The building blocks required for the synthesis of tubulysin derivatives can be purchased commercially or synthesized on gram-scale using standard protocols (Murray et al., 2015; Wang et al., 2019; Shankar et al., 2013; Wipf et al., 2004). However, the assembly of some of these blocks using solution-phase protocols can be challenging due to the steric hindrance of some of them, especially in the coupling step leading to the tertiary amide formation (Peltier et al., 2006; Balasubramanian et al., 2009). We recently developed a solid-phase protocol that enables the construction and multicomponent derivatization of the tubugi skeleton, including the combinatorial diversification of the Ugi-derived N-substituent (Ricardo et al., 2024). We applied this protocol to synthesize tubugi derivatives with N-substituents bearing a variety of functional groups suitable for conjugation, such as carboxylic acids, amines, and alcohols. However, their introduction led to a significant drop in the anti-proliferative activity, thus limiting their use as PDC payloads. As shown in Scheme 1, here we extend this strategy to generate novel tubugis functionalized with a thiol group at the N-substituent, thus paving the way for the subsequent conjugation to a homing peptide without overly reducing the payloads activity.

The synthetic strategy for tubugi analogs begins with the preparation of the diazo resin 1 following a reported protocol (Bhalay and Dunstan, 1998; Lévesque et al., 2017). The resin can be stored at −20°C until further use. A portion of this resin equivalent to a synthesis scale of 0.05 mmol was used to quantitatively incorporate the Tup amino acid 2 without the need to use an excess of this valuable building block. After incorporation of the Tuv residue 4 using standard coupling protocol, the dipeptide was deprotected for the two-step multicomponent protocol. The first step involved an organocatalytic transimination reaction that allowed the quantitative formation of the resin-linked imine 6, which was subjected to the Ugi-4CR by treatment with thiol-functionalized isocyanides 7 and the Fmoc-protected Ile under microwave irradiation (Rivera et al., 2021; Ricardo et al., 2018). The two isocyanides were synthesized with a thiol moiety protected with an acid-labile group (see the Supplementary Material), bearing geminal methyl groups at the α position or not. Because the final payloads were designed to form a disulfide bond as a linkage to the targeting peptide, the α-substitution was chosen to influence the conjugate stability. The two geminal methyls are known to form rather stable disulfide bonds that only get cleaved in the presence of high concentrations of free thiols like glutathione or similar reductants found under the reducing conditions in cancer cells and solid tumors. After the on-resin Ugi-4CR, N-methyl-D-pipecolic acid 9 was coupled and the resulting tubugis 10a and 10b were obtained in good crude purity after acidic cleavage from the resin. A final HPLC purification step rendered the pure thiol-functionalized tubugis in good isolated yields considering the number of synthetic steps implemented in the solid-phase protocol.

Tubulysin analogs 10a and 10b were tested for anti-proliferative activity on the human prostate cancer cell line PC-3 and the colon cancer cell line HCT-116. As illustrated in Scheme 1B, both 10a and 10b inhibited the proliferation of the cancer cell lines under investigation with IC₅₀ values in the low nanomolar range. Several remarks regarding the activity of the compounds were noted: i) the non-methylated analog 10a showed a 2-fold lower anti-proliferative activity toward the two cell lines than the α-di-methylated analog 10b (Scheme 1B); ii) the prostate cancer PC-3 cells were found to be slightly more sensitive towards both tubugi analogs than the colorectal cancer HCT-116 cells, and iii) the compounds’ effect on PC-3 cells seems to be rather cytostatic (lower curve plateau does not reach the 0% level of cell viability), whereas the effect on HCT-116 cells appears to be cytotoxic, reaching nearly 0% cell viability.

