Profiling Protein Markers Associated with the Sensitivity to Concurrent Chemoradiotherapy in Human Cervical Carcinoma Hong Zhu,†,# Hai-ping Pei,‡,# Shan Zeng,†,# Jia Chen,†,# Liang-fang Shen,† Mei-zuo Zhong,† Ruo-jing Yao,§ and Hong Shen*,| Departments of Oncology, Surgery, and Obstetrics and Gynecology, and Medical Research Center, Xiangya Hospital, Central South University, Changsha, Hunan, China 410078 Received March 27, 2009
Concurrent chemoradiotherapy (CCRT) is recently recommended as the primary and standard treatment modality for cervical cancer. The aim of this study is to investigate the protein biomarkers associated with CCRT sensitivity, so as to better understand the mechanisms underlying CCRT resistance. Fresh tumor tissues from five cases for each group of CCRT-highly sensitive (CCRT-HS) and CCRT-lowly sensitive (CCRT-LS) were analyzed by 2-D electrophoresis coupled with MALDI-TOF-MS, followed by Western blot for four candidate proteins including S100A9, galectin-7, nuclear matrix protein-238 (NMP238), and heat shock protein-70 (HSP-70). In randomly selected CCRT-HS (n ) 60) and CCRT-LS (n ) 35) cases, these four differentially expressed proteins were detected by tissue microarray with immunohistochemistry staining to explore the association between these interested proteins and CCRT sensitivity. Nineteen proteins differentially expressed more than four times between two groups were identified. An association was revealed between CCRT sensitivity and increased S100A9 and galectin7, but decreased NMP-238 and HSP-70 expression (p < 0.001, respectively). Although none of these four protein markers could be used as an independent predictive factor, a recurrence prediction model was generated by combining S100A9, galectin-7, NMP-238, and HSP-70 as a full predictive factor. The proteomic analysis combined with tissue microarray provides us a dramatic tool in predicting CCRT response. The increased expression of S100A9 and galectin-7, but decreased expression of NMP-238 and HSP-70, suggests a significantly increased sensitivity to CCRT in cervical cancer. Keywords: Protein markers • concurrent chemoradiotherapy • sensitivity • cervical carcinoma
Introduction Radical surgery is generally restricted to stages I and IIa of IFGO (the International Federation of Gynecology and Obstetrics) Classification, while radiotherapy may be applied to all stages of cervical cancer.1 However, most patients of cervical cancer in late stages will experience limited treatment options.2 The National Cancer Institute of the U.S. recently established concurrent chemoradiotherapy (CCRT) as the primary and standard treatment modality for cervical cancer.3 CCRT has shown its advantages to reduce treatment failure rates compared to radiotherapy alone, and thus improve both local control and cervical cancer overall survival by approximately 40%.4-10 However, the resistance to CCRT is an important clinical finding and the efficacy of CCRT is restricted in considerable number of patients.3,11,12 Identification of specific protein signatures associated with CCRT sensitivity in cervical * Address correspondence to: Prof. Hong Shen, Xiangya Hospital, Central South University, Changsha, Hunan, China 410008. Tel: 86-13574846576. E-mail:
[email protected]. † Department of Oncology. # These authors contributed equally to this paper. ‡ Department of Surgery. § Department of Obstetrics and Gynecology. | Medical Research Center. 10.1021/pr900287a CCC: $40.75
2009 American Chemical Society
cancer may provide novel biomarkers that allow for more accurate prognostic information, help to identify new molecular therapeutic targets, and provide clues for understanding the molecular mechanisms governing CCRT resistance. An identification of CCRT-associated proteins in cervical cancer has not reported yet to produce reliable candidate markers. A lack of this kind of proteomic profiles of CCRT resistance hinders our efforts to clearly understand and efficiently treat human cervical cancer. The recent application of two-dimensional electrophoresis (2-DE) coupled with mass spectrometry (MS) to cancer research allows the characterization of global alterations in protein expression in human cancers.13 With the advantages of being nonprejudicial, proteomic analysis can give us new unexpected insights in the mechanisms of CCRT resistance in human cervical cancer. Employing the technologies of 2-DE/MS identification and protein expression detection in combination, the aim of this study is to investigate the proteins associated with CCRT sensitivity, so as to better understand the mechanisms underlying CCRT resistance, as well as to provide clues for the selection standard of individualized therapy. Journal of Proteome Research 2009, 8, 3969–3976 3969 Published on Web 06/09/2009
