Characterization and Validation of Antibodies for Immunohistochemical Staining of the Chemokine CXCL12

Abstract

Chemokines and their receptors have been implicated in cancer biology. The CXCL12/CXCR4 axis is essential for the homing and retention of hematopoietic stem cells in bone marrow niches, and has a significant role in neonatal development. It is also implicated in multiple facets of cancer biology including metastasis, angiogenesis/neo-vasculogenesis, and immune cell trafficking at the tumor microenvironment (TME). Immunohistochemistry (IHC) is an ideal method for investigating involvement of CXCL12 in the TME. Three antibodies were evaluated here for their suitability to stain CXCL12. Both D8G6H and K15C gave apparent specific staining in both lymphoid and tumor tissue, but with converse staining patterns. D8G6H stained cells in the parafollicular zone whereas K15C showed staining of lymphoid cells in the interfollicular zone of tonsil tissue. Using a cell line with high CXCL12 expression, TOV21G, as a positive control, it was found that D8G6H gave strong staining of TOV21G cells whereas no staining was observed with K15C indicating that D8G6H specifically stains CXCL12. Significant staining of CXCL12 in the ovarian TME using tissue microarray was observed using D8G6H. These data demonstrate the importance of antibody characterization for IHC applications, and provide further evidence for the involvement of CXCL12 in ovarian cancer biology.

Introduction

Chemokines, chemotactic cytokines, are 8 to 10 kD proteins that signal and regulate the homing and trafficking of cells of the immune system to and from lymphoid organs and sites of inflammation. To date, more than 40 chemokines have been identified of which the majority fall into two major classes characterized by the relative position of N-terminal cysteines.^1,2 These are the CXC chemokines where the cysteines are separated by a single amino acid and the CC chemokines where the two cysteines are adjacent. CXCL12, previously named SDF-1, falls into the first category.

The receptors for chemokines are seven-transmembrane proteins of the class A G-protein coupled receptor family. CXCL12 is the sole cognate ligand for the receptor CXCR4. CXCR4 plays a critical homeostatic role in hematopoiesis. The CXCL12/CXCR4 axis regulates lymphocyte trafficking, and homing and retention of hematopoietic stem cells (HSC) in bone marrow niches.³ Disruption of the CXCL12/CXCR4 interaction in the bone marrow results in mobilization of HSCs into the circulation system. Plerixafor, a CXCR4 inhibitor, is approved for mobilization of HSCs for autologous transplant in patients with multiple myeloma and non-Hodgkin’s lymphoma.^4,5 In addition, CXCR4 plays a significant role in neonatal development on processes including hematopoiesis, vascularization, and neurological development. CXCL12 is also a ligand for the atypical chemokine receptor ACKR3 (formerly known as CXCR7 or RDC1).⁶

CXCR4 was first investigated as a therapeutic target for HIV as CXCR4 is one of the two main co-receptors used by HIV for viral cell entry.^7,8 CXCR4 is also expressed on cancer cells and cancer stromal cells including endothelial cells and infiltrating immune cells where it plays multiple roles in tumor progression including metastasis, angiogenesis/neo-vasculogenesis, survival, and immune cell trafficking in the tumor microenvironment (TME).^9,10 In addition, ACKR3 is expressed in cancer cells and cancer-associated endothelial cells and has been implicated in cancer progression, metastasis, and angiogenesis. There is an increasing focus on the CXCL12/ACKR3/CXCR4 axis as a target for oncology.^6,11,12 To fully understand the role of this chemokine axis in tumor biology, there is a need to be able to identify and demonstrate expression and localization of the ligand, CXCL12, within the TME.

CXCL12 mRNA expression has been frequently demonstrated in cancer tissue from cancers such as ovarian,¹³ pancreatic,¹⁴ neuroendocrine,¹⁵ and breast by RT-PCR,^16,17 and protein expression of CXCL12 has been measured by ELISA. However, neither of these techniques allows for identification of either cell localization or the cell type producing CXCL12, essential information for a full understanding of the role of the CXCL12/CXCR4 axis in both normal physiology and disease pathophysiology. Several groups have used immunohistochemistry (IHC) to demonstrate the presence of CXCL12 in the TME.^18–21 However, one of the challenges with an IHC approach is the little available antibody characterization data in reported studies. It has become increasingly apparent that there are major problems with reproducing published IHC data in part due to the lack of quality control over antibody production and in part due to increasing awareness of the poor specificity of many antibodies.^22–24 With the need to accurately assess expression of CXCL12 in the TME, we set out to identify and rigorously characterize three readily available commercial CXCL12 antibodies for IHC use. Only one evaluated antibody was found to satisfactorily stain CXCL12 specifically. Our studies emphasize the need to fully characterize antibodies for IHC use.