To design bombesin targeting peptides suitable for PDC applications, structural modifications are required to enable the conjugation without compromising the peptide’s receptor binding and activation. In this context, it has been described that a N-terminally truncated bombesin analog [Bn (Faintuch et al., 2008; Chen et al., 2014; Worm et al., 2020; Abd-Elgaliel et al., 2008; Ohki-Hamazaki et al., 2005; Begum et al., 2016; Mansi et al., 2021; Steinmetz et al., 2013; Bouchard et al., 2014): Asn-Gln-Trp-Ala-Val-Gly-His-Leu-Met] mimics the full-length bombesin’s ability to activate bombesin receptors, particularly the human bombesin-binding gastrin-releasing peptide receptor (GRPR) (Begum et al., 2016). This study also proved that adding a Lys residue at the N-terminus of this truncated sequence permits the subsequent derivatization at the Lys ɛ-amino group without affecting the internalization capacity. Thus, by tagging this Lys residue with a fluorescein label, significant agonist activity and high levels of internalization were observed, particularly in prostate cancer cells (Begum et al., 2016).

To validate this [K⁵]-Bn(6–14) sequence as a targeting peptide, we synthesized various truncated bombesins and compared their internalization in the prostate cancer PC-3 cells by conducting a flow cytometry analysis. Our approach was influenced by similar studies on other targeting peptides, e.g., the human pancreatic polypeptide (hPP), (Mäde et al., 2014), for which enhanced cell internalization was achieved by increasing the hydrophobicity through attachment of a fatty acid. Given that multicomponent reactions are powerful approaches for the one-pot incorporation of various functionalities, we relied on a multicomponent strategy for the on-resin derivatization of the bombesin analogs with a lipid and a fluorescent label simultaneously. To minimize a potential negative impact of a large hydrophobic label (e.g., fluorescein) on the ligand-receptor binding and cell internalization, we chose to incorporate the considerably smaller 7-nitro-benzofurazan (NBD) fluorophore.

As depicted in Scheme 2A, the truncated sequence Bn(6–14) 11 was assembled on a Tentagel resin (0.05 mmol scale), followed by the incorporation of Boc-Lys(Fmoc)-OH using the standard SPPS methodology to afford the resin-bound [K⁵]-Bn(6–14) 12. The non-lipidated analog 13 was obtained in 32% overall yield after a final reaction of NBD-Cl with the ɛ-amino group of Lys and subsequent standard cleavage with TFA/TIPS/H₂O (94:3:3). For the incorporation of both the fatty acid and the fluorescent label, we employed an on-resin Ugi-4CR comprising the aminocatalytic transimination approach followed by reaction of the resulting imine with various fatty acids and isocyanide 14, bearing an orthogonally (Alloc) protected amino group (Rivera et al., 2021; Ricardo et al., 2018). The parallel incorporation of four fatty acids of different chain lengths (C14, C16, C18, and C20) followed by Alloc deprotection, incorporation of the NBD label and cleavage from the resin led to bombesin analogs 15a, b, c, and d in good overall yield after HPLC purification (see the Supplementary Material).

The cellular internalization of the NBD-labeled bombesin derivatives without (13) and with lipid tails of varied lengths C14-C20 (15a, 15b, 15c, and 15d) into GRPR-expressing PC-3 prostate cancer cells were semi-quantified by flow cytometry. Trypan blue quenching of outer cell surface fluorescence was done immediately before and while measuring, aiming to inspect as much as possible truly intracellular, hence internalized fluorescence of the labeled bombesin analogs (Rennert et al., 2009). The results of the flow cytometry measurements are illustrated in Scheme 2B and Supplementary Figure S46. As shown, the non-lipidated bombesin analog 13 showed only a poor cell internalization within the 2 h of treatment, as indicated by the just slightly elevated fluorescence count (RFU 4.2 compared to 3.5 of untreated control cells, this base value is due to their autofluorescence) caused by the intracellularly located peptide. On the other hand, all lipidated analogs were internalized better than the non-lipidated one, even within a short treatment time of only 2 h. Interestingly, a reverse correlation between the alkyl chain length and the internalization potency was observed, making 15a (C14 chain) the most efficiently internalized bombesin analog with a 43.7-fold higher intracellular fluorescence compared to 13, followed by 15b (C16), 15c (C18), and 15d (C20) with decreasing 36.5-fold, 20.7-fold, and 12.2-fold higher intracellular fluorescence as compared to 13. Previously, other groups have reported a similar trend for membrane-active lipopeptides, where C6–C12 acyl chains improved the lipopeptide interaction with the cell membranes better than longer and shorter lipid tails (Toniolo et al., 1996). As summarized in Supplementary Figure S46, the time dependency of the bombesin analogs’ cellular internalization was investigated as well, indicating a fast cell internalization process for all lipidated bombesins, with substantial internalization already within the first minute of treatment, reaching the majority of internalized ligand after just 1 hour, and with little increase to a maximum within the second hour. Such a fast cell internalization kinetics is typical for several G protein-coupled receptor peptide ligands. For example, the data of another GPCR-ligand system, the neuropeptide Y Y₁ receptor (NPY1R), also shows the GPCR rapidly internalizing the peptide ligand NPY or agonistic analogs within minutes. (Gicquiaux et al., 2002; Babilon et al., 2013). However, for the GRPR receptor, at least in our hands, fast internalization is only achieved with the lipidated peptide analogs, and not with the non-lipidated ligand 13.

