Membrane-Type Matrix Metalloproteinases in Mice Intracorneally Infected with Pseudomonas aeruginosa

Abstract

purpose. To establish the presence of membrane-type matrix metalloproteinases (MT-MMPs) in the cornea and their expression in naive and immunized mice intracorneally infected with Pseudomonas aeruginosa.

methods. Naive (unimmunized) and immunized C57BL/6J mice were infected with P. aeruginosa, and gene expression of MT-MMPs were detected by RT-PCR. Immunoblot analysis and immunostaining were also used to characterize the MT-MMP response in both sets of animals.

results. Expression of MT1-MMP, MT2-MMP, and MT3-MMP (MMP 14, 15, and 16) was detected by RT-PCR and immunoblot analysis. Of the three MT-MMPs detected, MT1-MMP exhibited the greatest expression at protein levels. In general, a bell-shaped curve was obtained for each of the MT-MMPs in naive mice, but all of them showed much less expression in the immunized mice. MT1-MMP was localized in the epithelial tissue of the cornea, whereas MT2-MMP and MT3-MMP were mainly found in the interface between the epithelium and substantia propria.

conclusions. MT1-MMP was detected and expressed to a greater extent in naive mice than MT2-MMP and MT3-MMP. Peak expression of all three MT-MMPs showed a good correlation with the overall inflammatory response.

Pseudomonas aeruginosa is an opportunistic bacterial pathogen that produces a number of toxic substances and is capable of causing corneal necrosis, ulceration, and scarring. ^{1 2 3} Recent studies from our laboratory indicate that the ability or inability to restore corneal clarity by various naive (unimmunized) inbred mouse strains infected with P. aeruginosa is based on a regulatory genetic component. ^{3 4} Consequently, C57BL/6J mice were classed as susceptible because they were unable to restore corneal clarity but became resistant if immunized intracorneally or systemically and restored corneal clarity. ^{5 6 7 8} As a result of these studies, we have used naive and immunized C57BL/6J mice to study the relationship between the expression of matrix metalloproteinases (MMPs) and corneal destruction as well as corneal wound healing. ^{9 10} Thus, recent studies from our laboratory also indicate that both MMP-9 (gelatinase B) and bacterial alkaline protease expression is upregulated in naive mice infected by P. aeruginosa, and the enhancement corresponds to the inflammatory response on a temporal basis, resulting in a bell-shaped curve. However, in immunized mice capable of restoring corneal clarity, latent MMP-2 and alkaline protease expression was not detected by zymography, whereas MMP-9 expression was still detected but downregulated in comparison to the response of naive mice. ⁹  

MMPs are members of a multigene family of zinc-dependent enzymes. MMPs and three other groups of proteases—serine, cysteine, and aspartic proteases—play a major role in normal physiologic and pathologic tissue remodeling processes, such as trophoblastic implantation, wound healing, and tumor invasion. The MMPs have been classified into four broad categories based in part on their substrate specificity and domain organization. ¹¹ They include collagenases (MMP-1, 8, 13, and 18), stromelysins (MMP-3, 10, and 11), gelatinases (MMP-2 and 9), and membrane-type MMPs (MMP-14, 15, 16, 17, 24, and 25). Regulation of MMP expression occurs at several levels such as gene transcription, translation, pro-enzyme activation, and inhibition of activated enzymes by their endogenous inhibitors, known as tissue inhibitors of metalloproteinases (TIMPs). ^{12 13 14} The membrane-type MMPs (MT-MMPs) are characterized by the presence of a C-terminal transmembrane domain that anchors the molecule to the cell membranes. It has been found that pro-gelatinase A (pro-MMP-2) is a substrate for MT1-MMP (MMP-14) that cleaves the pro-peptide of pro-gelatinase A and generates active gelatinase A. Therefore, MT1-MMP plays a dual role in extracellular matrix (ECM) remodeling through activation of pro-MMP-2 and direct cleavage of ECM proteins, including type I collagen and fibronectin. ^{15 16} Furthermore, the secretion of various bacterial factors during infection may also play a significant role in ECM breakdown. ¹⁷  

In this study, we established the presence of MT-MMPs in the corneas of both naive and immunized C57BL/6J mice infected with P. aeruginosa and performed RT-PCR, immunoblot analysis, and immunohistochemical staining to determine the expression of MT-MMPs during wound development and the healing process. By combining these various experiments, we demonstrated the significant induction of MT1-MMP, MT2-MMP, and MT3-MMP expression at both mRNA and protein levels in naive mice, which temporally correlated with the inflammatory process over a 12- to 14-day period. Therefore, these MT-MMPs may play an important role in corneal destruction during the inflammatory response. 

