CLIN. CHEM. 37/4, 547-551 (1991)

CLINICALCHEMISTRY,Vol.37, No.4, 1991 547

GCDFP-70 Protein in Cyst Fluid Identified as Albumin and Used to Classify Cysts in Womenwith Breast Gross Cystic DiseaseMilagros Balbin,’ Francisco Vizoso,2 Luis M. Sanchez,’ Rafael Yenta,3 Alvaro Ruibal,4 Antonio Fueyo,’ andCarlos Ldpez-Otin”5

We used sodium dodecyl sulfate-polyacrylamide gelelectrophoresis to study cyst fluids from women withbreast gross cystic disease. The subjects could be clas-sified into two categories according to the concentrationsof protein GCDFP-70 in the cyst fluid: those with Type Icysts had a very low content of this protein; those withType II cysts had very high concentrations. Analysis of theamino acid sequence of GCDFP-70 from both cyst fluidtypes confirmed that this protein is human plasma albu-min. The average concentration of albumin found in TypeI cyst fluids was 0.32 g/L and that corresponding to TypeII was 10.16 g/L. Thus, albumin quantification from cystfluids or analysis by either polyacrylamide or agarose gelelectrophoresis provides a simple procedure for classify-ing these fluids, yielding results that correlate well withprevious classifications based on other measurementssuch as sodium, potassium, and chloride concentrations.This albumin-based quantification method may improvethe classification of breast cysts and might be useful infurther studies on functional changes in the cysts and theirrelationship to breast cancer.

Additional Keyphrases: electrophoresis, polyacrylamide gel -

electrophoresis, agarose gel - cancer

Gross cystic disease is the most common pathologicalcondition of the breast, affecting about 7% ofpremeno-pausal women (1). Although cystic lesions are onlyrarely precancerous, many studies in the last two dec-ades (2-6) indicate that these patients are at about afourfold greater risk of developing breast cancer thanare control populations.

Because of the possibility that benign and malignantbreast diseasehave some factors in common (7), severalstudies have been carried out to type cysts in terms ofthe risk of further development of breast carcinoma.These studies have mainly been focused on determiningthe biochemical composition of cyst fluid. At present,data are available for the hormone content (8, 9), ioniccomposition (9, 10), enzyme activities (11), and proteincomponents (12-15). In addition, these analyses haverevealed in cyst fluid in gross cystic disease the presenceof four major protein components: GCDFP-70, -44, -24,

‘Departamento de Biologla Funcional, Facultad de Medicina,Universidad de Oviedo,33006 Oviedo, Spain.2Sarviciode Cirugfa,Hospitalde Jove, GijOn.

‘ Servicio de An#{225}lisisClmnicos, Hospital San Agustmn, Avil#{233}s.Servicio de Medicina Nuclear, Hospital General de Asturias,

Spain.To whom correspondence should be addressed.

Received October 22, 1990; accepted January 21, 1991.

and -15, proteins of respective molecular masses (inkilodaltons) corresponding to the numbers in the name.Some ofthese proteins have been studied with respect totheir molecular characteristics and hormone bindingproperties. However, very little is known about theirsource, biological functions, or possible role in the in-duction of the pathological process characteristic ofgross cystic breast disease.

Recently, we have described how GCDFP-24, a pro-gesterone-binding protein that accounts for more thanhalf of the total protein present in cyst fluid, is apolipo-protein D, a component of high-density lipoproteins inhuman plasma, which are involved in cholesterol trans-port (16). In the course of studies directed towards thecharacterization of this major component of cyst fluid,we found that the concentration of GCDFP-70 in cystfluid from different patients varied widely. Severalgroups have tentatively characterized this protein asplasma albumin by using immunological analysis (12,15, 17-19), but no structural data to confirm this assig-nation are currently available. In addition, GCDFP-70was present at concentrations 25- to 100-fold lower thanthat found for albumin in plasma (12, 17-19).

