Kallikrein 7 enhances pancreatic cancer cell invasion by shedding E-cadherin




Pancreatic cancer (PaC) is characterized by local invasion and early metastasis. Serine proteases have been associated with invasion and metastasis of many cancers due to their ability to degrade extracellular matrix (ECM) proteins and to activate other proteases; thus, the serine proteases expressed in PaC were investigated.


An expression profile of serine proteases was generated from both normal and malignant pancreatic tissues using a polymerase chain reaction (PCR)-based screen and differential expression of kallikrein 7 was examined by reverse-transcriptase PCR (RT-PCR) and immunohistochemical analyses. The ability of human kallikrein 7 (hK7) to cleave the epithelial cell adhesion molecule E-cadherin was tested in vitro using both recombinant E-cadherin and BxPC-3 cells and the effects of hK7 proteolytic activity on pancreatic cell invasion and aggregation were examined.


Expression profiling revealed that kallikrein 7 (KLK7) was overexpressed in pancreatic adenocarcinomas and its differential expression was confirmed by RT-PCR analysis. hK7 was observed in neoplastic cells of all tumors examined with moderate-to-intense staining in 70% of tumors examined (16/23). In contrast, only 15% of nonmalignant tissue specimens (2/13) displayed moderate hK7 staining, whereas the remaining specimens yielded weak, if any, immunoreactivity. Using in vitro assays, hK7 was shown to cleave E-cadherin and the soluble E-cadherin fragment produced significantly enhanced Panc-1 cell invasion through ECM proteins with a corresponding reduction in Panc-1 cell aggregation.


These results suggest that aberrant expression of KLK7 plays an important role in PaC and provides novel insight into the effects of elevated hK7 proteinase activity in this, and perhaps other, adenocarcinomas. Cancer 2007. © 2007 American Cancer Society.

Pancreatic cancer is a devastating disease that has the lowest survival rate (≈2%) for any solid cancer.1 In the year 2006, an estimated 33,730 new cases of pancreatic ductal adenocarcinoma (PDAC) are projected to be diagnosed, of which 32,300 are expected to die from this disease.2 This dismal survival rate is attributed to a lack of early diagnosis and effective treatments and the aggressiveness of this particular cancer; thus, the median survival for all stages of pancreatic cancer is less than 3–5 months from diagnosis, with a 5-year survival of 0.4% to 3%.3

Pancreatic cancer is characterized by local invasion of adjacent structures, perineural invasion, early metastases to lymph nodes and liver, and an intense desmoplastic stromal reaction. One of the first events in tumor cell invasion is the loss of junctional contact between adjacent cells in epithelial tumor tissue by disruption of cell-cell and cell-extracellular matrix (ECM) associations. The loss of tissue cohesiveness involves alteration in the expression and function of adhesion molecules and active proteolysis and degradation of ECM barriers by proteases.4

Proteases are involved in most stages of tumor growth and metastasis both at the primary and metastatic sites. Local proteolysis facilitates movement into and out of the vasculature, invasion of surrounding tissue, and finally growth of the tumor through neovascularization or angiogenesis.5 Serine proteases have been associated with invasion and metastasis of many types of cancer, in particular the plasminogen activators whose physiological role is degradation of the fibrin clot. Plasmin, the active form of plasminogen, has a broad, trypsin-like substrate specificity and can degrade several ECM components.6, 7 In addition, plasmin also activates certain prometalloproteases that are also highly expressed in many forms of cancer, including pancreatic cancer.8

Using a polymerase chain reaction (PCR)-based method we generated an expression profile of serine proteases expressed in both normal and carcinoma pancreatic tissues and found that KLK7, encoding a serine protease previously known as stratum corneum chymotryptic enzyme (SCCE), is overexpressed in PDACs. Based on the observation that this protease is involved in shedding and desquamation of skin cells, we determined that human kallikrein 7 (hK7) can cleave the extracellular domain of E-cadherin and that the soluble fragment thus produced can decrease pancreatic cell aggregation and enhance cell invasiveness. These results, therefore, suggest that aberrant expression of hK7 may play an important role in pancreatic cancer.


