R E V I S T A P O R T U G U E S A D E P N E U M O L O G I A

Vol XIII Suplemento 2 Outubro 2007

S 35

Identification of bio-markers in sarcoidosis BAL usingmultidimensional Differential Display Proteome Analysiscoupled with nano-HPLC tandem mass spectrometry

Otto F VillaOscar E GuevaraAvelino TeixeiraKusuman JosephPaul DelaRipaMarc JudsonRobert P BaughmanMichael Iannuzzi

1 Corresponding author.2 The Catherine and Henry Gaisman Division of Pulmonary and Critical Care Mount Sinai School of Medicine. New York, NY 10029,3 Medical University of South Carolina. Charleston, SC.4 University of Cincinnati Medical Center, Cincinnati, OH, United States.5 Lahey Clinic Burlinghton, MA.

Sarcoidosis is a granulomatous disease withworldwide prevalence estimated as high as300 per million1, 2. Sarcoidosis affects all or-gans of the body, but the lungs and intra--thoracic lymph nodes are affected in approx-imately 90% of patients accounting for themajor morbidity and mortality3, 4.The clinical progression of sarcoidosis is welldocumented5. Indeed, it is known that patientswho present with hilar-adenopathy on chestroentgenogram as the sole abnormality andthose with erythema nodosum do well twoyears later6,7. In contrast sarcoidosis patientswith lupus pernio only had an 20% chance ofresolution at two years7. Although there isevidence of genetic predisposition8, no testhelps the clinician predict the outcome of sar-coidosis patients on presentation or at timesof relapse. Therefore, identification of sar-coidosis biomarkers remains an unmet need.The molecular pathogenesis of sarcoidosisremains unsettled, however it is clear that ac-tivated T cells and macrophages (mf) are key

players9,10,11,12. These cells accumulate in thetissues forming granulomas and can also beidentified in the bronchoalveolar lavage (BAL).Because proteins govern cell physiology, acomprehensive analysis of sarcoidosis BALcell-associated proteins may identify muchneeded biomakers. A comprehensive analysisof the protein equivalent of a gnome is knownas “the proteome”. Therefore, we studied thesarcoidosis CD14+ derived proteome.Proteomics workflow includes: protein separa-tion and assessment of fold differences followedby identification of selected proteins via massspectrometry (MS). Protein separation can beaccomplished by two dimensional gel electro-phoresis (2DE13) or multidimensional, high per-formance liquid chromatography (HPLC) pro-tein identification technology (MUDPIT14,15).Under MUDPIT, peptides are produced fromthe entire extract followed by several in-line,modes of HPLC [cation exchanges (SCX), andreverse phase (RP)]. The last step is always anRP, so that peptides elute in organic solvent

R E V I S T A P O R T U G U E S A D E P N E U M O L O G I A

Vol XIII Suplemento 2 Outubro 2007

S 36

Identification of bio-markers in sarcoidosis BAL using multidimensional DifferentialDisplay Proteome Analysis coupled with nano-HPLC tandem mass spectrometry

Otto F Villa, Oscar E Guevara, Avelino Teixeira, Kusuman Joseph, Paul DelaRipa, Marc Judson, Robert P Baughman, Michael Iannuzzi

which is easily vaporized at the front end of massspectrometer. Mudpit can identify several thou-sand proteins but does not allow up-front selec-tion of differentially expressed proteins; hencecalled “shotgun proteomics”.On the contrary 2DE produces a mathematicalmatrix for each experimental condition where aspot can be identified by its x-y coordinates cor-responding to specific Pi and Mr. values. Spotsare defined as gray pixel areas fitting a Gaussianmodel while and non-Gaussian pixels surround-ing spots are considered “background”. Afterbackground subtraction, spots are quantifiedusing a transform of diameter and pixel intensi-ty. For analysis two gels are overlaid at a time(termed differential display). To correct for var-iations in spot position due to (shift during scan-ning, swelling and shrinking during staining, in-homogeneities in the gel, temperature or currentvariations), matching vectors are created and awarping algorithm executed16.To address the sarcoidosis proteome, patientswho required bronchoscopy and BAL as partof the workup for an interstitial lung disease(ILD) were recruited. The protocol was ap-proved by IRBs, and all patients signed in-formed consent. BAL was performed as de-scribed17. A total of 1516 spots comprisedthe sarcoidosis proteome, of these 383 wereup-regulated over two-fold and 294 over 10fold when compared with the proteome ofthe patients without ILD (Fig. 1).Comparisons between two samples can also beperformed using differential gel electrophoresis(DIGE)18. Under DIGE experimental and con-trol samples are each covalently linked to a fluor-ochrome (Cy3 and Cy5, Amersham), mixed andresolved in a single gel, so that spots will be lo-cated at the same x-y coordinates regardless ofgel distortion. However, since only three dyesexist, the number of comparisons is limited. On