The present work was aimed at evaluating breastcysts according to the content of GCDFP-70 as well as atproviding further molecular characterization of thisprotein in the fluid filling the cysts. As we will show, wehave identified GCDFP-70 as plasma albumin and es-tablished two categories of cysts on the basis of theirconcentration of this protein, which can be related toprevious classifications of cysts based on ionic content,pH, glucose content, and cytological examination (9,20-23). The possible relationship between cyst type anddifferent functional stages of the epithelium lining thecysts are also discussed.

Materials and Methods

Samples. Forty-fivecystfluidsamples were obtained,with informed consent, by needle aspiration from pa-tients with gross cystic disease of the breast, ages 31 to50 years. Cancer was excluded by clinical, echographic,mammographic, and cytological studies. All women hadregular menstrual cycles, and none was taking anyhormonal medication at the time of the study or duringthe preceding six months. Cystic fluid aspiration wascarriedout in the mild luteal phase of the menstrualcycle,which was confirmed by measuring the concentra-tions of lutropin (luteinizing hormone), follitropin (fol-licle-stimulating hormone), and progesterone in bloodsamples. We quantified these hormones by immunora-diometric assay, using kits obtained from Ire-Medgenix

1 2 3 4 5

GCDFP

7Qi..

44.-. - - S

24j 4 $ 1 .

15.- - -

Fig. 1. SDS-polyacrylamide gel electrophoresis of total proteins fromcyst fluidsAliquots from cyst fluids (1-2 L) were treated with reducing SOSsamplebuffer and run on 10% polyacrylamide/SDSminigels (lanes Ito . Lane 5corresponds to standard Iow-M,proteins from Pharmacia LKB (94, 67, 43, 30,20, and 14 kDa). The gel was stained with Coomassie blue

548 CLINICAL CHEMISTRY, Vol. 37, No. 4, 1991

(Fleurus, Belgium) for lutropin and follitropin and fromCIS International (Gif-sur-Yvette, France) for proges-terone. Cyst fluids were centrifuged at 35 000 x g for 1h at 4 #{176}Cand stored at -20 #{176}Cuntil analyzed.

Biochemical determinations. Protein concentrationwas determined in each cyst fluid by the Bradfordtechnique (24).

Albumin was determined with the Beckman AlbuminTest in an Array rate nephelometer (Beckman Instru-ments Inc., Palo Alto, CA). The within-run coefficient ofvariation (CV) for control serum samples was <4%; thebetween-runs CV was <8%.

Sodium and potassium concentrations were measuredwith a flame photometer (Model 450; Corning MedicalLtd., Halstead, U.K.) with an internal standard of

lithium; chloride concentrations were determined witha chioridometer (Model 1200; Orejas y Maillo S.A.,Oviedo, Spain).

Protein purification. Cyst fluidcontainingabout 2 mg

of protein was applied to a 2.15 x 30 cm TSK-3000 SWGcolumn (Beckman Instruments Inc.), equilibrated, andeluted at a constant flow rate of 0.4 mL/min with 0.1mollL ammonium acetate, pH 5.0, containing 1 g ofsodium dodecyl sulfate (SDS) per liter. The sampleswere run at room temperature with a high-performanceliquid chromatograph (HPLC; Waters, Milford, MA)equipped with a Model 481 ultraviolet detector, a Model680 automated gradient controller, and two Model 510solvent-delivery systems (all from Waters). We mea-sured absorbance at 280 nm with a sensitivity setting of2.0, and 1.0-mL fractions were collected for furtheranalysis.

Polyacrylamide gel electrophoresis. Samples of cystfluid or purified proteins were analyzed by SDS-poly-acrylamide gel electrophoresisin 10% gels,at 50 mA for30 mm, according to the method of Laemmli (25). Theproteins were stained with Coomassie Blue and in somecases with silver nitrate.