RNA Isolation

Pancreatic tumor samples were procured from the Cooperative Human Tissue Network (CHTN), which is funded by the National Cancer Institute. Tumor RNA was prepared from snap-frozen gross PDAC tissue samples obtained during pancreaticoduodenectomy (Whipple procedure) that were verified by microscopic diagnosis. Duct RNA was prepared from bulk pancreatic ducts excised from fresh nonmalignant pancreatic tissues obtained from the Department of Surgery, University of Arkansas for Medical Sciences (UAMS), from organ donors without suitable recipients. Total RNA was isolated and processed as described previously.9 Human tissues were obtained with informed consent and their acquisition and use was reviewed and approved by the UAMS Human Research Advisory Committee.

Primer Design and Serine Protease PCR Amplification

Degenerate oligonucleotide primers SerProt/F, 5′-TGGGTIYTIACIGCIGCICAYTG-3′, and SerProt/R, 5′-ARIGGICCICCISWRTCICC-3′, (I = inosine, Y = pyrimi-dine, R = purine, S = G and C, and W = A and T) were synthesized that target the conserved domains flanking the histidine and serine residues of the catalytic triad of serine proteases.10 A pool of 4 duct RNA samples and 4 tumor RNA samples were prepared and reverse-transcribed using MMLV-RT and oligo (dT) and random primers. The cDNAs prepared from pools of duct or tumor RNA were amplified using the degenerate oligonucleotide primers, subcloned into pGEM-T Easy (Promega, Madison, Wis), and the plasmid DNA was sequenced. The resulting sequences were identified by searching the GenBank database and an expression profile was generated by enumerating the number of clones of each protease found in normal and tumor tissues.

Semiquantitative RT-PCR of KLK7

The relative expression levels of KLK7 were examined by semiquantitative PCR.11 First-strand cDNAs were synthesized from total RNA from 13 PDACs (median age, 68 years; range, 43–78 years; 4 females, 9 males) and 6 nonmalignant pancreatic duct samples (median age, 29.5 years; range, 13–59 years; 2 females, 4 males) using MMLV-RT and oligo (dT) and random primers. Primers used for amplification of KLK7 were 5′-AGTGATACGCTGGGCGACAG-3′ (forward) and 5′-CCTGGGTCATTGGGTTGG-3′ (reverse) corresponding to nucleotides 259 to 713 of the mRNA sequence (GenBank #L33404). The β-actin primers 5′-GCATGGGTCAGAAGGAT-3′ (forward) and 5′-CCAATGGTGATGACCTG-3′ (reverse) were included as an internal control (GenBank #XM_004814). PCR amplification was carried out with 35 cycles at 94°C for 30 seconds, 55°C for 30 seconds, and 72°C for 1 minute. Relative expression of KLK7 was calculated as the ratio KLK7/β-actin as measured by a phosphoimager. An unpaired, 2-tailed t-test was used to compare mean relative expression (KLK7/β-actin) in nonmalignant pancreatic tissue to pancreatic tumors (n = 4).


For immunohistochemistry, formalin-fixed, paraffin-embedded tissue blocks from 23 PDACs (median age, 66 years; range, 21–83 years; 10 females, 13 males) and 13 normal pancreas tissues (median age, 44 years; range, 19–66 years; 4 females, 9 males), obtained from donor organs without pancreatic pathology or malignancy. Representative hematoxylin and eosin-stained sections from each tissue were evaluated by microscopic analysis. Sections (4 μm) were deparaffinized and rehydrated in xylene followed by graded ethanol. Antigen retrieval was performed in a 95°C water bath using 10 mM citrate, pH 6.0, for 30 minutes. Endogenous peroxidase activity was quenched by hydrogen peroxide treatment followed by a protein block (Dako serum-free protein block, Carpinteria, Calif). Sections were incubated with goat anti-hK7 antibody (R&D Systems, Minneapolis, Minn, diluted 1:800) overnight in a humidified chamber at 4°C. Immunoreactive staining was detected using a DAKO LSAB+ peroxidase system followed by hematoxylin counterstain. Immunoreactivity intensity was evaluated as absent; very fine, diffuse cytoplasmic staining (+); more coarse, clumped cytoplasmic staining, not diffuse (2+); or diffuse, coarsely clumped staining (3+).

