INTRODUCTION

Cell cycle progression requires the coordinated performance of a series of regulating molecules that orchestrate cycle transitions through either mitogenic or antiproliferative signals (1).

The two crucial mechanisms used by cells to control the protein levels required in each step of the cycle are protein synthesis and degradation. Disruption of these two mechanisms leads to abnormal cell proliferation and oncogenesis, particularly if the derangement results in loss of control at the G1-S transition.

One of the many cell-cycle regulating molecules is cyclin A, whose role as an S-phase propeller is the consequence of its ability to complex with its catalytic partner cyclin-dependent kinase 2 (CDK2). Subsequently, the latter acts as a modulator of the behavior of effector proteins such as the Kip/Cip family CDK inhibitors p21CIP1 and p27KIP1, transcription factor E2F-1, DNA replication and repair factor proliferating cell nuclear antigen (PCNA), and S phase kinase-associated proteins 1 and 2 (p19SKP1 and p45SKP2), among others (2).

In recent years, the mechanisms of protein degradation have attracted much attention and intense efforts have been made to elucidate the intricacies of the machinery involved in the ubiquitin proteasomal pathway, which plays a paramount role in the degradation of short-lived regulatory proteins involved in the cell cycle (3). The two main steps followed by the ubiquitin proteasomal pathway are attachment of multiple ubiquitin molecules to a protein substrate and degradation through the 26S-proteasome complex. Transference of ubiquitin to the substrate requires at least the collaboration of an ubiquitin-activating enzyme (E1) and an ubiquitin-conjugating enzyme (E2). Oftentimes, however, improvement of substrate recognition requires the cooperation of a third component termed ubiquitin–protein ligase (E3).

Recently, a novel class of E3 ubiquitin ligases named SKP1/CDC53(cullin)/F-box protein complex (SCF) has been described. This E3 class is involved in the degradation of several key cell-cycle regulatory proteins such as p27KIP1, p21CIP1 (4), E2F-1 (5), cyclin A (2), cyclin D (4), and cyclin E (6). The SCF complex consists of the invariable components p19SKP1, cullin-1 (Cul1 or CDC53) and regulator of cullins/RING box protein (Roc1/Rbx1) as well as a variable component known as F-box proteins. The latter include p45SKP2, which by binding to p19SKP1 is responsible for substrate specificity (7). The SCF core complex is localized to the cytoplasm until p45SKP2 is expressed in late G1 and moves the SCF/p45SKP2 complex into the nucleus (Diagram 1), where it promotes the ubiquitination of selected proteins. Nevertheless, a frequent splice variant named C-terminal variant (p45SKP2-CTV) is found in the cytoplasm of various cell lines, in most cases associated with the usual form of p45SKP2. Probably owing to this cytoplasmic mislocation, p45SKP2-CTV is unable to achieve proper ubiquitination of the selected proteins (8). The C terminal domain may contain a cytoplasmic retention domain that is dominant over the nuclear localization signal of p45SKP2 failing to bind p19SKP1 (9).

Diagram 1
figure 4

General model of p45SKP2–p27KIP1–cyclin/CDK2 interaction. The SKP1/CDC53(cullin)/F-box protein complex (SCF) is involved in the degradation of several key cell-cycle regulatory proteins such as p27KIP1, p21CIP1, cyclin A, cyclin D, and cyclin E. The SCF core complex is localized to the cytoplasm until p45SKP2 is expressed in late G1 and moves the SCF/p45SKP2 complex into the nucleus, where it promotes the ubiquitination of selected proteins. p27KIP1 degradation is mediated by the SCF/p45SKP2 complex and by ubiquitin-independent processing during progression from G1 to S phase (6). The association between loss of p27KIP1 protein and uncontrolled proliferation of cancer cells is congruent with the function of p27KIP1 as a negative regulator of cyclins E and A, which in complex with CDK2 drive cells into the S-phase (15). P45SKP2 is frequently over-expressed in transformed cells, induces S phase in quiescent cells, and is a suspected proto-oncogene in human tumors. It has been postulated that low levels of p27KIP1 in aggressive human cancers may be caused by increased expression of p45SKP2 that targets p27KIP1 for ubiquitin-mediated degradation.