Materials and Methods

Antibodies

Three anti-CXCL12 antibodies were tested (Table 1). The rabbit polyclonal LS-B943 was obtained from LifeSpan BioSciences, the rabbit monoclonal antibody D8G6H and the mouse monoclonal antibody K15C were obtained from Cell Signaling Technology and Merck Millipore, respectively.

Table 1.

CXCL12 Antibodies Examined.

Anti-CXCL12 Antibody	Antigen	Heat Induced Antigen Retrieval Method for IHC	Antibody Concentration for IHC
Polyclonal (rabbit) LS-B943	C-terminus of CXCL12	EDTA buffer pH8.5	0.4 μg/ml
Monoclonal (rabbit) D8G6H	Recombinant CXCL12	EDTA buffer pH8.5	0.45 μg/ml
Monoclonal (mouse) K15C	N-terminus of CXCL12	EDTA buffer pH8.5	20 μg/ml

Abbreviations: IHC, Immunohistochemistry; EDTA, ethylenediaminetetraacetic acid.

The rabbit monoclonal anti-CXCR4 clone UMB2 (AbCam, Cambridge, MA) was used for detecting CXCR4.

Cell Lines and Human Tissues

Cell lines used are listed in Table 2. TOV21G was obtained from ATCC (Manassas, VA). HT-29 was obtained from DSMZ (Braunschweig, Germany). Caco-2 and A549 were obtained from Cell Line Service (Eppelheim, Germany).

Table 2.

Cell Line Information.

Cell Line Name	Organism/Tissue	Supplier/Order Number
TOV21G	Human/Ovary	ATCC (CRL-11730)
Caco-2	Human/Colon	Cell Line Service (300137)
HT-29	Human/Colon	DSMZ (ACC 299)
A549	Human/Lung	Cell Line Service (300114)

To prepare the cell pellets for staining, the cells were washed in cold PBS and the cell pellet resuspended in 1 ml 4% formaldehyde and incubated overnight at room temperature. The cells were then washed two times in 1 to 2 ml 70% ethanol. Embedding was performed using 200 µl of warmed Histogel (Thermo Scientific) for 1 × 10⁷ cells. Histogel pieces were placed in ethanol for dehydration and then paraffin-embedded.

Human tonsil tissue as well as the human lung tumor tissue used in the study were from Indivumed’s tissue bank (Indivumed GmbH, Hamburg, Germany). All biospecimens in the bank were obtained based on donor’s written informed consent and approved by IRBs or ethic committees.

Western Blot Analysis

Recombinant CXCL12 isoforms α and β at 50 ng each were run on NuPage 4% to 12% Bis-Tris gel (Thermo-Fisher Scientific, Dreieich, Germany), transferred to PVDF (polyvinylidene difluoride) membrane, blocked with 5% milk in TBST (25 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH7.5), probed overnight at 4C with each of the primary antibodies, probed for 2 hr at room temperature with secondary antibody and developed by ECL (Thermo-Fisher Scientific).

The polyclonal LS-B943 was used at a concentration of 1 μg/ml. The monoclonal antibodies D8G6H and K15C were used at concentrations of 0.1 mg/ml and 0.374 mg/ml, respectively. Rabbit anti-mouse IgG HRP (Invitrogen) and donkey antirabbit IgG HRP (Thermo Scientific) were used as secondary antibodies.