Activation of Pro-hK7

Recombinant, pro-hK7 (100 μg/mL) (R&D Systems) was proteolytically activated using 10 μg/mL thermolysin (R&D Systems) as per the manufacturer's instructions. Thermolysin activity was subsequently inhibited using 1.8 mM phosphoramidon (Sigma-Aldrich, St. Louis, Mo). Activated, recombinant hK7 (200 ng) was suspended in activity buffer (50 mM Tris-HCl, pH 8.5, 0.15 M NaCl, 5 mM CaCl2) and its proteolytic activity was verified using a fluorogenic substrate, ES002 (R&D Systems).

E-cadherin Cleavage by hK7

For in vitro cleavage of E-cadherin, recombinant pro-hK7 (6 ng) was activated with 0.65 ng of thermolysin in 50 mM Tris-HCl, pH 7.5, 0.15 M NaCl, 10 mM CaCl2, 0.05% Brij (Sigma-Aldrich) for 2 hours at 37°C. After proteolytic activation, the mixture was split into 2 fractions and thermolysin activity was inhibited in each fraction with phosphoramidon (15 μM final concentration). In 1 fraction, hK7 activity was inhibited by the addition of chymostatin (15 μM final concentration). Recombinant E-cadherin (100 ng) (R&D Systems) was added to each sample and incubated at 37°C. At 2-hour intervals an aliquot was removed from each reaction, added to SDS-PAGE sample buffer, heated at 95°C for 5 minutes, and stored at 4°C. After the 6-hour timepoint, all the samples were resolved by SDS-PAGE, transferred to a PVDF membrane, and E-cadherin was detected by immunoblot with a polyclonal antibody (R&D Systems).

For cleavage of E-cadherin from the surface of cultured pancreatic cancer cells, BxPC-3 cells (American Type Culture Collection [ATCC], Manassas, Va) were grown to 80% to 90% confluence, washed with phosphate-buffered saline (PBS), and incubated with thermolysin-activated hK7 (400 ng) in activity buffer containing phosphoramidon at 37°C in a CO2 incubator. As a control, BxPC-3 cells were treated similarly with phosphoramidon-inhibited thermolysin to monitor endogenous E-cadherin shedding. After 2 hours a protease inhibitor cocktail was added to the cell supernatants, cell debris was removed by centrifugation, and protein concentrations determined by BCA assay (Sigma-Aldrich). Equal amounts of total protein in each sample were resolved by SDS-PAGE and soluble E-cadherin (sE-CAD) was visualized by Western blot analysis using ECL plus reagent (GE Healthcare, Piscataway, NJ) and a VersaDoc image documentation system (Bio-Rad, Hercules, Calif). The amount of sE-CAD released in each sample was quantified using Quantity One image analysis software (Bio-Rad).

Preparation of Cell Supernatant Containing Soluble E-cadherin

BxPC-3 cells were grown to 80% to 90% confluence, washed with PBS, and incubated with thermolysin-activated rhK7 in activity buffer for 2 hours at 37°C. A protease inhibitor cocktail was added to the supernatant containing sE-CAD, cell debris was removed by centrifugation, and the supernatants containing sE-CAD from multiple experiments were pooled and concentrated using Microcon YM-10 filter devices (Millipore, Bedford, Mass).

Immunodepletion of Soluble E-cadherin

To supernatant containing sE-CAD, 2 μg of polyclonal anti-E-cadherin antibody (R&D systems) was added and mixed overnight at 4°C on a rotating mixer. Immunocomplexes were removed by addition of a 50% slurry of Protein G Sepharose 4 Fast Flow (GE Healthcare), mixed for 2 hours at 4°C, followed by centrifugation. Immunoprecipitation was repeated with the resulting supernatant to ensure complete removal of sE-CAD and verified by Western blot analysis of the supernatant obtained during each step of the procedure. In control reactions, the immunodepletion procedure was repeated with a second aliquot of BxPC-3 cell supernatant containing sE-CAD in which the polyclonal anti-E-cadherin antibody was replaced with activity buffer.