Agarose gel electrophoresis. The proteinbands fromdifferent cyst fluids were separated by electrophoresis

on agarose gel as described elsewhere (26). The reagentsand apparatus for electrophoresis were purchased fromCorning Medical Ltd. (Halstead, U.K.).

Amino acid sequence analysis. Samples containingabout 0.5nmol of protein were subjected to NH2-termi-nal analysisby sequential degradation on a Model 477Asequencer (Applied Biosystems, Foster City, CA) in thepresence of Polybrene. The amino acid anilinothiazoli-none derivatives were converted to phenylthiohydan-tomderivatives in the automated conversion flask of thesequencer and were identified and quantified with anon-line phenylthiohydantoin analyzer (Model 120A; Ap-plied Biosystems) (27). Direct sequencing of proteinsseparated by SDS-polyacrylamide gel electrophoresiswas accomplished by blotting proteins onto a polyvinyli-dene difluoride membrane, staining them withCoomassie blue, and putting the membrane segmentcarrying the relevant protein directly into the sequencer(28).

Statistical analysis. Student’s t-test for unpaired datawas performed to assess differences between the twogroups of cysts. Correlation coefficients were calculatedby the least-squares method. Significance was estab-lished at the 0.05 level.

ResultsEvaluation of human breast cysts by SDS-polyacryl-

amide and agarose gel electrophoresis. For a detailedcharacterization of the protein components in cyst fluidfrom human breast gross cystic disease, aliquots (1-2j.tL) of fluids obtained by needle aspiration were sub-jected to SDS-polyacrylamide gel electrophoresis. Thisallowed us to identif’ the four major protein compo-nents, GCDFP-70, -44, -24, and -15, with GCDFP-24being the most common in virtually all cases. However,the intensity of the band corresponding to GCDFP-70varied greatly from one cyst fluid to another (Figure 1).In light of these data, we considered two categories ofcyst fluids: one group with a very low content of thisprotein (lanes 3 and 4: Type I cysts) and another groupin which the concentration of this protein was very high(lanes 1 and 2: Type II cysts).

This classification of cyst fluids, based on SDS-poly-acrylamide gel electrophoresis, was confirmed also by asimpler procedure involving agarose gel electrophoresison commercially available gels (Figure 2). Althoughthis latter method did not allow the total separation ofthe different protein components present in cyst fluid, itwas good enough to identifr two different subpopula-tions according to the intensity of a protein band havingan electrophoretic mobility identical to that of plasmaalbumin.

Purification of protein GCDFP-70 from different cystfluids. Because the above results suggested that intra-cystic fluid concentrations of GCDFP-70 could be usedas a marker for different breast cyst subpopulations, wecarried out further studies for isolation and molecularcharacterization of this protein component obtainedfrom different types of cyst fluid.

Protein components in cyst fluids from the two above-described categories of subjects were separated by size-exclusion HPLC in the presence of SDS. As Figure 3

SERUM ar I $TYPE II CYST __________

TYPEIICYST - +

TYPE I CYST #{149}TYPE I CYST

Fig. 2. Agarose gel electrophoresis of total proteins from cyst fluidsAliquots (1-2 pL) from Type land Type II cyst fluids were sublectedto agarosegel electrophoresis. An aliquot from human serum was run as control. Thearrow indicates the position of albumin from serum

shows, the chromatographic profilesdifferclearlyin theregion corresponding to an elution volume of about50-51 mL. A major peak was detected at this position influids from Type II cysts but was practically absent inthose fluids belonging to Type I cysts. Calibration of thecolumn with serum albumin (data not shown) revealedthe presence of a single peak eluting at the samevolume. Analysis by SDS-polyacrylamide gel electro-phoresis showed that fractions corresponding to thesepeaks contained a single band, the electrophoretic mo-

I

I

mm.The presence of the band corresponding to protein

GCDFP-70 in the respective fractions of the chromato-graphic eluates from most of the Type I cyst fluids wasvisible only after prolonged silver staining of the poly-acrylamide gels (Figure 3, inset).

