Comparative oncology: ErbB-1 and ErbB-2 homologues in canine cancer are susceptible to cetuximab and trastuzumab targeting.

To facilitate comparative oncology trials we compared the biological and molecular homologies of canine (dog; Canis lupus familiaris) and human tumor-associated antigens ErbB-1 and -2. Further, we investigated whether they could serve as targets for anti-ErbB-1 (cetuximab) and anti-ErbB-2 antibodies (trastuzumab), which are highly relevant in human clinical oncology. Immunohistochemistry of canine mammary cancer showed ErbB-1 overexpression in 3/10 patients and ErbB-2 in 4/10. We report 91% amino acid homology for ErbB-1 and 92% for ErbB-2 between canine and human molecules. Modeling of canine on human ErbB-1 revealed that the cetuximab epitope only differs by 4 amino acids: Lys443 is replaced by Arg, Ser468 by Asn, Gly471 by Asp, and Asn473 by Lys in canines. The trastuzumab binding site is identical in human and canine ErbB-2 apart from a single amino acid change (Pro557 to Ser). Binding of cetuximab and trastuzumab to canine mammary carcinoma cells CF33, CF41, Sh1b and P114 was confirmed by flow cytometry. Both antibodies significantly inhibited canine tumor cell proliferation partly due to growth arrest in G(0)/G(1) phase. We explain the lower efficiency on the tested canine than on human SKBR3 and A431 cells, by a 2-log lower expression level of the canine ErbB-1 and -2 molecules. Our results indicate significant homology of human and canine Erb-1 and -2 tumor associated antigens. The fact that the canine homologues express the cetuximab and trastuzumab epitopes may facilitate antibody-based immunotherapy in dogs. Importantly, the striking similarities of ErbB-1 and -2 molecules open up avenues towards comparative strategies for targeted drug development. Copyright Â

Introduction

In human medicine antibodies against tumor associated antigens are applied for passive immunotherapy of cancer. Illustrative examples are trastuzumab (Herceptin®; Genentech, South San Francisco, CA, USA), a humanized IgG1 antibody which is clinically applied for the treatment of metastatic breast cancers overexpressing HER-2 (ErbB-2, Her2/neu) (Garnock-Jones et al., 2010), or cetuximab (Erbitux®, Merck, Darmstadt, Germany), a chimeric IgG1 antibody applied for the treatment of EGFR (ErbB-1) overexpressing metastatic colon carcinomas (Banerjee and Flores-Rozas, 2010), regionally advanced head and neck squamous carcinomas and other tumor types (Vincenzi et al., 2010). The overexpression of ErbB-1 and -2 antigens in human malignancies is associated to each other and leads to heterodimer formation (Citri et al., 2004). Their expression is indirectly correlated with hormone receptor levels, and directly with higher proliferation, genomic instability and poorer overall prognosis (Rimawi et al., 2010), making ErbB-2 expression a prognostic or possibly predictive factor (Ferretti et al., 2007). Both cetuximab and trastuzumab directly affect cellular proliferation of cancer cells: either by interfering with ligand binding (cetuximab), structure (Li et al., 2005) and heterodimerization of these membrane molecules (Patel et al., 2009), thereby inhibiting vital growth and survival signals (Lurje and Lenz, 2009); and possibly by affecting their internalization and degradation (trastuzumab) (Cuello et al., 2001; Gennari et al., 2004). In addition, effector functions of trastuzumab (Gennari et al., 2004; Clynes et al., 2000) or cetuximab (Kurai et al., 2007) are determined by their binding to Fc receptors on various immune effector cells, such as NK cells, monocytes, macrophages and granulocytes, which induce antibody-mediated cytotoxicity, phagocytosis, apoptosis or necrosis of the targeted tumor cells.

The understanding that companion dogs (Canis lupus familiaris) also develop similar tumors to humans initiated the concept of “comparative oncology”, which aims to simultaneously speed up the developments of anti-cancer therapies in human and veterinarian medicine. Like in humans, ageing is a contributing factor in the development of mammary cancer in dogs, as are nulliparity and inheritance (Mulligan, 1975), especially in purebreds (Vascellari et al., 2009). Moreover, dogs live under similar environmental conditions as pet owners including pollution or nutritional aspects which contribute to epigenetic risks (Owen, 1979; Perez Alenza et al., 2000). Therefore, it has been realized and accepted that clinical trials in dogs may bear close resemblance to clinical scenarios in human medicine (Paoloni and Khanna, 2008; Gordon et al., 2009).