the other hand, using our approach, a proteomesimply becomes a mathematical matrix, allow-ing unlimited comparisons.This feature is very practical for our studiesbecause the optimal control in sarcoidosis isnot clear. Indeed, in an attempt to control for“normal” cells, we utilized cells from patientswho presented with hemoptisis but had nor-mal bronchoscopy since this procedure is notstandard of care for healthy individuals. Be-

Fig. 1 – Differential display analysis between CDl4+

cellular proteomes of sarcoidosis and a patient withoutILD. Upon completion of bronchoscopy BAL was centrifugedat 400xg, and the cell pellet incubated with anti-Cd14+ para-magretic beads (Dynal). Positively selected cells were lyzedin Trizol (Invitrogen) and the protein pellet resuspended inIEF buffer (9.5M Urea, 05% CHAPS, 0.4% DTT, 0.5% (v/v)Pharmalyte 4-7). After ioselectric focusing on 24cm IPG strips(pH 4-7 L G&E Healthcare), DTT and iodoacetamidetreatments, PAGE was performed using custom-made 25x25cm gels. After scanning on an infrared instrument (Licor),quantification and warping using the software Delta 2D(decodon.com), spots sharing the same x,y coordinates werecompared. Spots upregulated in sarcoidosis (ratio over 2)were selected, and over 10 fold (shaded). Spots downregulated in sarcoidosis (ratio below 0.5) were colored ingreen but not studied. Yellow spots were present evenly inboth samples (ration 0.48 to 1.98). The control gel in thisanalysis comprised CD14+ cells isolated from a patientwithout ILD (5*10^6 cell equivalents) and bronchoscoped fivedays after an unexplained episode of hemoptisis. Gels havebeen repeated at least three times.

R E V I S T A P O R T U G U E S A D E P N E U M O L O G I A

Vol XIII Suplemento 2 Outubro 2007

S 37

Identification of bio-markers in sarcoidosis BAL using multidimensional DifferentialDisplay Proteome Analysis coupled with nano-HPLC tandem mass spectrometryOtto F Villa, Oscar E Guevara, Avelino Teixeira, Kusuman Joseph, Paul DelaRipa, Marc Judson, Robert P Baughman, Michael Iannuzzi

cause of this, the mφ may have been resting,thus accounting for the dissimilar proteomes.Therefore, to narrow the array of upregulat-ed sarcoidosis spots, the proteome of mφ nat-urally activated in the human lung were puri-fied from biopsy proven desquamativeinsterstitial pneumonia (DIP, Fig. 2).Surprisingly, the proteomes were diametricallydifferent. However the DIP mφ becomes stim-ulated by cigarette smoking and never form gran-ulomas suggesting that the activation pathwaysbetween the two diseases are different. To con-trol for activated normal human CD14+ cells,the procedure described in Fig. 1 was used topurify CD14+ from the peripheral blood (PBL)of healthy volunteers (n=10). However, neitherresting nor PMA/ionomycin stimulated PBL--derived mφ-proteomes changed the sarcoido-sis CD14+-BAL array (data not shown).However, alveolar mφ and PBL-mφ are obvi-ously different cells and it is not difficult to dif-

ferentiate sarcoidosis from a non-ILD patient.A more relevant clinical challenge is to excludeother ILDs, and cancer. Therefore proteomesfrom patients with biopsy proven ILD n=9 wereanalyzed. After 2DE, scanning, spot detection,background subtraction and spot quantification,a mathematical matrix was prepared for eachcondition and all spots integrated into a largematrix n=23960 spots. This matrix was next usedto identify spots shared between sarcoidosis andhypersensitivity pneumonitis (HP; both druginduced and caused by animal droppings) sincethese diseases could be clinically and sometimespathologically similar. Indeed a large number ofsarcoidosis spots had comparable x.y coordinatesand spot intensities in the HP proteomes (Fig. 3).

