2.1.

Fluorescent Probes

Herceptin was coupled to the NIRF dyes CypHer5E (GE Healthcare, Munich, Germany) and Alexa Fluor 647 (Invitrogen, Darmstadt, Germany) at different dye-to-protein (DP) ratios in cooperation with Squarix Biotechnology (Marly, Germany). For this purpose, different amounts of the amine-reactive $N$-hydroxysuccinimidyl (NHS) esters of these NIRF dyes (different ratios of dye NHS ester to antibody) were allowed to react with Herceptin in a buffer at basic pH, using identical reaction conditions for both labels (e.g., solvent, pH, temperature, reaction time). The resulting dye-Herceptin conjugates were purified chromatographically using a Sephadex column and a phosphate buffer as eluent. The DP ratios were calculated from the photometrically determined concentration of the dye and the protein using a calibrated Cary 5000 spectrometer (Varian Inc., Australia) following a previously described procedure.³² In the case of Alexa Fluor 647, the dye concentration was determined from the integral absorbance at the longest-wavelength absorption. For CypHer5E, the dye concentration was determined from the absorbance measured at 565 nm using a previously measured molar absorption coefficient of $15,335\mathrm{L}/\mathrm{mol}/\mathrm{cm}$.

2.2.

Spectroscopic Analysis

2.3.

Cell Lines

KPL-4 (provided by J. Kurebayashi, Kawasaki Medical School, Kurashiki, Japan)³⁰ and MDA-MB-231 (ATCC/LGC, Wesel, Germany) breast cancer cells were propagated in standard DMEM containing $4.5\mathrm{g}/\mathrm{L}$ glucose and l-glutamine. The medium was supplemented with 10% fetal calf serum (PAN Systems, Aidenbach, Germany). Cells were cultivated under standard cell culture conditions at 37°C in a humidified atmosphere under 5% CO₂. Tumor cells were harvested near confluence by brief trypsinization in 0.05% trypsine-EDTA solution, washed several times, and placed in sterile PBS shortly before implantation.

2.4.

Fluorescence Microscopy

Breast cancer cells were seeded on four-well chambered coverglasses coated with poly-l-lysine at a density of $2\times {10}^4$ KPL-4 cells and $1.2\times {10}^4$ MDA-MB-231 cells per well, respectively, in 500 µL DMEM. After 48 h, the medium was removed and the cells were supplemented with 500 µL DMEM and 0.1 M HEPES buffer containing 5 µg fluorescent probes or 0.2 nmol of the free fluorophores, CypHer5E (hydrolyzed CypHer5E NHS ester) or Alexa Fluor 647 (hydrolyzed Alexa Fluor 647 NHS ester). Cells were incubated with probes for 8 h at 37°C or 4°C. Afterward, the cells were washed several times with PBS and incubated with Hoechst 33342 ($2\mathrm{mg}/\mathrm{L}$; Invitrogen, Darmstadt, Germany) to stain the nuclei of the living cells. After 10 min, the Hoechst solution was removed and the cells were covered with 500 µL PBS and subsequently used for fluorescence microscopic studies with the Axiovert 200M (Carl Zeiss, Göttingen, Germany) equipped with a NIRF-sensitive ORCA-ER digital camera (Hamamatsu, Herrsching am Ammersee, Germany). Signals of the fluorescent probes were recorded with a $640\pm 15\mathrm{nm}$ excitation and a $690\pm 25\mathrm{nm}$ emission filter and Hoechst with a $365\pm 12.5\mathrm{nm}$ excitation and a $445\pm 25\mathrm{nm}$ emission filter, respectively. Image generation and processing were performed with the software systems AxioVision Rel.4.6. and ImageJ,³⁵ respectively.

2.5.

Animals

All animal experiments were performed according to the animal protection law, and all animal protocols were approved by the administration of Lower Saxony, Germany. Experiments were performed on female athymic nude mice, strain NMRI-Fox1nu/nu. Two weeks before in vivo imaging, mice received chlorophyll-depleted low fluorescent food, Regime 210 (SAFE, Augy, France). Mice were maintained in a sterile environment in special cages with filter huts placed in a Scantainer (Scanbur, Koge, Denmark).

2.6.