Methods

Bacteria

Stock cultures of P. aeruginosa 19660 (ATCC, Manassas, VA) were stored at 4°C on tryptose agar slants (Difco Laboratories, Detroit, MI) and were used for the inoculation of 50 to 75 ml of broth medium containing 5% peptone (Difco Laboratories) and 0.25% trypticase soy broth (BBL Microbiology Systems, Cockeysville, MD). Strain 19660 is hemolytic and lecithinolytic and produces exotoxin A, alkaline protease, and elastase under appropriate culture conditions. Cultures were grown on a rotary shaker at 37°C for 16 to 18 hours, centrifuged at 6000g at 4°C for 10 minutes, washed with normal saline (Travenol Laboratories, Cambridge, MA) and diluted to a concentration of 2 × 10¹⁰ colony-forming units per milliliter. A standard curve was developed to relate viable counts to optical density at 440 nm. 

Infection of Animals

All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Age-matched naive and immunized C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME), each weighing 18 to 22 g, were infected at 14 weeks of age. Before infection, they were lightly anesthetized with ether and placed beneath a stereoscopic microscope. The corneal surface was then gently incised with three 1-mm incisions using a sterile 26-gauge needle, taking care not to penetrate the anterior chamber or to damage the sclera. A bacterial suspension (5 μl) containing 10⁸ colony-forming units was topically delivered onto the wounded cornea using a micropipette with a sterile disposable tip. Controls consisted of scratched and unscratched mice that were uninfected. Mice were examined 24 hours later to verify infection. Naive C57BL/6J have previously been classified as susceptible because they are unable to restore corneal clarity, whereas immunized mice were considered resistant because the majority were able to restore corneal clarity within a few days to a few weeks. Immunization began at 6 weeks of age by administrating 0.1 ml of 10⁶ to 10⁷ heat-killed P. aeruginosa 19660 intraperitoneally weekly for 4 weeks and then “rested” for 4 weeks before corneal infection. 

Corneal Sample Collection and Processing

At selected time points after infection, mice were killed and corneas were excised. Individual samples for reverse transcriptase–polymerase chain reaction (RT-PCR), and immunoblot analysis consisted of 12 pooled corneas per time period. Immediately after isolation, corneas were rinsed in sterile saline to remove contaminating blood and then were processed for the purposes of the different assays. Control mice were treated similarly. 

Reverse Transcription–Polymerase Chain Reaction

The pooled corneas were washed once with RNase-free PBS and homogenized with TRIzol (1 ml; GIBCO, Grand Island, NY) in a 10-ml homogenizer (Wheaton, Millville, NJ). The homogenized cells were incubated at room temperature for 5 minutes. Two hundred microliters of chloroform was added to the extract, and the mixture was vortexed vigorously. The extract was centrifuged at 13,000 rpm at 4°C for 10 minutes. The aqueous phase (containing total RNA) was transferred to a new centrifuge tube. Six hundred microliters of isopropyl alcohol was added and mixed with the aqueous phase. The mixture was incubated at 4°C for 2 hours and then centrifuged at 13,000 rpm at 4°C for 5 minutes. The supernatant was removed, and the total RNA precipitate was washed once with 75% ethanol and saved (in 75% ethanol at −20°C) for RT-PCR analysis. 