Analyzing an Italian pet registry, the highest incidence rates of cancer were those of mammary cancer and for non-Hodgkin's lymphoma (Vascellari et al., 2009). Interestingly, expression of ErbB-1 as well as ErbB-2 homologues have been described in canine cancer cells including mammary carcinomas (Ahern et al., 1996; Dutra et al., 2004; Gama et al., 2009; Peruzzi et al., 2010), but the expression pattern, exact molecular structure and biological function of canine ErbB-1 and -2 remain to be resolved. Therefore, in this study we examined the differences and similarities between the human and canine ErbB-1 and -2 expression and biology. To the best of our knowledge, the question whether the canine molecules might be candidate targets for passive immunotherapy using cetuximab or trastuzumab specificities has not been investigated to-date. Indeed, the results of our study indicate that canine ErbB-1 and ErbB-2 homologues are recognized by trastuzumab and cetuximab antibodies, with targeting leading to growth inhibition in canine neoplastic mammary cell lines. This knowledge supports the development of novel immunotherapies and adjuvant strategies for the treatment of ErbB-expressing canine cancers but also for human patients in terms of comparative proof-of-concept (POC) studies in canine cancer clinical trials.

Materials and methods

Immunohistochemistry of canine ErbB-1 and ErbB-2 and quantitative analysis by Tissue FAXs

Serial 4-μm sections were cut from canine mammary tumor samples previously fixed in buffered 3.9% formalin and paraffin embedded. Paraffin was removed with xylene from sections and immunohistochemistry was performed using the FDA-approved in vitro diagnostic EGFR pharm Dx™ (DAKO, Glostrup, Denmark) and HercepTest® (DAKO, Glostrup, Denmark) (Dutra et al., 2004) kits according to the manufacturer's instructions. Specimen labeling assessments were conducted using TissueFAXS microscopy system (TissueGnostics, Vienna, Austria) and ErbB-1 and ErbB-2 positive cells in the mammary carcinoma tissues were analyzed with the HistoQuest® software module (TissueGnostics, Vienna, Austria), a cell-based staining intensity analysis of immunohistochemical experiments (Wallmann et al., 2010). Evaluation was performed according to the HercepTest® classification (Reis-Filho et al., 2005).

Homology alignments of human and canine ErbB-1 and ErbB-2 molecules

Amino acid sequences of human and canine ErbB molecules were searched in the NCBI protein (http://www.ncbi.nlm.nih.gov/protein) or UniProtKB/Swiss-Prot database (http://www.ebi.ac.uk/uniprot). Human ErbB-1 (accession no.: P00533) was aligned with canine ErbB-1 (XP_533073.2), human ErbB-2 (P04626) with canine ErbB-2 (NP_001003217) using BLAST programme (Basic Local Alignment Tool; http://blast.ncbi.nlm.nih.gov/Blast.cgi) (Altschul et al., 1990; Gish and States, 1993).

Modeling of human and canine ErbB-1 and ErbB-2 molecule structures with cetuximab and trastuzumab binding sites

The models of canine ErbB-1 and ErbB-2 were based on the human crystal structures (EGFR:PDB ID:1YY9 (Li et al., 2005), resolution: 2.61 Å and HER2:PDB ID:1N8Z (Cho et al., 2003), resolution: 2.52 Å). Modeling was carried out with MODELLER (version 9v8) (Sali and Blundell, 1993; Marti-Renom et al., 2000) using the automodel protocol with the refinement set to “very_slow”. Two hundred models were generated. Modeling of the 10 amino acid long loop at the C-terminus of the HER-2 (G602-P611) was not attempted as no electron density was observed in the template human HER-2 crystal structure and therefore, no template was available for this loop. Model quality was assessed using DOPE (Shen and Sali, 2006) and GA341 scores (Melo and Sali, 2007) ProQ and ProCheck (Laskowski et al., 1993). The models with the best DOPE score were selected and fitted to the structure of the human template for visualization. The FAB fragments of the same crystal template structure were retained and joined with the ErbB models, allowing in both cases to predict the structure of the complex between the ErbB and the antibody, as the interfaces are almost identical.