Full size image

The fact that p27KIP1 is rapidly degraded during late G1 phase in most cell types suggests that this cyclin-kinase inhibitor plays an important role in cell-cycle control, especially in regard to the G–S transition. In various tumors decreased CDK inhibitor p27KIP1 levels are associated with a poor prognosis (8, 9, 10, 11, 12). The low levels of p27KIP1 tumor suppressor may be due to either transcriptional and translational reduction, as in prostatic hyperplasia (11), or posttranscriptional protein degradation by the ubiquitin proteasomal pathway, as in numerous malignant neoplasms. Tumor-specific mutations in the p27KIP1 gene, on the contrary, seem to be exceptional events.

Recent studies have demonstrated that p27KIP1 degradation (Diagram 1) is mediated by the SCF/p45SKP2 complex and by ubiquitin-independent processing during progression from G1 to S phase (6). P45SKP2 specifically interacts with p27KIP1 only when the CDK inhibitor is phosphorylated by cyclin E-CDK2, thereby promoting the ubiquitination and degradation of p27KIP1. P45SKP2 is frequently over-expressed in transformed cells, induces S phase in quiescent cells (13), and is a suspected proto-oncogene in human tumors. In fact, there are recent reports of increased levels of p45SKP2 in association with reduced p27KIP1 levels in epithelial neoplasms (13, 14, 15, 16). On the other hand, cyclin E expression is a periodic event that reaches maximal levels at the G1-S transition and cyclin E degradation is mediated by ubiquitin-dependent proteolysis. Both CDK2-associated and free forms of cyclin E appear to be targets for ubiquitination and rapid degradation, whereas binding of p45SKP2 is one of the events involved in the proteolysis of free (but not of CDK2-associated) cyclin E (6). Apart from their potential action as mediators of ubiquitin-dependent cyclin A proteolysis, p19SKP1 and p45SKP2 may also directly regulate the kinase activity of cyclin A-CDK2 (17).

Kaposi’s sarcoma is an angiohyperplastic disease mediated by inflammatory cytokines and angiogenic growth factors in a setting of human herpesvirus type 8 infection. In advanced stages Kaposi’s sarcoma may behave as a multifocal neoplasm whose various lesions display monoclonality (18). Immunosuppression is a triggering factor for Kaposi’s sarcoma development, as demonstrated by the common occurrence of AIDS-associated Kaposi’s sarcoma, which usually exhibits a much more aggressive behavior than classic Kaposi’s sarcoma. In the skin, the organ primarily targeted by Kaposi’s sarcoma, lesions are clinicopathologically classified into macules, plaques and tumors in agreement with their progression in severity (19). In aggressive cases, usually of the AIDS-associated Kaposi’s sarcoma form, the lesions may also involve extracutaneous locations.

We have previously demonstrated that decreased immunoreactivity for the cell-cycle regulator p27KIP1 correlates with higher stage and extracutaneous involvement in Kaposi’s sarcoma (20). To the best of our knowledge, studies of p45SKP2 expression have never been performed in Kaposi’s sarcoma. Thus, the aim of this study was to determine whether p45SKP2 expression is altered in aggressive Kaposi’s sarcoma lesions and explore the possible relation of p45SKP2 expression to p27KIP1 down-regulation and other variables such as Ki67 expression, gender, and human immunodeficiency virus infection.

MATERIALS AND METHODS

Human Tissue Samples

Cutaneous and extracutaneous Kaposi’s sarcoma biopsy paraffin blocks were retrieved from the files of the Department of Pathology of Hospital Germans Trias i Pujol, Badalona, Barcelona, Spain. Paraffin blocks of extracutaneous cases of Kaposi’s sarcoma were also provided by the Departments of Pathology of Hospital de la Santa Creu i Sant Pau and Hospital Prínceps d’Espanya, Barcelona, Spain. We evaluated 47 cases of Kaposi’s sarcoma, of which 35 corresponded to AIDS-associated Kaposi’s sarcoma and 17 to classic Kaposi’s sarcoma. One of the 17 classic Kaposi’s sarcoma cases followed an aggressive course and showed early extracutaneous involvement. The 47 Kaposi’s sarcoma cases provided 32 skin biopsy specimens (8 macules, 10 plaques, and 12 tumors) and 15 biopsy specimens from extracutaneous locations (6 from lymph nodes, 4 from oral mucosa or conjunctiva, 2 from the gastrointestinal tract [including the 1 case of aggressive classic Kaposi’s sarcoma], 1 from the larynx, 1 from the lung, and 1 from soft tissues).