research articles Patients and Methods Patients and Tissue Preparation. All recruited cases assessed by the official IFGO staging system for cervical carcinoma14 were in the stages of IIb-IVa. The cases experiencing any conditions potentially confounding the results were excluded. The study was approved by the Research Ethics Committee of Central South University, China. Written informed consent was obtained from all study participants before entry into the study. On the basis of the criteria of evaluation as previously reported,15 the subjects’ responses to CCRT were classified as complete remission (CR), partial remission (PR), stable disease (SD), or progressive disease (PD). Without either SD or PD cases in the recruited subjects, the cases with CR were deemed as CCRThighly sensitive (CCRT-HS) group, while the patients with PR were defined as CCRT-lowly sensitive (CCRT-LS) group. For initial 2-DE/MS proteome approach and Western blot confirmation, the preliminary 10 cases including 5 CCRT-HS (age range 27-73 years, mean ( SD ) 51.3 ( 12.2 years) and 5 CCRT-LS (age range 32-76 years, mean ( SD ) 53.9 ( 13.5 years) subjects of biopsy-proved primary moderately differentiated squamous cell carcinoma of the cervix were randomly collected from the Department of Oncology, Xiangya Hospital, Central South University, China. Fresh tumor tissues were obtained before CCRT treatment and the protein from a total of 50 mg of frozen tissue was extracted by homogenization in lysis buffer.16 Following 2-DE/MS proteome analysis to profile the protein markers associated with CCRT sensitivity, we extended the sampling scale to all 285 patients of primary moderately differentiated squamous cervical carcinoma who received CCRT under the same treatment protocol from January 2005 to June 2006. A total of 95 cases were randomly selected by the method of completely random design. According to the clinical response to the treatment, 60 (age range 23-75 years, mean ( SD ) 52.3 ( 10.6 years, staging IIb-IVa) and 35 (age range 25-76 years, mean ( SD ) 53.1 ( 10.9 years, staging IIb-IVa) randomly collected cases were classified as CCRT-HS and CCRT-LS, respectively. The distribution of the patients’ age, and the disease’s IFGO stages including the tumor size and local invasion, had no statistical difference between two groups (p > 0.05, respectively). The paraffin-embedded specimens were tested by tissue microarray with immunohistochemistry staining. All tissues were examined by the same group of two senior pathologists experienced in cervical cancer diagnosis. The interpretation was done blinded to the outcome results. Radiotherapy and Chemotherapy. With the use of a 15-mV linear accelerator, X-ray radiotherapy was administered to the whole pelvic region in 23 fractions totalling 46 Gy, adding simultaneous integrated boost intensity-modulated radiotherapy (IMRT) to uterine lateral fields in 3-5 fractions totaling 6-10 Gy. Following 13-15 fractions of external radiotherapy, all subjects were scheduled to undergo P192PIr high dose rate intracavitary brachytherapy for a total of six times with 7 Gy for each fraction. The total dose delivered to point A (a reference location 2 cm lateral and 2 cm superior to the cervical os) was 68-72 Gy; the total dose delivered to point B (the pelvic wall) was 52-56 Gy. For the lateral fields, the anterior border was the anterior border of the pubic symphysis and the posterior border was the space between S2 and S3.6 3970
Journal of Proteome Research • Vol. 8, No. 8, 2009
Zhu et al. Chemotherapy started 2 weeks before radiotherapy and was given every 21 days thereafter for a total of four cycles. Cisplatin was given intravenously at a dose of per square meter of bodysurface area (75 mg/mP2P). Paclitaxel was given at a dose of 175 mg/mP2P on day 1 of each cycle.17 In case of severe toxicity unsuitable for coming course, chemotherapy was withheld until recovery. IPG-2-DE and Image Analysis. A total of 400 µg of total protein was loaded onto a 240 mm linear IPG strip (pH3-10, Amersham Biosciences, Piscataway, NJ) for first-dimensional isoelectric focusing and subsequently on the second-dimension SDS-PAGE (Bio-Rad, Hercules, CA) for protein separation. The gels were stained by Coomassie Brilliant Blue, scanned by an Imagescanner (Amersham Biosciences) and analyzed by the software of PD-Quest 7.3.1 (Bio-Rad). Each sample was subjected to 2-DE in triplicate and paired Student’s t test was used to evaluate the average protein abundance change corresponding to each target spot across the gels.16 In-Gel Trypsin Digestion of Target Protein and MALDITOF-MS. Fifty differential spots between CCRT-HS and CCRTLS were excised from the gels and digested in-gel. Briefly, the gel spots were destained in 100 mM Na2S2O3 and 30 mM K3Fe(CN)6 (v/v ) 1:1), incubated at 57 °C for 1 h in the reduction buffer (100 mM NH4HCO3 and 10 mM DTT), and subsequently alkylated in the alkylation buffer (100 mM NH4HCO3 and 55 mM iodoacetamide) in the dark for 30 min at room temperature. The vacuum centrifugation-dried gel pieces were digested for 24 h at 37 °C in the solution containing 50 mM NH4HCO3, 5 