Invasion Assays

Panc-1 cells (ATTC) were grown to at least 80% confluence, washed with PBS, then incubated in Versene solution (Invitrogen, La Jolla, Calif) for 5 minutes at 37°C in a CO2 incubator. Detached cells were suspended and washed once in serum-free Dulbecco modified Eagle medium (DMEM), resuspended in DMEM, and cell concentrations determined with a Z1 Coulter particle counter (Beckman-Coulter, Fullerton, Calif). Panc-1 cells (105 cells/mL final concentration) were added to immunodepleted BxPC-3 supernatant, or supernatant containing sE-CAD, and after thorough mixing an aliquot of each mixture was transferred to a cell culture insert with 8 μm pores (BD Biosciences, Bedford, Mass) coated overnight with 120 μg of Cultrex growth factor reduced basement membrane extract (Trevigen, Gaithersburg, Md) per insert. The inserts were transferred to a 24-well plate containing DMEM with 10% fetal bovine serum as a chemoattractant and incubated at 37°C in a 5% CO2 incubator for 24 hours. At the end of the incubation period, cells on the upper surface of the insert membranes were removed with cotton-tipped swabs and cells on the lower surface were fixed and stained with Diff-Quik stain (Dade Behring, Newark, Del). The number of cells in 10 random high-power fields/insert were counted and the results were normalized to the number of invading cells counted on inserts incubated with the control supernatant set at 100%. The experiments were done in triplicate and repeated thrice. Statistical analysis was performed using a paired, 2-tailed t-test using GraphPad Prism (GraphPad Software, San Diego, Calif).

Cell Aggregation Assays

Panc-1 cells were grown to at least 80% confluence, washed with PBS, then incubated with 0.01% trypsin in PBS for 10 minutes at 37°C in a 5% CO2 incubator. Detached cells were suspended in aggregation buffer (50 mM Tris-Cl, pH 8.5, 0.15 M NaCl, 5 mM CaCl2), pelleted by centrifugation, then washed once and resuspended in aggregation buffer. After addition of a protease inhibitor cocktail, the cell suspension was passed twice through a 26-gauge needle to obtain single-cell suspensions, and cell concentrations determined. Cells (2.5 × 104) suspended in aggregation buffer were mixed with immunodepleted supernatant or supernatant containing sE-CAD, then added to a 24-well plate and checked for the presence of aggregates (defined as clusters of 4 or more cells) using an inverted microscope at ×200 magnification. The cells were incubated at 37°C with shaking (115 rpm) and after 30 minutes the number of aggregates in each well was counted in 6 random fields and reported as the average number of aggregates per field. The experiments were performed in duplicate and repeated thrice. Statistical analysis was performed using an unpaired, 2-tailed t-test using GraphPad Prism software.


Serine Protease Profiling and RT-PCR Validation

To investigate the expression of serine proteases in PaC, we performed a PCR-based screen using degenerate oligonucleotides to the conserved catalytic domain of this protease family. The serine proteases thus obtained were identified by GenBank search and the number of clones of each protease found in normal and tumor tissues was enumerated. Fifteen unique transcripts were identified from the pancreatic tumor clones, of which 21% (25/119) corresponded to KLK7, previously known as SCCE.12 In contrast, none of the 123 clones generated from pancreatic duct RNA corresponded to KLK7. Thus, KLK7 mRNA appeared to be markedly elevated in pancreatic tumors.

To verify the differential expression of KLK7 in PDACs, its expression relative to the expression of β-actin was determined by RT-PCR from RNA isolated from pancreatic tumors and compared with expression in nonmalignant pancreatic duct samples (Fig. 1). Expression of KLK7 varied among the tumor tissue samples, but was significantly overexpressed in PDACs compared with nonmalignant pancreatic tissues.