The stage of cutaneous Kaposi’s sarcoma cases was determined by histopathologic study of hematoxylin-eosin stained sections. Macular stage lesions consisted of a superficial or mid-dermal proliferation of collagen-dissecting jagged capillary vessels that disposed themselves around normal dermal structures. In some instances the newly formed vessels were confluent, but the spindle-cell component was always inconspicuous. A diagnosis of plaque stage Kaposi’s sarcoma was made when the lesion consisted of a proliferation of malformed vascular channels that dissected collagen fibers and contained only isolated spindle-shaped cells or small groups of them. Cases were classified as nodular or tumor phase Kaposi’s sarcoma when the entire lesion or most of it showed a compact proliferation of spindle-shaped cells with an intersecting fascicle-like pattern alleviated only by some inflammatory cells, erythrocytes, or telangiectatic spaces. All tissue specimens had been fixed in neutral-buffered formalin and routinely processed.

Antibodies and Immunohistochemical Studies

Immunohistochemistry studies were performed using polyclonal anti-full-length SKP2 p45 antibody (Santa Cruz Biotechnology, Santa Cruz, California, USA, diluted 1:500 with phosphate-buffered saline (PBS)), anti-p27 protein mouse monoclonal antibody, clone 1B4, (Novocastra, Newcastle, UK, diluted 1:40 with phosphate-buffered saline (PBS)), and NCL-Ki67-MM1 mouse monoclonal antibody (Novocastra, diluted 1:50 with phosphate-buffered saline (PBS)). Five-micron-thick sections were deparaffinized, hydrated, immersed in buffered citrate and autoclaved. Afterwards, the sections were incubated for 30 minutes in rabbit serum. Incubations with primary antibodies were carried out for 22 hours at room temperature. Slides were washed and incubated with biotinylated rabbit anti-mouse Ig antibodies at a 1:700 dilution and then incubated in PBS/6% hydrogen peroxide for 15 minutes at room temperature before avidin-biotin peroxidase complex addition (Dakopatts, Glostrup, Denmark). The chromogen 3, 3′-diaminobenzidine tetrachloride (Serva, Heidelberg, Germany) was applied, and counter-staining was performed with Harris hematoxylin. A non-immune mouse serum was used as a negative control in this protocol.

Both elongated cells lining abnormal blood vessels and spindle-shaped cells unrelated to vessels were considered to be Kaposi’s sarcoma neoplastic cells. On evaluating p45SKP2 staining intensity, a sub-population of cells with strongly positive nuclei could be easily distinguished. The percentages of tumor cells showing p45SKP2 positive nuclei, p45SKP2 intensely positive nuclei, p27KIP1 positive nuclei and Ki67 positive nuclei were independently assessed by two researchers (RMP and MTFF) in 500 tumor cells. In macule and plaque stage lesions, where neoplastic cells were less abundant, 20 fields were evaluated. At least 200 cells were counted in every case. The quotients (positive tumor cells/total number of tumor cells counted) were converted to percentages and rounded to the nearest integer. The arithmetic mean of both observers’ scores was used for statistical evaluation. All cases in which inter-observer variation exceeded 10% were jointly re-evaluated. In addition, the intensity of cytoplasmic stain for p45SKP2 was jointly evaluated using the following score: 4, strong stain in at least 50% of cells; 3, strong stain in 25 to 50% of cells or moderate in more than 80% of cells; 2, strong stain in 5 to 25% of cells or moderate in 5 to 80% of cells; 1, moderate or strong stain in less than 5% of cells or weak in more than 5% of cells; 0, absent or weak stain in less than 5% of cells.