NH2-terminal sequence analysis of protein GCDFP-70

isolated from different cyst fluid types. From the above

resultsas well as from previous immunological data,itseems reasonableto assume that proteinGCDFP-70 isplasma albumin. However, in view of the recent discov-ery of the presence of active proteases in cyst fluid (29),

we considered it necessary to provide further character-ization of the structure of the purified proteins. Thesestudies would determine whether the proteins in cystfluid show structural modifications as a consequence ofproteolytic activities. Thus, purified GCDFP-70 fromType II cyst fluids was subjected to automated Edmandegradation in a gas-liquid phase sequencer, whichallowed the identification of the first 15 amino acidresidues (Table 1). The limited amount of GCDFP-70available from Type I cyst fluids made the sequencing ofits amino terminal very difficult. To overcome thisproblem, we blotted the electrophoretically separatedproteins from this cyst fluid onto polyvinylidene difluo-ride filters before subsequent protein sequencing. Theseanalyses allowed the identification of the first eightresidues of GCDFP-70 from Type I cysts. In both casesthe amino-terminal sequences were identical to that forhuman plasma albumin (Table 1).

Quantification of albumin concentrations in the dif-ferent cyst fluids and correlation with other variables.The classification of cysts on the basis of theirGCDFP-70 content proposed above was based on visualestimation of the electrophoretic patterns of the dif-ferent cysts. However, the discovery that GCDFP-70 isidentical to plasma albumin made it possible to defineaccurately the concentrations of albumin that wouldcorrespond to each cyst subpopulation. To do so, we useda micromethod that allowed the automated determina-tion of very low concentrations of albumin. As shown inTable 2, the average concentration of albumin found inType I cyst fluids was 0.32 gIL, whereas the valuecorresponding to the Type II fluids was 10.16 g/L. Inaddition, a positive correlation was found betweenNa/K ratio and albumin content (r = 0.71, P <0.001),and between C1 and albumin concentrations (r = 0.84,P <0.001).

40 50 60 70

FRACTIONNUMBER

Fig. 3. Fractionation of GCDFP species by size-exclusion HPLC inthe presence of SDSAliquots (50 ML)fromType1(a)or Type11(b)cyst fluids were applied to aTSK-3000SWGcolumn (2.15 x 30cm) in 0.1 mol/L ammonium acetate, pH5,containing SOS,1 gIL.Theflowratewas0.4mL/min,and 1.0-mL fractionswere collected. Inset: Fractionno. 50(20 rL) from both types was lyophilized,treatedwith reducingSOSsample buffer,and run on 10%polyacrylamidelSDSminigels.The gels were stainedwith silver nitrate (a) and Coomassieblue (b)

Table 1. Amino-TermInal Sequences of GCDFP-70Proteins from Cyst Fluids

GCDFP-70 (Type I) Asp-Ala-His-Lys-Ser-Glu-Val-AlaGCDFP-70 (Type II) Asp-AIa-His-Lys-Ser-Glu-Val-Ala-

His-Arg-Phe-Lys-Asp-Leu-GlySerum albumina Asp-AIa-His-Lys-Ser-Glu-VaI-AIa-

His-Arg-Phe-Lys-Asp-Leu-Glya Sequence data obtained from ref. 30.