Figure 1.

Semiquantitative reverse-transcriptase polymerase chain reaction (RT-PCR) analysis of KLK7 expression in pancreatic tissues. (A) cDNAs prepared from total RNA isolated from 6 nonmalignant pancreatic tissues and 13 pancreatic ductal adenocarcinomas (PDACs) were amplified using primers corresponding to KLK7 and β-actin in the presence of [α-32 P]dCTP and the PCR products were resolved by agarose gel electrophoresis and visualized with ethidium bromide. (B) The amount of each product was quantified with a phosphoimager and the relative expression was determined as the ratio of KLK7/β-actin. Statistical analysis of the relative expression of KLK7 revealed a highly significant difference in KLK7 expression between the nonmalignant and pancreatic tumor samples (n = 4).


Because protein abundance does not necessarily correlate with mRNA expression13 and overexpression of KLK7 was examined using bulk tumor tissues, we performed immunohistochemical analyses to examine hK7 protein levels in sections of PDACs compared with normal pancreatic tissues (Fig. 2). All of the PDACs sections evaluated exhibited hK7 adenocarcinoma cell immunoreactivity, with 70% of the tumors (16/23) yielding moderate (2+) or intense (3+) staining. This staining was distributed among a majority of the tumor cells in 65% of the PDAC sections evaluated (15/23). In contrast, only 15% of the normal pancreatic tissue sections (2/13) displayed moderate (2+) hk7 immunoreactivity within sporadic acinar cells, whereas 7 elicited only a weak (1+) signal. This is comparable to the intensity and distribution of hK7 staining observed in a chronic pancreatitis tissue sample. The remaining 4 normal tissue sections did not reveal the presence of any hK7 with the exception of some positive, scattered interstitial cells. Although hK7 protein levels varied among the tumor samples analyzed, the immunohistochemical analyses confirm the observation that this serine protease is overexpressed in pancreatic tumors and thus supports the notion that hK7 may play an active role in the development and/or dissemination of this neoplasm.

Figure 2.

Immunohistochemical verification and localization of human kallikrein 7 (hK7) protein expression in pancreatic ductal adenocarcinoma (PDAC) and nonmalignant pancreatic tissues. (A, B) Intense staining of hK7 in adenocarcinoma cells was observed in sections of PDACs, whereas (C, D) little hK7 immunoreactivity was observed in normal pancreas. Original magnification ×200.

In Vitro Cleavage of E-cadherin by Kallikrein 7

hK7 was initially characterized as an enzyme implicated in the degradation of intercellular cohesive structures in the stratum corneum preceding desquamation.14–16 Consistent with its perceived role in the desquamation process, hK7 has been shown to degrade the adhesive protein of the extracellular part of the corneodesmosomes, corneodesmosin17, 18 and desmocollin 1,18 thus facilitating cell shedding at the skin surface. Because classical cadherins (eg, E-cadherin) maintain intercellular adhesion in the adherens junction and have been reported to be lost during the development of pancreatic cancer,19, 20 we examined the ability of hK7 to act upon E-cadherin in an in vitro degradation assay (Fig. 3). After proteolytic activation of recombinant pro-hK7, the active protease was incubated for varying amounts of time with recombinant E-cadherin and the products of the reaction were monitored by Western blot analysis. Active hK7 was found to cleave the intact ≈120-kDa recombinant substrate (Fig. 3, solid arrow) into discrete products of approximately 80 (Fig. 3, open arrow) and 65 kDa in a time-dependent manner (Fig. 3, left). The recombinant E-cadherin appears to be exquisitely sensitive to proteolysis by hK7, as cleavage products were observed even when the reaction was terminated immediately after the components were mixed (Fig. 3A, left, 0 hour). Substrate site preferences in the recombinant E-cadherin are revealed through the rapid production of the ≈80-kDa fragment and the more gradual increase in the amount of the ≈65-kDa cleavage product over time. Upon extended periods of incubation, in vitro degradation continues until these larger fragments are completely proteolyzed (data not shown). The degradation assays were performed in the presence of phosphoramidon, an inhibitor of the thermolysin used to activate hK7; however, to verify that proteolysis of E-cadherin resulted from hK7 activity and not from residual thermolysin activity, a parallel set of reactions was performed that included chymostatin, an oligopeptide that inhibits serine proteases with chymostatin-like activity (Fig. 3A, right). Cleavage of E-cadherin was effectively eliminated by the presence of the hK7 inhibitor.