Statistical Study

The statistical significance of differences observed between classic Kaposi’s sarcoma and AIDS-associated Kaposi’s sarcoma lesions and between plaque lesions and tumor lesions in regard to p45SKP2, p27KIP1, and Ki67 immunostaining scores, Kaposi’s sarcoma clinical-epidemiological type, sex of patient, lesion location and clinicopathological type was determined using analysis of variance. When variances were not homogeneous or samples were not normally distributed, the Kruskal-Wallis test was used. In addition, p27KIP1 expression was compared in the same groups using the cut-off described above and the Fisher’s exact test for differences between proportions. Differences between groups were considered to be statistically significant when the p value was less than .05. The Pearson correlation coefficient with 95% confidence limits was used to test the strength of association between p27KIP1, p45SKP2 and Ki67 expressions.

RESULTS

The results are summarized Tables 1, 2, and 3. We have found p45SKP2 nuclear expression in all stages of Kaposi’s sarcoma. The average percentage of neoplastic cells expressing p45SKP2 in their nuclei was significantly increased in skin tumors (mean ± 95% confidence interval: 39.2 ± 18.8) and extracutaneous lesions (25.8 ± 17.3) as compared with macules (18.9 ± 8.2) and plaques (29.2 ± 12.0; P = .0199) (Table 1). The staining in some nuclei was distinctly intense, allowing separate quantification. The corresponding percentages were significantly increased in skin tumors (11.1 ± 12.7) and extracutaneous lesions (7.4 ± 5.8) with respect to macules (3.7 ± 2.1) and plaques (4.8 ± 3.7; P = .0126). The differences between macules and plaques were not significant and, when lesions in both stages were put together (Table 2), the differences between groups became even more significant (P = .0052). Interestingly, p45SKP2 expression in extracutaneous Kaposi’s sarcoma lesions was lower than in cutaneous tumors but higher than in plaques. As regards p45SKP2 cytoplasmic stain, we observed a marked tendency toward an increase in the advanced stages (Table 3), but this trend did not reach statistical significance (P = .0557 when macules and plaques were grouped together) owing to the non-continuous nature of the score we used. Application of the Kruskal-Wallis test to a non-continuous variable was not possible, but the corresponding means and results are shown in Tables 1 and 2 just for the purpose of comparison with nuclear staining scores.

TABLE 1 Summary of Mean Scores of p45SKP2, p27KIP1, and Ki-67 Immunostaining in Cutaneous and Extracutaneous Kaposi’s Sarcoma Lesions Showing Increasing Severity and Progression
Full size table
TABLE 2 Summary of Mean Scores of p45SKP2, p27KIP1, and Ki-67 Immunostaining in Cutaneous and Extracutaneous Kaposi’s Sarcoma Lesions Showing Increasing Severity and Progression
Full size table
TABLE 3 Comparison of the Number of Cases with Cytoplasmic p45SKP2 Immunostaining Scores (0 to 4) in Kaposi’s Sarcoma Lesions Showing Increasing Severity and Progression
Full size table

We have found Kaposi’s sarcoma progression to be associated with a decrease in p27KIP1 expression, thus confirming the results of our previous study (20) of a different set of cases. On the other hand, Ki67 expression levels were independent of the disease stage, as already demonstrated by a previous study of ours (20). Regression analysis showed no statistical correlation between p45SKP2 over-expression and loss of p27KIP1. In some Kaposi’s sarcoma skin lesions a reduced expression of p27KIP1 paralleled a high nuclear and cytoplasmic expression of p45SKP2 (Fig. 1), whereas in others high p45SKP2 levels were associated with p27KIP1 preservation (Fig. 2). Similar findings regarding p45SKP2 and p27KIP1 expression were also observed in some extracutaneous Kaposi’s sarcoma lesions (Fig. 3), while in some aggressive lesions both molecules were poorly expressed. Moreover, whereas the maximum degree of p27KIP1 down-regulation was observed in extracutaneous Kaposi’s sarcoma lesions and then in cutaneous Kaposi’s sarcoma tumors, the highest degree of p45SKP2 expression corresponded to cutaneous Kaposi’s sarcoma tumors and then to extracutaneous Kaposi’s sarcoma lesions.