mM CaCl2, and 0.1 g/L TPCK-trypsin (Sigma-Aldrich, St. Louis, MO). The tryptic peptide mixture was extracted three times with 50% acetonitrile (ACN), 0.1% trifluoroacetic acid (TFA), and dried in a vacuum centrifugation, dissolved in 0.1% TFA and desalted with a C18 ZipTip column (Millipore, Billerica, MA). The eluted peptides was analyzed using a Voyager-DETM STR Biospectrometry Workstation System 4307 (Applied Biosystems, Foster City, CA) in positive ion-reflector mode. The parameters were set up as follows: extraction voltage 20 kV, grid voltage 64.5%, mirror voltage ratio 1.12, N2 laser wavelength 337 nm, pulse width 3 ns, acquisition mass range 1000-3000 Da, delay 100 ns, and vacuum degree 4 × 10-7 Torr. Mass spectra were recorded using 50-200 shots, depending on the signal-to-noise ratio obtained from each sample. All mass spectra were externally calibrated with a standard peptide mixture (MH+: Angiotensin II, 1046.5420 Da; Angiotensin I, 1296.6853 Da; ACTH clip 18-39, 2465.199 Da). The auto digestion peaks of trypsin served as internal calibration. A list of the corrected mass peaks was the peptide mass fingerprinting (PMF).16 Protein Identification and Database analysis. With regard to the protein identification using PMF, the peak-picking for peptide mass fingerprint was performed by MASCOT Distiller (Version 2.2.1) using the search engine of MASCOT (Version 2.2, http://www.matrixscience.com/, Matrix Science Ltd., U.K.) against the Swiss-Prot protein database (Version 50.4, 230 133 sequences). The searching parameters were set up as follows: the taxonomy was Homo sapiens, the enzyme was trypsin, the number of missed cleavage sites was allowed up to 1, the fixed modification was carbamidomethylation of cysteine, the variable modification was oxidation of methionine, the peptide tolerance was 50 ppm, the mass value was MH+, searching range within the experimental pI value (0.5 pH unit and experimental mass range (Mr) (20%. The criteria for positive identification of proteins were set as follows: (1) the MS match
Chemoradiotherapy-Associated Proteins in Cervical Carcinoma
research articles
Figure 1. The representative 2-DE maps of human cervical cancer tissue highly sensitive and lowly sensitive to CCRT from five independent experiments. The numbered spots represent the differentially expressed proteins between CCRT-HS and CCRT-LS groups.
consisted of a minimum of five peptides; (2) the matched peptides covered at least 20% of the whole protein sequence; (3) the MASCOT score of a protein matching was higher than 63 (p < 0.05).16 Western Blot. A total of 50 µg of total protein was separated on a 10% SDS-PAGE, and the transferred membranes were incubated overnight at 4 °C with mouse monoclonal antibodies against human heat shock protein-70 (HSP-70) (Santa Cruz Biotechnology, Santa Cruz, CA), S100A9 (Novus Biologicals, Littleton, CO), goat polyclonal antibodies against human galectin-7 (R&D Systems, Minneapolis, MN), and rabbit polyclonal antibodies against human nuclear matrix protein 238 (NMP-238) (Proteintech Group, Chicago, IL), respectively. They were incubated with the corresponding horseradish peroxidase (HRP)-conjugated second antibodies (Amersham Biosciences) and were subsequently visualized.18 Tissue Microarray and Immunohistochemistry. Tissue microarray was performed using an ATA-27 automated arrayer (Beecher Instruments, Sun Prairie, WI). Following antigen retrieval and endogenous peroxidase blockage, the 5-µm tissue microarray slides were assessed by immunohistochemistry staining employing EnVision System (Dako Diagnostics, Zug, Switzerland). The tissues with obviously positive expression of individual interested proteins, which were ascertained by the initial proteomic approach and Western blot confirmation, were applied as the positive control. For negative controls, the primary antibodies were replaced with immunoglobulin (Ig)G depending on the different species of corresponding primary antibody.19 Following a hematoxylin counterstaining, the immunostaining was scored by two independent experienced pathologists and any discrepant scores were re-examined to achieve a consensus score. The immunostaining was scored on a scale from 0 to 3 (negative/weak/moderate/intense staining) as
follows: 0 ) less than 10% positive staining cells, 1 ) 10-25%, 2 ) 25-50%, 3 ) more than 50%. Cases with scores of 2+ or 3+ were designated as “positive”, whereas cases with scores of 0 or 1+ were designated as “negative”.20 Statistical Analysis. The Wilcoxon test was used to compare the difference between CCRT-HS and CCRT-LS groups, and χ2 test was applied to evaluate the association between CCRT sensitivity and the expression of each protein marker. All suspected proteins were tested by multiple stepwise regression analysis in a Logistic regression model. Subsequently, a predictive model was generated from these biomarkers. Differences were considered to be significant when the p-value was less than 0.05.