The total RNA was dissolved in water treated with diethyl pyrocarbonate (DEPC), and the concentration was measured by a spectrophotometer at 260 nm. RT-PCR was performed sequentially in the same 0.65-ml, RNase-free tubes under optimized conditions, and all the reagents for RT-PCR were purchased from Perkin Elmer (Norwalk, CT). To a final volume of 10 μl for reverse transcription reaction, the following reagents were added: 1 μl of 10× PCR buffer II; 2 μl of 25 mM MgCl₂; 1 μl of 10 mM dGTP; 1 μl of 10 mM dTTP; 1 μl of 10 mM dCTP; 1 μl of 10 mM dATP; 0.5 μl of RNase inhibitor (1 U/μl); 0.5 μl of MuLV reverse transcriptase (2.5 U/μl); 0.5 μl of 50 μM random primers; 0.5 μl sample RNA (500 ng/sample), and 1 μl DEPC-treated water. The reaction was carried out at 42°C for 1 hour. The whole product of reverse transcription in each tube was amplified by PCR. To a final volume of 50 μl, the following reagents were added: 2 μl of 25 mM MgCl₂; 4 μl PCR buffer II; 0.25 μl AmpliTaq;1> DNA polymerase (2.5 U/100 μl); 3 μl specific primers (0.05 nM); and 30.75 μl DEPC-treated water. There were two negative controls: one without reverse transcriptase and the other one without specific primers. Cycle parameters were generally a 1-minute melting step at 95°C, a 1-minute annealing step at 55°C, and a 2-minute extension step at 72°C. Thirty cycles were selected for MT-MMP amplification. The specific primers for mouse MT-MMPs were designed and prepared based on the available information for these mouse genes. ^{18 19} (Information on MT3-MMP was unpublished data from GenBank.) The primers are listed below (both forward and reverse, from 5′ to 3′), mouse MT1-MMP: ACA CCC TTT GAT GGT GAA GG and TCG GAG GGA TCG TTA GAA TG; mouse MT2-MMP: GAC CTT CTC CAG CAC TGA CC and TAC CAT CTG GGG AGC CAT AC; and mouse MT3-MMP: GGA GAC AGT TCC CCA TTT GA and CGT TGG AAT GTT CCA GTC CT. A housekeeping gene (18S rRNA; Ambion, Austin, TX) was also amplified with 30 cycles and used as an internal control for the comparison of all time samples. Finally, 1% agarose gels were prepared, and the amplified specific genes were revealed by electrophoresis. 

Immunoblot Analysis

Corneal samples were homogenized as described by Brown et al. ²⁰ After homogenization in 200 μl Tris-HCl, pH 7.4 (50 mM), containing 10 mM CaCl₂ and 1% Triton X-100, the samples were centrifuged at 9000g at 4°C for 30 minutes. The concentrations of the total protein were measured with the BCA protein assay. Equal amounts of individual samples (5 μg) were mixed with 5 μl of 4× sample loading buffer (0.125 M Tris-HCl, pH 6.8, 4% SDS, 40% glycerol, and 0.02% bromphenol blue) containingβ -mercaptoethanol and boiled for 5 minutes. The samples and a prestained molecular weight marker (Bio-Rad, Cambridge, MA) were electrophoresed on 12% SDS gels and subsequently transferred to nitrocellulose membranes. The membranes were blocked for 30 minutes in Blotto (TBS containing 0.5% Tween 20, 3% nonfat milk, and 2% bovine serum albumin) and then incubated with the specific primary antibodies that included a monoclonal anti-human MT1-MMP antibody (IM 57L, 4μ g/ml; Oncogene, Manhasset, NY), a monoclonal anti-human MT2-MMP antibody (IM 48L, 4 μg/ml; Oncogene), and a polyclonal anti-human MT3-MMP antibody (AB 853, 4 μg/ml; Chemicon, Temecula, CA), respectively, on a rocker at room temperature for 2 hours. These antibodies also recognize the respective mouse MT1-MMP, MT2-MMP, and MT3-MMP. Samples without primary antibody treatment were processed as negative controls. Afterward, the blots were incubated with secondary antibodies conjugated with horseradish peroxidase (0.5 μg/ml; Boehringer Mannheim, Indianapolis, IN) at room temperature for 1 hour. Finally, the blots were developed by chemiluminescence kit (Amersham, Arlington Heights, IL) and MT1-MMP, MT2-MMP, and MT3-MMP were visualized as dark bands with molecular weights of 65, 72, and 64 kDa, respectively. 

Histologic Study

The corneas harvested on selected time points were fixed with 10% buffered formaldehyde overnight, dehydrated with increased concentrations of ethanol, and 100% xylene, infiltrated with paraffin overnight, and embedded with fresh paraffin. The tissue blocks were sectioned (4 μm) to prepare slides that were then deparaffinized, rehydrated, stained with eosin-hematoxylin, and examined under a microscope. 