V bility of which was indistinguishable from plasma albu-

CLINICAL CHEMISTRY, Vol. 37, No. 4, 1991 549

Albumin, gIL Na’/K

Mean SEN Range Mean SEM Range Mean SEN Range

Type I (35) 0.32 0.05 0.05-1.7 0.9 0.3 0.05-6.0 21.4 3.6 1-72Type 11(10) 10.2” 1.3 4.7-15.6 322L 2.9 23-47 101.4” 1.3 94-108

#{149}The number of samples assayed are given in parenthesis.b P <0.001 vs Type I cysts (Students unpaired f-test).

ci-, mmoIIL

Table 2. ClassIfication of Cyst Fluids on the Basis of Albumin Concentration, and Correlation with Other Analytes

550 CLINICAL CHEMISTRY, Vol. 37, No. 4, 1991

DiscussionSeveral studies performed by this group and others

over the last few years have revealed the existence ofdistinctivepatterns in the concentrations of differentsubstances present in the cyst fluid of gross cysticdisease of the breast (9, 20-23). However, no data areavailable for the classification of cysts on the basis oftheir protein composition. Here, on the basis of poly-acrylamide and agarose gel electrophoresis analysis, wepresent evidence that cysts can be subdivided intodifferent categories according to the concentrations ofprotein GCDFP-70. In addition we confirm by aminoacid sequence determinations that this protein is hu-man plasma albumin.

The fact that women with breast gross cystic diseasedevelop breast cancer with about four times the normallyexpected frequency has focused the attention of severallaboratories on the evaluation of the possible relationshipbetween the presence of specific substances in cyst fluidand the further development of breast carcinoma. As partof our studies directed towards the identification of po-tential markers for risk of breast cancer, we performedsystematic analyses by polyacrylamide gel electrophore-sis of the aspirated fluid from patients with breast grosscystic disease. Although these analyses allowed detectionof the peculiar pattern of proteins described by severalauthors, a clear variation in the concentration of proteinGCDFP-70 was easily detectable. The analysis of a largenumber of fluids, by either polyacrylamide or agarose gelelectrophoresis, was useful to clearly establish two dif-ferent subpopulations of cysts. Subsequent molecularcharacterization of GQDFP-70 isolated from the differentcyst types showed that this intracystic protein is in allcases plasma albumin.

This identification of GCDFP-70 as human albuminallowed us to obtain the appropriate quantitative datarequired for a clear identification of separate popula-tions of cyst fluids on the basis of their concentration ofthis protein. Type I cysts have very low albumin con-centrations (about 0.3 g/L), whereas those designatedType II show high concentrations of albumin (around 10g/L). According to these results, we propose an albuminconcentration of 3 g/L as an arbitrary cutoff value toseparate the two categories of breast cysts.

Although other groups have also reported variationsin the concentration of intracysticGCDFP-70/albumin(12, 19), our results represent the first classification ofgross cystic breastdisease patients into different subsetsby considering the concentration of a protein compo-nent. In addition, the observation that albumin could be

a marker for different cyst subpopulations confirms andextends classifications defined on the basis of othermarkers, e.g., sodium and potassium concentrations,chloride content, or glucose concentration in the cysts (9,20-23). When we subdivided the breast cysts into twotypes according to their albumin content and deter-mined these biochemical markers, we obtained a highcorrelation of classifications. Thus, Type II cysts, withhigh albumin concentrations, showed a high content ofsodium and chloride and a low content of potassium. Onthe other hand, Type I cysts, with low albumin concen-trations, correlate well with low concentrations of so-dium and chloride, and with high potassium concentra-tions (Table 2).

From our results for albumin concentrations as wellas from the above-described correlations with otherprevious classifications, we propose a preliminary hy-pothesis concerning the mechanisms of the developmentof breast cysts (31-33). Type I cysts might represent aninitial stage characterized by the activity of the apocrineepithelium that surrounds the cyst. The very low con-centrations of albumin in these fluids would be theresult of a high degree of impermeability to the extra-

cellular fluid. On the other hand, Type II cysts could bethe last stage of development, with the apocrine epithe-hum being flattened. The high concentrations of albu-min found in these latter cysts could reflect a highdegree of transudative phenomena and therefore moreclosely resemble the composition of extracellular fluid.The finding of some cyst fluids with albumin concentra-tions slightly higher than the mean values determinedfor Type I cysts or with lower concentrations than thoseestablished for Type II cysts might be indicative of theevolution of the breast cystic disease by passage fromType I to Type 11(9).