FIGURE 1
figure 1

Classic Kaposi’s sarcoma tumor lesion with marked p27KIP1 down-regulation (left) and p45SKP2 over-expression in the same area (right). Slight cytoplasmic stain for p45SKP2 is also present.

Full size image
FIGURE 2
figure 2

Example of over-expression of both p27KIP1 (left) and p45SKP2 (right) in a classic Kaposi’s sarcoma tumor lesion.

Full size image
FIGURE 3
figure 3

Example of over-expression of p27KIP1 (left) and p45SKP2 (right) in a Kaposi’s sarcoma extracutaneous lesion (lung). This case also displays strong cytoplasmic positivity and variable nuclear intensities of p45SKP2 staining.

Full size image

Immunohistochemical expression of p45SKP2 was not correlated with Ki67 proliferation index, and no statistically significant differences were found in regard to patients’ sex and human immunodeficiency virus status (results not shown).

DISCUSSION

Numerous studies have shown that reduced levels of p27KIP1 protein, an inhibitor of cyclin-dependent kinases, are associated with a more aggressive course and a poorer prognosis in a large variety of carcinomas. Moreover, in the case of colon (10, 15), breast (12) and prostate cancers (11) low p27KIP1 expression provides independent prognostic information. The association between loss of p27KIP1 protein and uncontrolled proliferation of cancer cells is congruent with the function of p27KIP1 as a negative regulator of cyclins E and A, which in complex with their catalytic partner CDK2 drive cells into the S-phase (15). In the normal cell cycle, the G0/G1 phase is characterized by high p27KIP1 levels and low p45SKP2 levels. Subsequently, during the S-phase, p45SKP2 levels increase and p27KIP1 is rapidly degraded, thus allowing the promotion of cell proliferation by the conjoint action of cyclin E/CDK2 and cyclin A/CDK2 (15). P27KIP1 appears to belong to a recently recognized class of tumor suppressors in which reduced protein expression is usually not caused by genetic change. Recent studies have identified the machinery involved in p27KIP1degradation as an SCF type ubiquitin ligase complex that contains p45SKP2 as the specific substrate-recognition unit. Levels of p45SKP2 are rate-limiting for the degradation of p27KIP1 (21), and it has been postulated that low levels of p27KIP1 in aggressive human cancers may be caused by increased expression of p45SKP2 that targets p27KIP1 for ubiquitin-mediated degradation. Levels of p45SKP2 expression correlate directly with malignancy grade and inversely with p27KIP1 levels in human lymphomas (22), colorectal carcinomas (15) and oral squamous cell carcinomas (14, 16). In the normal cell cycle, p45SKP2 levels are very low in the G0/G1 phase, increase in the S-phase, and decline afterwards (2). High levels of p45SKP2 are not due just to increased proliferation, despite the direct correlation observed in lymphomas (22), inasmuch as the percentage of cells expressing high levels of p45SKP2 in colorectal carcinoma greatly exceeds the percentage of cells expected to be in the S-phase in a randomly dividing population (15). Increased p45SKP2 protein levels do not always correlate with increased cell proliferation (as assayed by Ki67 staining), which suggests that p45SKP2 alterations may contribute to the malignant phenotype without affecting proliferation (14). Our results indicate that the aforementioned observations also apply to Kaposi’s sarcoma, although the picture seems to be somewhat more complicated in this neoplasm. Specifically, the findings in need of alternative explanatory hypotheses in Kaposi’s sarcoma are the lack of an inverse correlation between p27KIP1 and p45SKP2 levels and the apparent paradox of p45SKP2 expression, which in extracutaneous Kaposi’s sarcoma lesions happens to be lower than in cutaneous Kaposi’s sarcoma tumors but higher than in Kaposi’s sarcoma macules/plaques.