Fig. 2 – Differential display analysis between sarcoi-dosis and DIP CD14+ cellular proteomes. CD14+ cells(5*10^6 cell equivalents) from a patient with DIP werepurified with paramagnetic beads resolved on 2D-gels andcompared with the sarcoidosis proteome shown in Fig. 1.Strikingly, 624 spots were upregulated over 5 fold insarcoidosis (red spots surrounded by white circles) whileonly a few spots were shared. Spots found only in DIP wereidentified in green but not studied further.

Fig. 3 – Multi-dimensional differential display analysis ofCDl4+ cellular proteomes and other granulomatosis. CDI4+

cells were purified from BAL of patiens with the following biopsyproven ILDs: Hypersensitivity pneumonitis (HP) induced by adrug and by animal droppings, DIP, pumonary alveolarproteinosis (pAP), pulmonary Langerhans cell histiocytosis(PLCH), Coccidioidomycosis pneumonia (Cocci as arepresentative of other granulomatosis) and patients withoutILD (hemoptisis, described in Fig. 1). Gels were prepared asdescribed in the preceding figures, mathematical matrixescomputed and integrated into a large matrix containing all spotsn=23960. This matrix was next used to compute fold rationsbetween 0.48 and 1.98 between sarcoidosis and HP causedby animal droppings (purple), sarcoidosis and drug-inducedHP (light purple) as well as sarcoidosis and DIP (blue). Blackspots did not fit the search criteria (were not shared betweenthese diseases). Spots Note that a high fold ratío does notequate with high abundance.

R E V I S T A P O R T U G U E S A D E P N E U M O L O G I A

Vol XIII Suplemento 2 Outubro 2007

S 38

These shared spots may relate to moleculesinvolved in granuloma formation and will bethe focus of further analysis. To narrow thisarray of “granuloma” spots three patients withtuberculosis were studied but their cells didnot resolve well on 2DE (data not shown).Interestingly, DIP and sarcoidosis did not havemany shared molecules (Fig. 3) confirming theanalysis presented in Fig. 2 and reinforcingthe conclusion that the DIP and sarcoidosismφ have different activation pathways.This large mathematical matrix was next usedto identify sarcoidosis spots upregulated over3 fold over all (Fig. 4). The importance ofthis analysis is that, it identified 1516 spotsin the sarcoidosis proteome, of which 383were not present in non-ILD proteome butonly 97 persisted after exclusion of spotspresent with ratios as low as 0.3 in 7 ILDs.Obviously the analysis is not complete be-

cause more proteomes from other granulo-matous and disease additional ILDs shouldbe added to fine-tune the sarcoidosis array.In conclusion, multi-dimensional differentialdisplay allows comparison of sarcoidosisspots with several ILD and granulomatousdiseases thereby providing robustness to theidentification of spots with potential impor-tance as biomakers of sarcoidosis over dis-eases with similar clinical presentation. Pro-tein identification of this array of spots andbiological validation is the focus of a sepa-rate manuscript.Future directions: Our current efforts aredirected towards a more comprehensive pro-teome analysis using MudPit since we rec-ognize that the current approach is limitedto the most abundant BAL species that wereresolved on 2DE-gels. Our long term goal isto identify biomarkers able to assist the cli-nician in the care of sarcoidosis patients.

AcknowledgementsWork supported by NHLBI-K01 grant (OV),Massachusetts Thoracic Society RT award(OV) and the Fennessey Foundation (OV).We thank Matthias Bert (Decodon.com) forcomputational advice.

References1. Siltzbach LE, James DG, Neville E, et al. Course andprognosis of sarcoidosis around the world. Am J Med1974;57(6):847-52.2. Hosoda Y, Sasagawa S, Yasuda N. Epidemiology ofsarcoidosis: new frontiers to explore. Curr Opin PulmMed 2002;8(5):424-8.3. Villa OF, Fanburg BL. Sarcoidosis. The Textbook ofRespiratory Medicine, JF Murray and JA Nadel, Eds WB,Saunders, Co, Philadelphia, PA 2000:1486-500.