Immunohistochemical Staining

The slides of corneal tissue were deparaffinized, rehydrated, and treated with proteinase K (20 μg/ml; Sigma). Normal horse serum and 3% hydrogen peroxide were applied to the slides separately to reduce nonspecific staining and to remove endogenous peroxidase. Then the slides were treated with specific primary antibodies recognizing mouse MT1-MMP, MT2-MMP, and MT3-MMP (same as used for immunoblot analysis) in a humidified chamber at 4°C overnight and washed with PBS three times, 5 minutes each time at room temperature. The M.O.M kit (Vector, Burlingame, CA), which contains nonspecific blocking solutions, a biotinylated anti-mouse IgG secondary antibody, and the VECTASTAIN ABC reagents, was applied to the slides for MT1-MMP and MT2-MMP detection following manufacturer’s instruction. A biotinylated anti-rabbit secondary antibody (Vector) and the VECTASTAIN ABC reagents were applied to the slides for MT3-MMP detection. Finally, positive staining was exhibited by diaminobenzidine (Vector) treatment as brown granules. 

Results

Corneal Inflammation Caused by P. aeruginosa

When mouse corneas were infected with P. aeruginosa, the typical inflammatory reaction previously described occurred with a peak inflammatory expression between the 5th to 7th days. Figure 1 shows the normal cornea and the infected corneas on the 6th and 9th days after infection. On the 6th day after infection, the cornea of the naive mouse was thickened because of the edema and inflammatory cell infiltration compared with the normal cornea. The majority of the inflammatory cells were polymorphonuclear leukocytes (PMNs) that were distributed throughout the entire cornea, especially in the substantia propria, accompanied by capillary formation in the cornea. On the 9th day after infection, PMNs began to diminish, but an increase in macrophages appeared. At the same time, angiogenesis was more apparent, and the number of fibroblasts also increased. The uninfected immunized mouse cornea appeared the same as the naive controls. Similar inflammatory events occurred in the corneas of immunized mice. However, the inflammation was much less intense than that of naive mice. On the 6th day after infection, although the inflammation was still apparent, the number of inflammatory cells was much lower and the thickness of the cornea did not change dramatically. On the 9th day after infection, the cornea of the immunized mouse appeared less inflamed, with only a few remaining inflammatory cells. 

MT-MMP mRNA Expression in Mouse Corneas

Initial studies were centered on determining whether normal and infected murine corneas contain mRNA for MT-MMPs. To do so, the primers for MT1-MMP, MT2-MMP, and MT3-MMP were designed and prepared. The corneas were harvested on the 0 (control), 2nd, 5th, 8th, and 11th days after the infection, and total RNA was prepared for RT-PCR analysis. When the naive mice were infected with P. aeruginosa, MT1-MMP mRNA expression in the corneas was induced and remained high from day 2 through day 11. The enhancement of MT1-MMP mRNA also appeared in the corneas of the immunized mice on the 2nd day, but it dropped quickly and was no longer detectable by day 8 (Fig. 2) . 

Normally, MT2-MMP mRNA expression was detectable but was relatively low in the corneas of the naive mice and was not detectable in the corneas of the immunized mice on day 0. Trace amounts of MT3-MMP mRNA in the corneas of the uninfected naive and immunized mice were detected. Induced MT2-MMP and MT3-MMP mRNA expression in the corneas of infected naive and immunized mice showed a pattern similar to that of MT1-MMP with the exception that the peak expression of MT2-MMP mRNA appeared a little earlier (the 5th day) and was maintained at high levels. 

There was no difference in MT-MMP expression between the scratched corneas and the unscratched corneas on day 0, indicating that corneal abrasion did not affect the mRNA levels. In addition, negative controls (samples not treated with RT or amplified without specific primers) did not exhibit any MT-MMP expression (data not shown). 