Optimization and TMA Staining

The three anti-CXCL12 antibodies were optimized for IHC on the Ventana Discovery XT staining platform (Roche Diagnostics/Ventana Medical Systems). The slides were deparaffinized within the staining instrument and immuno-stained using the Discovery DAB Map Detection Kit (Roche Diagnostics). Multiple antigen retrieval conditions were tested including no retrieval, heat-induced epitope retrieval (HIER) in EDTA (ethylenediaminetetraacetic acid) buffer (Cell Conditioning Solution 1, CC1) or citrate buffer (Cell Conditioning Solution 2, CC2) and Protease 1 (Roche Diagnostics). The optimal antibody dilution and antigen retrieval conditions are listed in Table 1.

The images were generated with the Axio Scan.Z1 automated slide scanner (Zeiss) using ZEN 2 (blue edition) slidescan software (Zeiss), and were prepared in ZEN 2 lite software (Zeiss) at a gamma value of 1.0 after brightness adjustment (white balance).

UMB2 monoclonal antibody was chosen for staining CXCR4 based on the study by Fischer et al.²⁵ CXCR4 IHC staining was performed at a final antibody concentration of 1.5 μg/ml on two ovarian cancer tissue microarrays (TMAs) containing a single core from each of more than 130 cases representing multiple histological subtypes and stages. CXCL12 IHC staining was performed on the adjacent section of the same ovarian cancer TMAs using the chosen rabbit monoclonal anti-CXCL12/SDF1 clone D8G6H (Cell Signaling Technology, Danvers, MA).

CXCL12 and CXCR4 staining was scored manually by providing the average intensity and percentage of positive tumor cells staining with each marker. In addition, the inflammatory cells, vessels, and stroma were scored as positive or negative for each core for both CXCL12 and CXCR4.

Gene Expression Analysis by qPCR

RNA purification, including a DNA removal step, was performed with the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to supplier’s instructions. RNA purity and concentration were measured with the NanoDrop 2000 spectrophotometer (Thermo-Fisher Scientific, Dreieich, Germany) and RNA quality was determined by using the Agilent 2100 bioanalyzer (Agilent Technologies, Berlin, Germany).

The cDNA synthesis from purified RNA was performed using the QuantiTect Reverse Transcription Kit (Qiagen, Hilden, Germany) according to supplier’s instructions. Reverse transcription reaction as well as non-reverse transcriptase controls (NRTC) of each cell line were generated. Each sample contained 1 μg of template RNA in a reaction volume of 20 μl. Generated cDNA was 10-fold diluted before performing gene expression analysis.

Quantitative Real-time PCR (qPCR) was performed using the 2× SsoAdvanced Universal SYBR Green Supermix as well as the CXCL12 (qHsaCID0012398) and GAPDH (qHsaCED0038674) PrimePCR SYBR Green Assays (both Bio-Rad, Munich, Germany) according to supplier’s instructions (Table 3). Primers are shown in Table 3; qPCR was carried out in a reaction volume of 20 µl with a concentration corresponding to cDNA generated from 10 ng of RNA. Samples as well as controls were measured in triplicates. The thermal cycling protocol was performed as described previously.²⁶ The ΔΔCq method was applied for quantification of respective mRNA expression of CXCL12. Gene expression data were displayed using GraphPad Prism 7.0 (GraphPad Software Inc., San Diego, CA).

Table 3.

Bio-Rad Primer Information.

PrimePCR SYBR Green Assay	Gene Name	Unique Assay ID	Amplicon Context Sequence
CXCL12 (human)	chemokine (C-X-C motif) ligand 12	qHsaCID0012398	TCCACTTTAGCTTCGGGTCAATGCACACTTGTCTGTTGTTGTTCTTCAGCCGGGCTACAATCTGAAGGGCACAGTTTGGAGTGTTGAGAATTTTGAGATGCTTGACGTTGGCTCTGGCAACATG
GAPDH (human)	glyceraldehyde-3-phosphate dehydrogenase	qHsaCED0038674	GTATGACAACGAATTTGGCTACAGCAACAGGGTGGTGGACCTCATGGCCCACATGGCCTCCAAGGAGTAAGACCCCTGGACCACCAGCCCCAGCAAGAGCACAAGAGGAAGAGAGAGACCCTCACTGCTGGGGAGTCCCTGCCACAC

