It was found previously that αB-crystallin, a small heat-shock protein, was overexpressed in a metastatic variant of the GI101A human breast carcinoma cell line. The objective of the current study was to determine whether the expression of αB-crystallin in primary breast carcinomas was associated with lymph node metastasis and survival.
Expression of αB-crystallin was measured in human breast carcinoma cell lines by immunoblotting. Expression in human breast carcinomas was evaluated by immunohistochemical staining of tissue microarrays that contained samples from 317 patients with lymph node–negative breast carcinoma and 291 patients with lymph node–positive breast carcinoma.
It was found that αB-crystallin was expressed constitutively in certain breast carcinoma cell lines, including those that were capable of metastasizing in immunodeficient mice. Expression of αB-crystallin in human tissue samples was associated strongly with lymph node involvement (P < 0.0001; chi-square test) and, to a lesser degree, with high nuclear grade (P = 0.05). Increased intensity of expression was correlated with shorter survival (P = 0.0091; log-rank test). Multivariate analysis indicated that αB-crystallin expression was not independent of lymph node status as a predictor of survival.
Axillary lymph node tumor involvement is the most important prognostic factor for patients with breast carcinoma.1 The concept on which the radical mastectomy procedure developed by Halstead in the 1890s was based was that breast carcinoma spread progressively over time to lymph nodes and, ultimately, to distant sites. The National Surgical Adjuvant Breast Project B04 trial1 found that axillary lymph node dissection was important for achieving local control and obtaining prognostic information but that prophylactic removal of the lymph nodes did not improve survival. These findings suggest that breast carcinoma is systemic at the outset and that axillary lymph node metastases are markers of tumor biology rather than a step in dissemination of the disease. Accurate prognostic information based on molecular characteristics of primary breast tumors could eliminate the need for lymph node dissection in many patients. Moreover, studies of markers of poor prognosis are likely to identify new therapeutic targets, such as Her2/neu, a marker of poor prognosis in patients with breast carcinoma and a molecule that is targeted effectively by the drug trastuzumab.2
High-throughput technologies have enhanced prognostic marker identification in breast carcinoma by identifying gene expression patterns correlated with clinical outcomes.3, 4 Another approach involves the use of experimental tumor models to identify genes responsible for metastatic progression.5, 6 We investigated the gene expression profiles of isogenic cell populations that differ in their metastatic ability. Variants of the GI101A human breast carcinoma cell line7 were compared via cDNA microarray analysis (unpublished data). One of the genes that was overexpressed (by a factor of ∼8) in the more metastatic variant was αB-crystallin. Crystallins, including αB-crystallin, are soluble proteins that are found primarily in the lens of the eye; αB-crystallin also is found in normal and diseased nonlenticular tissues.8 It is a small heat-shock protein (hsp) that functions as a molecular chaperone,9 and like hsp27, it has a cytoprotective role and has been shown to inhibit apoptosis.10, 11 Unlike hsp27, however, αB-crystallin has not been reported on frequently in the literature on malignant disease. One study reported the presence of αB-crystallin in tumors arising from organs that expressed it in nonpathologic conditions, including colorectal, thyroid, renal cell, and breast carcinomas.8 The objectives of the current study were to examine αB-crystallin expression in breast carcinoma cell lines and to assess the value of this protein as a molecular staging marker using breast carcinoma tissue microarray analysis.12
MATERIALS AND METHODS
Cell Culture and Immunoblotting
αB-crystallin expression was measured in breast carcinoma cell lines that were obtained from the American Type Culture Collection (Rockville, MD; SKBR3, T47D, and MCF-7), the Goodwin Institute (Plantation, FL; GI101A), Dr. Stephen Ethier (University of Michigan, Ann Arbor, MI; SUM149), and Dr. Relda Cailleau (University of Texas M. D. Anderson Cancer Center, Houston, TX; MDA-MB-231, MDA-MB-435, MDA-MB-468, and MDA-MB-361). The GILM2, MDA-435Lu2, MDA-435Br1, and MDA-435LvBr cell lines were derived from metastases in nude mice, as reported previously.7, 13 Cells were maintained in medium (minimum essential medium or Dulbecco minimal essential medium/Ham F12) containing 5% or 10% fetal bovine serum and L-glutamine in a humidified incubator with 5% CO2.