Cell lines and monoclonal antibodies

The canine mammary carcinoma cell lines designated CF33 and CF41 were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA. Cat. No.: CRL-6227 for CF33 and CRL-6232 for CF41) and cultivated according to the manufacturer's instructions in DMEM medium supplemented with 10% FCS, 2 mM l-glutamine, penicillin (100 U/ml) and streptomycin (100 μg/ml). The canine mammary carcinoma cell lines Sh1b and P114 were a kind gift of Dr. Gerard Rutteman (Department of Clinic Science and Companion Animals, University of Utrecht, The Netherlands) and were maintained in DMEM/F12 supplemented with 10% FCS, 2 mM l-glutamine and 10 μg/ml gentamicin sulfate. All cells were maintained at 37 °C in a humidified atmosphere of 5% CO2. Cetuximab (Erbitux®), a chimeric IgG1 anti-ErbB-1 (EGFR) monoclonal antibody, was obtained from Merck KGaA (Darmstadt, Germany), and trastuzumab (Herceptin®), a humanized IgG1 monoclonal anti-ErbB-2 (HER-2) antibody, was from Roche (Hertfordshire, United Kingdom). Rituximab (MabThera®), a chimeric IgG1 anti-CD20 monoclonal antibody, employed as isotype control, was purchased from Roche (Hertfordshire, United Kingdom). The murine precursor of trastuzumab, antibody 4D5, was kindly provided by Genentech (South San Francisco, CA, USA) and 225 antibody, the murine precursor of cetuximab, was bought from Sigma–Aldrich (St. Louis, MO, USA).

Flow cytometric assessments of antibody binding to ErbB-receptors on cells

For flow cytometric assessments of human antibody binding to the tumor-associated antigens ErbB-1 and ErbB-2 on canine tumor cells, cells were incubated with 200 μl of 10 μg/ml cetuximab or trastuzumab, respectively, for 30 min at 4 °C, followed by two washes in FACS buffer (PBS, 5% normal goat serum) and incubation by anti-human IgG FITC antibodies (200 μl of 10 μg/ml) (DAKO, Glostrup, Denmark) for 30 min at 4 °C. After washing with FACS buffer, analysis was performed on a dual laser FACSCalibur™ (BD Biosciences, Franklin Lakes, NJ, USA).

Cell viability assay

Tumor cell viability was assessed by means of the EZ4U® assay (EZ4U® the 4th Generation non radioactive cell proliferation & cytotoxicity Assay kit, Biomedica, Vienna, Austria). Cells were seeded in 96-well plates at 3 × 104 cells per well and allowed to adhere overnight under standard culture conditions prior to assays. Cells were exposed to 5 μg/ml trastuzumab or 5 μg/ml cetuximab antibodies over a period of 24 and 48 h. Control groups received media alone, 5 μg/ml human IgG1 isotype control or 0.9% Triton-X-100 (to account for total cell death), for 10 min prior to addition of tetrazolium compound. Following treatments, tetrazolium solution, prepared according to the manufacturer's instructions, was added at 20 μl per well and cells were incubated for a further 1 h prior to recording absorbance at 450 nm with 620 nm as a reference using a 96-well plate reader. The quantity of formazan product as measured by the amount of 450 nm absorbance is directly proportional to the number of living cells in culture.

Cell cycle analysis

The effects of cetuximab and trastuzumab on the cell cycle of canine mammary carcinoma cells were determined by flow cytometric analysis. Cells were seeded in 6-well plates at 2 × 105 cells per well and allowed to adhere overnight under standard culture conditions prior to assays. Cetuximab and trastuzumab antibodies were added to the respective wells at a concentration of 5 μg/ml. After 48 h of incubation, cells were harvested and stained with the CycleTEST™ PLUS DNA Reagent Kit (BD Biosciences, Franklin Lakes, NJ, USA) according to the manufacturer's instructions and analyzed on a dual laser FACSCalibur™ (BD Biosciences, Franklin Lakes, NJ, USA).

Quantitative analysis of indirect immunofluorescence staining in flow cytometry

For quantitative assessment of ErbB receptor surface expression on human and canine tumor cells, cells were incubated with 200 μl of 10 μg/ml suspension of the murine monoclonal anti-human ErbB-1 receptor antibody 225 or the murine anti-human ErbB-2 antibody 4D5, respectively (kindly provided by Genentech, South San Francisco, CA, USA). Quantitative measurement of bound anti-ErbB antibodies was achieved using the flow cytometry based QIFIKIT® assay (DAKO, Glostrup, Denmark) according to the manufacturer's instructions. To account for nonspecific background fluorescence, cells were also incubated with an control murine IgG1 antibody (Abcam, Cambridge, United Kingdom) and background fluorescence intensity was substracted from specific signals.

Data handling and statistical analysis

For histological analysis of ErbB-1 and ErbB-2 expression, 10 samples of canine mammary cancer patients were analyzed. Flow cytometric experiments of receptor binding were repeated at least three times, and histograms display one representative example. In vitro cell growth inhibition assays were performed with n = 9, statistical analyses of assays were performed by means of the Mann–Whitney U test and significance was accepted at p < 0.05 (*), p < 0.01 (**), and p < 0.001 (***).

