Study of Free and Complexed Prostate-SpecificAntigen in Mice Bearing Human Prostate

Cancer Xenografts

Kent R. Buhler,* Eva Corey, James E. Stray, and Robert L. Vessella

Genitourinary Cancer Research Laboratory, Department of Urology, School of Medicine,University of Washington, Seattle, Washington

BACKGROUND. Our objective was to evaluate five preclinical prostate cancer (CaP) xeno-graft models to determine whether (1) prostate-specific antigen (PSA) formed complexes inmurine serum, (2) the percentage of free PSA (f-PSA) was characteristic of a given xenograftline, and (3) the percentage of f-PSA was similar to that in the patient at time of tumor harvest.Our fourth objective was to identify which murine serpin(s) bind(s) to PSA in vivo.METHODS. Xenografts were established from metastatic foci. The percentage of f-PSA, andtotal PSA (t-PSA) in serum of animals bearing CaP xenografts was determined by immuno-assay. Size exclusion high-performance liquid chromatography and Western blots were usedto evaluate the presence of PSA complexes in murine serum. Edman degradation was used todetermine the N-terminal sequence of complexed proteins.RESULTS. PSA was detected as both free and complexed forms in murine serum from allmice bearing the CaP xenografts. Three xenografts (related sublines) produced PSA thatresulted in low mean percentages of f-PSA (1.9–6.4%). In sera from the other two xenografts,the mean percentages of f-PSA were high (>25%); patient sera, where available at time oftumor acquisition, were in agreement. Western blots showed that murine protease inhibitorsformed complexes with PSA. Edman degradation yielded a sequence with 80% homologyover 15 amino acids with that of murine a1-protease inhibitor (a1-PI).CONCLUSIONS. Our data have shown that the majority of PSA secreted by these CaPxenografts complexes in murine serum with a protease inhibitor with high homology tomurine a1-PI and that the percentage of f-PSA is a characteristic of each xenograft line tested,which is in agreement with patient values at time of tumor harvest. These CaP xenograftsoffer opportunities for study of human PSA biology and phenomenology. Prostate 36:194–200,1998. © 1998 Wiley-Liss, Inc.

KEY WORDS: prostate cancer; prostate-specific antigen; serine proteases; serpins; pros-tate cancer xenografts

INTRODUCTION

Prostate-specific antigen (PSA) is widely used as aserum marker for assisting in the diagnosis and themonitoring of adenocarcinoma of the prostate (CaP),the leading cancer diagnosed and the second cause ofcancer death in American men [1]. Recently, there hasbeen great interest in the protease inhibitor complexesof PSA formed in serum [2–6]. Three major forms ofPSA are present in human serum: free-PSA (f-PSA),the PSA-a1-antichymotrypsin complex (PSA-ACT),and the PSA-a2-macroglobulin complex [2]. Of these

forms, only f-PSA and PSA-ACT can be measured bycurrent antibody-based serum assays. The sum ofthese two species is defined as total PSA (t-PSA). Data

Contract grant sponsor: George M. O’Brien Center Program; Con-tract grant number: NIDDK #1 P50 DK/CA 47656-03; Contract grantsponsor: Department of Veterans Affairs; Contract grant sponsor:Richard M. Lucas Foundation.*Correspondence to: Kent Buhler, Department of Urology, Box356510, University of Washington, Seattle, WA 98195. E-mail:buhler@u.washington.eduReceived 17 February 1998; Accepted 7 April 1998

The Prostate 36:194–200 (1998)

© 1998 Wiley-Liss, Inc.

from several large clinical diagnostic studies supportthe general contention that the percentage of f-PSA((f-PSA/t-PSA) × 100) is lower in patients with CaPthan in those with benign prostatic hyperplasia (BPH);this finding is especially important in patients whoset-PSA levels are in the diagnostic gray range of 4 to 10ng/ml [3–5].

The phenomena of the existence of f-PSA in serumand the lower percentages of f-PSA in CaP patients arenot yet understood. One of the hindrances to studythese phenomena has been the lack of xenografts andcell lines that clearly portray the biological features ofCaP in man. Accordingly, we [7] and others [8–11]have intensified efforts, and have developed newprostate cancer xenografts. We have begun to charac-terize our new xenografts, both phenotypically in vivoand at the molecular level, to establish suitability as aCaP biological model and permit comparisons be-tween CaP models [12,13].