Figure 3.

E-cadherin is a substrate for human kallikrein 7 (hK7) proteolytic activity. (A) Thermolysin-activated hK7 was incubated with recombinant E-cadherin (solid arrow) for indicated times in the absence (left) or presence (right) of the hK7 inhibitor chymostatin. Reaction products were separated by SDS-PAGE and visualized by Western blot using an anti-E-cadherin antibody. Sizes of protein markers indicated on the right. Open arrow indicates cleavage product similar in size to soluble E-cadherin (sE-CAD) released from pancreatic cancer cell surface by hK7. (B) BxPC-3 pancreatic cancer cells were incubated with (lane 3) or without (lane 2) thermolysin-activated hK7. After 2 hours the cell supernatant was collected and equal amounts of total protein were separated by SDS-PAGE and sE-CAD was detected by Western blot. The relative amount of sE-CAD in each sample was quantified using ImageQuant software and indicated more than a 2-fold increase in sE-CAD released by hK7 (lane 3) compared with its constitutive release from BxPC-3 cells in the absence of exogenous hK7 (lane 2). Sizes of protein markers indicated on the left.

Cell-Based Cleavage of E-cadherin by Kallikrein 7

To examine the ability of hK7 to cleave E-cadherin from the surface of cultured pancreatic carcinoma cells, active hK7 was added to BxPC-3 cells and the release of sE-CAD into the culture supernatant was monitored by Western analysis and compared with the level of sE-CAD endogenously shed without hK7 addition (Fig. 3B). hK7 cleavage of cell-surface resident E-cadherin, rather than recombinant E-cadherin, produced a more distinct ≈80-kDa sE-CAD fragment, which was similar in size to the E-cadherin ectodomain constitutively shed from the surface of BxPC-3 cells in culture. Application of hK7 enhanced sE-CAD release more than 2-fold (Fig. 3B, compare lanes 2 and 3).

Cell Invasion Assays

To investigate the effects of sE-CAD released from pancreatic carcinoma cells by hK7 on pancreatic carcinoma cell invasion, Panc-1 cells were seeded on top of ECM-coated cell culture inserts in the presence of supernatant from BxPC-3 cells treated with hK7 (Fig. 4A, lane 4) or BxPC-3 cell supernatant immunodepleted of sE-CAD (Fig. 4A, lane 5). After 24 hours, cell migration through the ECM was evaluated and revealed that elimination of sE-CAD significantly reduced Panc-1 invasion (Fig. 4B). This observation is consistent with the hypothesis that overexpression of hK7 in PDACs enhances tumor invasiveness.

Figure 4.