Results

Tissue FAXS-based profiling of ErbB-1 and ErbB-2 in canine mammary cancer

ErbB-2 overexpression in human cancer is uniform, in both primary and secondary lesions (Paik et al., 1990), and it has been shown to correlate with ErbB-1 rather than with estrogen receptor expression (Rimawi et al., 2010). To elucidate the situation in dogs, formalin-fixed paraffin-embedded tissue samples of canine mammary carcinoma were analyzed immunohistochemically using tests routinely applied in human medicine to evaluate ErbB expression (Fig. 1). Histological staining was recorded via TissueFAXS microscopy system and analyzed using the HistoQuest® software module. The results of single cell counts are summarized in Table 1 and demonstrate strong ErbB-1 expression in 3 out of the 10 dog patients’ mammary cancer samples and strong ErbB-2 expression in 4 samples out of 10 canine mammary cancer samples. Only strong membrane-specific immunohistochemical staining was considered positive for either ErbB molecule in line with the HercepTest® classification (0 = no membrane staining or <10% of cells stained; 1+ = incomplete membrane staining in >10% of cells; 2+ = >10% of cells with weak to moderate complete membrane staining; and 3+ = strong and complete membrane staining in >10% of cells) (Reis-Filho et al., 2005). All three ErbB-1 positive samples ranged from 11.8 to 36.4% for complete membrane staining indicating “3+” in HercepTest® classification. With respect to ErbB-2 expression, 4 positive samples could be detected, ranging between 10.1% and 12.2% complete membrane-stained cells. Furthermore, 2 samples could be identified with moderate ErbB-2 membrane staining in 6.2 and 7.6% of cells, respectively, meaning “2+” classification.

Fig. 1

Qualitative and quantitative assessment of ErbB receptor expression in representative tissue samples of canine mammary carcinoma patients using the diagnostic tests EGFR pharm Dx™ and HercepTest®, respectively. (a and b) Immunohistochemistry shows a pronounced membranous expression of ErbB-1 antigen in tumor lesions of canine patient No. 9. (c) Respective scattergram by tissue FAXs; x-axis: counted cells according to the staining of their nuclei; y-axis: staining intensity of cells positive for ErbB-1 antigen. Patient No. 9 has 11.81% of cells in tumor lesions positive for ErbB-1 antigen. This corresponds to a human “+++” classification. (d and e) Immunohistochemistry shows a pronounced membranous expression of ErbB-2 antigen in tumor lesions of canine patient No. 2. (f) Respective scattergram by tissue FAXs; x-axis: counted cells according to the staining of their nuclei; y-axis: staining intensity of cells positive for ErbB-2 antigen. For canine patient No. 2 12.25% of cells in tumor lesions have tested positive for ErbB-2 antigen. This corresponds to a human “+++” classification.

Table 1

Quantitative assessment of ErbB receptor expression on cells of canine mammary carcinoma lesions.

(a) 3 of 10 analyzed samples show in more than 10% of cells strong membrane specific staining for ErbB-1, corresponding to a “+++” grading in human HercepTest®-classification
Pat #	No of analyzed cells	ErbB-1 positive cells (%)	Classification
1	23,620	19.43	+++
2	41,649	36.41	+++
3	77,811	0.18	−
4	76,313	2.42	−
5	109,155	0.08	−
6	7178	0.89	−
7	9217	0.19	−
8	125,643	0.55	−
9	218,182	11.81	+++
10	67,373	3.26	−

Homology alignments, modeling and epitope analysis of human and canine ErbB-1 and ErbB-2 molecules

The BLAST alignments of canine ErbB-1 and ErbB-2 revealed aa sequence homologies of 91% and 92% respectively to the equivalent human homologues, and identities of 95% for both molecules with the human counterparts. Models of canine ErbB-1 and ErbB-2 were based on the human crystal structures (EGFR:PDB ID:1YY9 (Li et al., 2005) and HER-2:PDB ID:1N8Z (Cho et al., 2003) and carried out with MODELLER (version 9v8) (Sali and Blundell, 1993; Marti-Renom et al., 2000)). Model quality was assessed using the DOPE (Shen and Sali, 2006) and GA341 scores (Melo and Sali, 2007) of MODELLER, ProQ (Wallner and Elofsson, 2003) and ProCheck (Laskowski et al., 1993). The best scoring models were selected and fitted to the structure of the human template for visualization (Table 2).

Table 2

Quality of models shown in Fig. 2 as assessed by Modeller, ProQ and ProCheck.

	ErbB-1	ErbB-2
Modeller
DOPE	−62,008	−63,337
GA341 score	1.0	1.0
ProQ
ProQ LGscore	4.966	4.809
ProCheck
Most favorable	88.9%	89.6%
Additionally allowed	10.2%	9.4%
Generously allowed	0.8%	0.6%
Disallowed	0.2%	0.4%

The MODELLER DOPE-score is a statistical potential energy-score designed to recognize correctly folded protein. The more negative the score, the higher is the probability of a correctly folded model. The GA341 score is a statistical potential for correct local model geometry. The score ranges from 0 to 1, where 1 indicates a correctly folded model. A ProQ LG score above 1.5 indicates a correct, above 2.5 of a good and above 4.0 of a very good model. ProCheck analyses local model quality, the Ramachandran plot classifies backbone dihedral angles in favorable, allowed, generously allowed and disallowed regions.