In this study, we have shown that (1) PSA producedby CaP xenografts form a complex in murine serum,(2) the percentage of f-PSA is a distinguishing charac-teristic for each tumor line, (3) the percentage of f-PSAin xenograft serum reflects patient sera at time of tu-mor harvest, and (4) a murine protein with high ho-mology to a1-protease inhibitor (a1-PI) binds to PSA.

MATERIALS AND METHODS

Animals

All animal use and procedures were approved bythe University of Washington’s Animal Care Commit-tee and performed according to National Institutes ofHealth guidelines. Five CaP xenograft lines that pro-duce PSA were examined: LuCaP 23.1, LuCaP 23.8,LuCaP 23.12, LuCaP 35, and LNCaP [7,10,14]. All Lu-CaP lines were maintained in BALB/c nu/nu (athy-mic) mice (Simonsen Laboratories, Gilroy, CA) by se-rial transplantation of tumor bits as previously de-scribed [7]. LNCaP xenografts were produced bysubcutaneous injection of LNCaP cells (2 × 106 cells/100 ml, 1:1 in Matrigel) (Becton Dickinson, Bedford,MA) [15]. Blood samples from xenograft-bearing ani-mals were obtained by means of the tail vein for smallvolumes of blood or exsanguination by cardiac punc-ture for large volumes after anesthesia with a keta-mine/xylazine mixture (130 mg/8.8 mg per kg bodyweight).

PSA and Formation of PSA-ACT Complex

PSA was purified from human seminal plasma aspreviously described [16]. PSA-ACT complex wasformed in vitro by mixing 0.1 mg of PSA with 0.4 mg

of human ACT (Athens Research & Technology, Ath-ens, GA), in a volume of 1 ml of 20 mM sodium phos-phate, 150 mM sodium chloride, pH 7.4 (PBS). Pro-teins were incubated at 37°C for 120 min; the mixturewas used without further purification as a standardfor the Western blots.

For the size exclusion chromatography and for im-munoaffinity purification (vide infra), PSA was addedto murine serum, human female serum, or 1% bovineserum albumin (BSA) in PBS, to a level of 0.5 mg/ml.The samples were first incubated for 4 hr at roomtemperature with rotary inversion, and then at 4°Covernight.

Determination of PSA Levels

Determination of f-PSA and t-PSA in murine andhuman sera was performed with the Abbott AxSYMPSA Assays (Abbott Laboratories, Abbott Park, IL).Determination of t-PSA levels in high-performanceliquid chromatography (HPLC) fractions was per-formed with the Abbott IMx PSA assay, after 1:1 di-lution of the fractions.

Size Exclusion Analysis of PSA-Spiked Samples

The analysis of PSA-spiked samples was performedon a ZORBAX G-250 size exclusion column (MacmodAnalytical, Chadds Ford, PA), and a Gilson HPLC sys-tem (Middleton, WI). Samples of 20 ml were loadedonto the column and eluted with 50 mM sodium phos-phate buffer (pH 6.8), at a flow rate 0.75 ml/min. Frac-tions of 0.4 ml were collected, and PSA content wasdetermined as described above.

Isolation of PSA and PSA Complexes FromSerum Samples

Free-PSA and PSA-protease inhibitor complexeswere isolated from xenograft sera, and PSA-spikedmurine sera using an anti-PSA monoclonal antibodyaffinity column, which was prepared from CNBr-activated Sepharose 4B gel (Pharmacia Biotech, Piscat-away, NJ) according to the manufacturer’s protocol.8.3 mg of the anti-PSA monoclonal antibody 15-19A2[16], which binds to f-PSA as well as PSA-ACT, wascoupled to 1.0 g of resin in 0.1 M borate buffer, 0.5 MNaCl, pH 7.4. Resin, 0.5 ml, was poured into a columnand equilibrated with 15 column volumes of PBS. Se-rum aliquots were diluted 1:5 in PBS and passed overthe column twice. After washing with 15–20 ml of PBS(AUF280 < 0.01), PSA and PSA complexes were elutedwith 2 ml of 100 mM glycine buffer, pH 2.3; eluteswere collected in tubes containing 0.4 ml of 1.0 M Tris-HCl, pH 8.0. Samples were then desalted and concen-

PSA Complexes With Serpin in CaP Xenografts 195

trated in Centriplus concentrators 10 (10-kDa MWCO,Amicon, Inc., Beverly, MA).