Tumor Cell Implantation

KPL-4 and MDA-MB-231 breast tumor cells were implanted in the right abdominal mammary fat pad of female nude mice at the age of 6 to 12 weeks. For this purpose, mice were anesthetized by intraperitoneal injection of $15\mathrm{mg}/\mathrm{kg}$ xylazine and $75\mathrm{mg}/\mathrm{kg}$ ketamine. Solutions containing the respective amount of tumor cells ($5\times {10}^6$ KPL-4 cells or $1\times {10}^6$ MDA-MB-231 cells) suspended in 30 µL sterile PBS were implanted with an insulin syringe with integrated needle (30G, BD, Heidelberg, Germany) as described.³⁶ Afterward, mice were inspected twice a week for weight loss, general condition, and tumor formation.

2.7.

In Vivo Imaging

The time domain fluorescence imager, Optix MX2 (ART, Montreal, Canada),³⁷ was used for intensity and LT imaging in the NIRF region. This device scans in a raster modus and measures the time between every laser excitation pulse and the detection of the first fluorescence photon using a single photon counting detector. The resulting photon time-of-flight histogram, termed temporal point spread function (TPSF), obtained for each scan point, is used to calculate the LT of the emissive species as an exponent of the slope of the decay by underlying a single exponential model.³⁸ The quality of the fits was judged by the value of ${\chi }^2$ for each TPSF that should be between 0.8 and 1.2 for reliable fits (OptiView software, User’s Manual, Version 2.02).

The in vitro fluorescence LTs of pH-Her and Alexa-Her (3 µg IgG) and their parent fluorophores were measured in 150 µL PBS and in PBS containing 5% (mass) bovine serum albumin (BSA; PBS/BSA 5%; Sigma-Aldrich, Seelze, Germany) at pH values of 7.5 and 5.5, respectively. To determine the LT of the fluorescent probes in vivo, 10 µg pH-Her and Alexa-Her dissolved in 50 µL 0.9% NaCl were each administered subcutaneously (s.c.) in healthy nude mice.

Mice were used for in vivo fluorescence imaging experiments 1 to 4 months after tumor implantation when the tumors had reached sizes of approximately $0.3{\mathrm{cm}}^3$. Mice received i.v. injections of 25 µg pH-Her or Alexa-Her or 0.8 nmol of the respective free fluorophore in 0.9% NaCl. KPL-4 tumor-bearing mice were scanned before and 1, 2, 4, 24, 48, and 72 h after probe injection, and control mice, bearing MDA-MB-231 tumors, were scanned before and 24, 48, and 72 h after probe injection. During injection and scanning, mice were anesthetized with vaporized isoflurane at concentrations of 0.8% to 1%.

Imaging was carried out using an excitation wavelength of 635 nm and a $670\pm 20\mathrm{nm}$ band pass emission filter. Scans were performed with a 1.5 mm (whole body scan) or 1.0 to 1.5 mm (excised organs) raster, and an integration time of 0.7 to 1.0 s per scan point. Data were analyzed with OptiView software 1 00 00 (ART, Montreal, Canada). Intensity values were determined in normalized counts (NC), which are independent of laser power and integration time. For the subtraction of autofluorescence, the average fluorescence intensity over the tumor region of the prescan was subtracted from all intensity maps of the respective mouse. The contrast-to-noise ratio (CNR) was calculated as the difference between the average fluorescence intensity over the tumor and over a background region divided by the standard deviation of the background signals. As a background region, the area over the lung was selected, as no specific fluorescent signals from the HER2-bound probe and no unspecific signals (from fluorescent food, for example) are to be expected.

After the in vivo experiments, the animals were sacrificed, and autopsies of mice were performed. Subsequently, the tumors were excised and the tumor sizes were measured by a caliper. Furthermore, the abdomen and the thoracic cavity were opened and lung, heart, liver, kidneys, spleen, stomach, and gut were isolated and imaged ex vivo with the Optix MX2.

