, pre-invasive, and invasive epithelial tissue specimens from the skin and oral mucosa (
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
). p16 proved to be expressed heterogeneously in cells of some premalignant lesions and consistently in areas of microinvasion and at superficial margins of invasive SCC.
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
) that the p16-expressing cells in such lesions also express greatly increased levels of Laminin 5 (Lam5), previously identified as a marker of invasion in many types of epithelial cancers (
Laminin-5 is a marker of invading cancer cells in some human carcinomas and is coexpressed with the receptor for urokinase plasminogen activator in budding cancer cells in colon adenocarcinomas.
). The trimer of α3, β3, and γ2 chains is synthesized by keratinocytes as an Mr∼460 kDa precursor form and processed by cleavage of the α3 and γ2 chains at specific sites Figure 3 by BMP1/Tld and other proteases (
Mammalian tolloid metalloproteinase, and not matrix metalloprotease 2 or membrane type 1 metalloprotease, processes laminin-5 in keratinocytes and skin.
) to an Mr∼400 kDa mature form. In their normal, undisturbed state in vivo, basal keratinocytes maintain an intact, continuous layer of mature Lam5 beneath them, to which they adhere via α6β4 integrin to form stable, anchoring hemidesmosome (HD) structures. In HD, α6β4 integrin forms clusters and associates closely with the transmembrane BP180 protein; the cytoplasmic domains of these proteins form stable connections with the cytokeratin filament network, mediated by BPAG1 and plectin (
). Keratinocytes at the migrating front of wounds express high levels of Lam5 and display an increased expression and unpolarized distribution of α6β4, α3β1, and α2β1 integrins (
reported that exogenously applied Lam5 promoted α3β1 integrin-dependent keratinocyte motility as measured by a Boyden chamber/transwell migration assay. Subsequent studies by
provided evidence that reduced ligation of mature Lam5 to α6β4 integrin and increased ligation of precursor Lam5 to β1 integrins, principally α3β1, results in keratinocyte motility at the leading edge of healing wounds.
Figure 3Laminin 5 (Lam5) structure. Lam5 is secreted as a three-chain heterotrimer, which is normally processed proteolytically soon after secretion in the γ2 chain and α3 chain at the sites indicated by large blue arrows. The γ2 chain is subject to further processing during some developmental processes and in some carcinomas by cleavage at the site indicted by the small black arrow.
Interestingly, Lam expression is increased in p16-positive, senescent normal keratinocytes in late passage cultures, associated with a remarkable directional hypermotility (
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
). As described below, we have found that coordinate Lam5 and p16 expression occurs in normal wound healing. We have identified an experimental system for inducing this response acutely and synchronously in culture and have used this system to characterize the fundamental elements of a keratinocyte motility/growth arrest (“KMA”) mechanism, in which certain changes in cell–substratum adhesion induce directional hypermotility followed by an increase in p16 expression and growth arrest (Natarajan et al., submitted).
Natarajan E, Omobono JD II, Guo Z, Hopkinson S, Lazar AJF, Brenn T, Jones IC, Rhein Wald JG: A keratinocyte hypermotility growth arrest response activated by Lamin5 and dependent upon the TGFB receptor and p16INK4A (submitted).
Results
p16-related keratinocyte senescence mechanism and immortalization barrier
Our initial investigations confirmed and extended the findings of
, that the expression of TERT in keratinocytes via stable retroviral transduction results in telomere maintenance but is insufficient by itself to produce an immortalized line. We found (see
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
) that rare, immortalized variants frequently emerged from a TERT-transduced population following either a short and subtle, or a prolonged and obvious, “slow growth phase” (SGP), during which most of the cells underwent p16-enforced senescence (as in the example in Figure 4a of the emergence of N/TERT-1 from the primary newborn foreskin keratinocyte line strain N). In some TERT-transduced keratinocyte populations, rapidly dividing immortalized (RDI) variants emerged. Some of these were associated either with complete loss, by deletion or other mechanisms, of p16 expression. Other TERT-transduced, immortalized keratinocyte lines we generated proved to evade p16-related arrest by markedly reducing the frequency with which p16-expressing, growth-arrested cells are produced during serial passage, rather than by mutational loss or promoter silencing of the p16 locus (
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
). All the TERT-immortalized lines that we generated retained a normal p53-dependent DNA damage arrest response, but some had p14ARF deletions.