As shown in Fig. 2a, the canine ErbB-1 molecule shows indeed a great structural homology to the human counterpart. Even more importantly, the cetuximab binding site of human ErbB-1 is highly conserved in the canine molecule, as can be seen in Fig. 2b, where there is a difference of only 4 amino acids between the two proteins. Furthermore, the human lysine of position 443 which turns into an arginine in the canine molecule, and the human serine of position 468 which is replaced by asparagine are both conservative amino acid changes maintaining the character of amino acids (basic or neutral, respectively). The residual two amino acid differences are non-conservative, as a glycine in the human molecule is substituted with aspartic acid (471) and asparagine is replaced by lysine (473) in the canine molecule. In the case of ErbB-2, the canine molecule also revealed a striking structural similarity (Fig. 2c). The trastuzumab binding site on human ErbB-2 is highly conserved in the canine molecule as can be seen in Fig. 2d, where there is only 1 amino acid difference between the two proteins, namely proline 557 of the human homologue, which is replaced by a serine in the canine protein.

Fig. 2

Modeling of human and canine ErbB molecules. (a) Modeling of human (blue) and canine (red) ErbB-1 molecules based on crystal structures in complex with cetuximab Fab (heavy chain shown in lime, light chain in green). (b) The cetuximab binding site: the backbones of human (blue) and canine (red) ErbB-1 molecules are shown as ribbon diagrams with side chains contacting the cetuximab Fab added (heavy chain shown in lime, light chain in green). Amino acids that differ between the human and canine molecule are labeled: in position 443, the human lysine is in the canine molecule arginine, in 468, which is hidden by cetuximab light chain the human serine turns to canine asparagine, in 471 glycine is replaced by aspartic acid and in 473 asparagine turned to lysine. (c) Modeling of human (blue) and canine (red) ErbB-2 molecules based on crystal structures in complex with trastuzumab Fab (heavy chain shown in lime, light chain in green). (d) The trastuzumab binding site: the backbones of human (blue) and canine (red) ErbB-2 molecules are shown as ribbon diagrams with side chains contacting the trastuzumab Fab added (heavy chain shown in lime, light chain in green). The amino acid that differs between the human and canine molecule is labeled: in position 557, human proline is replaced in the canine molecule by serine.

Flow cytometric assessments of antibody binding to ErbB-1 and ErbB-2 receptors

To confirm that the cetuximab and trastuzumab epitopes on canine ErbB-1 and ErbB-2 on tumor cells are recognized by the equivalent antibodies as predicted by the homology alignment results above, we performed flow cytometric measurements of canine mammary carcinoma cell lines Sh1b, P114, CF33 and CF41. Our results demonstrate that all four cell lines tested expressed both ErbB family members. Sh1b, P114 and CF33 showed relatively higher capacity for trastuzumab binding than cetuximab binding. CF41 was the only cell line with relatively higher cetuximab binding. Highest binding of both antibodies could be observed in Sh1b and P114 cells. The IgG1 isotype showed no specific staining in any cell line tested (Fig. 3).

Fig. 3

Flow cytometric analysis of binding of trastuzumab and cetuximab on the surface of canine mammary carcinoma cell lines. All four tested cell lines (CF33 top line, CF41 2nd line, P114 3rd line, Sh1b bottom line) showed expression of both receptors of the ErbB-family. CF41 cells were the only ones expressing more ErbB-1 than ErbB-2, highest ErbB-2 expression could be observed in P114 and Sh1b cells. (Monoclonal antibodies: solid lines; secondary antibody controls: grey histograms.)

Effects of cetuximab and trastuzumab on cell viability and cell cycle of canine mammary carcinoma cells

Canine Sh1b and P114 cells, which have been demonstrated to have the highest levels of cetuximab and trastuzumab binding by flow cytometric assessments, were employed to investigate the functional effect of the antibodies. Fig. 4 shows that no significant difference in cell growth could be measured after 24 h of incubation with either cetuximab or trastuzumab, although a trend towards slower growth could be observed in both cell lines. However, after 48 h of incubation P114 cells showed significant growth inhibition with cetuximab (19.94% inhibition, p = 0.0019) and trastuzumab (31.55%, p = 0.0006) in comparison to the untreated group, and a p-value <0.05 with trastuzumab compared to cells treated with the isotype control antibody. In Sh1b cells the differences in cell growth after 48 h of treatment with either antibody were even higher: cetuximab (p = 0.0106) and trastuzumab (p < 0.0001) both compared to the untreated and trastuzumab (p = 0.0091) compared to the isotype control treated group.

