In vivo imaging of immunotoxin treatment using Katushka-transfected A-431 cells in a murine xenograft tumour model

Introduction

Cancer accounts for 13% of all deaths worldwide (in 2004). Although common treatment like surgery and radiotherapy (WHO Cancer, Fact Sheet No. 297, 2009) yields increasing success, there is still room for improvement, especially in preventing the development of metastasis.

Development of novel therapeutics and therapeutic strategies for the treatment of cancer nowadays is an elaborate and expensive procedure. Therefore, in vivo immunodeficient mouse models using human tumour cell lines have been developed to mimic different human cancers. An effective and sensitive monitoring system to determine the in vivo effect in such animal models prior to start clinical trials is of great importance. Such system should be cost-effective, easy of use and have high predictive value. Classically, tumour growth is measured with a vernier calliper, and animals are killed at different time points, and ex vivo examination like pathology and specific staining is performed.

The major drawback of this approach is the necessity for separate groups per individual time point and consequently the intermittent nature. This inherently implies the use of large numbers of animals to achieve statistical relevance [1].

For continuous monitoring, in vivo imaging techniques are greatly gaining interest, in particular optical imaging, as it has the advantage that relatively simple equipment can be used. One groundbreaking development was the use of green fluorescent protein (GFP) [2], as it can easily be transfected into different cell types or used in fusion proteins. Due to its emission in the green, the use of GFP in fluorescence-based molecular imaging approaches, however, has several disadvantages like low penetration depth, high tissue absorption, as well as emission in the same spectral range as many endogenous autofluorescent molecules. These together lead to low signal and high background, limiting the use of GFP [3]. This can largely be avoided by using the far red (FR) to near infra red (NIR) window [1, 3–5]. The fluorescent protein Katushka (Kat2) has a very strong emission maximum at 635 nm. This in combination with high photostability compared to other red and FR proteins makes it a powerful tool for in vivo imaging [6].

In this study, we established the use of Kat2 for monitoring therapy in a human xenograft tumour model in hairless immunodeficient mice. To this end, we imaged the growth of subcutaneous tumours of epidermal growth factor receptor (EGFR)-expressing A431scM3 cells transfected with Kat2. A431scM3 are A-431-derived epidermoid cancer cells, which have been passaged in vivo to select for fast-growing cells. Because of their high EGFR expression, A-431 are suitable target cells to study the effect of therapeutic approaches directed to EGFR, which is overexpressed on a variety of cancer cells [7, 8]. Due to this, A-431 cells are often used in preclinical research [9, 10]. Already approved therapeutics targeting the EGFR are Gefitinib (Iressa) and Erlotinib (Tarceva) [11, 12] belonging to the class of tyrosine kinase inhibitors. Cetuximab [13] and Panitumumab [14] are monoclonal antibodies binding to the EGFR-blocking natural ligand binding sites. All these agents inhibit proliferation, differentiation and angiogenesis of EGFR-positive tumours improving the efficiency of conventional chemotherapy. However, development of resistance to anti-EGFR therapy is a huge problem for many patients [15]. Therefore, the need for other EGFR targeting therapeutics is high.

As EGFR-specific therapeutic, we used the 425(scFv)-Pseudomonas aeruginosa Exotoxin A’ (425(scFv)-ETA’) immunotoxin (IT), which has previously shown to have anti-tumour activity in a disseminated human pancreatic cancer mouse model [16, 17]. ITs are fusion proteins consisting of a cell-specific binding domain coupled to a toxic entity [18–22]. Due to their cell specificity, ITs show greatly reduced side effects in comparison with chemotherapeutics [23]. As toxic moieties, many molecules of different origins have been tested, but as yet, only two have been used in clinical studies: Diphtheria toxin and ETA [24–26]. Both toxins inhibit protein biosynthesis by ADP ribosylation of eukaryotic elongation factor 2, leading to cell death.

Cancer cells are known to adapt different cellular programmes coordinating regulatory circuits, allowing them to, for example, metastasise and invade tissues, sustain angiogenesis or proliferate infinitely [27, 28]. Additionally, receptor downregulation can be observed in some tumours as a mechanism to evade immunogenicity, which becomes a major problem when targeting those antigens. Recently, it was shown that Cetuximab treatment resulted in translocation of the EGFR to the nucleus [29]. Another study reports that treatment with Erlotinib leads to lower EGFR amounts, while not affecting gene expression levels [30]. Hence, it is important to check on antigen expression during therapy.