Results Proteome Differential Expression between CCRT-HS and CCRT-LS Cervical Cancer. There were 781 ( 74 and 766 ( 52 protein spots recognized, respectively, on CCRT-HS and CCRTLS gels (n ) 5), with an average matching rate of 87.6% and 105 differential protein expression spots between the two groups (Figure 1). A total of 35 spots differentially expressed more than 4 times were excised from gels and analyzed by Mr, pI, the sequence coverage for protein identification and MASCOT scores for protein matching (score > 75, p < 0.05). Compared with CCRT-LS tissues, CCRT-HS tissues had nine proteins with higher expression and 10 proteins with lower expression, respectively, which were finally identified (Figure 1) and listed in Table 1. Western Blot Confirmation of Differentially Expressed Proteins. On the basis of the importance of the suspected proteins, the availability of commercial antibodies and consistent differential expression, S100A9 (spot 4), galectin-7 (spot 5), NMP-238 (spot 15), and HSP-70 (spot 19) (Figure 2) were Journal of Proteome Research • Vol. 8, No. 8, 2009 3971
research articles
Zhu et al.
Table 1. Characterized Differentially Expressed 19 Proteins in Cervical Cancer Tissues between CCRT-HS and CCRT-LS Groupsa Mr (Da)
pI
sequence/ covered
protein ID
1 2 3 *4 *5 6 7 8 9
P68871 P31944 Q5SQI3 P06702 2GALB CAA30534 CAA00999 KRHU9 P60709
Highly Expressed in CCRT-HS but Lowerly Expressed in Hemoglobin subunit beta (Hemoglobin beta chain) 132 15857 Caspase-14 precursor (EC 3.4.22.-) (CASP-14) 81 27662 Calmodulin-like 5 130 15883 S100A9 protein 75 13234 galectin-7, chain B 187 14635 HSKERC4 115 45479 MRP-14 protein 98 12770 keratin 19 306 44065 Actin (Beta-actin) 106 41710
CCRT-LS 6.81 5.44 4.34 5.71 7.52 5.14 5.55 5.04 5.29
10 11 12 13 14 *15 16 17 18 *19
BAB18261 LMNB1 Q502Y0 BLVRB DHE3 JE0334 AL1A1 AAL50327 Q805Y7 1ATR
Lowerly Expressed in CCRT-HS but Highly Expressed in Anti HBs antibody light-chain Fab (Fragment) 90 23307 Lamin-B1 83 66237 WARS protein 98 53145 Flavin reductase (NADPH-dependent diaphorase) 90 21974 Glutamate dehydrogenase 1 114 61359 nuclear matrix protein (NMP-238) 92 50196 Retinal dehydrogenase 1 97 54696 AF165172 139 6803 Subunit of replicative DNA polymerase 82 49937 Heat shock protein 70 122 42081
CCRT-LS 7.75 5.11 5.83 7.31 7.66 6.02 6.29 5.89 5.31 6.62
a
protein description
MASCOT/ score
no.
matched/ mass no.
not matched/ mass no.
78% 40% 76% 65% 78% 40% 88% 80% 54%
11 8 9 12 12 15 8 29 18
14 10 20 22 11 32 40 36 29
55% 36% 43% 51% 43% 32% 32% 30% 34% 59%
7 15 14 7 18 12 11 20 8 17
21 54 39 18 21 38 27 20 17 47
The protein numbers with “*” represent the differentially expressed proteins selected for Western blot confirmation and further tissue microarray.