Disulfide bond formation is widely used in the field of peptide derivatization, especially in dimerization processes and side chain crosslinking involving Cys residues (Danial and Postma, 2017; Deng et al., 2020). However, the formation of asymmetric disulfide bonds is of greater synthetic challenge and often requires thiol exchange in electrophilic disulfides, typically using either Elman’s reagent or 2-pyridyl disulfides (Bernardes et al., 2012; Pillow et al., 2016). As shown in Scheme 3A, the bombesin peptide was functionalized at the Lys side chain using 2-pyridyl disulfide containing a carboxylic acid (Beck et al., 2017) derived from 3-mercaptopropionic acid. The integrity of the 2-pyridyl disulfide component remained unaffected by the acidic conditions used during cleavage from the resin, leading to the activated bombesin analog 18 in 82% overall yield and 87% crude purity without the need for HPLC purification. Because of the superior internalization performance of the lipidated bombesin in cancer cells, i.e., derivative 15a with a C14 chain, we aimed to functionalize the truncated Bn with both a 2-pyridyl disulfide linker and a C14 lipid chain. For this purpose, lipidic isocyanide 17, paraformaldehyde and linker 16 were simultaneously incorporated at the Lys side chain using the Ugi-4CR (Scheme 3A), yielding bombesin derivative 19 in 17% isolated yield after RP-HPLC purification.

Integrating the activated disulfide into the bombesin peptide streamlined the solution-phase coupling with the thiol-functionalized tubugi. As depicted in Scheme 3B, bombesin analog 18 was ligated to the thiol-tubugi 10a in MeOH/DIPEA, with full consumption of starting materials after 2 h, leading to bombesin-tubugi conjugate 20 in 46% yield after purification. Following a similar conjugation protocol, conjugate 21 was produced in 52% isolated yield, while the preparation of the lipidated bombesin-tubugi conjugate required 12 h to reach complete consumption of the starting materials, ultimately giving 22 in 41% isolated yield.

The bombesin-tubugi conjugates 20, 21, and 22 were tested on human cell lines representing various receptor expression levels of GRPR, a human receptor with a high affinity for bombesin. We sought to investigate the correlation between the anti-proliferative efficacy of the conjugates and the GRPR expression levels of the cells, hypothesizing a better anti-proliferative effect in the cells with the higher expression level. For this purpose, a panel of human, primarily cancer, cell lines was initially selected to encompass a broad spectrum of GRPR expression levels. The selection process was based on differences in mRNA expression levels of GRPR as documented in the publicly available RNA-Seq database, GRCh38.p12 (O’Leary et al., 2024). Differential expression visualization was performed using the Genevestigator® gene expression analysis tool (Hruz et al., 2008), aiming to cover a wide range of GRPR expression levels. Accordingly, the following cell lines were selected for testing: T-47D (HR+ breast cancer), MDA-MB-231 (triple-negative breast cancer, TNBC), PC-3 (prostate cancer), MCF-10A (non-cancer, “healthy” breast epithelium), and HCT-116 (colorectal cancer). Of notice, T-47D has been reported with relatively high, MDA-MB-231 and PC-3 with medium, and HCT-116 and MCF-10A with relatively low GRPR expression levels (decreasing in the order of that list).