Figure 4The p16INK4A-dependent barrier to human keratinocyte immortalization in culture can be bypassed by spontaneous mutations or by genetic engineering to evade p16INK4A and p14ARF/p53-dependent arrest mechanisms. Panel a summarizes the results described in
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
. The inset photographs show examples of cultures immunostained for p16 at various stages in the lifespan of primary keratinocytes and TERT-transduced keratinocytes. Normal primary human keratinocytes, such as strain N depicted in this panel, can be serially passaged in suitable culture medium formulations for as many as 60 population doublings (PD) before senescence (asterisk), associated with induction of p16 expression in all cells. Cells stably transduced to express TERT at the point in their lifespan indicated by the downward red arrow maintain long telomeres but enter senescence at the same time as untransduced control cells. TERT-expressing keratinocyte populations exhibit a “slow growth phase (SGP)” of indefinite length (nine passages over 2 mo in this experiment) in which only a small subpopulation remains p16-non-expressing and proliferative, and may produce rare variants in which a second event has occurred, associated with complete loss (upper right photograph, N/TERT-1) or greatly reduced frequency (lower right photograph, N/TERT-2G) of p16 expression occurs, permitting the cells to divide rapidly and without limit as an immortalized line. Panel b summarizes the results described in
. Strain N cells transduced to express either the p16-non-binding mutant cdk4R24C (cdk4R) or a dominant-negative p53 (p53DD) senesce (asterisk) after about the same number of population doublings as the untransduced control cells, but doubly transduced (p53DD+cdk4R) cells evade senescence and divide for an additional 25–35 PD before being limited by very short telomere crisis. These cells become “directly” immortalized after transduction to express TERT, without requiring the acquisition of further complementary second events.
Interestingly, TERT-transduced keratinocytes cultured from a p53 (+/-) individual (Li–Fraumeni hereditary predisposition to cancer syndrome) proved to select for loss of the wild-type p53 allele following attenuation of p16 expression (
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
), in which we genetically engineered normal keratinocytes to bypass senescence and continue dividing to short telomere crisis. Those experiments showed that, in order to evade their senescence arrest mechanism, keratinocytes require abrogation of both p16 and p53 function, accomplished by sequential retroviral transduction to express a p16 non-binding mutant form of cdk4 (cdk4R24C (
) Figure 4b. Keratinocytes in which both p16 and p53 were missing or rendered nonfunctional proved to be directly immortalized following transduction to express TERT. Supporting this result, we found that keratinocytes cultured from a p16 (+/-) individual (familial melanoma hereditary predisposition to cancer syndrome) underwent selection for a spontaneously arising, extended lifespan variant that lost the wild-type p16 allele, dependent upon experimental p53DD expression to abrogate the other, p53-dependent component of p16-related senescence (
Two types of experiments showed that the p16-dependent component of the dual, p16- and p14ARF-dependent keratinocyte senescence mechanism is activated first and enforces growth arrest. First, whether cultured in K-sfm or in the fibroblast feeder system, senescence arrest proved to correlate tightly with high levels of p16, identified by BUdR/p16 double immunocytochemical staining (
) Figure 5a. Second, two-channel flow cytometry, measuring DNA content by propidium iodide fluorescence and cycling vs. non-dividing state by BUdR immunofluorescence following a 24 h labeling period, showed that keratinocytes arrest with a G1 DNA content, as expected for p16-enforced arrest (Gray and Rheinwald, unpublished) Figure 5b. This result was very different from that of senescing fibroblasts, which arrested in either G1 or G2 (in a ratio similar to their G1/G2 distribution during exponential growth) as a result of telomere erosion-triggered, p53-dependent, p21cip1-enforced arrest (data not shown).
Figure 5Keratinocytes cultured in optimized culture systems senesce in G1, following their induction of p16INK4A expression. Panel a shows mid-to-late lifespan cultures of strain N keratinocytes cultured in GIBCO ker-sfm medium (a, b) or in the 3T3 fibroblast feeder layer system (c, d). Bromodeoxyuridine (BUdR) was added to the medium 24 h before cultures were fixed and enzymatically immunostained for BUdR (reddish-brown) and p16INK4A (blue) in a, c (compare with phase-contrast images in b, d). Note the inverse correlation between p16 expression and BUdR incorporation, with only rare p16 non-expressing cells found to be non-cycling. Panel b shows the results of two channel flow cytometric analysis of a mid-lifespan culture of strain N that was BUdR labeled for 24 h before it was trypsin suspended, fixed, permeabilized, and incubated with the DNA intercalating dye propidium iodide to assess DNA content and fluorescent-labeled anti-BUdR antibody to permit determination of the cycling or non-cycling character of each cell. The non-cycling subpopulation (red) of this culture (to the lower left of the dashed blue line) was found to be predominantly of 2n DNA content, consistent with G1 arrest, as expected for cells arresting as a result of inhibition of the G1/S restriction point-specific cyclin-dependent kinases cdk4 or cdk6.
