ARTICLE pubs.acs.org/ac
Multiplexed Immunohistochemical Detection of Tumor Markers in Breast Cancer Tissue Using Laser Ablation Inductively Coupled Plasma Mass Spectrometry Charlotte Giesen,*,†,‡ Thomas Mairinger,§ Lina Khoury,§ Larissa Waentig,‡ Norbert Jakubowski,‡ and Ulrich Panne†,‡ †
Humboldt-Universitaet zu Berlin, Department of Chemistry, Brook-Taylor-Strasse 2, 12489 Berlin, Germany BAM Federal Institute for Materials Research and Testing, Richard-Willstaetter-Strasse 11, 12489 Berlin, Germany § HELIOS Klinikum Emil von Behring, Walterhoeferstrasse 11, 14165 Berlin, Germany ‡
bS Supporting Information ABSTRACT: We optimized multiplexed immunohistochemistry (IHC) on breast cancer tissue. Up to 20 tumor markers are routinely evaluated for one patient, and thus, a common analysis results in a series of time consuming staining procedures. As an alternative, we used lanthanides for labeling of primary antibodies, which are applied in IHC. Laser ablation (LA) ICPMS was elaborated as a detection tool for multiplexed IHC of tissue sections. In this study, we optimized sample preparation steps and LA ICPMS parameters to achieve a sufficient signal-tobackground ratio. The results prove the high selectivity of applied antibodies, which was sustained after labeling. Up to three tumor markers (Her 2, CK 7, and MUC 1) were detected simultaneously in a single multiplex analysis of a 5 μm thin breast cancer tissue at a laser spot size of 200 μm. Furthermore, the LA ICPMS results indicate a significantly higher expression level of MUC 1 compared to Her 2 and CK 7, which was not obvious from the conventionally stained tissue sections.
O
ne third of all cancers in women are breast and ovary carcinoma. Together, they account for approximately 1/4 of cancer related deaths in females.1 Immunohistochemistry (IHC) is the method of choice in cancer diagnostics to assess the level of expression of tumor markers in tissue sections. An antibody reacts with the tumor marker, e.g., an oncoprotein, with high selectivity. Three important tumor markers for breast cancer diagnosis are discussed in the following. The c-erbB-2 oncoprotein is overexpressed in 2530% of human primary breast cancers.1 Antihuman epidermal growth factor receptor 2 (Her 2) recognizes an epitope on the c-erbB-2 protein and is used as a primary antibody in IHC. The results of IHC analysis determine further medical treatment concepts. Another antibody used for IHC in breast cancer diagnosis is anticytokeratin 7 (CK 7). Cytokeratin 7 is a type II cytokeratin, which is specifically expressed in glandular epithelia. The majority of breast cancers are positive for CK 7, and it is used to determine the origin of metastatic breast carcinoma.2 Mucin 1, a membrane-associated protein found on the luminal surface of many columnar epithelia, is also known to be up-regulated in a subset of breast cancers but is expressed at very low levels in a normal mammary gland. Thus, antimucin 1 (MUC 1) is used as a tumor marker for breast cancer diagnosis.3 Commonly, an indirect method is applied for IHC using a primary antigen-specific antibody and a second incubation with r 2011 American Chemical Society
an enzyme-labeled secondary antibody for signal amplification.4 This procedure impedes screening for simultaneous evaluation of more than one antigen at the same time, resulting in several subsequent staining procedures until a tumor-identifying antigenprofile can be determined. Therefore, the analytical tool of choice for imaging of cancer tissue samples should be able to detect many biomarkers simultaneously (multiplexing) with high sensitivity and high local resolution. Laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) is such a powerful tool and has been widely used over the past decade for bioimaging of tissue samples.5,6 It was shown in the literature that tumor boundaries were clearly marked by imaging of 31P in lymph node biopsies.7 Becker et al.8 used matrix-matched laboratory standards to receive information on the quantitative distribution of copper, zinc, uranium, and thorium in 20 μm thin tissue sections of human brain (hippocampus). More recently, the same group produced images of quantitative element distribution of metals (zinc, copper, iron, manganese, and titanium) in mouse heart tissue sections, in addition to secondary ion mass spectrometry (SIMS) images of alkali metals and biomolecules.9 For imaging of biomolecules in Received: July 5, 2011 Accepted: September 17, 2011 Published: September 18, 2011 8177