Results

Rabbit Polyclonal Antibody LS-B943 Lacks Specificity for CXCL12

Three anti-CXCL12 antibodies were selected for characterization and optimization for detecting CXCL12 protein by IHC, a rabbit polyclonal antibody LS-B943, and two monoclonal antibody clones D8G6H and K15C (Table 1). The reactivity of the three antibodies toward recombinant CXCL12 was evaluated by Western Blot analysis using the two most predominantly expressed isoforms of the chemokine, CXCL12α and CXCL12β. The rabbit polyclonal LS-B943 reacted with the isoform CXCL12β. The monoclonal antibody clone D8G6H reacted strongly with the isoform CXCL12β. The monoclonal antibody clone K15C detects CXCL12β as well, although weaker than clone D8G6H. Both clones reacted very weakly with isoform CXCL12α (Fig. 1).

Figure 1.

Western Blot analysis of recombinant CXCL12 isoforms α (lanes 1, 3, and 5) and β (lanes 2, 4, and 6) using the polyclonal antibody LS-B943 (lane 1 and 2) and the monoclonal antibody clones D8G6H (lane 3 and 4) and K15C (lane 5 and 6).

CXCL12 is expressed in lymphoid tissue; therefore, tonsil was used as a positive control for preliminary IHC analysis. To obtain optimal staining results, different HIER procedures were performed using all tested anti-CXCL12 antibodies. Two types of pretreatment resulted in no anti-CXCL12 staining, which includes no epitope retrieval (Fig. 2A, E, and I) staining after protease treatment (Fig. 2D, H, and L). Only very weak signals were observed when staining were done after HIER in citrate buffer (Fig. 2C, G, and K). The optimal epitope retrieval condition is determined at pH 8.5 in EDTA buffer for all three antibodies (Fig. 2 and Table 1).

Figure 2.

Anti-CXCL12 staining of tonsil tissue using the polyclonal antibody LS-B943 (A–D) and the monoclonal antibody clones D8G6H (E–H) and K15C (I–L). A, E, and I: Staining without epitope retrieval. B, F, and J: Staining after HIER in EDTA buffer. C, G, and K: Staining after HIER in citrate buffer. D, H, and L: Staining after protease treatment. B and C: Tonsil tissue showed weak to moderate staining of the whole section after HIER in EDTA (B) and citrate buffer (C), respectively. F: Tonsil tissue showed moderate to strong staining of various cell types of the parafollicular zone (arrow) after HIER in EDTA buffer. G: After HIER in citrate buffer tonsil tissue showed very weak staining (arrow). J: Tonsil tissue showed membranous (arrow) and cytoplasmic (arrowhead) staining of lymphoid cells in the interfollicular zone after HIER in EDTA buffer. The germline center of a follicle is marked with an asterisk. K: After HIER in citrate buffer tonsil tissue showed very weak cytoplasmic staining of lymphoid cells (arrow). Abbreviations: HIER, heat-induced epitope retrieval; EDTA, ethylenediaminetetraacetic acid.

The rabbit polyclonal antibody LS-B943 showed a moderate to strong staining of the entire tonsil tissue section after HIER in EDTA buffer (Fig. 2B). Since the staining of antibody LS-B943 was nonspecific, this polyclonal antibody was not investigated further.

The monoclonal antibody clone D8G6H showed anti-CXCL12 staining of various cell types of the parafollicular zone in tonsil tissue as well as of vascular endothelial cells and squamous epithelial cells after HIER in EDTA buffer (Fig. 2F). In contrast, K15C showed mainly cytoplasmic and membranous staining of lymphoid cells in the interfollicular zone of tonsil tissue (Fig. 2J).

Staining conditions were optimized for monoclonal antibody clones D8G6H and K15C. The best staining results were achieved with DCS antibody diluent (DCS Diagnostics), and an antibody incubation time of 60 min at room temperature for both clones. An optimum antibody concentration was determined at 0.45 μg/ml for D8G6H and 20 μg/ml for K15C (Table 1).

Intriguingly though both K15C and D8G6H showed apparent specific staining, they gave opposite staining patterns with K15C staining the interfollicular zone and with D8G6H staining the parafollicular zone, therefore additional tests were conducted for further characterization.