In summary, albumin quantification in cyst fluidsprovides a simple procedure for classifying them. Be-cause several studies indicate that the natural history ofbreast disease varies according to the subtype of the cyst(6, 29), the characterization of subgroups of patientswith breast gross cystic disease may be of future clinicalinterest. From previous data, the risk of developingcancer would be greater in the apocrine cyst group,which corresponds to the Type I cysts with very lowalbumin concentrations. Further studies and clinicalfollow-up of patients belonging to different subgroupsare in progress to evaluate the clinical significance ofthe use of albumin concentrations in cyst fluid as indic-ative of risk of epithelial proliferative changes, includ-ing breast carcinoma.

CLINICAL CHEMISTRY, Vol. 37, No. 4, 1991 551

This work was supported by Research Grants PB 87-1045 fromComisiOn Interministerial para elDesarrollo de Ia InvestigaciOnCientfflca y T#{233}cnicaand DF 90/1661 from Universidad de Oviedo,Spain. We are grateful to Dr. Francisco V. Alvarez for suggestionsand critical reading of the manuscript. M.B. and L.M.S. arerecipients offellowshipsfrom the Fondo de Investigaciones Sani-tarias de Ia SeguridadSocialand Fundaci#{244}npara la CientfficaAphicada y Ia Tecnologf a (FICYT), respectively.

References1. Haagensen CD, Bodian C, Haagensen DE. Breast carcinoma:risk and detection. Philadelphia: WB Saunders, 1981.2. Davis HH, Simons M, Davis JB. Cystic disease of the breast:relationship to carcinoma. Cancer 1964;17:957-78.3. Monson RR, Yen 5, MacMahon B, Warner S. Chronic mastitisand carcinoma of the breast. Lancet 1976;ii:224-6.4. Page DL, Zwaag RV, Rogers LW, Williams LT, Walker WS,Hartman WH. Relation between component parts of fibrocysticdisease complex and breast cancer. J Natl Cancer Inst1978;61:1055-63.5. Haagensen DE, Mazoujian G. Relationship of glycoproteins inhuman breast cystic disease fluidto breast carcinoma. In: AngeliL, Bradlow HL, Dogliotti L, eds. Endocrinology of cystic breastdisease. New York: Raven Press, 1983:149-68.6. Dixon JM, Lumsden AB, Miller WR. The relationship of cysttype to risk factors for breast cancer and the subsequent develop-ment of breast cancer in patients with breast cystic disease. Eur JCancer Clin Oncol 1985;21:1047-50.7. MacMahon B, Cole P, Brown J. Etiology of human breastcancer: a review. J NatI Cancer Inst 1972;50:21-42.8. Bradlow HL, Schwartz MK, Fleisher M, et a!. Accumulation ofhormones in breast cyst fluid. J Clin Endocrinol Metab1979;49:778-82.9. Vizoso F, Fueyo A, Allende MT, Fern#{225}ndezJ, Garcfa-Mor#{225}nM,Ruibal A. Evaluation of human breast cysts according to theirbiochemical and hormonal composition, and cytologic examina-tion. Eur J Surg Oncol 1990;16:209-14.10. Bradlow HL, Breed CN, Nisselbaum JS, Fleisher M, SchwartzMK. pH as a marker of breast cyst fluid type. Eur J Surg Oncol1987;13:331-4.11. BradlowHL, FleisherM, Schwartz MK, Nisselbaum J, BreedCN. Classification of patients with gross breast disease accordingto the biochemical composition of breast cyst fluid. In: Dogliotti L,Mansel RE, eds. Fibrocystic breast disease. Berlin: Cantor,1986:9-20.12. Yap PL, Miller WR, Roberts MM, et al. Protein concentrationsin fluid from gross cystic disease of the breast. Clin Oncol1984;10:35-43.13. Zangerle PF, Spyratos F, LeDoussal V, et al. Breast cyst fluidproteins and breast cancer. Ann NY Acad Sci 1986;464:331-49.14. Collette J, Hendrick JC, Jaspar JM, Franchimont P. Presenceof a-lactalbumin, epidermalgrowth factor, epithelial membraneantigen and gross cystic disease fluid protein (15 000 Daltons) inbreast cyst fluid. Cancer Res 1986;46:3728-33.15. Haagensen DE, Mazoujian G. Biochemistry and immunohis-