Preferably, one would like to determine multiple parameters like growth and marker expression in one go, using optical imaging.

Labelling of molecules with different fluorescent dyes can be used to track these molecules in vivo. With the SNAP-tag technology, small antibody fragments like scFvs or antibodies fused to O⁶-alkyl-DNA alkyltransferase (SNAP-tag) can be coupled to various fluorescent dyes or nanoparticles [31] and toxins. We recently published the coupling of 425(scFv)SNAP to the NIR dye BG-747 using this technology to efficiently detect EGFR-positive tumour cells in vivo. An additional advantage is the rapid systemic clearance of the fusion protein [32].

In this study, we established a system to test and quantify the therapeutic potential of a candidate drug, using optical imaging. To this end, we set up a model in immunocompromised mice using the Kat2-transfected, high EGFR-expressing epidermoid cancer cell line A431scM3 in combination with the EGFR-specific 425(scFv). This scFv was both used for detection, as imaging agent conjugated to BG-747 via SNAP-tag, as well as an IT linked to ETA’.

Materials and methods

Results

Discussion

The use of in vivo molecular imaging, which allows for the real-time continuous monitoring of therapeutic effects over a prolonged period in a single animal, has considerably improved therapeutic evaluation. Due to its ease of use and versatile application to small animal in vivo experiments, especially optical imaging is increasingly used to address several questions in terms of molecule targeting or pharmacokinetics. Optical imaging techniques mainly comprise bioluminescence imaging and fluorescence imaging [1, 3]. For fluorescence optical imaging, a broad range of fluorophores has been developed today, which offers the possibility to track multiple probes simultaneously [4].

In this study, we report the feasibility of visualising treatment effects by Kat2 fluorescence-based molecular imaging. For a first proof-of-principle, we used cells with a high antigen expression as target, which were passaged through the mouse to select for cells that grew fast in vivo and have a high tumour take rate. By transfection of these A431scM3 tumour cells with DNA encoding for the highly stable and fluorescent protein Kat2, we established a method for easy in vivo monitoring based on the fluorescence signal, which is only dependent on the expression of Kat2 in the cells and not on external factors.

Transfected A431scM3 cells showed very strong fluorescence that even persisted for several weeks in culture. Continuous Kat2 expression does not alter the EGFR antigen expression nor apoptosis induction; both transfected and non-transfected cells showed similar sensitivity towards the 425(scFv)-ETA’ IT in vitro (Online Resource 3). Implantation of A431scM3-Kat2 cells induced tumour growth in 100% of BALB/c nu/nu mice. After determining the signal parameters, a spectral library was established that allowed for the unmixing of the images. Due to the strong fluorescence and the very low background signal from the emission of the Kat2 in the FR, this resulted in a clear delineation of the tumour area, which enables the exact measurement of the tumour and thereby increases statistical significance, compared to, for instance, measurements with a vernier calliper. This is also a clear difference to experiments, where GFP-transfected tumour cells are used, as in that case the entire body emits background fluorescence. This makes it more difficult to completely separate the specific signal from the background signal in, for example, feet or ears of the animals (results not shown).

Treatment efficacy was detectable by measuring the fluorescent area of the tumour. With this approach, no additional dye had to be injected, thus preventing the risk of potentially unbound fluorescent molecules circulating in the mice’s vasculature, which decreases the tumour-to-background ratio. This set-up was subsequently used to evaluate the effect of two different therapeutics against EGFR-expressing tumour cell lines in vivo.

Tumour-bearing mice were treated with 425(scFv)-ETA’ or Doxorubicin, a chemotherapeutic successfully used in the clinic for the treatment of cancer [38, 39], as positive control.

While treatment with 425(scFv)-ETA’ showed decelerated tumour growth compared to the PBS control group, treatment with Doxorubicin resulted in a shrinkage of the tumours.

The 425(scFv)-ETA’-treated mice, however, did not show any signs of the physical weakness or limitations observed in the Doxorubicin-treated animals. The arrest in tumour growth, combined with the lack of side effects, clearly demonstrates the high specificity of the 425(scFv)-ETA’ treatment. Changes in the treatment schedule would be required to optimise the therapeutic effect. Possibly, a combination therapy with a lower dose of Doxorubicin could combine both effects resulting in less side effects for the treatment of EGFR-positive tumours.