MT-MMP Protein Detection in Mouse Corneas

To confirm MT-MMP expression at the protein level and to compare MT-MMP expression in the corneas at different time points during the infection and between the naive and immunized mice, immunoblot analysis of corneal extracts was carried out. The corneal samples were collected on the 0 (control) 3rd, 5th, 7th, and 12th days after infection in naive and immunized groups. As shown in Figure 3 , there was moderate MT1-MMP expression in the corneas of normal, naive mice. When infected with P. aeruginosa, increasing levels of MT1-MMP were expressed. The peak of MT1-MMP expression was reached around the 7th day after the infection. Although MT1-MMP expression began to decrease slowly after the 7th day, the level was still higher than that of uninfected controls. Immunized mice also expressed MT1-MMP, however, the peak of MT1-MMP expression shifted to an earlier time (about 4th day) and then quickly returned to normal levels. These results corresponded to the inflammatory events occurring in the naive mice compared with immunized mice and were also consistent with the expression of MT1-MMP at the mRNA levels. 

MT2-MMP expression in the corneas was also induced by P. aeruginosa infection. The level of this enzyme in naive mouse corneas was quite low but increased with time after the infection and remained high after the 5th day. In immunized mice, this enzyme increased earlier than that in naive mice but quickly abated as MT1-MMP did in the same group of mice. MT3-MMP expression in the corneas of naive and immunized mice showed the same pattern as that of MT1-MMP, but the level of MT3-MMP was much lower at each time point. 

MT-MMP expression in the scratched corneas was comparable to the unscratched corneas on day 0. No signals were detected in any negative controls (data not shown). 

Localization of MT-MMP Expression

To localize MT-MMP expression in the mouse corneas, immunohistochemical staining was performed using the corneal samples of normal and infected naive mice. As shown in Figure 4 , positive staining of MT1-MMP, MT2-MMP, and MT3-MMP was observed in the cornea on the 6th day after infection. The distribution of MT1-MMP was primarily found in the epithelial tissue, whereas MT2-MMP and MT3-MMP were mainly localized in the interface between epithelia and substantia propria and also in the substantia propria. 

Discussion

Normally, the cornea is a transparent and nonvascular membrane. Histologically, the cornea consists of five different types of tissues including stratified squamous nonkeratinizing epithelium, Bowman’s membrane, substantia propria, Descemet’s membrane, and endothelium. The substantia propria comprises approximately 90% of the thickness of the cornea and contains fibroblasts and extracellular matrix composed of various collagens. Corneal infection caused by P. aeruginosa is characterized by corneal edema, inflammatory cell infiltration, angiogenesis, and eventually corneal perforation caused by corneal ulceration and stromal dissolution. ^{1 2 3} Clinically, despite the rapid sterilization of the ocular tissues by optimum conservative therapy, the symptoms of the infection frequently do not improve and the illness may persist for days or even weeks. One possible explanation is that a combination of host-derived inflammatory processes and extracellular bacterial factors cause the corneal destruction in P. aeruginosa infections. ^{17 21 22 23 24 25} Consequently, degradation of basement membranes (Bowman’s membranes) and ECM components lead to severe corneal damage. Evidence for this hypothesis comes from demonstration that purified proteases cleave laminin, fibronectin, proteoglycans, and various collagens. ²⁶ Furthermore, Twining et al. ²⁴ has demonstrated that bacterial proteases can indirectly contribute to keratitis through the activation of host proteases in corneal tissue. 

In addition, several types of host cells such as infiltrating PMNs, macrophages, and resident corneal cells can elicit tissue degrading proteases. Fini et al. ²⁷ described the expression of both MMP-2 and MMP-9 in normal corneal tissues, whereas studies in a excimer laser keratectomy wound model of corneal ulceration suggested that increased levels of the activated form of MMP-9 mediated the breakdown of the basement membrane underlying the epithelium. ²⁸ The studies of Matsubara et al. ²⁹ also suggested that increased levels of the activated form of MMP-2 may be important in remodeling of new matrix components in generation corneal tissue. Girard et al. ³⁰ demonstrated synthesis of collagenases and stromelysins from fibroblasts during long-term remodeling. 

The purpose of the current studies was to use RT-PCR, immunoblot analysis, and immunohistochemical staining to demonstrate and characterize the expression of MT-MMPs in naive and immunized animals during the inflammatory response associated with P. aeruginosa infection. At the present time, little is known regarding the identity and putative role of MT-MMPs in the cornea despite a recent nonocular review of MT-MMPs by Seiki. ³¹ The results described herein showed that MT1-MMP appeared to be the dominant protease, whereas smaller amounts of other MT-MMPs were also detected. The results also indicated that corneal MT-MMP expression during infection exhibited excellent correlation with our previous ocular inflammatory studies with naive mice as quantified by bacterial numbers and PMNs measured by myeloperoxidase, arachidonic acid metabolites, and cytokines. ^{32 33 34} Of particular significance is the fact that the corneal response in immunized animals showed a lower and shorter expression of all these MT-MMP enzymes. These results suggest that in a complex and dynamic host–bacterial system such as this, there appears to be a direct correlation between the ability to restore corneal clarity in immunized animals and a dampened MT-MMP response. 