As the CXCL12/CXCR4 axis has been shown to be involved in the biology of lung cancer progression, two cancer tissues, squamous cell lung carcinoma and lung adenocarcinoma, were stained with either D8G6H or K15C. The clone D8G6H stained vascular endothelial cells in both tumor tissue samples, compatible with the role of CXCL12 in vascularization. Staining of stromal and inflammatory cells was also noted using clone D8G6H. By contrast, membranous staining of the tumor cells was observed in both tissues using clone K15C (Fig. 3), whereas no staining of vascular endothelial cells was observed. As with the staining of tonsil tissue, apparent specific but different staining patterns were observed with the two monoclonal antibodies in the tumor tissue samples.

Figure 3.

Anti-CXCL12 IHC of human FFPE squamous cell carcinoma (A, B) and adenocarcinoma (C, D) of the lung using clone D8G6H (A, C) and clone K15C (B, D). A: Anti- CXCL12 staining was detected in vascular endothelial cells of squamous cell carcinoma tissue (arrow), stromal cells (arrowhead), and inflammatory cells (open arrowhead) using clone D8G6H. B: Membranous anti-CXCL12 staining was detected in tumor cells of squamous cell carcinoma tissue (arrow) using clone K15C. C: Anti-CXCL12 staining was detected in vascular endothelial cells of adenocarcinoma tissue (arrow) using clone D8G6H. D: Membranous anti-CXCL12 staining was detected in tumor cells of adenocarcinoma tissue (open arrow) using clone K15C. Abbreviations: IHC, Immunohistochemistry; FFPE, Formalin-fixed paraffin-embedded.

Identification of a CXCL12 High Expressing Cell Line

To elucidate which of the two monoclonal antibodies is able to detect specifically CXCL12, a panel of cell lines including Caco-2, HT-29, A549, and TOV21G, with differing reported expression of CXCL12, was analyzed for CXCL12 expression by qPCR with the aim of identifying cell lines that could be used as positive and negative controls. The cell lines were chosen based on literature reports. CXCL12 was detected at such a low level in the cell line A549, it was not quantifiable. The cell line HT-29 did not show detectable CXCL12 expression. The A549 cell line is reported to be negative for CXCL12 expression,²⁷ whereas, HT-29 cells have been reported in the literature to be positive for CXCL12 both by RT-PCR for mRNA and protein by Western Blot analysis.²⁸ Conversely, we detected CXCL12 mRNA in both the ovarian cancer cell line TOV21G and the colon epithelial cell line Caco-2. CXCL12 expression in TOV21G cells was 5-fold higher than the level in Caco-2 cells (Fig. 4). The Caco-2 cell line has been reported to express CXCL12 mRNA by RT-PCR²⁸ and the TOV21G cell line is reported to produce CXCL12 as measured by ELISA assay.²⁹ These data confirm that TOV21G expresses high levels of CXCL12, and thus can be utilized as a positive control for antibody validation.

Figure 4.

Gene expression of CXCL12 in TOV21G, Caco-2, A549, and HT-29 cells. Expression levels of CXCL12 were analyzed by qPCR and normalized to GAPDH expression. CXCL12 expression changes in cell lines were depicted as fold change normalized to TOV21G. Results are shown as scatter plots including mean of triplicates with SD. Abbreviations: qPCR, Quantitative Real-time PCR; n.q., not quantifiable; n.d., not detectable.

D8G6H Specifically Stains CXCL12

Formalin-fixed paraffin-embedded (FFPE) TOV21G cell pellets were used to stain for CXCL12 protein with either clones D8G6H or K15C. Strong cytoplasmic staining of TOV21G cells was observed using clone D8G6H. On the contrary, no staining of TOV21G cells was observed with K15C. These data indicate that clone D8G6H specifically detects CXCL12 (Fig. 5).

Figure 5.

Anti-CXCL12 IHC of human FFPE TOV21G cells using clone D8G6H (A) and clone K15C (B). A: Cytoplasmic anti-CXCL12 staining was detected in TOV21G cells using clone D8G6H (arrow). B: No anti-CXCL12 staining was detected in TOV21G cells using clone K15C. Abbreviations: IHC, Immunohistochemistry; FFPE, Formalin-fixed paraffin-embedded.