Sodium Dodecyl Sulfate Polyacrylamide GelElectrophoresis and Western Blotting

Proteins were separated under reducing conditionson 4–15%, 0.1% sodium dodecyl sulfate (SDS)-polyacrylamide gradient gels (Bio-Rad Laboratories,Hercules, CA) and were transferred to nitrocellulosemembranes (0.45 mm) using standard immunoblottingprocedures. A rabbit anti-PSA polyclonal antibody(DAKO Corporation, Carpinteria, CA) or a mousemonoclonal anti-human ACT antibody (Athens Re-search & Technology) was used to detect PSA, ACT-like proteins, and/or their complexes. Confirmationwas obtained using an alkaline phosphatase conju-gated goat anti-rabbit polyclonal antibody or an alka-line phosphatase conjugated horse anti-mouse poly-clonal antibody (Vector Laboratories, Burlingame,CA) and BCIP/NBT phosphatase substrate (Kirke-gaard & Perry Laboratories, Gaithersburg, MD). Con-trol blots were performed in parallel with nonspecificrabbit IgGs or mouse monoclonal MOPC-21 antibody(Sigma Chemical Co., St. Louis, MO) substituted forthe primary detection antibody.

N-Terminus Sequencing of PSA Complexes FromXenograft Sera

PSA complexes isolated by immunoaffinity columnfrom xenograft sera were submitted for Edman deg-radation sequencing. Sequence searches were con-ducted using the FASTA program on the Swiss-Protein database of the GCG program version 9.0UNIX (Genetics Computer Group, Madison, WI).

Statistical Analysis

Statistical analysis was performed with GraphPadPRISM, version 2.0 (GraphPad Software, Inc., SanDiego, CA).

RESULTS

Determination of f-PSA Percentages in CaPXenograft Sera

The percentage of f-PSA for each xenograft line wasanalyzed by univariate analysis and the results aresummarized in Table I. Descriptive growth statisticsfor each line are described in Table I also; these statis-tics include take rate in intact males, tumor volumedoubling time for logarithmic growth in intact males,androgen sensitivity, and PSA index–a relative gauge

of PSA production as detected in serum from eachline, defined as ng PSA/ml serum/mm3 of tumor. Sta-tistical comparisons were made between the LuCaP 23sublines using Welch’s unpaired t-test for unequalvariance. The mean percentage of f-PSA of the sublineLuCaP 23.1 (6.4%) versus either LuCaP 23.8 (2.3%) orLuCaP 23.12 (1.9%) was found to be significantly dif-ferent (P < 0.001, P < 0.001, respectively), whereas nosignificant difference was observed between the sub-lines LuCaP 23.8 and LuCaP 23.12 (P > 0.06). In con-trast to all three LuCaP 23 sublines, the LuCaP 35 andLNCaP xenografts exhibited more than twofoldhigher percentages of f-PSA. Furthermore, these twoxenografts had a lower PSA index. For all five xeno-graft lines tested, there was no observed correlationbetween tumor volume and percentage of f-PSA (r2,0.00–0.15). Although t-PSA and tumor volume in-creased over time as expected, the percentage of f-PSAdid not vary significantly upon serial bleeding of in-dividual animals (slope, −0.0014 to −0.0005; r2, 0.62–0.84). However, the percentage of f-PSA seems to havean inverse relationship to PSA index in these xeno-grafts.