). Keratinocytes sequentially transduced to express cdk4R and TERT were not immortalized, but an immortalized variant arose from within this population that proved to have acquired a function-impairing mutation in p53. Furthermore, immunocytochemical staining and western blot analysis showed that, in keratinocytes engineered to resist p16 and to block p53 function, both p16 and ARF levels increased in cells just at the time when they began dividing beyond their normal replicative lifespan limit. These results support the sequential, two-stage model of the p16-enforced keratinocyte senescence arrest mechanism and immortalization barrier shown in Figure 2.
p16 expression during neoplastic progression to squamous cell carcinoma
We next sought to determine the setting in which p16 becomes expressed in keratinocytes in vivo to function as a tumor suppressor mechanism. We examined sets of normal, benign hyperplastic, dysplastic, carcinoma in situ, and invasive SCC lesions of human oral mucosa and epidermis (
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
). Our initial hypothesis, based on our observation of increased frequency of p16 expression with serial passage of normal primary keratinocytes, was that excessive cell division in vivo, as would occur in chronic benign epithelial hyperplasia or premalignant dysplasia, would trigger p16 expression. We found, however, that benign hyperplastic lesions never contained p16-positive cells, but that p16 was expressed by cells engaged in or about to begin neoplastic invasion through the basement membrane into superficial connective tissue Figure 6. As expected from results published by others that the p16 locus is typically lost or silenced by promoter hypermethylation in advanced, deeply invasive SCC (reviewed by
), p16 protein expression proved to be lost at some point between the stages of early invasive and deeply invasive SCC in both epidermal and oral epithelial neoplastic progression Figure 6.
Figure 6p16INK4A expression during neoplastic progression of stratified squamous epithelia toward squamous cell carcinoma (SCC) occurs at the histopathologically classified stages of severe dysplasia to early invasion. Summarizing the principal results of
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
, the panels show hematoxylin/eosin-stained sections (above) and p16 immunostained sections (below) of specimens of benign hyperplasia, severe dysplasia, carcinoma in situ (non-HPV-related), and actinic keratosis overlying a deeply invasive squamous cell carcinoma. Note absence of p16 expression in benign hyperplasia, variegated pattern of p16 expression in late-stage pre-malignant lesions, and loss of p16 expression by deeply invading, progressively growing SCC (asterisk).
Coordinate p16 and Lam5 expression in neoplastic progression and normal wound healing
Interestingly, we found that the same regions and many of the same cells that expressed p16 also expressed greatly increased levels of Lam5 Figure 7a–c. Lam5 had been reported previously to be associated with regions of early invasion (e.g.,
Laminin-5 is a marker of invading cancer cells in some human carcinomas and is coexpressed with the receptor for urokinase plasminogen activator in budding cancer cells in colon adenocarcinomas.
). We found that coordinate induction of p16 and Lam5 expression occurs in vivo in keratinocytes at the edge of skin wounds Figure 7d–f, and is also displayed by keratinocytes at the migrating front of partial thickness wounds made in pieces of surgically resected human skin and subsequently maintained in organ culture for two days Figure 7g–i. Lam5 and p16 also proved to be coordinately upregulated at the edge of experimental wounds made in confluent cultures of early passage keratinocytes (
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
) Figure 7j–l. Taken together, our neoplastic and wound tissue analyses and our studies of scratch wounded, early-passage keratinocyte cultures strongly suggest that, in vivo, a p16- and Lam5-related KMA mechanism is triggered by an acute abnormality in cell–cell or cell–substratum interaction, coupling growth arrest with directional hypermotility. The scratch wound result demonstrates that the KMA response is cell-autonomous and therefore amenable to study using conventional culture conditions.
Figure 7Coordinate induction of p16INK4A and Laminin 5 (Lam5) expression in keratinocytes of premalignant dysplasias and in normal keratinocytes migrating during wound healing. p16 immunostaining (b, e, h, k) and Lam5 immunostaining (c, f, i, l). Panels a–c: microinvasive region of severe epithelial dysplasia. Panels d–e: edge of a healing skin wound (arrow indicates edge of migrating tongue of epithelium; asterisk indicates fibrin scab). Panels g–i: wound made in an excised piece of newborn foreskin and maintained in organ culture for 2 d before immunostaining (N indicates normal, intact epidermis; W indicates the margin of the original wound). Panels j–l: confluent culture of strain N keratinocytes scratch wounded and maintained for 2 d before immunostaining.
Hypermotility and growth arrest induced by plating cells on a precursor form of Lam5
These results led us to conduct a series of experiments to determine whether KMA can be induced synchronously in early-passage keratinocyte cultures, thereby providing a useful experimental system with which to determine the time course of activation and the temporal and causal relationships among Lam5 expression, p16 expression, hypermotility, and growth arrest. Following studies by Jones et al, described above, which concluded that the keratinocyte wound response is associated with a change in cell–substratum adhesion from (α6β4 integrin: mature Lam5) to (α3β1 integrin: precursor Lam5), we asked whether plating keratinocytes on culture dishes coated with precursor Lam5 would induce KMA. In our first experiments, we plated keratinocytes at a low density onto dishes that had been precoated with medium conditioned by the rat bladder carcinoma cell line 804G, known to secrete Lam5 having a processed α3 chain and unprocessed γ2 chain (
) (“Lam5γ2pre”). After 2 d at 37°C, immunostaining with a human Lam5-specific antibody revealed that keratinocytes plated on control surfaces had formed small pads of Lam5 and had undergone little or no net migration. In contrast, cells plated on 804G CM-coated dishes became hypermotile, as revealed by the tracks of Lam5 each cell left behind. Composite low-magnification images of immunostained wells were used to measure the lengths of these tracks. As indicated in the scatterplot graph in Figure 8b, the rate of migration stimulated over the 2 d period after plating averaged ∼1500 μm per d—remarkable for these ∼12 μm diameter cells.