dx.doi.org/10.1021/ac2016823 | Anal. Chem. 2011, 83, 8177–8183
Analytical Chemistry tissue sections, other mass spectrometric techniques such as matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) were used as well.10 The mapping of metallodrugs by LA-ICPMS is also possible.11 A critical review on the most common metal imaging techniques, LA-ICPMS, secondary ion mass spectrometry (SIMS), and synchrotron X-ray fluorescence (SXRF), was published by Qin et al.12 The first application of an element-tagged immunoassay coupled with ICPMS detection was described by Zhang et al.13 for the detection of thyroid-stimulating hormone in human serum. Tanner and co-workers further pioneered the determination of proteins by ICPMS in their work on elemental-tagged immunoassays.14,15 Hu et al.16 employed elemental tagged antibodies and reported on the detection of multiple proteins on a single microarray spot by means of LA-ICPMS. For analysis of the Mre11 protein in crude lysates of CHO-K1 fibroblasts, M€uller et al.17 employed LA-ICPMS of Western Blot membranes using gold-clusterlabeled antibodies. The multielement capability of LA-ICPMS directly on a Western blot membrane has been investigated by Waentig et al.18 using lanthanide labeled antibodies. Terenghi et al.19 have elaborated a multiplexed ICPMS determination of cancer biomarkers in serum and tissue lysates using size exclusion chromatography (SEC) to separate the antibodyantigen complex from unbound antigen and antibody. For signal amplification in ICPMS with liquid sample introduction, Tanner and co-workers developed a malemide-functionalized polymer tag,20 which is especially useful for the analysis of single cells.21,22 Recently, Bendall et al. reported on the simultaneous detection of 34 cellular parameters in the primary human hematopoietic system, applying these reagents.23 The direct detection of cancer biomarkers in tissue sections was already reported in the literature. Hutchinson et al.24 applied Eu- and Ni-coupled secondary antibodies for LA-ICPMS imaging of β-amyloid deposits in mouse brain tissue. Furthermore, the distribution of MUC 1 or Her 2 in breast cancer tissue sections was studied by employing gold nanoparticle labeled secondary antibodies and silver enhancement for signal amplification.25 An overview on elemental tagging in inorganic MS is given by Bettmer et al.26 In modern cancer diagnostics, it is urgently needed to screen for several tumor markers simultaneously, but the use of labeled secondary antibodies hampers a multiplexed IHC. There are only very few attempts of double or triple staining in IHC literature due to the risk of cross reactions among individual staining steps and the difficulty of simultaneous visual evaluation of stained tissue. A commercial triple staining cocktail (PIN-4 cocktail, BioCare) was employed for prostate carcinoma diagnosis. Each marker stained different cell compartments, but visualization of all three markers was accomplished by 3,30 diamino benzidine (DAB) in brown.27 Therefore, an independent assessment of three tumor markers was impeded. For the use of multiple dyes, secondary antibodies with different selectivity have to be employed.28 This approach hampers multiplexing by the limited availability of different antibody types applicable to IHC. Furthermore, dyes overlap for colocalized tumor markers,28 and there is a danger of cross-reactivity. Hence, time-consuming subsequent staining procedures of parallel thin sections are still the gold standard in IHC. Additionally, quantification is hampered for this approach. The use of elemental labeled primary antibodies in combination with LA-ICPMS for IHC detection offers three advantages: (i) a possibility for quantification, provided suitable standards are available, (ii) a simultaneous detection of
ARTICLE
several tumor markers within one tissue section, even if they are colocalized, and (iii) a shorter analysis time due to simultaneous detection of several tumor markers. Concerning metallabeled primary antibodies, only a few are commercially available. To overcome this drawback, we employed LA-ICPMS as a detection tool for IHC in combination with a labeling technique established by Waentig et al.18 using p-SCN-Bn-DOTA for labeling of primary antibodies. The optimization of this novel approach and first applications for breast cancer tissue samples will be described in the following sections.