The expression of CXCL12 has been reported to be a prognostic factor related to poor survival in ovarian cancer.^30,31 We therefore examined two ovarian cancer TMAs for CXCL12 expression using clone D8G6H and compared with CXCR4 expression by immunohistochemical staining using the anti-CXCR4 antibody clone UMB2.²⁵ A total of 132 subjects were represented with one 1.5 mm core per tissue sample on two TMAs. The TMAs were composed of a broad array of ovarian cancer subtypes (including adenocarcinomas, clear cell carcinomas, and squamous cell carcinomas) and stages. It was found that 87% of the tumor cells express CXCR4, and only 16% express CXCL12 (Tables 4 and 5). Expression of CXCL12 was frequently found in stromal cells of the TME. Figure 6A shows the representative images of CXCR4 expression on tumor cells and CXCL12 expression on stromal cells. CXCL12 expression was observed on cells comprising the TME including 10% of inflammatory cells, 27% of stromal cells, and 40% on tumor blood vessels (Table 4). The tumors showing a high CXCR4 expression tend to have high CXCL12 expression in the TME (Fig. 6B).

Table 4.

CXCL12 Expression in Ovarian TMA.

Cancer Stage	Frequency of CXCL12-Positivea Tumor Cells	Frequency of CXCL12-Positive Inflammatory Cells	Frequency of CXCL12-Positive Blood Vessels	Frequency of CXCL12-Positive Stromal Cells
All stages	16/132 (12%)	11/112 (10%)	53/132 (40%)	36/132 (27%)
Stage I	5/56 (9%)	4/43 (9%)	26/56 (46%)	16/56 (29%)
Stage II	7/24 (29%)	4/22 (18%)	7/24 (29%)	9/24 (38%)
Stage III	3/25 (12%)	1/21 (5%)	11/25 (44%)	8/25 (32%)
Stage IV	1/22 (5%)	1/21 (5%)	8/22 (36%)	2/22 (9%)
Normal	NA	0/1 (0%)	10/19 (53%)	4/19 (21%)

Abbreviation: TMA, tissue microarray.

^aAt least 5% of cells are positive.

Table 5.

CXCR4 Expression in Ovarian TMA.

Cancer Stage	Frequency of CXCR4-Positivea Tumor Cells	Frequency of CXCR4-Positive Inflammatory Cells	Frequency of CXCR4-Positive Blood Vessels	Frequency of CXCR4-Positive Stromal Cells
All stages	114/131 (87%)	50/112 (45%)	75/131 (57%)	18/132 (14%)
Stage I	47/56 (84%)	19/43 (44%)	29/56 (52%)	10/57 (18%)
Stage II	24/24 (100%)	9/22 (41%)	17/24 (71%)	1/24 (4%)
Stage III	21/24 (88%)	12/21 (57%)	16/24 (67%)	5/24 (21%)
Stage IV	17/22 (77%)	8/21 (38%)	11/22 (50%)	1/22 (5%)
Normal	NA	1/1 (100%)	9/19 (47%)	15/19 (79%)

Abbreviation: TMA, tissue microarray.

^aAt least 5% of cells are positive.

Figure 6.

IHC of a human FFPE ovarian TMA using clone UMB2 for the detection of CXCR4 and clone D8G6H for the detection of CXCL12. A: CXCR4 expression was detected in tumor cells of ovarian carcinoma tissue. B: CXCL12 expression was detected in stromal cells of the same tissue sample. C: Comparison of CXCL12 expression frequency between CXCR4 high tumor specimen and CXCR4 low tumor specimen in all three compartments, inflammatory cells, vessels, and stroma. Abbreviations: IHC, Immunohistochemistry; FFPE, Formalin-fixed paraffin-embedded: TMA, tissue microarray.