The actual GRPR expression levels of these cell lines were checked by conducting RT-qPCR, as described in the SI. Indeed, as shown in Figure 2A, the five cell lines ranked in the proposed order described above. The cell viability and proliferation assays were carried out by using a resazurin-based fluorometric assay. The cells were exposed to the bombesin conjugates 20, 21, and 22 for initial 6 h and 48 h, and regardless of the initial incubation time, the cells were then allowed to grow until 72 h in sum were finalized, after which the cell viability was determined (see the SI).

The results of the cell viability assays indicating the anti-proliferative and cytotoxic effects of the bombesin-tubugi conjugates are shown in Figures 2B–D (conjugate 20–22, respectively) for the initial 6 h treatment, and in the SI (Supplementary Figures S7, S48, S49) for the 48 h treatment. Because of the rather short metabolic half-lives of peptides in vivo, usually minutes up to just a few hours, the 6 h initial treatment was thought to be of higher relevance for potential anticancer therapeutics. The 48 h treatment was intended to show a proposed maximum effect of the sample. As illustrated in Figure 2B, after only 6 h initial cell treatment, conjugate 20 showed an anticancer activity in the high nanomolar range, as proven by an IC₅₀ ∼ 600 nM in T-47D breast cancer cells, and just slightly higher IC₅₀ values in PC-3 prostate and MDA-MB-231 TNBC cancer cells. Furthermore, IC₅₀ values approximately 2-fold and 5.5-fold higher were detected in the GRPR-lower expressing HCT-116 (IC₅₀ ∼ 1.3 µM) and GRPR-lowest expressing, healthy MCF-10A cells (IC₅₀ ∼ 3.2 µM), respectively. Although the anticancer activity of conjugate 20 is not remarkable, the data illustrate a direct correlation between the cells’ GRPR expression levels and the anti-proliferative efficacy. The results of conjugate 21 after 6 h initial treatment (as shown in Figure 1C) are quite comparable to those described above for conjugate 20. The IC₅₀ values of 21 are slightly higher in the different cell lines than for 20, except for the MDA-MB-231 cell line. In this latter cell line, conjugate 21 showed an IC₅₀ ∼ 360 nM, whereas the IC₅₀s in the higher expressing T-47D and PC-3 cancer cells were determined to be 741 nM and 716 nM, respectively. However, compared to these values, IC₅₀ values of 21 in the lower and lowest GRPR-expressing HCT-116 cancer and MCF-10A healthy breast cells were approximately 2.5- and 5.5-fold higher, respectively. Compared to the lowest IC₅₀ of 21 in MDA-MB-231 (∼360 nm), the IC₅₀s of the conjugate in HCT-116 and MCF-10A were even ∼10-fold higher.

Interestingly, in the case of both 20 and 21 treatments for the initial 6 h, a significant difference in IC₅₀s between the three highest and medium GRPR-expressing cell lines - i.e., T-47D, PC-3, and MDA-MB-231 – was not detected, which is in contrast to what was expected. Seemingly, there might be a certain GRPR expression level that allows a maximum targeting effect of the PDC, here IC₅₀ ∼ 600–700 nM after 6 h initial treatment, that is already reached at least with the medium GRPR expression level of MDA-MB-231 cells. This could imply that even far higher GRPR levels would not have a substantial benefit for the targeting and thus selective cytotoxic effect.

After the cell treatment for 48 h with both conjugates 20 and 21, all IC₅₀ values were measured in the nanomolar range, as well as those against the lower GRPR-expressing cell lines HCT-116 and MCF-10A. The reason for this could be the time that cells have to import the conjugates via GRPR-mediated endocytosis in multiple receptor internalization and recycling cycles. This could also be the reason for the lack of a clear correlation between GRPR expression levels and the conjugates’ anti-proliferative impact after 48 h initial incubation, as indicated by a difference of the lowest (in T-47D; IC₅₀ ∼ 200 nM) and highest IC₅₀ (in HCT-116; IC₅₀ ∼ 570 nM) of at most ∼2.5-fold. This has, of course, consequences for the cancer cell selectivity of the bombesin conjugates, and should be kept in mind in future developments of bombesin-toxin conjugates. Considering the high similarity of the conjugates 20 and 21 in terms of chemical structure, in vitro cytotoxicity (IC₅₀ values) and selectivity, the intended stabilization of the disulfide bridge by the dimethylation in 21 seems to have only a very modest effect on the PDC efficacy, at least in vitro.