’ EXPERIMENTAL SECTION LA-ICPMS: Instrumentation and Measurements. Experiments were conducted using a commercial laser ablation system (New Wave 213) coupled to an ICP sector field mass spectrometer (Element XR, Thermo Fisher Scientific, Germany). The New Wave 213 is equipped with a beam expander, and the laser spot size is adjustable between 4 and 250 μm. Typical experimental parameters are summarized in Table S1 in the Supporting Information. The ICP was tuned daily for maximum ion intensity, keeping the oxide ratio (ThO/Th) below 1% during ablation of a microscopic glass slide. Thin cuts of tissue samples, mounted on microscopic glass slides, were inserted into the two volume cell (New Wave) and ablated line by line. The ICPMS was synchronized with the LA unit in external triggering mode. Laser ablation parameters including laser energy, laser spot size, scan speed, and repetition frequency were optimized to ablate tissue samples completely without cracking the sample and causing damage to adjacent lines. LA-ICPMS parameters were tuned for optimum signal-to-background ratios, and the total analysis time of a tissue (6 mm 5 mm) was 1 h at 200 μm laser spot size and 200 μm s1 scan speed. LA-ICPMS data were exported to Origin 8.5 (Originlab Corporations, Northampton), where either intensity time profiles or color coded images can be produced. Sample Preparation for Immunostains. Formalin-fixed, paraffin-embedded (FFPE) tissue blocks of human breast cancer were sectioned using a conventional sliding microtome (Leica SM 2010R, Leica Microsystems, Wetzlar, Germany) with a thickness setting of 3 μm. The sections were mounted onto Superfrost Plus slides (Thermo Fisher Scientific, Braunschweig, Germany) for IHC staining. Thin sections for IHC staining were processed by the fully automated BenchMark XT slide preparation system (Ventana Medical Systems, Inc., Tucson, AZ). The sequence of automated events for immunohistochemical analyses is specified in Table S2 in the Supporting Information. Briefly, paraffin was removed from tissue samples, followed by epitope retrieval. Tissues were then incubated with primary and secondary antibody, which are applicable to paraffin-embedded sections. The last step involves visualization of binding events. Optimized Sample Preparation for LA-ICPMS. Labeling of anti-Her 2, anti-CK 7, and anti-MUC 1 with holmium, thulium, and terbium, respectively, was performed as described in more detail by Waentig et al.18 Anti-Her 2 and anti-MUC 1 used for IHC staining were purchased in purified form. Anti-CK 7 used for IHC staining was delivered as cell culture supernatant by the manufacturer. Slight modifications were applied for anti-CK 7, which was labeled at a higher excess rate of n[SCN-DOTA(Tm)]/n[Ab] = 275:1 to guarantee a ligand excess even in those cases where the purification (Supporting Information section S3) is not sufficient and cell culture proteins remained in solution. 8178
dx.doi.org/10.1021/ac2016823 |Anal. Chem. 2011, 83, 8177–8183
Analytical Chemistry
ARTICLE
Figure 1. Optimization of laser ablation parameters on a 5 μm breast cancer tissue section, incubated with ∼1 μg mL 1 MUC 1 (Tb) for 3 h: (a) laser spot size 200 μm, scan speed 200 μm s 1, repetition rate 10 Hz, laser energy 35%; (b) laser spot size 25 μm, scan speed 25 μm s 1, repetition rate 10 Hz, laser energy 35%; (c) laser spot size 200 μm, scan speed 150 μm s 1, repetition rate 20 Hz, laser energy 35%; (d) laser spot size 200 μm, scan speed 200 μm s 1, repetition rate 20 Hz, laser energy 35%.
FFPE tissues were sectioned at 5 μm for LA-ICPMS measurements by a microtome and mounted onto Superfrost Plus slides. Thin sections were deparaffinized in xylene, rehydrated through a series of alcohols, and incubated in target retrieval solution pH 6 (Ventana) for 20 min at 90 °C. For the multiplex experiment, after rinsing with wash buffer, an aliquot (100 μL) of all diluted primary antibodies (∼1 μg mL1 anti-Her 2 (Her 2 (Ho), antiCK 7 (CK 7 (Tm), and anti-MUC 1 (MUC 1 (Tb)) at the same time was placed onto the sections for 3 h. Incubation was performed in a hybridization chamber to prevent drying of the tissue. In the following step, sections were rinsed with wash buffer to remove unbound antibodies and dehydrated through graded alcohols prior to laser ablation.