Furthermore, it is now well established that all these MT-MMPs described herein, are capable of activating other proteases such as MMP-2, MMP-9, and MMP-13. ^{15 16 17 30} Therefore, the induction of MT-MMP expression in the mouse corneas infected with P. aeruginosa may be also responsible for the damage of the corneas caused by other proteases that are activated by MT-MMPs. It is significant that Ye et al. ³⁵ suggested that MMP-2 and MMP-9 along with TIMP-1 and TIMP-2 play an important role in the early stages of wound healing in rats treated with excimer laser keratectomy. Lu et al. ³⁶ also reported that stromelysin 1 may be involved in the repair of the wound bed after excimer keratectomy in rat corneas, whereas matrilysin may play a role in the epithelial wound remodeling. At present, the clinical significance of these findings as they relate to corneal infections by P. aeruginosa awaits future study. 

 Supported by National Institutes of Health Grant P30 EY04068 and National Eye Institute Grant EY11757, Bethesda, Maryland.

 Submitted for publication May 24, 2000; revised August 25, 2000; accepted September 8, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Histologic study of the mouse corneas infected with P. aeruginosa. Naive mice and immunized mice were infected with P. aeruginosa, and the corneas were harvested on the 0 (control), 6th, and 9th days after infection. Slides were prepared and stained with hematoxylin and eosin. On day 6, lots of inflammatory cells, mainly PMNs (arrows), infiltrated into the cornea, accompanied by angiogenesis (arrowheads). The infected cornea became thickened compared with the normal one. On day 9, infiltrated PMNs began to diminish while macrophages became the major inflammatory cells in the cornea. More fibroblasts formed and angiogenesis was maintained at a high level. In contrast to naive mice, immunized mice showed much less inflammation in the corneas on the same day. Magnification, ×400.

Figure 2.

 

MT-MMP mRNA Expression in the mouse corneas infected with P. aeruginosa. Naive and immunized mice were infected with P. aeruginosa, and total RNA samples were prepared on day 0 (uninfected), day 2, day 5, day 8, and day 11. Equal amounts of total RNA from individual samples (500 ng) were used for RT-PCR. After reverse transcription, specific primers for mouse MT1-MMP, MT2-MMP, and MT3-MMP were applied to the reaction system to amplify respective MT-MMPs. The cDNA for 18S RNA was also amplified and used as an internal control. Amplified genes were stained by ethidium bromide as bright bands in the dark background. M, molecular marker (100-bp ladder).

Figure 3.

 

MT-MMP Protein expression in the mouse corneas infected with P. aeruginosa. Naive and immunized mice were infected with P. aeruginosa, and samples were prepared on day 0 (uninfected), day 3, day 5, day 7, and day 12. Equal amounts (5 μg) of total protein from individual samples were loaded for electrophoresis. Then the blots were incubated with specific antibodies recognizing mouse MT1-MMP, MT2-MMP, and MT3-MMP. MT-MMPs were visualized as dark bands corresponding to their molecular weights.

Figure 4.

 

Localization of MT-MMP Expression in the mouse corneas infected with P. aeruginosa. Naive mice were infected with P. aeruginosa, and the corneas were harvested on the 0 (control) and 6th days after infection. Slides were prepared and stained with specific antibodies recognizing mouse MT1-MMP, MT2-MMP, and MT3-MMP. All three MT-MMPs were positively stained on day 6, whereas no MT-MMPs were stained on day 0. MT1-MMP was localized in the epithelial tissue (arrows). MT2-MMP and MT3-MMP, however, were found mainly expressed in the interface between corneal epithelia and substantial propria (arrowheads). Magnification,× 400.

The authors thank Julianna M. Berk, Sarah Alousi, Markku Kurkinen, and Mao Yang for their technical assistance.