Fig. 4

Cell growth arrest due to growth signal inhibition. Canine mammary carcinoma cell lines with highest levels of cetuximab and trastuzumab binding were evaluated in a tetrazolium based cell viability assay for susceptibility to cetuximab and trastuzumab treatment. No significant difference in cell growth could be measured after 24 h of incubation with either cetuximab or trastuzumab. After 48 h of incubation with cetuximab (19.94% inhibition, p = 0.0019), and trastuzumab (31.55%, p = 0.0006), P114 cells showed significant growth inhibition in comparison to the untreated group, and the trastuzumab compared to the isotype control group (18.51% inhibition, p < 0.05). In Sh1b cells after 48 h of treatment with cetuximab (23.76%, p = 0.0106) or trastuzumab (p < 0.0001) showed highly significant differences to the untreated and the trastuzumab to the isotype control treated groups (p = 0.0091). Data points show mean % cell viability + SD (n = 9 for untreated, cetuximab and trastuzumab treated groups, n = 3 for isotype control treated group).

To further investigate the growth inhibitory effects of cetuximab and trastuzumab antibodies on canine mammary carcinoma cells, P114 and Sh1b cells were also examined in cell cycle analysis. After 48 h of incubation, the time point, when both antibodies mediated significant growth inhibition on both cell lines, cells were harvested and stained with the CycleTEST™ PLUS DNA Reagent Kit. In P114 cells it could be clearly demonstrated, that both cetuximab and trastuzumab mediated G0/G1 phase arrest. As can be seen in Table 3, displaying the percentage of cells in respective cell cycle phases, the percentage of untreated P114 cells in G0/G1 after 48 h was 41.58%. In the cetuximab treated group, this population rose to 48.28% and in the trastuzumab treated group to 49.28%. However, this effect could not be observed in Sh1b cells, were the antibody treated groups showed almost no difference in cell cycle compared to the untreated group (39.16% of cells in G0/G1 in the untreated group vs. 38.25% in the cetuximab and 39.19% in the trastuzumab treated group).

Table 3

Cell cycle analysis of canine mammary carcinoma cells P114 and Sh1b. Cell cycle analysis demonstrates G0/G1 arrest of P114 cells after 48 h of cetuximab as well as trastuzumab treatment, but no effect of antibody treatment on cell cycle in Sh1b cells.

	G0/G1	S	G2/M	Apoptosis
P114 cells
Untreated	41.58%	22.67%	21.91%	2.55%
Cetuximab	48.28%	22.76%	19.59%	2.22%
Trastuzumab	49.82%	18.58%	21.37%	2.19%
Sh1b cells
Untreated	39.16%	16.53%	26.86%	5.59%
Cetuximab	38.25%	14.42%	28.60%	4.23%
Trastuzumab	39.19%	14.05%	28.03%	4.73%

Quantitative analysis of immunofluorescense staining by flow cytometry using the QIFIKIT® assay revealed that ErbB-receptor expression was decreased by 2 logs on the tested canine cancer cells in comparison to human reference cells (Fig. 5). ErbB-1 positive canine cells express low levels of the tumor antigen in the range of 900 and 4500 molecules per cell, whereas for the human reference cell line A431, well known for substantial overexpression of ErbB-1, we measured more than 80,000 molecules per cell.

Fig. 5

Quantitative assessment of ErbB-receptor expression on canine mammary carcinoma cells. ErbB-1 is expressed on canine cells in the range between 900 and 4500 molecules per single cell, with specific average numbers depicted below bars. ErbB-2 is presented in levels between 200 and 1400 molecules per single cell on canine cancer cells. Human reference cell lines A431 for ErbB-1 and SKBR-3 for ErbB-2 show 2 log higher levels. Data points show mean % cell viability + SD.

Similarly to the above findings for ErbB-1, ErbB-2 antigen expression levels ranged between 200 and 1400 molecules per single cell on canine cancer cells, compared to more than 60,000 in average for the human reference cell line SKBR-3 which is known to over express this antigen. In summary, we measured low but detectable levels of the two receptors on canine mammary carcinoma cell lines.

Discussion

In human medicine the development and especially the clinical approval of novel cancer therapies proves to be lengthy, cumbersome and extensive (DiMasi and Grabowski, 2007). One of the limiting aspects in conventional drug development is that murine cancer models do not fully represent the features that define malignant diseases in humans including long periods of latency, genomic instability and the heterogeneity of tumor cells and their microenvironment (Gordon et al., 2009).

