Our investigations revealed that the distinct spacer compositions of the probes QE and bivQ influenced all in vivo specificity, stability and kidney uptake above. Compared to other optical minigastrin probes, in vivo tumour-to-background contrast was markedly increased. The negatively charged d-glutamic acids of the probe QE led to the highest TBRs and best in vivo specificity and stability against proteases. The bivalent nature of the bivQ probe, however, was not adequate to increase target affinity, as observed for other multimeric small molecules.
In vitro probe characteristics
We could confirm that binding characteristics of optical minigastrin probes are influenced by the spacer sequence, which connects the imaging moiety with the binding moiety, as it was published for radiolabelled minigastrin probes e.g.
[
14,
28‐
30]. The
in vitro experiments demonstrated that the structure of CCK2R-targeted NIRF probes QE and bivQ mainly influenced metabolic stability, whereas
in vitro specificity was not altered. Both probes showed a high affinity binding (QE,
K
d = 5.8 ± 0.6 nM; bivQ,
K
d = 5.8 ± 1.6 nM). However, the minigastrin peptide with a spacer of six uncharged
d-glutamines (dQ-MG-754,
[
20]) revealed even a higher binding affinity (
K
d = 1.8 ± 0.6 nM,
[
20]). This is an indication that this spacer is slightly advantageous in terms of affinity compared to a spacer with a combination of three
d-glutamines and three
d-glutamic acids (QE) or a spacer with three
d-glutamines and a bivalent peptide construct (bivQ).
Surprisingly, no increased affinity was detected for the bivalent probe bivQ, albeit multimeric variants of small molecule probes have been reported to be of superior target affinity before
[
31]. For example, an affinity increase was reported for RGD peptides for nuclear medicine applications, which was proportional to the number of oligomerically combined RGD units
[
32,
33]. Also, a radiolabelled bivalent minigastrin, different from the one in this paper, showed a sixfold higher tumour uptake compared to its monovalent variant
[
19]. A possible influence of the dye moiety on affinity is discussed below.
Possibly, an increase in affinity by multimerization of CCK2R-targeted peptides can only be achieved by following defined basal design principles which consider that a multivalent probe has more than one possibility to bind to the target structure. It might be advantageous to choose a low-affinity minigastrin for multimerization because this increases the possibility that more than one binding event per molecule can occur before internalization prevents further multivalent interactions. Alternatively, the use of non-internalizing minigastrin antagonists might also provide a high potential to increase affinity for multivalent high-affinity probe constructs
[
34].
Regarding the influence of the dye conjugation on probe affinity, previously published results on analogous minigastrin variants as radioligands revealed dissociation constants in a comparably low nanomolar range. Aloj et al. 2011
[
13] reported a
K
d value of 9.0 ± 4.2 nM for the radiotracer analogue of QE and 4.8 ± 4.2 nM for the monovalent radiotracer version of bivQ. Hence, no loss of affinity occurred through the conjugation of the fluorochrome DY-754, opposed to former optical probes, where affinity decreased after dye conjugation compared to unconjugated or radioactively labelled peptides
[
11,
35,
36]. This is ascribed to the particular features of the fluorochrome DY-754, such as high hydrophilicity, low non-specific binding to tissue and renal elimination properties
[
20]. Therefore, the dye conjugation complemented probe characteristics by sterical stabilization of the molecule construct.
As we could show, the optical probes QE and bivQ bound specifically to CCK2R-expressing cells. Moreover, they were actively internalized at a temperature of 37°C but restricted to the cell membrane at 4°C. Thus, internalization behaviour was not qualitatively influenced by the different spacers and was not impeded by the dye conjugation. Successful internalization of ligand-receptor complexes has been described as a hallmark for successful tumour targeting, since retention time is prolonged and cumulative accumulation leads to signal multiplication in tumour tissue
[
37], while non-internalized probe molecules are cleared from blood and tissues
[
38,
39]. These effects ultimately lead to an augmented sensitivity.
The probe QE was evaluated to be highly stable against degradation by liver proteases. After systemic application into the blood system where the protease density is lower than in liver tissue
[
17], the probe is likely to show up a comparatively higher stability. The probe bivQ, however, elicited only a limited metabolic stability in the presence of liver proteases. This is an indication that the metabolic stability of the probes was sufficient to reach the target site in its original constitution in our
in vivo experiments, where we demonstrated specific targeting of CCK2R-expressing tumours. Compared to a previously published optical minigastrin probe
[
20] (Additional file
4), the negatively charged
d-glutamic acids of the probe QE strongly improved enzymatic stability, whereas the bivalent constitution seems to increase the vulnerability of the molecule to proteases. Furthermore, the long spacer of six hydrophilic amino acids in QE seemed to be superior over the shorter three amino acid spacer in bivQ.