Figure 8Directional hypermotility induced in keratinocytes by plating on a culture dish pre-coated with the γ2 precursor form of Laminin 5 (γ2Lam5pre). Panel a shows Lam5 immunostaining of strain N keratinocyte cultures 2 d after plating at a low density in control conditions (dishes pre-coated with fresh culture medium) or on a dish pre-treated for 30 min with medium conditioned by a confluent culture of 804G rat bladder carcinoma cells, which secrete the1 γ2 precursor form of Lam5. Note that in control conditions, cells wander little from the point at which they first attached, as evidenced by the distribution of secreted Lam5, in contrast to the remarkable hypermotility of cells plated on a dish coated with Lam5γ2pre. Panel b shows migration distances of individual cells, determined by measuring path lengths from composite pictures assembled from a series of adjacent microphotographic images of the immunostained culture. Panel c shows the migration distances traveled during successive 2 h intervals by keratinocytes plated on a coverslip coated with recombinant human Lam5γ2pre-dissolved in fresh culture medium and tracked by time-lapse photography beginning at the time of plating. Note that most cells migrate at or near their maximum rate, typically ∼150 um per h, as soon as they have attached to the surface.
Using purified recombinant human Lam5 (the α3-processed, γ2-precursor form: Lam5γ2pre), we have confirmed that coating dishes with 0.4 μg per mL of this protein dissolved in culture medium induces the same hypermotility response as 804G CM. We do not have available a source of purified recombinant Lam5 mature form to compare; thus, we do not know whether the γ2 precursor form is essential for eliciting this response. We can conclude, however, that the α3 precursor is not essential for triggering hypermotility. Time-lapse photography has disclosed how quickly cells become hypermotile after plating on a recombinant human Lam5γ2pre-coated surface. Such experiments have shown that cells begin to migrate within 20 min of attachment and spreading on a coated substratum, suggesting that a change in gene expression or new protein synthesis is not necessary to initiate hypermotility. By measuring the distance traveled during successive 2 h intervals of filming (i.e., 0–2, 2–4, 4–6 h, etc), the behavior of individual cells could be determined. As shown in Figure 6c, cells typically maintained rather constant rates of motility, averaging about 125 μm per h during the first day after plating.
Hypermotility stimulated by precursor Lam5 is followed by p16 expression and growth arrest
We next asked whether cells induced to become hypermotile by plating on Lam5γ2pre are growth inhibited by plating cells in control and precoated wells and counting the cells 7 d later. As shown in Figure 9a, cell growth was markedly inhibited on Lam5γ2pre-coated surfaces. Note that plating keratinocytes on dishes coated with culture medium alone or with fibronectin dissolved in medium resulted neither in increased directional motility nor in growth inhibition. Examining the cells by immunostaining for p16 expression Figure 9b showed that by 1 and 4 d after plating on Lam5γ2pre, ∼25%–30% and ∼60%–70% of the cells were expressing p16, respectively, and by 7 d were unable to reinitiate growth when subcultured onto a control surface Figure 9c.
Figure 9The hypermotility response of normal keratinocytes to Laminin 5 (Lam5)γ2 precursor (Lam5γ2pre) is accompanied by p16INK4A induction, growth inhibition, and ultimate irreversible arrest. Panel a shows the average population growth rate (during 7 d after plating) and the motility response (measured 2 d after plating) of strain N keratinocytes plated on fresh culture dishes or dishes pre-coated with serum-containing medium, fibronectin, or Lam5γ2pre (804G CM). Note substantial growth inhibition and hypermotility from Lam5γ2pre but not from fibronectin. Panel b shows induction of p16INK4A protein expression, determined in individual cells by immunostaining, during the first 4 d after plating cells on surfaces coated with 804G CM or with purified human recombinant Lam5γ2pre. Panel c shows the colony-forming ability on fresh dishes of 1000 cells subcultured from a 7 d control culture or from a 7 d Lam5γ2pre-coated dish culture, revealing the irreversibility of the growth arrest ensuing from long-term adhesion to Lam5γ2pre.