Soluble E-cadherin (sE-CAD) enhances Panc-1 cell invasion and decreases Panc-1 cell aggregation. (A) sE-CAD cleaved by human kallikrein 7 (hK7) from BxPC-3 cells was removed from cell supernatants by two cycles of immunodepletion. BxPC-3 cell supernatant was incubated with buffer or anti-E-cadherin antibody followed by protein G-Sepharose and the resulting supernatants (lanes 2 and 3, respectively) were again incubated with buffer or anti-E-cadherin antibody followed by protein G-Sepharose and supernatants (lanes 4 and 5, respectively) were used for invasion and aggregation assays. sE-CAD immunodepletion in each of the supernatants was monitored by Western blot analysis. Size of protein markers indicated on the left. (B) sE-CAD enhances Panc-1 cell invasion. Panc-1 cells were seeded on top of growth factor-reduced, extracellular matrix-coated cell culture inserts in the presence of BxPC-3 cell supernatant containing sE-CAD shed by hK7 without or with immunodepletion of E-cadherin, as indicated. After 24 hours, cells on the upper surface of the insert were removed and those cells that had migrated to the lower surface were stained and counted in 10 random fields. The invasion assays were performed in triplicate and repeated thrice. For each replicate the average number of cells counted on the control inserts were normalized to 100 and the average number of cells on the immunodepleted inserts were reported relative to the control value. Vertical line represents standard deviation (SD) (n = 3). (C) sE-CAD reduces Panc-1 cell aggregation. Single-cell suspensions of Panc-1 cells were prepared and 50,000 cells were incubated in a shaking incubator in the presence of BxPC-3 cell supernatant containing sE-CAD shed by hK7 without or with immunodepletion of E-cadherin, as indicated. After 30 minutes, cell aggregates (clusters of 4 or more cells) were counted in 6 random fields for each sample. The aggregation assays were performed in duplicate and repeated thrice. For each replicate the average number of aggregates was reported for each sample. Vertical bars represent SD (n = 3).

Cell Aggregation Assays

Because E-cadherin participates in cell-cell adhesion, we sought to determine whether the sE-CAD produced by hK7 would alter cell aggregation (Fig. 4C). Single-cell suspensions of Panc-1 cells were incubated with supernatant from BxPC-3 cells treated with hK7 and cell aggregation was evaluated after 30 minutes. The formation of Panc-1 cell aggregates was then compared with cells treated with BxPC-3 cell supernatant after sE-CAD immunodepletion. The elimination of sE-CAD resulted in a significant increase in the number of Panc-1 aggregates that formed, which supports the notion that the sE-CAD produced by hK7 can disrupt cell-cell adhesion. Because removal of sE-CAD reduced cell invasion by almost half and elicited a 2-fold increase in cell aggregation, these results suggest that hK7 may contribute to pancreatic cancer tumorigenesis by enhancing the release of sE-CAD, thus disrupting epithelial cell adhesion and facilitating tumor cell invasion.


Using a PCR-based method, we generated an expression profile of serine proteases expressed in both normal and carcinoma pancreatic tissues. By this procedure, we made the novel observation that KLK7 is overexpressed in PDACs compared with nonmalignant pancreatic tissues. The significant elevation in the levels of KLK7 transcript was verified by RT-PCR analysis of individual tissue samples and a corresponding increase in hK7 protein levels was confirmed by immunohistochemical analysis of tissue sections prepared from both PDACs and unaffected pancreatic tissues. SCCE is a chymotrypsin-like serine protease that was originally purified and characterized in extracts prepared from human skin.12, 21 Initial immunohistochemical studies indicated that it was specifically expressed in keratinizing squamous epithelia, thus supporting the notion that it played a role in the turnover and/or formation of the stratum corneum.22 With the recent characterization of the entire human kallikrein gene family,23, 24 it became apparent that SCCE is a member of this serine protease subfamily located at the chromosomal locus 19q13.3-q13.4, and is now named kallikrein 7.

The prominent expression of hK7 in human skin was confirmed in a survey of tissue extracts and biological fluids using an immunofluorometric assay and its expression was also detected in other normal tissues, particularly in esophagus and kidney.25 hK7 was not detected in normal pancreatic tissue extracts in this survey of human tissues (or the concentration was below the detection limit of the assay), which is consistent with the essential lack of hK7 immunoreactivity we observed in our immunohistochemical analyses of normal pancreatic tissue sections (Fig. 2). In addition, similar to our PCR-based profiling and RT-PCR results (Fig. 1), the KLK7 transcript was not detected in normal pancreas in other studies.12, 24, 26

We observed the expression of hK7 in pancreatic tumors, which extends the aberrant expression profile of this serine protease in neoplastic tissues. Previously, the KLK7 transcript and/or hK7 protein have been found to be overexpressed in ovarian, squamous cervical, and breast cancer.26–28 Interestingly, KLK7 expression was found to be significantly down-regulated in lung adenocarcinomas compared with noncancerous lung tissue29; thus, it appears that transcriptional activation of the KLK7 gene is not a generalized consequence of epithelial tumorigenesis.