’ RESULTS AND DISCUSSION Optimization of LA-ICPMS Based IHC Detection. It should be mentioned that up until now, metal tagged antibodies have been employed in our work for Western Blot assays only, and we have achieved limits of detection in the subpicomole range.18 Antigen concentration in tissue samples is unknown to a large extent, and this fact also applies to the tissue samples investigated in this work. Thus, it was certainly questionable if the sensitivity of metal tagged antibodies using our Western Blot procedure could be transferred to an IHC application. Various parameters of the IHC procedure (labeling, incubation, and laser ablation conditions) have a direct impact on the selectivity of the labeled primary antibody and on the sensitivity in LA-ICPMS, respectively. They determine the signal-to-background ratio and consequently local resolution in LA-ICPMS imaging. Hence, sample preparation steps and LA-ICPMS parameters were investigated in this work. The optimization strategy was aimed at sufficient signal-to-background ratios with concomitant high local resolution and complete ablation of the tissue. An optimized tissue thickness is crucial for assessment of
tumor markers in immunohistochemically stained thin tissue sections. The same applies to LA-ICPMS measurements. During target retrieval, tissues stick to the glass surface due to adhesive effects of the Super Frost microscopic slides. This is true for 3 μm thin slices employed in conventional IHC. For LA-ICPMS detection, signal intensities increase linearly with tissue thickness. Thus, several tissue thicknesses were tested during the IHC procedure and during laser ablation. From 8 μm onward, tissue sections were no longer removed from the glass surface during target retrieval, and LA-ICPMS detection could be performed. However, the 8 μm tissue was prone to cracking during laser ablation at laser energies above 35%, which caused damage to adjacent lines and inhibited imaging of the tissue. On the other hand, the tissue was not ablated completely at lower energies. Hence, a thickness setting of 8 μm was not suitable for tissue imaging in this case. Furthermore, particles being torn out of the tissue during cracking most likely did not reach the ICP and did not contribute to a detectable signal. In contrast, the 5 μm thin section was ablated without cracking of the tissue. Drying of the tissue with a graded series of alcohols as a last step of sample preparation reduced cracking during laser ablation even further. Since 3 μm tissue sections resulted in lower signal intensities, the optimal tissue thickness for LA-ICPMS was 5 μm. Laser energy was optimized on behalf of a clean ablation of the whole sample without cracking the tissue, and thus the maximum laser energy was limited to not more than 35% for our experiments. Furthermore, incubation time and antibody concentration were optimized at 3 h and ∼1 μg mL1, respectively. Compared to conventional IHC, primary antibody concentrations employed for LA-ICPMS application were in the same range or lower, together with longer incubation times (for details please refer to Table S2 in the Supporting Information). These conditions generated high signal intensities on the one hand but minimized nonspecific binding on the other hand. 8179
dx.doi.org/10.1021/ac2016823 |Anal. Chem. 2011, 83, 8177–8183
Analytical Chemistry
Figure 2. Single line scans of 5 μm breast cancer tissue section (a,b), and 5 μm palatine tonsil tissue section (ce) incubated with ∼1 μg mL1 primary antibodies for 3 h. Her 2 (Ho) (a,c) and CK 7 (Tm) (b,d) were detected at a laser spot size of 200 μm, scan rate of 150 μm s 1, repetition rate of 10 Hz, and laser energy of 35%. MUC 1 (Tb) (e) was detected at a laser spot of 200 μm, scan rate of 200 μm s 1, repetition rate of 20 Hz, and laser energy of 35%. The positive control corresponding to part e is given in Figure 1d.
Laser spot size and scan speed are the most important parameters for laser ablation measurements since they determine signal-to-background ratio, local resolution, and analysis time. Single line scans were performed on a 5 μm breast cancer tissue section incubated for 3 h with ∼1 μg mL 1 MUC 1 (Tb). The laser spot size was varied from 25 to 200 μm while keeping the repetition rate at 10 Hz, the laser energy at 35%, and the ratio of the laser spot size to scan speed constant. Terbium intensities are depicted in Figure 1a,b. They were highest for large laser spot sizes and decreased for smaller spot sizes. Terbium intensities up to approximately 1000 cps could be detected for laser spot sizes as small as 25 μm and are shown in Figure 1b. The scan speed was also varied, with 200 μm laser spot size, 20 Hz repetition rate, and 35% energy. Examples of single line scans are presented in Figure 1c,d. The signal intensities increased with increasing scan speed since more tissue aerosol was produced within the same time frame. Laser ablation with repetition rates of 20 Hz yielded higher intensities than with 10 Hz for the same reason, as illustrated in Figure 1a,d. Therefore, 200 μm laser spot size, 200 μm s1 scan speed, 20 Hz repetition rate, and 35% laser energy were applied for an optimized analysis. Immunostain vs LA-ICPMS Detection of Tissue Sections. Primary antibodies Her 2, CK 7, and MUC 1 applied in this work were labeled with Ho, Tm, and Tb, respectively, via a DOTA linker. A prerequisite for pathological analyses is the high selectivity of antibodies which needs to be sustained after labeling
ARTICLE
and thus was carefully evaluated. Therefore, incubation on a breast cancer tissue section was always accompanied by a parallel incubation on a palatine tonsil tissue section, which served as a negative control. Positive and negative controls were cut at the same thickness and mounted onto the same glass slide, and all sample processing steps were performed identically. The tissue was cut at 5 μm and simultaneously incubated with labeled antibodies, as described in the Experimental Section. Examples of single line scans of breast cancer tissue sections (positive controls) are presented in Figure 1 for MUC 1 and in Figure 2a,b for Her 2 and CK 7, respectively. Single line scans of identically processed palatine tonsil tissue (negative controls, Figure 2ce) were analyzed during the same ICPMS sequence as positive controls. Holmium and Tm intensity time profiles shown in parts a,b and parts c,d of Figure 2 were from the same single line scans due to a parallel incubation of Her 2 (Ho) and CK 7 (Tm), respectively. In the case of Her 2 (Ho) negative controls, measured intensities were hardly distinguishable from the background noise in the tissue (