Identification of bio-markers in sarcoidosis BAL using multidimensional DifferentialDisplay Proteome Analysis coupled with nano-HPLC tandem mass spectrometry

Otto F Villa, Oscar E Guevara, Avelino Teixeira, Kusuman Joseph, Paul DelaRipa, Marc Judson, Robert P Baughman, Michael Iannuzzi

Fig. 4 – Multi-dimensional differential display analysisof CDl4+ cellular proteomes and other ILD. A large matrixcontaining alI spots was used to calculate fold ratios foreach ILD over the entire matrix and plotted on the matrix.Out of 1516 sarcoidosis spots, 97 (red) remainedupregulated over all. Likewise 8 HP (violet), 19 DIP (purple),39 PAP (green) and 61 EG (PLCH light blue) appear to bepreferentially expressed in these ILDs. Of note for analysisof PAP proteome three patients were used because thePAP BAL contains very few cells, albeit of its large volumeand lipid content. Black spots did not fit the search criteria(were not uniquely up regulated in the diseases studied).

R E V I S T A P O R T U G U E S A D E P N E U M O L O G I A

Vol XIII Suplemento 2 Outubro 2007

S 39

Identification of bio-markers in sarcoidosis BAL using multidimensional DifferentialDisplay Proteome Analysis coupled with nano-HPLC tandem mass spectrometryOtto F Villa, Oscar E Guevara, Avelino Teixeira, Kusuman Joseph, Paul DelaRipa, Marc Judson, Robert P Baughman, Michael Iannuzzi

4. Baughman RP, Teirstein AS, Judson MA, et al. Clinicalcharacteristics of patients in a case control study of sar-coidosis. Am J Respir Crit Care Med 2001;164(10 Pt1):1885-9.5. Judson MA, Baughman RP, Teirstein AS, Terrin ML,Yeager H, Jr. Defining organ involvement in sarcoidosis:the ACCESS proposed instrument. ACCESS ResearchGroup. A Case Control Etiologic Study of Sarcoidosis.Sarcoidosis Vasc Diffuse Lung Dis 1999;16(1):75-86.6. Nagai S, Shigematsu M, Hamada K, Izumi T. Clinicalcourses and prognoses of pulmonary sarcoidosis. CurrOpin Pulm Med 1999;5(5):293-8.7. Neville E, Walker AN, James DG. Prognostic factorspredicting the outcome of sarcoidosis: an analysis of 818patients. Q J Med 1983;52(208):525-33.8. Iannuzzi MC, Maliarik MJ, Poisson LM, Rybicki BA.Sarcoidosis susceptibility and resistance HLA-DQB1 al-leles in African Americans. Am J Respir Crit Care Med2003;167(9):1225-31.9. Zorzetto M, Bombieri C, Ferrarotti I, et al. Comple-ment receptor 1 gene polymorphisms in sarcoidosis. AmJ Respir Cell Mol Biol 2002;27(1):17-23.10. Semenzato G, Chilosi M, Ossi E, et al. Bronchoalve-olar lavage and lung histology. Comparative analysis ofinflammatory and immunocompetent cells in patientswith sarcoidosis and hypersensitivity pneumonitis. AmRev Respir Dis 1985;132(2):400-4.

11. Forman JD, Silver RF, Klein JT, et al. T cell recep-tor variable beta-gene expression in the normal lungand in active pulmonary sarcoidosis. Chest 1993;103(2Suppl):78S.12. Forman JD, Klein JT, Silver RF, Liu MC, GreenleeBM, Moller DR. Selective activation and accumulationof oligoclonal V beta-specific T cells in active pulmona-ry sarcoidosis. J Clin Invest 1994;94(4):1533-42.13. Klose J. Protein mapping by combined isoelectric fo-cusing and electrophoresis of mouse tissues. A novelapproach to testing for induced point mutations in mam-mals. Humangenetik 1975;26(3):231-43.14. Yates JR, 3rd. Mass spectrometry and the age of theproteome. J Mass Spectrom 1998;33(1):1-19.15. McDonald WH, Yates JR, 3rd. Shotgun proteomicsand biomarker discovery. Dis Markers 2002;18(2):99-105.16. Luhn S, Berth M, Hecker M, Bernhardt J. Using stan-dard positions and image fusion to create proteome mapsfrom collections of two-dimensional gel electrophoresisimages. Proteomics 2003;3(7):1117-27.17. Haslam PL, Baughman RP. Report of ERS TaskForce: guidelines for measurement of acellular com-ponents and standardization of BAL. Eur Respir J1999;14(2):245-8.18. Unlu M, Morgan ME, Minden JS. Difference gel elec-trophoresis: a single gel method for detecting changes inprotein extracts. Electrophoresis 1997;18(11):2071-7.