Figure 2. PMF of protein spots 4, 5, 15, and 19, respectively, representing S100A9 (panel A), galectin-7 (panel B), NMP-238 (panel C), and HSP-70 (panel D), which expressed differentially between the groups of CCRT-HS and CCRT-LS (MASCOT scores are 75, 187, 92, and 122, p < 0.05, respectively) in five independent experiments. Ions score is -10P*P Log(P), where P is the probability that the observed match is a random event. The peptides are sorted by the residue number.
further assessed by Western blot to confirm our MALDI-TOFMS findings. The protein expression of S100A9 and galectin-7 3972
Journal of Proteome Research • Vol. 8, No. 8, 2009
was significantly higher, but that of NMP-238 and HSP-70 was significantly lower, in the CCRT-HS group than that in the
Chemoradiotherapy-Associated Proteins in Cervical Carcinoma
research articles
Figure 3. Protein expression of S100A9, galectin-7, NMP-238, and HSP-70 in CCRT-HS and CCRT-LS cervical cancer tissue identified by Western blot. There are five samples in each group to verify the results obtained from 2-D gel and MALDI-TOF-MS analysis, which demonstrate that the expression levels of S100A9 and galectin-7 significantly increase in CCRT-HS group (the cases numbered 1, 3, 5, 7, 9), while those of NMP-238 and HSP-70 significantly increase in CCRT-LS group (the cases numbered 2, 4, 6, 8, 10). GAPDH is used as the internal loading control in three independent experiments. The “*” in the histogram indicates the significantly different expression levels of the target proteins between CCRT-HS and CCRT-LS.
CCRT-LS group (Figure 3). The results of MALDI-TOF-MS and Western blot were consistent with each other. Tissue Microarray and Immunohistochemistry Analysis. The expression of S100A9, galectin-7, NMP-238, and HSP-70 was detected by immunohistochemistry staining in randomly selected 60 CCRT-HS cases and 35 CCRT-LS cases. A representative staining of individual protein in two groups was presented in Figure 4. The positive expression of S100A9 and HSP-70 was mainly located in cytoplasm, while galectin-7 and NMP-238 were mainly expressed in cellular nucleus with small amounts in cytoplasm. In CCRT-HS and CCRT-LS groups, the positive rates of S100A9 expression were 88.3% and 28.6%, the positive rates of galectin-7 expression were 91.7% and 40.0%, the positive rates of NMP-238 expression were 35.0% and 85.7%, and the positive rates of HSP-70 expression were 21.7% and 85.7%, respectively. CCRT-HS group had significantly higher positive expression rates of S100A9 and galectin-7, but significantly lower positive
expression rates of NMP-238 and HSP-70 than the CCRT-LS group (Table 2, p < 0.0001, respectively). The Prediction Analysis of Interested Proteins in CCRT Treated Cervical Cancer. The recurrence rate in CCRT-HS group (5/60, 8.3%) was significantly lower than that in CCRTLS (24/35, 68.6%) (χ2 ) 37.82, p < 0.01). None of these four proteins could be used as an independent prediction factor for the recurrence of cervical cancer after CCRT. However, four proteins of S100A9, galectin-7, NMP-238, and HSP-70 could be well fitted into a Logistic regression model following a multiple stepwise regression analysis (p ) 0.189) by combining these four markers as a full predictive factor. A recurrence predictive model was generated as: recurrence ) 0.808 - (2.311 × S100A9) - (2.740 × Galectin-7) + (0.606 × NMP-238) + (2.064 × HSP-70). The sensitivity, specificity and predictive accuracy of this prediction model was 97.0%, 89.7%, and 91.6%, respectively (Table 2). Journal of Proteome Research • Vol. 8, No. 8, 2009 3973
research articles
Zhu et al.
Figure 4. Immunohistochemical detection of S100A9, galectin-7, NMP-238, and HSP-70 in CCRT-HS and CCRT-LS groups. The arrow indicates the positive staining of interested protein in brown-yellow color in cervical cancer tissue (40×). Table 2. The Protein Expression of S100A9, Galectin-7, NMP-238, HSP-70 and the Prediction Analysis of Interested Proteins by Logistic Stepwise Regression in CCRT-HS and CCRT-LS Cervical Squamous Carcinomaa S100A9
galectin-7
NMP-238
HSP-70
group
case
N
P
rate
N
P
rate
N
P
rate
N
P
rate
CCRT-HS CCRT-LS χ2-value p-value Sensitivity Specificity
60 35
7 25
53 10 35.343