Although KLK7/hK7 has been shown to be overexpressed in several cancers, the role it plays in tumorigenesis has not been investigated. hK7 has been shown in vitro to directly cleave corneodesmosin and desmocollin 1, adhesive proteins of the extracellular part of the corneodesmosomes.18 As extracellular components of the cellular junctions derived from desmosomes, this proteolytic activity supports a role for hK7 in desquamation.

In epithelial organs cell-cell connections present constraints to cell migration. In invasive tumors, these constraints are overcome by decreasing the number of intercellular connections, which are primarily composed of adherens junctions and desmosomes. Thus, loss of these structures correlates with and appears to contribute to tumor invasion.30 E-cadherin is a key mediator of cell-cell adhesion in epithelial cells and loss of its activity is correlated with increased tumor cell invasiveness.31 For example, when E-cadherin expression was evaluated in pancreatic tumors, loss of membranous E-cadherin immunoreactivity was strongly correlated with high-grade and advanced-stage disease.20 Loss of E-cadherin activity can result either by suppression of cadherin gene expression or by posttranslational loss of function, including proteolysis by various proteases expressed by the tumor.31 After finding that KLK7/hK7 was overexpressed in PDACs, we initiated a series of studies to determine the potential role of hK7 in pancreatic cancer.

Using an in vitro assay, we found that hK7 could effectively cleave recombinant E-cadherin, resulting in a proteolytic fragment that resembled soluble E-cadherin shed from the surface of a number of cancer cell lines, both constitutively and as a result of proteolysis.32–35 In a similar fashion, cultured BxPC-3 pancreatic cancer cells, which produce high levels of E-cadherin (data not shown), also constitutively release sE-CAD. Because neither the KLK7 transcript nor hK7 protein could be detected in BxPC-3 cells by RT-PCR or Western analysis (data not shown), respectively, the constitutive release of sE-CAD from this cell line must be dependent on the action of another protease(s) or mechanism. Addition of hK7 to the cells, however, enhanced by more than 2-fold the release of sE-CAD. To examine the effects of the sE-CAD produced in this manner on tumor cell invasion, cell supernatants containing sE-CAD were applied to Panc-1 cells and found to significantly enhance cell migration through an extracellular protein matrix using a modified Boyden chamber assay. These findings are similar to those observed with sE-CAD produced by other proteases, includingmatrilysin,32, 34 stromelysin,34 and plasmin35 with other cancer cell types. Consistent with the observed enhanced cell migration after exposure to sE-CAD, Panc-1 cells exhibited significantly reduced cell aggregation in the presence of sE-CAD. These results, therefore, suggest that aberrant expression of KLK7/hK7 in pancreatic cancer facilitates the invasiveness of tumor cells by the cleavage and release of soluble E-cadherin.

In vitro studies have also shown that hK7 can catalyze the activation of interleukin-1β (IL-1β), an inflammatory cytokine that is normally activated by a highly specific cysteine protease, IL-1β converting enzyme, from pro-IL-1β.36 Some pancreatic cancer cell lines have been reported to express and secrete high levels of IL-1β, which participates in an autocrine loop to enhance cell growth and induce chemoresistance37; thus, overexpression of hK7 may also result in increased activation of IL-1β in pancreatic cancer. Therefore, in addition to increasing tumor cell invasiveness via the production of sE-CAD, hK7 may contribute to pancreatic tumorigenesis by other mechanisms that have yet to be elucidated.

Although this study provides insight into a potential mechanism by which hK7 may contribute to pancreatic tumorigenesis, and perhaps other cancers, future studies will be needed to determine whether hK7 may also directly degrade extracellular matrix proteins, activate proteins that are capable of matrix degradation, or promote tumorigenesis and metastasis by other undefined mechanisms.


We thank Dr. Gary Barone for acquisition of normal pancreatic tissues, Mr. Eric Siegel for assistance with the statistical analyses, and Dr. Maria Brattsand for generously providing SCCE antibodies.