3.1.

pH-Her Conjugates at Low DP Ratios Show Favorable Spectroscopic Properties

For the choice of a suitable pH-sensitive probe for in vitro and in vivo studies, dye-Herceptin conjugates of different DP ratios were spectroscopically studied with respect to their pH sensitivity and QYs. As shown in Fig. 1(a) and 1(b), a decrease in pH led to an increase in fluorescence only in the case of the pH-sensitive dye conjugate pH-Her, whereas the emission of Alexa-Her did not undergo any apparent pH-induced changes ($n=3$). No apparent change in the spectral position of the emission maxima of either probe was observed for pH values of 5.5 and 7.5. The protonation-induced fluorescence enhancement factors of different dye-Herceptin conjugates and their QYs are summarized in Fig. 2. The pH sensitivity of the fluorescent conjugates was calculated as the ratio of the fluorescence intensities at the emission maximum obtained at pH values of 5.5 and 7.5 and thus presents a measure for the protonation-induced fluorescence enhancement of the probe. The maximum enhancement factor of 5.6 was found for pH-Her, with a DP ratio of 1.6, whereas pH-Her with a DP ratio of 4.9, for example, showed a 3.9-fold increase in fluorescence upon protonation. Due to its good pH sensitivity and comparably high QY, the pH-Her conjugate with a DP ratio of 1.6 [Figs. 1(a) and 2] was applied for further experiments shown in Figs. 3 to 6. The Alexa-Her conjugate with a DP ratio of 1.6 [Figs. 1(b) and 2] was used as control and as an example for an always-on probe.

Fig. 1

Fluorescence emission spectra of probes at different pH. Representative uncorrected fluorescence emission spectra ($n=3$) of pH-Her (a) and Alexa-Her (b) measured in PBS at pH of 7.5 (black curve) and of 5.5 (red dashed curve) shown as examples for conjugates with a DP ratio of 1.6 ($n=3$); excitation was at ${\lambda }_{\mathrm{ex}}$ 635 nm.

Fig. 2

Fluorescence QYs and pH sensitivities of Herceptin conjugates. Fluorescence QYs (left y-axis) of different pH-Her conjugates (gray shaded columns) and a representative Alexa-Her conjugate (gray crossed columns) in PBS at a pH of 5.3 and pH sensitivities (right $y$-axis) of these conjugates. The pH sensitivities are given as protonation-induced fluorescence enhancement factors (red column) calculated from the ratio of the relative fluorescence intensities at the emission maximum obtained at pH 5.5 and 7.5 ($n=3$), respectively, as shown in Fig. 1(a) and 1(b).

Fig. 3

Fluorescence microscopy demonstrates internalization-dependent activation of pH-Her. Breast cancer cells grown on culture slides were incubated for 8 h with pH-Her or Alexa-Her. On the left panel, counterstain of cell nuclei with Hoechst 33342, in the middle, probe-derived signals, and on the right panel, merged images of the cell nuclei (blue) and the probe (red) are illustrated. (a), When incubated with KPL-4 cells at 37°C, pH-Her shows fluorescence only after receptor-mediated internalization (green arrow). (b), At 4°C, no signals from the pH-sensitive probe presumably bound to the cell membrane can be detected. (c), Alexa-Her shows fluorescence from the internalized probe (green arrow) and also membrane-derived fluorescence can be observed after 8 h of incubation at 37°C (c) and 4°C (d) (orange arrow, no internalization). Representative images of three independently performed experiments are presented. Bars represent 50 µm.

Fig. 4

In vivo intensity maps after probe application and subtraction of autofluorescence. KPL-4 breast tumor–bearing mice were imaged in a ventral position before (prescan) and 1 to 72 h after i.v. probe injection. Representative ($n=3$) fluorescence intensity images after subtraction of autofluorescence are shown in NC. (a), In a mouse receiving pH-Her, probe-derived signals are mainly detected in the tumor. The enhanced autofluorescence near the forelegs is most likely caused by a higher fat content of tissue. (b), In a mouse receiving Alexa-Her, probe-derived signals are obtained in the tumor and in the background. Background and tumor areas are indicated with white circles. Note the different scale bars in (a) and (b).

Fig. 5

Improved tumor detection sensitivity in vivo by use of the pH-sensitive tumor-specific probe. CNRs calculated from in vivo intensity scans after subtraction of autofluorescence are shown. CNRs were calculated in scans of KPL-4 tumor–bearing mice before (0 h) and 1, 2, and 4 h ($n=3$) and 24, 28, and 72 h ($n=5$) after i.v. injection of pH-Her (a) or Alexa-Her (b) (red columns). As controls, CNRs were also calculated from scans of KPL-4 tumor–bearing mice 0 to 72 h after i.v. injection of the free parent fluorophores CypHer5E (a) ($n=3$) and Alexa Fluor 647 (b) ($n=3$; gray shaded columns) as well as from scans of MDA-MB-231 tumor-bearing mice 0, 24, 48, and 72 h after injection of pH-Her (a) ($n=3$) or Alexa-Her (b) ($n=3$; blue columns with dots). Standard deviations are indicated as bars.