Size Exclusion Analysis PSA-Spiked Samples

The results in Figure 1 show that there are two PSAimmunoreactive peaks in the murine serum. In thehuman serum, two immunoreactive peaks with simi-lar retention times to murine serum were observed. APSA-containing species with a higher molecularweight than f-PSA was evident in the human and mu-rine serum samples containing added PSA. From pro-tein standards, we determined that the molecularweight of the first immunodetected peak was withinthe range of 158 to 66 kDa, corresponding to the PSA-ACT standard, and the second peak was between 45and 25 kDa, corresponding to the seminal plasma PSAstandard. Only one immunodetectable peak corre-sponding to PSA was seen in the BSA control.

Western Blot Analysis

The Western blot analysis using anti-PSA poly-clonal antibody of the affinity purified PSA-containingspecies confirmed that murine sera (PSA-spiked se-rum and pooled xenograft sera) contained complexedPSA. The complexed PSA migrated slightly furtherthan human PSA-ACT (Fig. 2A), and the majority ofPSA complexed in murine serum to one protein. How-ever, there were two additional less abundant anti-PSA reactive bands present (Fig. 2A). The PSA-murineserpin complexes, whether produced in vivo or invitro, were approximately 10 to 15 kDa smaller thanhuman PSA-ACT. A f-PSA band was detected in all

196 Buhler et al.

TABLE I. Descriptive Statistics of the Percentage of f-PSA Detected in Serum and Phenotypic Characteristics of thePSA-Producing CaP Xenografts Tested*

Parameter

Xenograft line

LuCaP 23.1 LuCaP 23.8 LuCaP 23.12 LuCaP 35 LNCaP

DescriptivePatient seraa (percent f-PSA) 7.6 7.6 7.6 48.9 NAPatient t-PSA (ng PSA/ml) 7113 7113 7113 11.4 NAMean 6.4 2.3 1.9 25.7 25.8Median 5.7 2.1 1.9 25.3 25.8Standard deviation 3.0 0.8 0.6 5.4 5.0Sample number 56 20 26 31 31Coefficient of variation 0.48 0.35 0.30 0.21 0.20Range 1.2–16.5 1.4–4.1 1.0–3.3 17.2–38.2 16.3–34.6PSA serum range (ng/ml)b 5–893 82–856 228–1100 2–233 1–382

Phenotypic characteristicTake ratec (%) 100 90 67 87 75Doubling time (days) 11 15 21 18 8PSA indexd

Mean 1.1 1.6 5.5 0.02 0.2Median 0.8 1.3 5.6 0.02 0.2Standard deviation 0.6 1.0 2.2 0.01 0.1Range 0.3–2.8 0.5–6.3 1.1–10.5 0.002–0.05 0.09–0.3

Androgen sensitivitye

PSA + + + + +Tumor volume +/− +/− + +/− −

*PSA, prostate-specific antigen; f-PSA, free PSA; t-PSA, total PSA; CaP, adenocarcinoma of the prostate; NA, not available. Percentageof f-PSA is calculated as (f-PSA/t-PSA) × 100.aPercentage of f-PSA measured in patient donor serum at time tissue was taken for establishment of the xenograft line.bPSA serum range (ng PSA/ml) is the range of t-PSA levels of xenograft serum used in the percentage of f-PSA analysis.cTake rate for intact males.dPSA index is nanograms of PSA per milliliter of serum per cubic millimeter of tumor.eAndrogen sensitivity is classified by a regressive response to castration: +, positive (regression); +/−, static; −, no response.

Fig. 1. Size exclusion high performance liquid chromatographyfractions versus immunodetected PSA content in mouse serum.Total prostate-specific antigen (PSA) content of each fraction wasdetermined. PSA complex was determined to run in the range of158 to 66 kDa and f-PSA in the range of 45 to 25 kDa.

Fig. 2. Western blots of prostate-specific antigen (PSA) and PSAcomplexes isolated from murine serum. Anti-PSA polyclonal anti-body stained (A) and anti-human antichymotrypsin (ACT) mono-clonal antibody stained (B). Lane 1: comparative standards, hu-man PSA-ACT and f-PSA. Lane 2: PSA (and complexes) isolatedfrom xenograft serum. Lane 3: PSA (and complexes) isolatedfrom PSA-spiked mouse serum.