Furthermore, in contrast to human medicine, targeted therapies including immunotherapy of cancer is limited in veterinary medicine. Apart from surgery, chemotherapy options for canine mammary cancer involve cyclophosphamide, 5-fluorouracil, and doxorubicin (Sorenmo, 2003), agents long-established for clinical use in human medicine. Only very limited research efforts have to-date been invested into the generation of novel medicines for the treatment of canine cancer, which has resulted in poor clinical outcomes, e.g. 57 days median survival time of canine patients with inflammatory mammary cancers treated with chemotherapy compared to 35 days for those given only palliative treatment (Clemente et al., 2009).

Hence, several groups have recently started comparative oncology trials with promising results, such as that of the orally administered receptor tyrosine kinase inhibitor toceranib phosphate (Palladia®, Pfizer, NY, USA) investigated in a multi-center, placebo-controlled, double-blind, randomized study in dogs with recurrent mast cell tumors. This study reported significantly better response rates in the toceranib phosphate-treated compared to the placebo-treated group and may demonstrate that the dog may constitute a relevant model with which one can evaluate therapeutic indices of targeted therapeutics in a clinical setting (London et al., 2009). Furthermore, a second receptor tyrosine kinase inhibitor, masitinib (Masivet®, Paris, France) has been approved for the treatment of advanced mastocytoma in dogs (Hahn et al., 2008). More recently a xenogenic DNA vaccine against canine oral melanomas using human tyrosinase as immunogen (Oncept™, Merial, Duluth, GA, USA) to overcome immunotolerance in canine cancer patients has been approved by the FDA as the first approved therapeutic tumor vaccine for dogs (Bergman et al., 2006). Taken together, the dog may serve as a surrogate of human malignancy benefiting canine and potentially human patients.

ErbB-1 and ErbB-2 are two well characterized tumor-associated antigens, harnessed for targeted therapy of cancer in human medicine. There is evidence from immunohistochemistry that as in human carcinomas, both ErbB-1 and ErbB-2 canine homologues also play a role in canine carcinomas (Ahern et al., 1996; Dutra et al., 2004; Gama et al., 2009; Martin de las Mulas et al., 2003; Peruzzi et al., 2010). To estimate the prevalence in canine carcinomas, we screened mammary carcinoma samples using tests developed for the diagnosis of human carcinomas, namely EGFR pharm Dx™ for EGFR detection or HercepTest®; the latter was previously described to detect canine ErbB-2 (Dutra et al., 2004). Using the TissueFAXS technology as a novel and accurate metric, we found that 3 of 10 canine patient mammary carcinoma samples displayed strong membrane specific ErbB-1 expression, and 4 of 10 canine patient mammary carcinoma samples expressed cell surface ErbB-2. The 3 ErbB-1 positive samples ranged from 11.8 to 36.4% membrane staining allowing a 3+ classification following the guidelines of HercepTest®. With respect to ErbB-2 expression, 4 positive samples could be detected, ranging between 10.1% and 12.2% membrane-stained cells. Furthermore, 2 samples could be identified showing moderate ErbB-2 membrane staining in 6.2 and 7.6% of cells, respectively, which corresponded to 2+ classification. In human clinical cases of ErbB-2 overexpression by mammary carcinomas, 2+ and 3+ classified carcinomas could in principal be considered for treatment with trastuzumab. However, clinical benefit and response rates seem to strongly correlate with the intensity of ErbB-2 overexpression (2+ or 3+) with response rates of 35% in grade 3+ expressors compared to lower benefit in 2+ positives (Singer et al., 2008).

The dog genome suggested a common set of functional elements across the three analyzed mammalian species (mouse, dog, human) with 19,300 predictions for canine genes, most of them corresponding to known human genes (Lindblad-Toh et al., 2005). Furthermore, there is significant correlation between coding with noncoding sequences between human and dog (Joy et al., 2006). Moreover, an analysis of human and dog gene expression data derived from tumor and normal mammary glands indicated a significant overlap of genes (Uva et al., 2009). Further to the above previously reported findings, our homology search of human compared to canine ErbB-1 and ErbB-2 exceeded these expectations. Therefore, the high amino acid homologies and similarities resulted in a strikingly similar structure. Most importantly, the antigens contain highly conserved epitopes for the antibodies cetuximab and trastuzumab, both of which have been therapeutically applied in human medicine. Indeed we could demonstrate by flow cytometry that cell lines derived from canine mammary carcinoma express ErbB-1 and ErbB-2, and can be stained with the respective chimeric/humanized antibodies. The tumoristatic effects of cetuximab and trastuzumab antibodies on canine mammary carcinoma cells were demonstrated by a tetrazolium based cell proliferation assay, which indicated significant susceptibility of both canine carcinoma cell lines. Subsequent cell cycle analysis demonstrated for canine P114 cells G0/G1 phase arrest upon 48 h of treatment of both cetuximab as well as trastuzumab antibodies, but neither treatment induced an effect in Sh1b cells. This finding is in line with human data, as Raben et al. could demonstrate G0/G1 phase arrest in non small cell lung cancer cells (adenocarcinoma, bronchoalveolar and squamous cell carcinoma) upon cetuximab treatment in the majority of, but not in all cell lines (Raben et al., 2005). Similar results are described for ErbB-1-overexpressing squamous cancer cells of head and neck (Nestor, 2010). For breast cancer cells and trastuzumab treatment, Mayfield et al. demonstrated that the highly ErbB-2 overexpressing SKBR-3 and BT474 cell lines could be arrested in G0/G1 phase, whereas the lower expressing MCF-7 and MDA-MD-231 cells were unaffected (Mayfield et al., 2001).