Cell lysates were prepared from cultures at approximately 80% confluence; some cultures were exposed to a temperature of 42 °C or to 1% O2 for 16 hours before lysate collection. Aliquots of total protein were separated on 15% sodium dodecyl sulfate (SDS)-polyacrylamide gels; transferred to nitrocellulose membranes; and hybridized with monoclonal anti-αB-crystallin, monoclonal anti-hsp27 (both from Stressgen Biotechnologies, Victoria, British Columbia, Canada), or polyclonal anti-hsp70 (Santa Cruz Biotechnology, Santa Cruz, CA). Membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies, which were detected with the Amersham ECL system (Amersham, Arlington Heights, IL). The αB-crystallin antibody detects a single band at approximately 20 kDa, the predicted molecular weight of the protein, in SDS-polyacrylamide gels; the intensity of this band is greater in samples obtained from heat-stressed cells (Fig. 1).
Tissue Microarray Construction and Immunohistochemistry
Tissue microarrays were constructed as described previously.12 Cores measuring 0.6 mm were spaced 0.8 mm apart. αB-crystallin expression was evaluated on an array of 328 lymph node–positive specimens and 379 lymph node–negative specimens that was constructed from paraffin-embedded, formalin-fixed blocks from the archives of the Department of Pathology at Yale University (New Haven, CT). Estrogen receptor staining was positive in 52% of specimens, progesterone receptor staining was positive in 46% of specimens, and HER2/neu staining was positive in 14% of specimens. Nuclear Grade 3 (on a scale of 1 to 3) was observed in 28% of specimens, and 59% of tumors measured > 2 cm. Histologic subtypes included invasive ductal carcinoma (72%), lobular carcinoma (1%), and mixed or other subtype (14%). Specimens were resected between 1962 and 1980, with follow-up ranging from 4 months to 53 years (mean follow-up, 12.6 years). Patient age at diagnosis ranged from 24 years to 88 years (mean age, 58 years). Complete treatment history was not available for the entire cohort. Most patients were treated with local irradiation. No patient with lymph node–negative disease received adjuvant systemic therapy. A minority of patients with lymph node–positive disease (approximately 15%) received chemotherapy, and approximately 27% received tamoxifen (all after 1978). Data on the time between tumor resection and tissue fixation were not available. Slides were reviewed by a pathologist, who selected areas of invasive tumor to be placed on the tissue microarray using a Tissue Microarrayer (Beecher Instruments, Silver Spring, MD). Five-micrometer-thick sections were cut and placed on glass slides using an adhesive tape system (Instumedics, Hackensack, NJ) and ultraviolet cross-linking. Slides were deparaffinized, endogenous peroxidase activity was blocked using the peroxidase block reagent included in the Envision+ system (DAKO, Carpinteria, CA), and slides were incubated for 30 minutes at room temperature. Slides were then boiled in a pressure cooker with a sodium citrate buffer (pH 6.0) for antigen retrieval, washed in water, and incubated for 30 minutes in Tris-buffered saline (TBS) with 0.3% bovine serum albumin to block nonspecific staining. The αB-crystallin antibody was diluted 1:100 in the blocking buffer and incubated with the slides at 4 °C for 16 hours in a humidified chamber. HRP-conjugated anti-mouse immunoglobulin G was applied for 1 hour, and sections were washed in TBS with 0.05% Tween (Sigma Chemical Co., St Louis, MO). Antibodies were visualized via incubation with diaminobenzidine (DAKO). Sections were counterstained with hematoxylin, and slides were mounted with Immunomount (Shandon, Pittsburgh, PA).
Evaluation of Tissue Microarray Immunohistochemical Staining
The most intensely stained regions were scored by eye for each spot. Only cytoplasmic/membranous staining of malignant cells was observed and subsequently scored. Because histospots were small and lacked variability in terms of staining, no area variable was included in the scoring. Staining was graded as follows: 0, no staining; 1, weak staining; 2, moderate staining; and 3, intense staining. Specimens that did not contain infiltrating carcinoma and specimens that were not interpretable were excluded from the analysis. Slides were scored separately by two observers (H.M.K. and A.J.B.), with a very high correlation between the results obtained by the scorers (P < 0.0001). Consensus scores were determined for spots for which there were scoring discrepancies between observers.