p16- and p14ARF/p53-deficient keratinocytes retain the hypermotility component but evade the growth arrest component of the keratinocyte motility/arrest response
The immediacy of the hypermotility response, in contrast to the slower induction of p16 and subsequent irreversibility of growth arrest, indicates that p16 expression does not induce hypermotility. Instead, some aspect of the substratum that induces hypermotility, or the act of hypermotility itself, induces p16 expression and growth arrest. This conclusion was supported by the behavior of keratinocytes deficient in p16 and either p14ARF or p53. The advanced, invasive epidermal SCC-derived cell line SCC13 and normal primary keratinocytes engineered by transduction to express cdk4R24C and p53DD (N/P/C)—thereby evading p16- and p14ARF/p53- dependent arrest mechanisms, responded to plating on Lam5γ2pre with directional hypermotility but were not growth arrested Figure 10. Thus, they maintained the hypermotility component but evaded the growth arrest component of KMA. Interestingly, one of our TERT-immortalized lines (N/TERT-1), which had lost p16/p14ARF expression following transduction to express TERT as a rare heritable event associated with the emergence of an immortalized line (
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
; Figure 4a), also displayed a substantial resistance to Lam5γ2pre-induced growth arrest Figure 10. These results prove that neither the expression of p16 nor its ability to cause growth arrest is essential for the hypermotility response, and also suggest that the p16/p14ARF arrest mechanism that is activated asynchronously in serially passaged keratinocytes is identical to the growth arrest component of the Lam5γ2pre-induced, acute KMA response.
Figure 10p16 and p14ARF/p53 expression and function is essential for the growth arrest component, but not the hypermotility component, of the keratinocyte motility/arrest response toγ2 precursor form of Laminin 5 (Lam5γ2pre). Growth inhibition and hypermotility responses of keratinocyte cell lines plated on dishes pre-coated with Lam5γ2pre (804G CM). The normal primary keratinocyte line strain N at sixth passage (∼30 population doublings (PD)/mid-lifespan) showed the normal response. Note that strain N at 13th passage (∼50 PD, near-senescence) showed little proliferation under control conditions and significant constitutive hypermotility (as reported in
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
), which was not increased further by plating on Lam5γ2pre. The p16 expression-reduced TERT-immortalized strain line N/TERT-2G gave results similar to that of strain N, in contrast to the p16/p14ARF/p53-deficient lines marked by the box on the left, which exhibited a hypermotility response to Lam5γ2pre but were not substantially growth inhibited.
We are currently seeking to understand the signal pathways responsible for inducing the hypermotility and growth arrest components of KMA, which at present remain a “black box”. Our initial studies have used, as a point of departure, observations by
that cultured keratinocytes treated with the growth-inhibitory polypeptide factor transforming growth factor beta (TGFβ) increase synthesis and secretion of the α2pre, γ2pre form of Lam5 and exhibit increased motility as assessed using a porous filter trans-migration, Boyden chamber assay. We have confirmed that TGFβ treatment induces a directional hypermotility response similar to that induced by plating cells on Lam5pre Figure 11. Lam5 track measurement and time-lapse studies have disclosed that the hypermotility begins about 16–24 h after exposure of cells to TGFβ (data not shown), consistent with a requirement for a change in gene expression and new protein synthesis.
Figure 11Transforming growth factor beta (TGFβ) induces directional hypermotility in keratinocytes plated on control surfaces. Strain N keratinocytes were plated on fresh culture dishes in medium containing 1 ng per mL TGFβ1. Two days later, cultures were immunostained for Laminin 5 (Lam5) to reveal migration tracks (panel a), which were measured and compared (panel b) with those induced in the same experiment by plating cells in medium without TGFβ but on surfaces pre-coated with human recombinant Lam5±2pre.
The experiments described here have characterized the fundamental elements of a KMA response that is activated in keratinocytes in vivo and in culture when they encounter certain types of abnormal substrata. In retrospect, we recognize that we first detected KMA as it is activated heterogeneously in normal keratinocytes during serial culture, where its detectable effect is to cause senescence arrest. Keratinocyte replicative potential in culture is limited by two independent mechanisms. One, common to all somatic cell types, senses progressive telomere shortening and consequent exposure of telomere overhang sequences by the cells' p53-dependent DNA damage response mechanism, leading to expression of the cdk inhibitor p21cip1 and growth arrest either in G1 or G2. Telomere-related senescence arrest is the sole determinant of replicative lifespan/expansion potential in culture of fibroblasts and a number of other cell types, with p16 playing no essential role in this process (
). The other senescence mechanism, which is p16-dependent and telomere status-independent, is typically activated in serially cultured human keratinocytes before their telomeres have shortened sufficiently to trigger that arrest mechanism. Therefore, the p16-dependent mechanism is responsible for determining the replicative lifespan/expansion potential of keratinocytes in culture. As summarized above and described in
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
, serially cultured keratinocytes exhibit an abrupt induction of p16INK4A expression and irreversible growth arrest in G1 with increasing frequency at each passage. Considering that p16 induction is triggered in early-passage keratinocytes when plated on surfaces coated with the γ2 precursor form of Lam5, keratinocytes in serial culture may gradually lose their ability to process secreted Lam5 quickly enough to prevent engagement of precursor Lam5 with their α3β1 integrin and consequently instigate the KMA response.