tochemistry of fluid proteins of the breast in gross cystic disease.In: Haagensen CD, ed. Diseases of the breast. Philadelphia: Saun-ders, 1986:474-500.16. Balbfn M, Freije JMP, Fueyo A, Sanchez LM, Lopez-Otfn C.Apolipoprotein D is the major protein component in cyst fluid fromwomen with human breast gross cystic disease. Biochem J1990;271:803-7.17. Gatzy JT, Zaylown MP, Gaskins K, Perlman WH. Electrolytesof breast cyst fluid. Clin Chem 1979;25:745-8.18. Gairard B, Gros CM, Koehl C, Renaud R. Proteins and ioniccomponents in breast cyst fluids. Op. cit. (ref. 5):191-5.19. Rosner W, Khan MS, Breed CN, Fleisher M, Bradlow HL.Plasma steroid-binding proteins in the cysts of gross cystic diseaseof the breast. J Clin Endocrinol Metab 1985;61:200-3.20. Miller WR, Dixon JM, Scott WN, Forrest APM. Classificationof human breast cysts according to electrolyte and androgenconjugate composition. Clin Oncol 1983;9:227-32.21. Dixon JM, Miller WR, Scott WN, Forrest APM. The morpho-logical basis of human breast cyst populations. Br J Surg1983;70:604-6.22. Dogliotti L, Orlandi F, Torte M, et al. Cations and dehydro-epiandrosterone-sulfate in cyst fluid of pre- and menopausal pa-tients with gross cystic disease of the breast. Evidence for theexistence of subpopulations. Eur J Clin Oncol 1986;22:1301-7.23. Beccati D, Grilli N, Schincaglia P, et al. Apocrine cells inbreast cyst fluid and their relationship to cyst type: a morphomet-nc study. Eur J Cancer Clin Oncol 1988;24:597-602.24. Bradford MM. A rapid and sensitive method for the quantita-tion of microgram quantities of protein utilizing the principle ofprotein-dye binding. Anal Biochem 1976;72:248-54.25. Laemmli UK. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)1970;227:680-5.26. Jeppsson JO, Laurell CB, Franzen B. Agarose gel electropho-resis [Proposed Selected Method]. Clin Chem 1979;25:629-38.27. Hewick RM, Hunkapiller MW, Hood LE, Dreyer W. A gas-liquid solid phase peptide and protein sequenator. J Biol Chem1981;256:7990-7.28. Matsudaira P. Sequence from picomole quantities of proteinselectroblotted onto polyvinylidene difluoride membranes. J BiolChem 1987;262:10035-8.29. Kesner L, Yu W, Bradlow L, Breed C, Fleisher M. Proteases incyst fluid from human gross cyst breast disease. Cancer Res1988;48:6379-83.30. Minghetti PP, Rufl’ner DE, Kuang WJ, et al. Molecular struc-ture of the human albumin gene is revealed by nucleotide se-quence within qll-22 of chromosome 4. J Biol Chem1986;261:6747-57.31. Dixon JM, Scott WN, Miller WR. Natural history of cysticdisease: importance of cyst type. Br J Surg 1985;72:190-2.32. Dixon JM, Scott WN, Miller WR. An analysis of the contentand morphology of human breast microcysts. Eur J Surg Oncol1985;11:151-4.33. Page DL, Dupont WD. Are breast cysts a premalignantmarker? Eur J Cancer Clin Oncol 1986;22:635-6.