The Statview software package (Version 5.0.1; SAS Institute, Cary, NC) was used. Correlations between αB-crystallin expression and clinicopathologic parameters were analyzed using chi-square tests. Parameters were assessed for predictive value using the Cox proportional hazards model, with overall survival serving as an endpoint. Survival curves were generated using the Kaplan–Meier method, with significance evaluated using the Mantel–Cox log-rank test.
αB-Crystallin Expression in Human Breast Carcinoma Cells
Cell lysate immunoblotting revealed variable αB-crystallin expression in the panel of breast carcinoma cells that was investigated (Fig. 1A). The strongest expression was seen in the GILM2, MDA-435LvBr, and MDA-435Br1 cell lines, which were isolated from metastases in immunodeficient mice injected with GI101A and MDA-MB-435 cells.7, 13 Expression was weaker in the original cell lines. Figure 1B shows that expression levels of αB-crystallin and two other heat-shock proteins were responsive to heat stress in GI101A cells. However, the metastasis-derived GILM2 cells exhibited high constitutive expression of αB-crystallin under normal culture conditions (temperature, 37 °C).
Immunohistochemical Staining of Tissue Microarrays
Of the 672 tumors included in the microarrays, 608 (90%) were interpretable for αB-crystallin staining (317 lymph node–negative tumors and 291 lymph node–positive tumors). Spots that were deemed uninterpretable had insufficient tumor cells, loss of tissue in the spot, or an abundance of necrotic tissue. Overall, 536 tumor cores (88%) also had associated patient survival information. Examples of scores 0 and 3 are shown in Figure 2. Seventy-four tumors (12%) had no cytoplasmic αB-crystallin staining, 145 (24%) had weakly positive staining (score 1), 214 tumors (35%) had moderate cytoplasmic staining (score 2), and 175 tumors (29%) had strong cytoplasmic αB-crystallin staining (score 3). Of the 536 patients for whom survival information was available, 259 had lymph node–positive disease, whereas 277 had lymph node–negative disease. Figure 3A shows the distribution of αB-crystallin staining in these two populations. There were significantly more αB-crystallin-positive tumors among patients with lymph node–positive disease compared with patients with lymph node–negative disease (P < 0.0001; chi-square test). Cytoplasmic αB-crystallin expression was found to be correlated with breast carcinoma–specific overall survival after 20 years of follow-up. Kaplan–Meier survival curves stratified according to ordinal αB-crystallin expression score (Fig. 3B) revealed an association between increased expression and decreased survival (P = 0.0091). Although these data are only semiquantitative, the curves suggest a splitting of the data at 5 years of follow-up to define scores of 0 and 1 as being indicative of negative or low expression and scores of 2 and 3 as being indicative of high or positive expression. These designations are used for the remainder of the analyses.
Clinicopathologic Correlations and Multivariate Analyses
Using the Cox proportional hazards model, we performed multivariate analyses to assess the independent predictive value of cytoplasmic staining for αB-crystallin. The variables used included tumor size, lymph node status, estrogen and progesterone receptor staining, HER2/neu staining, nuclear grade, and patient age. αB-crystallin was not found to be an independent predictor of breast carcinoma–specific survival, and the only three independent predictors of survival were lymph node status, nuclear grade, and tumor size. Correlations of clinicopathologic variables with αB-crystallin staining are shown in Table 1. αB-crystallin staining was associated strongly with lymph node involvement (P < 0.0001) and also was associated with nuclear Grade 3 disease (P = 0.05); however, αB-crystallin staining was not associated with any of the other variables examined.
Table 1. Associations between Cytoplasmic αB-Crystallin Expression and Clinicopathologic Variablesa
The number of patients with high αB-crystallin levels (n = 389) and the total number of patients (n = 608) associated with each clinicopathologic variable are shown, with percentages indicated in parentheses. High αB-crystallin expression refers to results that were scored as 2 or 3.
High nuclear grade was defined as a nuclear grade of 3 on a scale of 1–3.
HER2-positive results were scored as 2 or 3 on a scale of 0–3.