The phenomenon of p16-related senescence in culture led us to test the hypothesis that induction of p16 expression occurs in stratified squamous epithelia in vivo as a mechanism to restrict the proliferative potential of normal keratinocyte stem cells. Our study of hyperplastic and dysplastic lesions of the epidermis and oral mucosa did not support this hypothesis but, instead, revealed that p16 expression is specifically induced in pre-malignant oral and epidermal keratinocytes that have progressed to the stage of incipient or initial invasion into underlying connective tissue as SCC (
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
). The finding of coordinate p16 and Lam5 expression at sites of severe dysplasia and microinvasion, as well as in normal keratinocytes at the migrating front of healing wounds, suggests a common response by keratinocytes to these two pathologic situations. We have conjectured (
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
) that the KMA response evolved as part of the wound-healing process to aid re-epithelialization but that it is also activated at a specific stage of neoplastic progression. KMA is marked (1) by increased synthesis of Lam5 and (presumably) its deposition in an incompletely processed, γ2 precursor form that makes the cells directionally hypermotile and (2) by subsequent p16 expression that arrests their growth. We hypothesize that sacrifice of the proliferative potential, by p16 induction, in the band of cells that are leading migration to reepithelialize a wound may enhance the directional motility of these cells so that they can more efficiently pull their still-proliferative neighbors behind them to cover the wound surface. Figure 12 illustrates our hypothetical model of how KMA is activated during epithelial neoplastic progression. Rare preneoplastic cells that evade the growth arrest component of KMA, typically by ceasing to express p16, would then be able to grow progressively as invasive SCC. Recent studies from other laboratories have reported an increase in p16 expression at the leading edge of healing oral mucosal epithelial wounds (
Invade or proliferate? Two contrasting events in malignant behavior governed by p16(INK4a) and an intact Rb pathway illustrated by a model system of basal cell carcinoma.
Figure 12A model for the p16/laminin 5 (Lam5)-related keratinocyte motility/arrest response in epithelial neoplastic progression. The diagram shows normal engagement of the basal layer of stratified squamous epithelia with mature, processed Lam5 via α6β4 integrin, in contrast to dysplastic cells that have progressed to a stage at which they are breaking down and beginning to penetrate the basement membrane. We hypothesize that this stimulates increased expression of Lam5, some of which remains in its precursor form and engages the motility-related integrins α3β1, and possibly also α2β1 (see
). The increasingly darker shades of brown color indicate induction of p16 protein expression in these cells, which results in cell cycle arrest and acts as a tumor suppressor to prevent progressive growth. If a rare p16-deficient variant arises within the lesion, it can continue dividing while invading as a squamous cell carcinoma.
Identification of the cell surface and extracellular matrix molecules that determine whether keratinocytes remain sessile or display migratory behavior has been a subject of active investigation for almost 20 years. Some of the earliest work in this field recognized that some cell types increase their motility in response to exposure to some forms of Lam5 (
) in which keratinocytes convert from a more sessile state favored by α6β4 integrin adhesion to mature Lam5 to a more motile state favored by α3β1 integrin adhesion to precursor Lam5 (or, as recently suggested (
), syndecan-1 adhesion to Lam5). Further proteolytic cleavage of the γ2 chain at a specific site indicated by the small black arrow in Figure 3 has been found in some cancer cell lines in culture, as a result of MMP-2 and/or MT1-MMP action, thereby releasing soluble fragments of γ2-containing epidermal growth factor (EGF)-like repeats that have been reported to stimulate motility measured by the transwell migration assay (
). It is important to note that, since we do not currently have a source of purified recombinant mature Lam5 (i.e., with both the α3 and γ2 chains processed), we do not know whether an abnormally high concentration of Lam5 in any form might be able to trigger the KMA response.
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
) and by early-passage cells plated on Lam5γ2pre surfaces (results presented here) as “persistent, processive” motility. Either term is suitably descriptive and serves to distinguish this type of cell movement from that of “sessile” cells in culture, which time-lapse photography reveals are actually very active and motile, but move within a confined area by frequently changing direction or by rotating. Studies using Lam5 immunostaining or colloidal gold displacement to detect migration, such as used by
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
, have the potential, depending upon the method used for quantitation, of distinguishing directional migration from non-persistent, non-directional motility. The method we used, Lam5 immunostaining to reveal 2 d track lengths of individual cells, can detect and quantitate the specific type of motility characteristic of the KMA response. This method, therefore, has an advantage over Boyden chamber transwell migration assays, colloidal gold area clearance measurements, or visual scoring for a scattered appearance of colonies for accurately assessing effects of experimental agents or altered gene expression on wound- and invasion-related directional hypermotility.
In a small survey of soluble factors that might be expected to influence keratinocyte migration, we found that high concentrations of EGF and hepatocyte growth factor did not induce hypermotility (data not shown), but that TGFβ, a potent keratinocyte growth inhibitor (for example, see
had reported that TGFβ-treated keratinocytes reduce their display of α6β4 integrin, predominantly begin depositing completely unprocessed Lam5 (i.e., the α3pre, γ2pre form) on their substratum, and become more motile.