PSA Complexes With Serpin in CaP Xenografts 197

the samples (including smaller bands of degradedPSA). Western blot analysis of xenograft sera frommice bearing each xenograft line showed the presenceof equivalent bands for the PSA complexes in eachline, as detected by the anti-PSA polyclonal antibody.

Results of Western blot analysis using the mouseanti-human ACT monoclonal antibody detected bandsfor the major PSA complex equivalent to those de-tected by the anti-PSA antibody (Fig. 2B). This findingsuggests that the serpin homolog that complexes withPSA in murine serum exhibits at least partial homol-ogy with human ACT based on the observed cross-reactivity with the anti-human ACT monoclonal anti-body. This result was observed, whether the complexwas formed in vitro by adding PSA into serum, orproduced in vivo.

N-Terminus Sequencing of PSA Complexes FromXenograft Sera

The N-terminal amino-acid sequence of the murineserpin that binds to PSA was determined. This se-quence was queried using the FASTA programagainst murine proteins. The best match to the querysequence was murine a1-protease inhibitor (a1-PI)precursor, which had an 80% amino-acid homologyover 15 amino acids. Alignment of this match corre-sponds to amino-acid residues 30–47 (Swiss-ProteinBank sequence p22599): TDTSQKDQSPASHEIAT(matching amino acids are in bold). This sequence cor-responds to amino acid 5 from the N-terminal of theactivated murine serpin.

DISCUSSION

Murine xenografts are important models of manyhuman cancers, including CaP. In this study, we de-cided to investigate the presence of PSA forms in CaPxenografts. We have shown that PSA exists in thesefive xenografts in mice, in two dominant immunode-tectable forms as in man: f-PSA and PSA-serpin com-plex. Based on our results, we have arbitrarily dividedthe CaP xenografts into two groups using the meanpercentage of f-PSA. The first group with low percent-ages of f-PSA (1.9–6.4%; LuCaP 23 series) is similar tothe predictive clinical range in man; and a secondgroup with higher mean percentages of f-PSA (>25%;LuCaP 35 and LNCaP). High percentage of f-PSA isnormally associated with benign disease in man, whenthe t-PSA level is within the diagnostic gray zone be-fore treatment. However, in this study the PSA wasproduced in preclinical models that represent ad-vanced disease, where all of the PSA is derived frommalignant cells. We have made numerous observa-tions in which the percentage of f-PSA in patients with

metastatic disease after radical prostatectomy can ex-ceed 25% [17].

The first group consists of the LuCaP 23 sublines,which exhibited consistently low percentage of f-PSA(<13 %; 95th percentile) and high PSA index (>1.1). Allthree LuCaP 23 sublines originated from the same pa-tient, but from different metastatic sites; LuCaP 23.1and LuCaP 23.8 from two different lymph node me-tastases, and LuCaP 23.12 from a liver metastasis. Al-though LuCaP 23.1 and LuCaP 23.8 share many simi-lar characteristics (e.g., growth rate, PSA index, andresponse to androgen ablation) the mean percentageof f-PSA may be a distinguishing characteristic be-tween these two sublines (6.4% and 2.3%, respec-tively). We found statistical significance in this com-parison due to the large data set (LuCaP 23.1, n = 56).However, the range of percentage of f-PSA seenamong individual mice bearing either LuCaP 23.1 orLuCaP 23.8 xenografts overlap; therefore, the percent-age of f-PSA could not be used to discriminate be-tween individual mice of these lines. In contrast tothese two sublines, the LuCaP 23.12 subline exhibits aslower growth rate, a higher PSA index, a significantlylower take rate after serial passage, and a greater re-sponse to androgen ablation, but the mean percentageof f-PSA of LuCaP 23.12 is nearly identical to thatproduced by LuCaP 23.8. These observations supportthe theory that metastatic foci within a given patientcan be heterogeneous in phenotype and this heteroge-neity can yield different responses to therapy [18,19].