Fig. 6

LT-gating does not improve tumor detection sensitivity in vivo. LTs of pH-Her (a) and Alexa-Her (b) were measured in PBS and PBS/BSA 5% at pH values of 7.5 and 5.5, respectively, as well as after s.c. injection in nude mice. All measurements were performed in triplicate. The mean LT of measurements in PBS/BSA 5% and after s.c. injection for each probe was used as predicted LT for subsequent LT-gated imaging analysis in tumor-bearing mice. (c and d), Representative examples of LT-gated in vivo intensity maps of KPL-4 tumor-bearing mice obtained 24 and 48 h after i.v. injection of Herceptin conjugates are shown as examples. Background and tumor area are indicated with white circles at 48 h after probe injection. Fluorescence intensities with LTs of $1.3\pm 0.2\mathrm{ns}$ for a mouse receiving pH-Her (c) and with LTs of $1.5\pm 0.2\mathrm{ns}$ for a mouse receiving Alexa-Her (d) are illustrated ($n=5$) in NC. (e and f), In vivo average LTs in KPL-4 tumors of mice (M1 to M5) 1, 2, and 4 h ($n=3$) and 24, 48, and 72 h ($n=5$) after i.v. injection of pH-Her (e) and Alexa-Her (f). The predicted LTs of the probes [determined in (a) and (b)] are indicated as black dashed lines, and the LT ranges used for LT gating of in vivo intensity images are shown in gray. Standard deviations are indicated as bars.

3.2.

Functionality of the pH-Sensitive Conjugate In Vitro

As follows from Fig. 3, the pH-sensitive probe pH-Her showed fluorescence only after receptor-mediated internalization into HER2-positive KPL-4 cells after 8-h incubation at 37°C [Fig. 3(a), green arrow]. In contrast, after 8-h incubation at 4°C, when internalization should have been strongly reduced,¹¹ no signals could be detected from pH-Her that was most likely only membrane-bound [Fig. 3(b)]. These results demonstrate that pH-Her becomes emissive only after internalization, presumably when localized in the acidic endosomes and lysosomes of the target cell.^9,10,39 In contrast, the always-on probe Alexa-Her showed fluorescence after internalization [Fig. 3(c), green arrow] as well as when bound to the cell membrane [Fig. 3(c) and 3(d), orange arrow].

No fluorescence signals were detected on HER2-negative MDA-MB-231 cells after incubation with either pH-Her or Alexa-Her at 37°C and 4°C (data not shown). Incubation of both cell types for 8 h with the corresponding free fluorophores CypHer5E and Alexa Fluor 647 at 37°C and 4°C did not yield any detectable fluorescence, underlining the absence of unspecific binding and/or internalization of the dyes (data not shown).

3.3.

Improved Detection Sensitivity of Tumors by the Combined Use of a pH-Sensitive Probe and Subtraction of Autofluorescence

Representative fluorescence intensity maps of KPL-4 tumor-bearing mice over time obtained after subtraction of autofluorescence are shown in Fig. 4. After i.v. application of pH-Her [Fig. 4(a), $n=3$], the fluorescence in the tumor (white circle, shown at 24 h) can be clearly detected. The background fluorescence (white circle, shown at 24 h) is low and comparable to the fluorescence of the mouse prescan, except for some signals deriving from autofluorescence of the gastrointestinal tract.³ In contrast, after injection of Alexa-Her [Fig. 4(b), $n=3$], the fluorescence in the tumor (white circle, shown at 24 h) is also clearly visible, but simultaneously, the background signals resulting for this probe (white circle, shown at 24 h), are high compared to the signals from the prescan.

Hence, as rationalized with our probe design concept, use of the pH-sensitive probe pH-Her in combination with subtraction of autofluorescence enabled the elimination of major contributions of background fluorescence in scans of tumor-bearing mice. In contrast, the intensity of the background strongly increased after i.v. injection of the always-on probe Alexa-Her and could not be eliminated by subtraction of autofluorescence.