However, our data indicated that the tested canine cancer cells generally responded less to trastuzumab or cetuximab stimulation than previously observed for human cells. Principally, this could be due to lower overall affinity, however, there are two major arguments against: (i) when we performed an in vitro competition assay, even molar excess of human ErbB-antigens was insufficient to remove cetuximab and trastuzumab from the canine cells, suggesting a low off-rate of the antibody from the dog orthologue (data not shown); (ii) there are only few point mutations within the epitope regions of both antibodies. For instance, in the trastuzumab binding site one prolin is changed to a serine. One may argue that generally proline residues are not well tolerated in secondary structure elements as α-helices and β-sheets. Proline residues are known to show a high propensity for being found within structural elements consisting of loops and turns (Nakashima et al., 1986; Monne et al., 1999). The position in the structure of the serine to proline mutation, however, is found in a loop segment, which shows a turn/bend exactly at the position of this amino acid. Therefore a mutation to proline would be well tolerated, given the bend in the backbone of the loop structure as well as the fact that the molecule resides at the margins of the epitope. Both facts rather argue for comparable affinities of both antibodies to ErbB1 and -2 molecules of human and dog origin.

This prompted us to determine the number or ErbB-1 and -2 antigens on the cellular membranes of the canine mammary cancer cells. In this experiment the murine precursors of cetuximab and trastuzumab, 225 and 4D5, respectively, were used to target the canine ErbB homologues. It is known and should be noted that trastuzumab acquired increased affinity towards HER-2 during humanization as a positive side effect (Carter et al., 1992). For this reason, the results of humanized antibody binding and definition of molecules per cell with the murine precursor antibodies may vary, although we could conclude, that the expression of the canine ErbB homologues was in general two logs lower than the expression of their human counterparts. These lower expression levels of ErbB-1 and -2, at least on the tested canine cells could provide an explanation for the relatively higher titers of antibody required to produce an inhibitory effect in the in vitro proliferation assay (5 μg/ml, compared to 0.5 μg/ml of antibody required to inhibit proliferation of human SKBR-3 cells) (Karagiannis et al., 2009). Nevertheless, the results of these cell viability assays indicate that ErbB-1 and -2 targeting and silencing by passive immunotherapy could be an efficient strategy for the treatment of mammary canine cancers. In the human setting, cetuximab and trastuzumab can also recruit immune cells via their constant domains and mediate ADCC and ADCP of tumor cells (Karagiannis et al., 2009; Vincenzi et al., 2008; Hudis CA, 2007). We suggest that if a “caninization” of these antibodies is performed, these antibodies may constitute safe and efficacious treatments for companion dogs suffering from mammary carcinomas. Engineering these antibodies with canine constant regions will exploit not only the direct tumoricidic aspects of these antibody treatments, but also those mechanisms of action mediated by Fc receptor-expressing immune effector cells, which contribute to the efficacy of cetuximab and trastuzumab in human clinical applications.

Of similar importance, the knowledge on the great similarity of canine and human ErbB-1 and -2 may prompt comparative oncology studies in novel targeting or combinatorial strategies for human cancer patients.

Conclusion

Our results on canine ErbB-1 and ErbB-2 molecules in comparison with the human counterparts surmounted all expectations on homology: modeling of both targets indicated an almost perfect evolutionary conservation including the epitopes for trastuzumab and cetuximab antibodies. Indeed, binding and biological efficacy could be confirmed in vitro for both immunoglobulins. Canine cancer is thus excellently suited for proof-of-concept studies in clinical comparative oncology settings and our findings will have important implications for the great number of researchers working on the ErbB-family.

Conflict of interest

The authors declare that they have no conflict of interest.