Increased p27KIP1 degradation (Diagram 2) may be the result of a defective ubiquitination and degradation of p45SKP2 by either SCF/p45SKP2 action during the G0-G1 phase (23) or an alternative pathway independent of cell-cycle phase (24). In addition to p45SKP2 over-expression, increased p27KIP1 proteolysis in Kaposi’s sarcoma and other neoplasms may be caused by Myc oncogenic activation leading to Cul1 over-expression (25). P27KIP1 control may be also achieved by a second proteolytic pathway that is activated by mitogens through Ras and Myc and is operative during the G1 phase (26).

Diagram 2
figure 5

Putative explanations for the lack of inverse correlations between p45SKP2 and p27KIP1 expressions in Kaposi’s sarcoma. 1: Increased p27KIP1 degradation may be the result of a defective ubiquitination and degradation of p45SKP2 by either SCF/p45SKP2 action during the G0-G1 phase (23) or an alternative pathway independent of cell-cycle phase (24). 2: Nedd8 enhances SCF/p45SKP2 effect on p27KIP1 (27) and may regulate p27KIP1 turnover independently of p27KIP1 phosphorylation (28). 3: Rbx2 is also able to bind p45SKP2 and mediate p27KIP1 degradation (29). 4: p27KIP1 is also processed rapidly by an ubiquitination-independent mechanism that exhibits higher activity in the S phase than during the G0-G1 phase (30). 5: Increased p27KIP1 proteolysis in Kaposi’s sarcoma may be caused by Myc oncogenic activation leading to Cul1 over-expression (25) or 6: by a proteolytic pathway activated by mitogens through Ras and Myc (26). 7: Increased expression of the inactive p45SKP2–CTV splice variant (8) might provide an explanation for nuclear p45SKP2 expression levels being lower in extracutaneous Kaposi’s sarcoma lesions than in cutaneous Kaposi’s sarcoma tumors with similar cytoplasmic p45SKP2 expression. 8: human herpesvirus type 8-encoded K cyclin, which is resistant to p27KIP1 CDK inhibition, would facilitate p27KIP1 phosphorylation and down-regulation, thus enabling activation of endogenous cyclin/CDK2 complexes (31, 32). 9: p45SKP2 may also directly regulate cyclinA-CDK2 kinase activity (17) and contribute to the malignant phenotype without affecting proliferation (14).

Full size image

Nedd8, an ubiquitin-like protein expressed in proliferating cells, acts on Cul1, enhances SCF/p45SKP2 effect on p27KIP1 (27), and may regulate p27KIP1 turnover independently of p27KIP1 phosphorylation (28). Also able to bind p45SKP2 and mediate p27KIP1 degradation is Rbx2, which is the product of the sensitive-to-apoptosis gene (SAG) and the second member of the RING box protein family (29). In parallel with its ubiquitin-dependent degradation, p27KIP1 is processed rapidly by an ubiquitination-independent mechanism that exhibits higher activity in the S phase than during the G0-G1 phase (30).

On the other hand, increased expression of a p45SKP2 splice variant that localizes to the cytoplasm and fails to direct cyclin D1 (and supposedly p27KIP1) ubiquitination and degradation (8) might provide an explanation for our finding that nuclear p45SKP2 expression levels are lower in extracutaneous Kaposi’s sarcoma lesions than in cutaneous Kaposi’s sarcoma tumors with similar cytoplasmic p45SKP2 expression levels.

The fact that Kaposi’s sarcoma is related to human herpesvirus type 8 infection suggests a plausible explanation for the lack of an inverse correlation between p27KIP1 and p45SKP2 expression levels in this neoplasm. Specifically, human herpesvirus type 8-encoded K cyclin, which is resistant to the actions of p16INK4A, p21CIP1 and p27KIP1 CDK inhibitors, would bypass a p27KIP1-imposed G1 arrest by facilitating p27KIP1 phosphorylation and down-regulation and thus enabling activation of endogenous cyclin/CDK2 complexes (31, 32). Indeed, the occurrence of a p27KIP1-phosphorylating CDK6 complex in cell lines derived from primary effusion lymphoma and Kaposi’s sarcoma may well indicate that virally induced p27KIP1 degradation takes place in human herpesvirus type 8-related tumors (32).