Discussion

A major acknowledged challenge in cancer research is the lack of reproducibility of published findings, particularly in target identification and validation, and clinical pathology.³² It is now apparent that one of the causes of this challenge is the lack of characterization of antibodies used for identification of target proteins.²² Antibodies are used in multiple bioanalytical techniques including ELISA, Western blot, and IHC. An antibody may be useful and validated for one methodology but may not be applicable to another. IHC is a valuable technique for identifying target proteins and localization in vivo, both in animal models of disease and in the clinical setting. Frequently there is limited information available on the method development and characterization of commercially available antibodies, particularly for IHC. Frequently the only reported control for IHC is an isotype control. It is therefore essential to rigorously characterize antibodies for IHC. This issue has been highlighted for the membrane G-protein-coupled receptor proteins.³³ One example of this has been the identification and staining of CXCR4, the receptor for CXCL12, in cancer tissue. CXCR4 is a member of the chemokine receptor family and is a G-protein-coupled receptor. Many of the antibodies developed against CXCR4 have been raised against cells expressing CXCR4 rather than against specific epitopes. Furthermore, many of the staining patterns show nuclear staining of CXCR4 even though it is known that CXCR4 is a transmembrane protein. This prompted the development and extensive characterization of a new CXCR4 antibody for IHC. Staining patterns obtained with this antibody were more consistent with the known membranous localization of CXCR4.²⁵

Given the importance of the CXCL12/CXCR4 axis in tumor progression, we sought to characterize the expression of CXCL12 in tumor tissue. However, a major challenge was the choice of a suitable antibody given both the variety of commercially available antibodies, and the known issues with antibody specificity for IHC described above. We therefore decided it was necessary to rigorously characterize a suitable antibody for CXCL12 IHC. We chose three antibodies for this evaluation based on their use in published papers, commercial availability, and vendor evidence for application in IHC: a rabbit polyclonal antibody LS-B943 and two monoclonal antibody clones D8G6H and K15C. A multistep approach was adopted for antibody characterization: (1) recognition of CXCL12 by Western blot; (2) preliminary staining of tonsil, a positive control tissue; (3) staining of selected tumor tissues; and (4) verification of specificity for CXCL12 using cell lines with demonstrated CXCL12 expression including a high and low expresser. Finally, the utility of the selected antibody was demonstrated in ovarian cancer where there is a known link between CXCL12/CXCR4 axis and tumor progression.

All three antibodies reacted with CXCL12 isoforms in the Western blot analysis. Tonsil tissue was chosen for preliminary comparison of the three antibodies as it is a mucosa-associated, secondary lymphoid tissue and therefore rich in chemokines and chemokine receptors associated with lymphocyte development including CXCL12 and CXCR4. Interestingly, in preliminary IHC, evaluation of the rabbit polyclonal antibody LS-B943 showed extensive nonspecific staining of tonsil tissue and was thus not investigated further.

The remaining antibodies were further evaluated for staining of tumor tissue. Both monoclonal antibodies, D8G6H and K15C gave apparent specific staining patterns that could be justified by proposed mechanisms of CXCL12 in tumor biology. However, converse staining patterns were observed in samples of lung cancer tissue with clone K15C showing tumor cell staining in contrast with clone D8G6H showing CXCL12 staining of vascular endothelial cells. To resolve this contradiction, we identified a cell line, TOV21G, that expresses high levels of CXCL12 confirmed by qPCR analysis as a positive control to ascertain if either antibody was able to specifically detect CXCL12 protein. Clone D8G6H detected CXCL12 in TOV21G in IHC analysis in contrast to clone K15C showing no staining in TOV21G cells. D8G6H was then chosen to assess CXCL12 in two panels of ovarian cancer TMA. We observed significant staining of CXCL12 in the ovarian TME using clone D8G6H. Staining correlated with high expression of CXCR4 on ovarian tumor cells consistent with the reciprocal relationship of the CXCR4 receptor with its ligand CXCL12.

Collectively these data emphasize the necessity for adequate antibody characterization for IHC application to achieve reliable and reproducible in situ protein expression information. In particular, the study highlighted the necessity to confirm specificity of an antibody for the target protein to be investigated. Western blot is commonly used, but as our experience has demonstrated, it can be limited by sensitivity of detection. This study illustrates the use of cell lines with confirmed differential expression of the target protein as a valuable tool to investigate specificity. Optimization of staining methodology and conformation of specific staining in a number of appropriate tissues are further recommended.

In addition, by using the chosen antibody for IHC after the multistep characterization, these data demonstrate and confirm the expression of CXCL12 in ovarian cancer, supporting a role for the CXCR4/CXCL12 axis in tumor biology and progression and its potential as a therapeutic target.