Our cell culture model experiments, summarized here and described more completely elsewhere (Natarajan et al, submitted), have elucidated the temporal and causal relationships among Lam5γ2pre exposure, directional hypermotility, p16 expression, and growth arrest. Of central importance is that KMA is a sequential, two-step process: hypermotility occurs quickly, followed later by p16 expression and a growth inhibition that ultimately becomes irreversible. Neither p16 expression nor growth arrest is essential to the hypermotility response. SCC cells, normal keratinocytes engineered to resist the inhibitory effects of p16 and p14ARF, and some TERT-immortalized keratinocyte lines that acquired loss of p16 and p14ARF expression still become hypermotile in response to Lam5pre-induced KMA but are not growth inhibited. Thus, neither receptor tyrosine kinase-initiated mitogen signaling pathways nor expression of essential cell cycling proteins such as cyclin D1, c-myc, etc., are shut down, blocked, or impaired in hypermotile cells. Our results are consistent with the invasive behavior of p16-deficient SCC cells in vivo.
The working model shown in Figure 13 summarizes our own findings to date about the KMA response in the context of discoveries by others about the mechanisms of Lam5γ2pre:integrin engagement and TGFβ-instigated growth arrest. The box outlined by a dashed line indicates the (many) signaling molecules and pathways that remain to be characterized. These include the events downstream of integrin engagement by Lam5γ2pre that lead to immediate hypermotility, the events that induce expression of p16 and p14ARF to cause growth arrest, and the events downstream of TGFβ receptor activation that result in hypermotility and growth arrest. We are currently investigating the possibility of cross-talk and shared signal pathway events between Lam5γ2pre-induced and TGFβ-induced KMA. We have begun studies using signal pathway kinase-specific inhibitory drugs, seeking to identify signaling steps that are either unique to hypermotility or else can be modulated by concentrations of drugs that impair motility without causing an acute, p16-unrelated growth inhibition. We are augmenting this approach with the design and expression of specific RNAi vectors to knock down expression of candidate signal pathway proteins and assess the consequence to the hypermotility and growth arrest components of the keratinocyte motility/arrest response.
Figure 13A working model of γ2 precursor form of Laminin 5 (Lam5γ2pre)-induced and transforming growth factor beta-induced hypermotility and growth arrest in keratinocytes.
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
) with 25 μg per mL bovine pituitary extract (BPE), 0.2 ng per mL EGF, and 0.4 mM CaCl2. To generate near-confluent or confluent cultures, cells were grown to about 40% confluence in K-sfm, then refed with a medium consisting of a 1:1 (vol:vol) mixture of calcium-free DMEM medium (GIBCO/Invitrogen), and 0.4 mM CaCl2-supplemented complete K-sfm, which permits formation of healthy, confluent, modestly stratified cultures. For some experiments, keratinocytes were cultured using the fibroblast feeder layer system (
), in which keratinocytes are co-cultured with mitomycin-treated 3T3J2 cells in FAD medium, consisting of DMEM/F12 (1:1 vol/vol) medium (GIBCO, Invitrogen)+5% calf serum (CS) (Hyclone, Logan, Utah), 10 ng per mL EGF, 0.4 μg per mL hydrocortisone (HC), 5 μg per mL insulin, 10 × 10-10 M cholera toxin (CT), 2 × 10-11 M triiodothyronine, and 1.8 × 10-4 M adenine. 804G cells (
The human p16INK4A-specific mouse monoclonal antibody G175-405 (Pharmingen, San Jose, California) was used at 2 μg per mL and the Lam5 γ2 chain-specific mouse monoclonal antibody D4B5 (
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
. Images were captured on a NIKON E600 Microscope (Nikon, Melville, New York) with a SPOT2 digital camera using SPOTcam v.3.5.5 software (Diagnostic Instruments, Sterling Heights, Michigan).
Purification of recombinant human Lam5, γ2 precursor form
The 293 cells were sequentially transfected with pcDNA3.1 plasmids expressing the complete coding sequences of each of the three chains of human Lam5. The β3 subunit sequence contained a C-terminal 6-His tag. Transfected 293 cells were grown to confluence and then switched to serum-free medium for two days, conditioned medium was then harvested and laminin-5 purified over a His-Bind column (Novagen, San Diego, California). The eluted Lam5 consisted of α3 chain in its cleaved, mature form but the γ2 chain in its precursor, uncleaved 155 kDa form, as determined by western blotting with the D4B5 antibody (data not shown).
Assay for acute KMA induction and modulation by extracellular matrix proteins and polypeptide factors
A six-well tissue culture plates (∼9 cm2 area per well per 1.5 mL medium volume) (Costar Corning, New York) were pretreated by incubation with conditioned medium (CM) from confluent cultures of 804G cells or with solutions of purified extracellular matrix proteins dissolved in 10% calf serum-supplemented DMEMαF12 medium for 30 min at room temperature, and then were rinsed and used to plate cells in K-sfm medium. To measure directional hypermotility, wells plated with 500 cells were fixed in 4% paraformaldehyde 40 h later and immunostained with a Lam5γ2-specific monoclonal antibody using the ABC peroxidase method (
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
). Six to twelve adjoining fields were photographed using a 2 × objective and the path lengths of at least 30 cells were measured. To assess growth inhibition, wells plated with 2,000 cells were permitted to grow for 7 d, with refeeding on days 3 and 5. Cells were trypsinized and counted, and the average growth rate was calculated as log2 [# cells counted/#cells plated]/7 days=population doublings (PD)/day. The average (mean) migration path length and degree of growth inhibition (as measured by PD/day growth rate) were compared between control and experimental conditions.