To quantify the tumor detection sensitivity in vivo after injection of the fluorescent probes and subtraction of autofluorescence, CNRs were calculated. As shown in Fig. 5, i.v. injection of pH-Her [Fig. 5(a)] in KPL-4 tumor-bearing mice ($n=5$) resulted in a continuous increase of CNRs within 24 h up to a ratio of 171 that remained comparably high for 72 h. In contrast, application of Alexa-Her [Fig. 5(b), $n=5$] led to an increase in CNR in the mice over time of at most only 82. In consequence, although the size of the fluorescence signal resulting from Alexa-Her exceeds that of pH-Her as indicated by the different range of the scale bars in Fig. 4(a) and 4(b), the average CNR over all times in KPL-4 tumor-bearing mice treated with pH-Her was about 2.5-fold higher than for mice treated with Alexa-Her.

Control experiments performed in HER2-negative MDA-MB-231 tumor-bearing mice injected with pH-Her [Fig. 5(a), $n=3$] revealed CNRs that are in average nine times lower than in HER2-positive KPL-4 tumor-bearing mice treated with the same probe. In the case of Alexa-Her [Fig. 5(b), $n=3$], the average CNR after probe application was only four times higher as found in MDA-MB-231 tumor-bearing mice. This illustrates that the detection specificity of the signals deriving from HER2-positive tumors, in comparison to HER2-negative tumors, can be enhanced by use of the pH-sensitive probe.

Further control experiments in KPL-4 tumor-bearing mice with the free dyes CypHer5E [Fig. 5(a), $n=3$] and Alexa Fluor 647 [Fig. 5(b), $n=3$] revealed that after i.v. dye injection, CNRs were also low after 24 to 72 h (maximum ratio of 21 and 20 for CypHer5E and Alexa Fluor 647, respectively) and comparable to CNRs in HER2-negative tumor-bearing mice treated with the Herceptin conjugates. These results indicate a low nonspecific accumulation of the fluorescent probes in the tumor. This further demonstrates that the tumor-targeting moiety Herceptin is necessary for efficient accumulation of the probe in the tumor and for the achievement of high CNRs. Surprisingly, Alexa Fluor 647 administered into KPL-4 tumor-bearing mice revealed a relatively high nonspecific accumulation in the tumor with comparably high CNRs at early times after injection [Fig. 5(b), CNR of 72 after 1 h].

To analyze in more detail whether the fluorescent probes accumulate nonspecifically in other organs, KPL-4 and MDA-MB-231 tumor-bearing mice were sacrificed 72 h after injection of pH-Her, Alexa-Her, or their free parent dyes and the organs were scanned ex vivo. In comparison to organs of untreated mice ($n=4$), no increased fluorescence could be detected in the liver, kidneys, lung, heart, spleen, stomach, or gut (data not shown).

In summary, subtraction of autofluorescence in vivo in combination with the use of the pH-sensitive probe pH-Her clearly improved the tumor detection sensitivity in breast tumor-bearing mice compared to the always-on probe Alexa-Her.

3.4.

LT-Gated Imaging in Combination with Herceptin Conjugates Does Not Improve Tumor Detection

To assess whether LT imaging in combination with pH-Her can also provide an increased tumor detection sensitivity, LT-gated image analysis was performed with scans of KPL-4 tumor-bearing mice receiving either pH-Her or Alexa-Her. For the application of LT gates as an extra method for data illustration only, principally the LT of the probe can be determined from the in vivo scans in the tumor and employed for gating. However, for the identification of unknown tumor sites not macroscopically visible at the time of imaging and/or for the detection of metastases, the probe’s LT must have been reliably determined from LT measurements with appropriate model systems before data analysis with LT gating can be performed with the in vivo scans. This very challenging approach requires identical or at least closely matching LTs (within the chosen LT gate) for the probe in the tumor and in the model system(s). This may not be necessary fulfilled, as the LT of a fluorophore can be affected by dye environment (proticity, polarity, viscosity, refractive index) in a dye-specific manner.