LuCaP 35 and LNCaP xenografts compose the sec-ond group. In these xenografts, the mean percentageof f-PSA is significantly higher (>25%), whereas thePSA index is 1–3 magnitudes lower compared withthe LuCaP 23 series. It is interesting that in these twoxenografts with a low PSA index, there were highmean percentages of f-PSA. The high percentage off-PSA could be due to the type of PSA produced by thexenografts, or due to interactions of the PSA with hostfactors. For example, the f-PSA may be the zymogenform [20,21], a nicked form [22], an aberrant form, orother form resulting from posttranslation modifica-tion. We have observed at least two forms of f-PSAproduced by LNCaP in vitro; one has the characteris-tics of the zymogen form (i.e., activated by trypsin)and the other form is not activated by trypsin and isreferred to as a stably inactive form [20]. Clearly, morePSA-producing CaP xenografts need to be developedand studied to define this further.

An important feature of these xenografts, whichshows that these models retained at least some of thecharacteristics of the original tissue, is that the per-centage of f-PSA in serum from mice bearing thesexenografts corresponds approximately to the patientpercentage of f-PSA. For example, in the LuCaP 23

198 Buhler et al.

series, the mean percentage of f-PSA ranged from 1.9to 6.4% compared with the patient serum, which hada value of 7.6% at the time of tumor acquisition. InLuCaP 35, the mean percentage of f-PSA was 25.3%versus the patient value of 48.9%. Unfortunately, thefifth xenograft was derived from the LNCaP cell line,and a patient serum sample for f-PSA assay was notavailable. In addition, preliminary characterization ofone of our new xenografts, LuCaP 58, seems to followthis trend–mean percentage of f-PSA in xenograft is33.6% ± 10.5%, versus patient serum value of 32.4%.

The Western blot data and size exclusion analysisclearly demonstrated the presence of f-PSA and PSAcomplexes in serum of mice bearing CaP xenografts.However, for the Western blot experiments, we usedan anti-PSA polyclonal antibody that exhibits cross-reactivity with human glandular kallikrein 2 (hK2)[23]. Therefore, we exclude from consideration thatpart of the detected band is due to this cross-reactivity.We have determined that the xenografts express thehK2 mRNA by RT-PCR (data not shown), but thepresence of protein was not evaluated due to the lackof hK2-specific antibodies. There were no PSA frag-ments detected in serum from these xenografts. Thisobservation is in agreement with the findings of Mi-kolajczyk et al. [21] in human sera. However, we can-not rule out that these PSA fragments were present atlow levels and were not detected by the techniquesused.

After examining the percentage of f-PSA as a pa-rameter of characterization of CaP xenografts, wewere also interested in identifying the murine proteinsthat complex with PSA, because the description/existence of murine ACT has not been reported. Inmice, the serpin family of protease inhibitors includestwo variants of a1-PI (E and T) and their isoforms,contrapsin, and countertrypsin [24–32]. These reportsshow that the murine serpins differ in specificity fromtheir human counterparts, but together cover a similarrange of inhibitory activity [28]. Our data suggest thatthe major species of complexed PSA in murine serumcontains a protein that is murine a1-PI, or proteinhighly homologous to murine a1-PI. If the protein isa1-PI, differences between our sequence and murinea1-PI sequence p22599 could be due to variances be-tween animal strains: the murine a1-PI sequencep22599 was determined from the inbred mouse strainC57BL/6, whereas our xenografts were grown in theinbred mouse strain BALB/c.

CONCLUSIONS

Here we report that in mice, as in man, PSA pro-duced by human CaP xenografts exists in two majorimmunodetectable forms: f-PSA and a complexedPSA. The percentage of f-PSA varies widely (1.0–

46.7%) among the xenografts, but is a consistent char-acteristic of each individual xenograft line. These pre-clinical models had similar values of percentage off-PSA compared with their tissue donors. The majorPSA-containing complex in murine serum is formedbetween PSA and a serpin, which is either identical orhighly homologous to murine a1-PI. These CaP xeno-grafts offer opportunities for study of human PSA bi-ology in phenomenology.

ACKNOWLEDGMENTS

Abbott Laboratories (Abbott Park, IL) generouslyprovided reagents for t-PSA and f-PSA determinationsin these studies. We thank Sarah C. Whitney for run-ning the PSA assays and Dr. William J. Ellis for con-tributing to the development of the LuCaP series xe-nografts.