Parallel with the optical imaging, tumour size was also determined classically, using a vernier calliper, and we observed that this resulted in much wider variances and therefore the optical imaging data led to better statistics. This is in accordance with the data from Jensen et al. [40], who also found a systematic bias increasing with tumour size. Not only small tumours are difficult to grasp, but the measurement of large tumours can be difficult as they might grow into an irregular shape and extend deep into the tissue. Thus, with optical imaging, we were able to easily monitor treatment-induced changes in tumour growth, also of deeper localised parts of the tumour tissue. It is therefore preferred over vernier calliper measurements.

Next to the mere monitoring of tumour growth, assessment of additional parameters would be beneficial to specify treatment effects or detect changes in tumour physiology. Targeting a receptor on tumour cells may lead to downregulation of this receptor by the tumour cell as escape mechanism. We therefore looked at the expression of the EGFR using the same 425(scFv) as in the 425(scFv)-ETA’ fusion construct for treatment but fused to a SNAP-tag coupled to a fluorescent dye, considering a washout period of 11 days to prevent blocking of the binding by previously bound 425(scFv)-based IT. In particular, the 425(scFv)SNAP fusion protein has been shown to be useful by our laboratory in targeting EGFR-positive tumours in vivo [32].

In vivo optical imaging requires the use of fluorochromes that emit in the 650–900 nm range, and with normal optical imaging techniques using filters, it is difficult to discriminate between closely adjacent signals. Spectral analysis, as used by the Maestro, however, makes it possible to discriminate between two optical signals, not based on the colour, but on the exact shape of the spectrum [41]. The combination of the fluorochromes here used, Kat2 and BG-747, both emitting in the same range, could be used after determining the appropriate spectral library using the multifilter acquisition mode. This library could subsequently be used to detect a second, NIR fluorescent BG-747 dye next to the FR Kat2-labelled tumour, in the same mouse. The two signals could completely be separated even though they were in the same physical location, showing expression of the EGFR also after therapy. We could hereby affirm that the EGFR as a target on the A431scM3 cells is not downregulated during therapy and can be used as therapeutic target. Due to the renal clearance, the signal was also detected in the kidneys, which could be shown earlier [32].

Multicolour imaging is a method, which allows the simultaneous use of differentially labelled molecules. Due to the possibility to unmix different colours, there is a great flexibility in colour selection [42]; also, dyes with similar emission wavelengths can be distinguished. The use of differentially labelled molecules plays an important role, for example, in the diagnosis of different tumour species by their specific receptor expression pattern. After administration of a cocktail of different targeting moieties, subtraction of unspecific signals from the specific signal is possible and a huge advantage over radiolabelled imaging, where simultaneously acquired signals from different binders cannot be distinguished [43]. Activatable fluorescent probes have the advantage that they can be activated by different cancer-related factors, like endosome-specific enzymes or changes in pH upon internalisation. In this case, free dye can be distinguished from internalised probes. Thereby, these probes are helpful for localising specific tumour tissue, for example, metastases. Furthermore, when using activatable probes that are reversible, the discrimination between viable and dead tumour cells is possible. However, there are at present few activatable fluorescent probes suitable for in vivo imaging [44].

In our experiments, we employed solid tumours for a first, preclinical evaluation of new therapeutics enabling us to apply an easy method using “standard” commercially available dyes, which can rapidly be coupled to our targeting construct. This approach will be even more interesting for orthotopic and disseminated models, when the binding of fluorescently labelled molecules to tumours, which are not visible from outside, or metastases is examined.

In summary, we established an in vivo imaging method that allows for the monitoring of tumour growth and thereby therapeutic effects. We developed a robust and feasible method to monitor treatment for subcutaneous tumours by optical imaging using target cells, transfected with the strongly fluorescent, FR-emitting protein Kat2. Furthermore, we show that also additional parameters, such as target receptor expression, can be assessed even though the signal is in a similar spectral range and location, using the spectral analysis of the Maestro system.

Additionally, we could show that treatment with an EGFR-specific IT resulted in decelerated tumour growth. Improvement of 425(scFv)-ETA’ treatment schedule or a combination therapy with Doxorubicin might thus be more effective in tumour reduction as well as in overall health of the mice.

With molecular imaging, each animal represents its own statistical control throughout the whole experiment [1]. The results have less statistical variance, as the individual mice are monitored at different time points, as opposed to different animals for each time point. This also leads to a strong reduction in total animals required per experiment.

Electronic supplementary material

Below is the link to the electronic supplementary material.