The data obtained in the current study support the hypothesis that the small heat-shock protein αB-crystallin serves as a marker of breast carcinoma progression. Constitutive αB-crystallin expression in human breast carcinoma cells in vitro was associated with the ability to metastasize in nude mice, with the highest expression levels observed in cell lines that were established from metastatic cells. αB-crystallin expression also has been observed in xenograft tumors and in lymph node and lung metastases of breast carcinoma in immunodeficient mice (data not shown). Most notably, elevated expression in a large cohort of breast carcinoma specimens was correlated with lymph node metastases, with high–nuclear grade disease, and with poor overall survival. The association between αB-crystallin expression and lymph node status was highly significant (P < 0.0001). The association between αB-crystallin and breast carcinoma–specific survival also was statistically significant (P < 0.0091), but it did not retain statistical significance in a multivariate analysis in which lymph node status was included. This finding suggests that the effect of αB-crystallin expression on patient survival is related to the strong association with lymph node metastasis.
The primary advantage associated with investigating tumors resected 20–40 years ago is that they are likely to reflect the natural history of the disease. Tumors that predated the routine use of screening mammography were detected due to the presence of palpable masses and, on average, are larger than tumors diagnosed using modern techniques. However, > 50% of the tumors that were included in the current study did not metastasize to the lymph nodes, with these tumors exhibiting a more benign biology than that of their lymph node–positive counterparts. Moreover, none of the patients in the lymph node–negative population and < 30% of patients with lymph node–positive disease were treated with adjuvant therapy. Hence, outcome data reflect biologic aggression, rather than the effects of systemic therapy. Specimens collected more recently would require stratification of patients based on the type of therapy received, and many of these therapy options would be obsolete at 20 years of follow-up.
The discovery of a potential new marker for breast carcinoma progression warrants a thorough investigation of the functions of the protein and its potential role(s) in metastasis. αB-crystallin expression has been reported in several pathologic conditions (primarily those of the nervous system)14 and in tumors, including breast carcinoma, that arise from tissue types that normally express the protein.8, 15 To our knowledge, the current report is the first to document a connection between αB-crystallin and breast carcinoma progression.
αB-crystallin shares 40% homology with other small heat-shock proteins. We confirmed the induction of αB-crystallin in breast carcinoma cells that were exposed to heat shock. There is growing evidence that heat-shock proteins play a major role in the ability of tumor cells to survive stress induced by external stimuli, enhancing resistance to apoptosis.16, 17 Similarly, it has been shown that the expression of αB-crystallin protects cells from apoptosis induced by DNA-damaging agents, tumor necrosis factor α, and Fas ligation.10 Such protection may result from the binding of αB-crystallin to partially processed caspase-3 and the resulting inhibition of the autoproteolytic maturation of the latter protein, a key effector molecule in the apoptotic cascade.18
Resistance to apoptosis has been defined as one of the hallmarks of malignant disease.19 Aberrant expression of small heat-shock proteins, including αB-crystallin, may contribute to resistance to apoptosis. Proteomic analysis of matched normal ductal/lobular units and ductal carcinoma in situ identified increased expression of six chaperone and heat-shock proteins, one of which was αB-crystallin, in the tumor samples.20 Ongoing studies are further investigating the contribution of αB-crystallin to the malignant progression of breast carcinoma.
Although not all lymph node–positive specimens expressed αB-crystallin, and although some of the lymph node–negative specimens did express this protein, αB-crystallin expression still may be a helpful predictor of the likelihood of lymph node involvement, particularly in cases in which complete lymph node sampling is not possible. Examples of such situations include cases in which neoadjuvant chemotherapy (which affects the ability to perform accurate surgical staging) is used and cases in which patients have comorbidities or financial considerations that prohibit lymphadenectomy. Moreover, because surgical staging has limited therapeutic utility (with the exception of providing local control for patients who have massive lymphadenopathy),1 accurate molecular predictors of poor outcome based on information that is obtainable at the time of initial biopsy may eventually obviate the need for lymph node dissection. Because αB-crystallin expression is not accurate enough on its own to predict poor outcome, ongoing studies are focusing on the coexpression of αB-crystallin together with other predictors of lymph node metastases that, collectively, may be clinically useful in predicting lymph node involvement and survival.