REFERENCES

1. Parker SL, Tong T, Bolden S, Wingo PA: Cancer statistics, 1996.CA Cancer J Clin 1996;65:5–27.

2. McCormack RT, Rittenhouse HG, Finlay JA, Sokoloff RL, WangTJ, Wolfert RL, Lilja H, Oesterling JE: Molecular forms of pros-tate-specific antigen and the human kallikrein gene family: Anew era. Urology 1995;45:729–144.

3. Luderer AA, Chen Y, Soriano TF, Kramp WJ, Carlson G, CunyC, Sharp T, Smith W, Petteway J, Brawer MK, Thiel R: Measure-ment of the performance of prostate-specific antigen in the di-agnostic gray zone of total prostate-specific antigen. Urology1995;46:187–194.

4. Oesterling JE, Jacobsen SJ, Klee GG, Pettersson K, Piironen T,Abrahamsson P, Stenman U, Dowell B, Lovgren T, Lilja H: Free,complexed and total serum prostate-specific antigen: The estab-lishment of appropriate reference ranges for their concentra-tions and ratios. J Urol 1995;154:1090–1095.

5. Stenman UH, Leinonen J, Alfthan H, Rannikko S, Tuhkanen K,Alfthan O: A complex between prostate-specific antigen anda1-antichymotrypsin is the major form of prostate-specific anti-gen in serum of patients with prostatic cancer: Assay of thecomplex improves clinical sensitivity for cancer. Cancer Res1991;51:222–226.

6. Christensson A, Laurell CB, Lilja H: Enzymatic activity of pros-tate-specific antigen and its reactions with extracellular serineprotease inhibitors. Eur J Biochem 1990;194:755–763.

7. Ellis WJ, Vessella RL, Buhler KR, Bladou F, True LD, Bigler SD,Curtis D, Lange PH: Characterization of an androgen sensitive,PSA producing prostatic carcinoma xenograft: LuCaP 23. ClinCancer Res 1996;2:1039–1048.

8. van Steenbrugge GJ, van Weerden WM, de Ridder CMA, vander Kwast TH, Schroder FH: Development and application ofprostatic xenograft models for the study of human prostate can-cer. In: Motta M, Serio M (eds): ‘‘Sex Hormones and Antihor-mones in Endocrine Dependent Pathology: Basic and ClinicalAspects.’’ Amsterdam: Elsevier; 1994:11–22.

9. Nagabhushan M, Miller CM, Pretlow TP, Giaconia JM, Edge-house NL, Schwartz S, Kung H, de Vere White RW, GumerlockPH, Resnick MI, Amini SB, Pretlow TG: CWR22: The first hu-man prostate cancer xenograft with strongly androgen-dependent and relapsed strains both in vivo and in soft agar.Cancer Res 1996;56:3042–3046.

10. Horoszewicz JS, Leong SS, Kawinski E, Karr JP, Rosenthal H,

PSA Complexes With Serpin in CaP Xenografts 199

Chu TM, Mirand EA, Murphy GP: LNCaP model of humanprostate carcinoma. Cancer Res 1983;43:1809–1818.

11. Klein KA, Reiter RE, Redula J, Moradi H, Zhu XL, BrothmanAR, Lamb DJ, Marcelli M, Belldegrun A, Witte ON, Sawyers CL:Progression of metastatic human prostate cancer to androgenindependence in immunodeficient SCID mice. Nat Med 1997;3:402–408.

12. Bladou F, Vessella RL, Buhler KR, Ellis WJ, True LD, Lange PH:Cell proliferation and apoptosis during prostatic tumor xeno-graft involution and regrowth after castration. Int J Cancer 1996;67:785–790.

13. Williams BJ, Jones E, Kozlowski JM, Vessella RL, Brothman AR:Comparative genomic hybridization and molecular cytogeneticcharacterization of two prostate cancer xenografts. Genes Chro-mosomes Cancer 1997;18:299–304.

14. Gleave ME, Hsieh J, Wu H, von Eschenbach AC, Chung LW:Serum prostate-specific antigen levels in mice bearing humanprostate LNCaP tumors are determined by tumor volume andendocrine and growth factors. Cancer Res 1992;52:1598–1605.