The high affinity binding of the probes QE and bivQ to the CCK2R in combination with internalization and hence signal amplification is an important prerequisite for low dosage for the prospective detection of CCK2R-expressing tumours in the clinical application due to a high contrast between malignant and healthy tissue. Particularly for small molecule probes, a dissociation constant of 0.1 to 10 M is needed for maximal tumour accumulation. Small molecule probes quickly achieve high tumour concentration levels but require high affinity to be retained, as unbound molecules are cleared rapidly from the tumour
[
40]. Due to its comparatively high metabolic stability, the probe QE elicits better perspectives for CCK2R imaging probes over the probe bivQ.
In vivo imaging capabilities
In vivo small-animal imaging revealed that the new optical probes QE and bivQ were capable of high-contrast imaging of CCK2R-expressing xenografts (TBR > 6). They showed a strong specificity of tumour accumulation. The corresponding TBRs were around one-third higher TBRs than that of the previously published probe dQ-MG-754 (TBR > 4,
[
20]) but a comparably weak non-specific accumulation in A431/WT tumours (TBR < 2). Hence, both spacer modifications, the introduction of negatively charged
d-glutamic acids in QE and the bivalency of bivQ, increased TBRs.
In terms of specificity, competitive conditions for CCK2R binding diminished the TBRs significantly in both cases, for QE and bivQ, by 30% to 40%, a high value for an only 10-fold excess of unlabelled peptide. Compared to dQ-MG-754
[
20], where TBRs were reduced by 50%, competition was less complete for QE and bivQ, indicating elevated unspecific binding
in vivo; the reasons of which remain to be identified. However, this might partly be due to the lower initial TBR of dQ-MG-754 of around 4 (compared to > 6 for QE and bivQ).
Importantly, we observed an absence of fluorescence signals in the gastrointestinal tract for QE and bivQ, resulting from the hydrophilicity-induced renal elimination pattern. The strongly negative ALogP values further support the high hydrophilicity of the molecules. A former optical minigastrin probe with a spacer of hydrophilic glutamic acids, but the less hydrophilic fluorescent dye DY-676, displayed predominantly a hepatobiliary elimination
[
41], underlining once more that probe characteristics are always a summation of peptide and dye properties and that selection of the dye can fundamentally change probe characteristics. With consideration of fluorescence-guided endoscopy, hepatobiliary elimination could lead to strong non-specific fluorescence signals in areas contaminated by faeces remaining after incomplete colonic irrigation, which would obscure specific fluorescence signals
[
36].
The renal elimination pattern that was successfully achieved by the probe design was accompanied by prominent renal retention. This was unexpected, since the radioligand variants of the probe QE as well the dye DY-754 showed a renal elimination as well but only a minimal kidney retention
[
20,
23]. Therefore, we assume that the combination of the four negatively charged sulfo groups of the fluorochrome DY-754 together with the hydrophilicity of the peptides fostered their kidney retention. Several studies have shown that an alteration of positively or negatively charged amino acids can influence the renal uptake pattern of peptides
[
23,
30,
42]. Possibly, after glomerular filtration followed by receptor-mediated endocytosis (e.g. megalin receptor) in the proximal tubules, the probe constructs exhibit different accessibility to lysosomal degradation and hence different retention times
[
43]. The observation that the renal fluorescence of the probe QE was twice as high as that from the probes bivQ and dQ-MG-754
[
20] implies that additional
d-glutamic acids are degraded slower than
d-glutamines. As long as areas other than the kidney are investigated, kidney retention will not negatively influence the imaging process. The biocompatibility of the optical minigastrin dQ-MG-754 has been already proven, where kidney cell growth
in vitro was not diminished by the exposure to high concentrations of the probe. Accordingly, therefore, nephrotoxicity is unlikely to occur
in vivo
[
20].
Our investigations have unveiled the influences of the probe constitution on imaging properties. In detail, the d-glutamines influenced especially affinity, specificity and biodistribution in a positive way. Additional d-glutamic acids can be applied to increase tumour-to-background contrast and metabolic stability, as shown here for the probe QE. Bivalency of two high-affinity minigastrins did not lead to an enhanced affinity but decreased stability. Whether these results are of general significance for peptide probes or are confined to optical minigastrins only should be elucidated in the future using other optical peptide constructs, for example for bombesin or somatostatin probes. Today, a large variety of near-infrared dyes are available, which can also be customized to possess a defined number of negative or positive charges. Hence, by selection of a defined dye and spacer, affinity and pharmacokinetic properties of small molecule probes can be selectively influenced.