As illustrated in Figure 1D (see also Supplementary Figure S49) the C14-lipidated conjugate 22 was 4- to 10-fold more potent in three cancer cells overexpressing the GRPR than the non-lipidated conjugates 20 and 21, respectively, with IC₅₀ values in the range of ∼70–110 nM. We also conducted a direct comparison of conjugate 22 and tubugi 10b with an initial treatment of the cells for 6 h, as it makes sense to compare a PDC with its free payload. Besides the cancer cells, both compounds were tested (Supplementary Figure S50) in another healthy breast cell line, 184B5, which has a GRPR (mRNA) expression level that is comparable to that of MCF-10A, according to the Genevestigator^® gene expression analysis tool (Hruz et al., 2008). As summarized in Table 1, the bombesin conjugate 22 exhibited a higher cytotoxicity than free tubugi 10b in the three cancer cells, but compound 10b also proved to be three times more toxic than conjugate 22 in the 184B5 healthy cells. In this case, the lack of selectivity of tubugi 10b among the cell lines tested is expected, given that this small compound can have a similar internalization efficiency in all cells after a short initial treatment of 6 h. It is also noteworthy the selectivity of conjugate 22, which exhibited a 3.5- to 5.5-fold higher cytotoxicity for cells overexpressing the GRPR receptor (Figure 1D) than for healthy 184B5 cells (Supplementary Figure S50). While this level of selectivity is not yet satisfactory for therapeutic applications, it is a clear indication of what can be achieved using this strategy.

While this study represents the first report on the development of bombesin-derived PDCs incorporating tubulysin analogs, we acknowledge that our approach has some limitations derived from both the type of receptor targeted and the methodology employed (Heh et al., 2023). For example, the initial cytotoxicity evaluation of the tubulysin analogs 10a and 10b was conducted using only the PC-3 and HCT-116 cells, with 48 h of initial treatment, as part of a general screening program in our laboratory. Accordingly, these results have limited comparability to the data obtained for the PDCs 20, 21 and 22, for which additional cancer cells and an initial treatment of 6 h were employed (as recommended for PDC evaluation). Fortunately, we were able to compare PDC 22 and its parent unconjugated payload 10b using 6 h of initial treatment and the same cell lines, which provided valuable information regarding the selectivity achieved by the bombesin PDC compared to the free toxin (discussed above). A further limitation is the lack of sufficient data concerning the role of lipidation on the PDC behavior. In addition to addressing PDC disadvantages such as poor metabolic stability and rapid renal clearance, lipidation also increases the PDC’s interaction with the cell membrane. The impact of this can be beneficial or detrimental, depending on whether it increases the PDC concentration on the cell membrane where the receptors are localized or also boosts an undesired, non-receptor-mediated internalization process. We did a comparative internalization analysis with the labeled bombesin peptide bearing various lipid chain lengths or no lipid chain, but assessing the internalization behavior of the final PDCs on different cell lines is further required.

Another limitation can be associated with the use of GRPR as target, since its level of overexpression in specific cancer cells might be insufficient to achieve a suitable selectivity index. The GRPR is known to be overexpressed in various cancer cells (D’Onofrio et al., 2023; Belge et al., 2024; Zhang et al., 2024), which was corroborated by our RT-qPCR data, but - like other GPCRs - it is not absent in healthy cells and tissues. It is also known that agonistic GPCR activation results in a subsequent desensitization and downregulation of the receptor, albeit with a certain delay. This also applies to the GRPR targeted by our PDCs (Benya et al., 1994). Furthermore, we don’t have data demonstrating the agonistic behavior of our PDCs or the lipidated bombesins used in the internalization study. The flow cytometry analysis in PC-3 cells of the bombesin derivatives (Scheme 2) indicated a fast and efficient cellular internalization of the lipidated bombesins, but other internalization mechanims cannot be ruled out. Overall, as expected, lipidated derivatives showed better internalization and thus higher bioactivity (lower IC₅₀s). Indeed, new experiments focus on assessing both the capacity of our PDCs to functionally activate GRPR and the effectiveness of a receptor-mediated payload delivery into the cancer cells. This will help either to confirm or exclude alternative mechanisms of internalization for the lipidated PDC.