15. Lim DJ, Liu XL, Sutkowski DM, Braun EJ, Lee C, Kozlowski JM:Growth of an androgen-sensitive human prostate cancer cellline, LNCaP, in nude mice. Prostate 1993;22:108–118.

16. Corey E, Wegner SK, Stray JE, Corey MJ, Arfman EW, LangePH, Vessella RL: Characterization of 10 new monoclonal anti-bodies against prostate-specific antigen by analysis of affinity,specificity, and function in sandwich assays. Int J Cancer 1997;71:1019–1028.

17. Lin DW, Noteboom JL, Blumenstein BA, Ellis WJ, Lange PH,Vessella RL: Serum percent free prostate-specific antigen (f-PSA) in metastatic prostate cancer. Urology 1998 (in press).

18. Ware JL: Prostate cancer progression: Implications of histopa-thology. Am J Pathol 1994;145:983–993.

19. Warzynski MJ, Soechtig CE, Maatman TJ, Goldsmith LC, Grob-bel M, Carothers GG, Shockley KF: DNA heterogeneity deter-mined by flow cytometry in prostatic adenocarcinoma–necessitating multiple site analysis. Prostate 1995;27:329–335.

20. Corey E, Brown LG, Corey MJ, Buhler KR, Vessella RL: LNCaPproduces both putative zymogen and inactive, free form ofprostate-specific antigen. Prostate 1998;35:135–143.

21. Mikolajczyk SD, Grauer LS, Millar LS, Hill TM, Kumar M, Rit-tenhouse HG, Wolfert RL, Saedi MS: A precursor form of PSA(pPSA) is a component of the free PSA in prostate cancer serum.Urology 1997;50:710–714.

22. Noldus J, Chen Z, Stamey TA: Isolation and characterization offree form prostate-specific antigen (f-PSA) in sera of men withprostate cancer. J Urol 1997;158:1606–1609.

23. Young CY, Seay T, Hogen K, Charlesworth MC, Roche PC, KleeGG, Tindall DJ: Prostate-specific human kallikrein (hK2) as anovel marker of prostate cancer. Prostate 1996;7(suppl):17–24.

24. Nathoo SA, Finlay TH: Immunological and chemical propertiesof mouse a1-protease inhibitors. Arch Biochem Biophys 1986;246:162–174.

25. Inglis JD, Hill RE: The murine Spi-2 proteinase inhibitor locus:A multigene family with a hypervariable reactive site domain.EMBO J 1991;10:255–261.

26. Hill RE, Shaw PH, Boyd PA, Baumann H, Hastie ND: Proteaseinhibitors in mouse and man: Divergence within the reactivecentre regions. Nature 1984;311:175–177.

27. Sinohara H, Takahara H: Mouse plasma trypsin inhibitors: Iso-lation and characterization of a-1-antitrypsin and contrapsin, anovel trypsin inhibitor. J Biol Chem 1982;257:438–446.

28. Takahara H, Sinohara H: Inhibitory spectrum of mouse contrap-sin and a-1-antitrypsin against mouse serine proteases. J Bio-chem 1983;93:1411–1419.

29. Morii M, Travis J: Amino acid sequence at the reactive site ofhuman a1-antichymotrypsin. J Biol Chem 1983;258:12749–12752.

30. Ohkubo K, Ogata S, Misumi Y, Takami N, Sinhara H, Ikehara Y:Cloning, structure and expression of cDNA for mouse contrap-sin and a related protein. Biochem J 1991;276:337–342.

31. Irvine J, Newlands GFJ, Huntley JF, Miller HRP: Interaction ofmurine intestinal mast cell proteinase with inhibitors (serpins)in blood: Analysis by SDS-PAGE and western blotting. Immu-nology 1990;69:139–144.

32. Inglis JD, Lee M, Davidson DR, Hill RE: Isolation of two cDNAsencoding novel a1-antichymotrypsin-like proteins in a murinechondrocytic cell line. Gene 1991;106:213–220.

200 Buhler et al.