Time-lapse photography
Keratinocytes were inoculated into a temperature-controlled heated chamber (Mike's Machine Company, Boston, Massachusetts) assembled with a 25 mm diameter Nalge/Nunc Thermonox (Rochester, New York) plastic coverslip, which was pre-coated with control medium or with medium containing Lam5γ2pre. A Zeiss (Zeiss, Göttingen, Germany) IM35 inverted microscope (Diagnostic Instruments) equipped with a Uniblitz Model VMM-D1 Shutter Driver (Vincent Associates, Rochester, New York), and a Spot Insight QE camera was used to capture images through a × 10 phase objective and × 10 eyepiece (total magnification × 100). Images collected at 5 min intervals using Spot version 4.0.4. software were converted into movies, which were examined to trace and measure the path lengths of cells.
Culture wound models
Scratch wound assay: Linear “wounds”∼2 mm wide were made in 1d post-confluent keratinocyte cultures (prepared as described above) by scraping the point of a sterile Eppendorf pipette tip across the cells, after which the cultures were returned to the incubator for 40 h until fixation and immunocytochemical analysis.
Organ cultured human skin wound assay: 1 cm2 pieces of recently resected human newborn foreskin were rinsed in medium to remove residual blood components and ∼2 mm diameter wounds were made through the epidermis and partially into the dermis with the aid of a fine forceps and curved iris scissors. The wounded pieces were submerged in medium and placed in the incubator for 2 d, after which they were fixed, sectioned, immunostained, and analyzed.
ACKNOWLEDGMENTS
It is a pleasure to recognize the contributions of current and past members of the Rheinwald lab who have contributed to this research: Zongyou Guo, Jay Omobono, Paula Hercule, Sarah Browne, Mark Dickson, Matt Ramsey, Laurie Gray, Jenny Wu, and Kathleen O'Toole. We also thank Barry Alpert of Microvideo Instruments, Inc. (Avon, Massachusetts) for assisting with the design of our time-lapse microscopy system and for optimizing our digital microphotography system.
The research described here has been aided in very important ways by past collaborators, including William Hahn, Robert Weinberg, David Louis, Yasu Ino, Vincent Ronfard, Fred Li, Philip McKee, Sook-Bin Woo, Christopher Crum, Hensin Tsao, Michele DeLuca, and Caterina Catricala. We wish to especially acknowledge our current collaborators Alex Lazar and Thomas Brenn, Brigham, and Women's Hospital, for providing pathology file specimens of wounds.
The human research materials were collected under appropriate, IRB-approved protocols including informed consent. This work has been supported by grants from NIDCR, NIAMS, and NIDDK.
References
Allen-Hoffmann B.L.
Rheinwald J.G.
Polycyclic aromatic hydrocarbon mutagenesis of human epidermal keratinocytes in culture.
Phosphatidylinositol 3-kinase function is required for transforming growth factor beta-mediated epithelial to mesenchymal transition and cell migration.
Human keratinocytes that express hTERT and also bypass a p16(INK4a)- enforced mechanism that limits life span become immortal yet retain normal growth and differentiation characteristics.
Concurrence of replicative senescence and elevated expression of p16(INK4A) with subculture-induced but not calcium-induced differentiation in normal human oral keratinocytes.
EGF-R signaling through Fyn kinase disrupts the function of integrin alpha6beta4 at hemidesmosomes: Role in epithelial cell migration and carcinoma invasion.
Co-expression of p16INK4A and laminin 5 γ2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture.
Laminin-5 is a marker of invading cancer cells in some human carcinomas and is coexpressed with the receptor for urokinase plasminogen activator in budding cancer cells in colon adenocarcinomas.
Invade or proliferate? Two contrasting events in malignant behavior governed by p16(INK4a) and an intact Rb pathway illustrated by a model system of basal cell carcinoma.
Mammalian tolloid metalloproteinase, and not matrix metalloprotease 2 or membrane type 1 metalloprotease, processes laminin-5 in keratinocytes and skin.
52567-0&id=gr1.jpg){kind=link}
52567-0&id=gr2.jpg){kind=link}
52567-0&id=gr3.jpg){kind=link}
52567-0&id=gr4.jpg){kind=link}
52567-0&id=gr5.jpg){kind=link}
52567-0&id=gr6.jpg){kind=link}
52567-0&id=gr7.jpg){kind=link}
52567-0&id=gr8.jpg){kind=link}
52567-0&id=gr9.jpg){kind=link}
52567-0&id=gr10.jpg){kind=link}
52567-0&id=gr11.jpg){kind=link}
52567-0&id=gr12.jpg){kind=link}
52567-0&id=gr13.jpg){kind=link}