Preevaluation of LTs of pH-Her [Fig. 6(a)] and Alexa-Her [Fig. 6(b)] under several application-relevant conditions (in vitro at different pH values and in the presence of BSA as model for plasma proteins, and in vivo after s.c. injection in living mice) revealed only small environment-induced changes in the fluorescence LTs of both conjugates. The presence of proteins (PBS/BSA 5%) resulted in slightly increased LTs for both probes compared to LTs in PBS. Surprisingly, the LTs of both probes obtained after s.c. injection in vivo were shorter compared to the LTs measured in PBS and PBS/BSA 5%, although a solution of PBS/BSA 5% has been described as a good model for the prediction of the in vivo LTs of several NIR cyanine dyes.⁴⁰ The mean LTs of measurements in PBS/BSA 5% at pH 7.5 and 5.5 and after s.c. injection expected to reflect the probe environment in the in vivo situation⁴⁰ were 1.3 ns for pH-Her and 1.5 ns for Alexa-Her [Fig. 6(a) and 6(b)]. These LTs (referred to as predicted LTs in the following) were used for probe identification in subsequent LT-gated in vivo imaging studies of tumor-bearing mice.

Subsequently, fluorescence intensity maps of KPL-4 tumor-bearing mice after i.v. injection of Herceptin-conjugates were gated with the predicted LTs of pH-Her [Fig. 6(c)] and Alexa-Her [Fig. 6(d)] as shown for scans obtained after 24 h and 48 h after probe administration ($n=5$). Only signals with the predicted LT for the respective probe within a LT range of $\pm 0.2\mathrm{ns}$ are shown. Interestingly, with this approach, the fluorescence signals of pH-Her in the tumor (white circle, shown at 48 h) are partly eliminated at most times as follows from Fig. 6(c), whereas in the case of Alexa-Her, the major part of the background fluorescence is still visible [Fig. 6(d), white circle, shown at 48 h]. Similar results were obtained for scans performed at different times after probe application (data not shown). A comparison of the LT-gated images with the images obtained after subtraction of autofluorescence in the same mice (see Fig. 4) revealed that the tumor detection sensitivity could not be increased for either pH-Her or Alexa-Her with LT-gated imaging under these conditions.

Precise analyses of the average LTs in the tumor area at each scan time and in different mice showed that the LTs of pH-Her in the tumor in vivo with a mean value of 1.5 ns [Fig. 6(e)] are all slightly longer than the predicted LT (black dashed line). Accordingly, they are only partly covered by the LT range used for gating ($1.3\pm 0.2\mathrm{ns}$, gray area) of the pH-Her signals. This increase in LT of pH-Her in the tumor accounts for partial elimination of tumor signals. On the contrary, the average LTs of Alexa-Her in the tumor of 1.5 ns [Fig. 6(f)] agree with the predicted LT of 1.5 ns (black dashed line) and thus lie well within the chosen LT gate ($1.5\pm 0.2\mathrm{ns}$, gray area).

As to be expected, consideration of the tumor-related LT changes in the case of pH-Her and consequently, the use of a LT gate of $1.5\mathrm{ns}\pm 0.2\mathrm{ns}$ (Fig. 7) for image analysis, allowed a clear visualization of the entire tumor signals at all times as well as a partial elimination of background fluorescence.

In summary, neither for pH-Her nor for Alexa-Her could the tumor detection sensitivity in vivo be improved by LT-gated imaging using LTs previously determined in model systems. As the predicted LT of pH-Her differed from the LT actually measured in the tumor, LT gating resulted in a considerable loss in signal. In the case of Alexa-Her, background signals deriving from unbound Alexa-Her could not be efficiently eliminated from tumor-derived fluorescence by LT gating.

Fig. 7

LT gating of scans from KPL-4 tumor-bearing mice to the tumor-characteristic LT of pH-Her. Representative examples of LT-gated in vivo intensity maps of KPL-4 tumor-bearing mice obtained 24 and 48 h after injection of pH-Her are shown ($n=5$). As an example, the same mouse as shown in Figs. 4(a) and 6(c) is depicted. In contrast to Fig. 6(c), here fluorescence intensities with LTs of $1.5\pm 0.2\mathrm{ns}$ (corresponding to the LT of pH-Her in the tumor area) are illustrated. The tumor (indicated with a white circle) is clearly visible, and parts of the background fluorescence could be eliminated.