A non-internalised CD38-binding radiolabelled single-domain antibody fragment to monitor and treat multiple myeloma

Non-targeting control sdAbs cAb-BcII10 and R3B23 were generated as described before [13]. To obtain the four hexahistidine (His₆)-tagged sdAbs (i.e. #551 [5], #375 [5], #1053 [5] and #2F8 [14]), their genes were cloned into the expression vectors pHEN2 or pHEN6 and subsequently transformed into the E. coli WK6 cells [15]. These His-tagged sdAbs were purified by metal chelate affinity chromatography [10]. Residual imidazole was removed by gel filtration (Sephadex G25). A stop-codon was introduced by mutation before the His-tag coding region to produce untagged nanobodies #551 and #2F8. Untagged sdAbs were produced and subsequently purified by combining ionic exchange (high TRAP Q HP and Capto S resin) and size exclusion chromatography (Superdex 75) in 50 mM HEPES, 150 mM NaCl. The purity and integrity of sdAbs were evaluated by SDS-PAGE and by mass spectrometry analysis.

The human MM cancer cell line RPMI 8226 (GFP⁺/Luciferase⁺) was obtained from A. Martens (Haematology department, VU University Medical Center, The Netherlands). RPMI 8226 cells were cultured using Roswell Park Memorial Institute medium (RPMI 1640, Lonza, Belgium) enriched with 10% foetal bovine serum, 100 U/mL penicillin and 0.1 mg/mL streptomycin. K562 cells were purchased from ATCC (Manassas, Virginia, US), double-hit lymphoma cell line LB5871-LYMP was obtained from N. Van Baren (UCL, Brussels) [17], and finally, U266, OPM2 and LP-1 cell lines were obtained from H. Jernberg-Wiklund (Uppsala University, Sweden), respectively.

The CD38 knockout (KO) cells used as specific controls were obtained by genetic engineering using the CRISPR/Cas9 technique as previously described [18].

Tumour xenografts were obtained by subcutaneously (s.c.) inoculating 6- to 12-week-old NOD.Cg-Prkdc^sci Il2rg^tm1Wjl1wʲˡ / SzJ (NSG) mice with RPMI 8226 cells (0.5 × 10⁶) in 150 μL PBS-Matrigel (ratio 1:1) into the left hind. Mice were irradiated with 2 Gy 24 h before cell injection to favour a homogenous tumour uptake. Tumour development was followed from day 14 (palpable from day 19) using an IVIS Spectrum In Vivo Imaging System (PerkinElmer). Briefly, after s.c. injection of D-luciferin (150 mg/kg), anaesthetised mice were imaged (scan and reading of luminescence emitted during substrate degradation by cancer cells expressing the luciferase). All animal experiments were approved by the Ethical Committee for Animal Experiments of the University of Liège and of Vrije Universiteit Brussel.

Anti-CD38 sdAbs were radiolabelled with ^99mTc at their C-terminal His₆-tag using the Isolink labelling kit (Mallinckrodt Medical BV), as previously described [19]. ^99mTc-sdAbs were separated from unreacted [^99mTc(H₂O)₃(CO)₃]⁺ via NAP-5 size exclusion chromatography (Sephadex, GE Healthcare). The eluate was passed through a 0.22 μm filter (Millipore), and radiochemical purity (RCP) was determined by instant thin-layer chromatography (iTLC) using acetone as running buffer (iTLC-SG, Pall Corporation) and was > 95%. Mice bearing s.c. CD38⁺ RPMI 8226 tumours (n = 3) were intravenously (i.v.) injected with ^99mTc-labelled sdAbs (about 4 μg and 51.1 ± 3.7 MBq). At 1 h post-injection (p.i.), mice were imaged using pinhole single-photon emission computed tomography (SPECT)/ computed tomography (CT) with a Vector + /CT MILabs system [20]. Images were obtained using a rat SPECT-collimator (1.5-mm pinholes) in spiral mode, 7 positions with 128 s per position for whole-body imaging. Images were reconstructed with 0.4 mm³ voxels with 2 subsets and 2 iterations, without post-reconstruction filter. For CT, a normal scan mode of two positions was used. Images were fused and corrected for attenuation based on the CT scan. Image analysis was performed using a Medical Image Data Examiner (AMIDE) software and data expressed as % injected activity per cm³ (% IA/cm³) [21]. After imaging, mice were killed, and organs and tissues were isolated and weighed. The radioactivity in each sample was measured using a Wizard2 γ-counter (PerkinElmer). Tracer uptake was expressed as % injected activity per gram organ (% IA/g).

Untagged #2F8 and R3B23 (3 mg/mL) reconstituted in 50 mM sodium carbonate buffer pH 8.5 were reacted with a tenfold molar excess of bifunctional chelator CHX-A’’-DTPA (Macrocyclics) for 3 h at RT, as described previously [22]. The conjugation reaction was quenched by reducing the pH of the mixture to pH 7.0. DTPA-2F8 was purified on Superdex 75 10/300 GL (GE Healthcare) in 0.1 M ammonium acetate buffer pH 7.0. The necessary amount of ¹¹¹In (54 to 270 MBq) or ¹⁷⁷Lu (74 MBq to 1 GBq) was incubated for 30 min at RT or 37 °C, respectively, with the DTPA-2F8 in 200 mM ammonium acetate pH 5. Resulting ¹¹¹In- and ¹⁷⁷Lu-DTPA-2F8 were purified on NAP-5 column (GE Healthcare) using 0.9% NaCl as eluent. In case of high radioactive labelling for preclinical therapy, 0.9% NaCl with 5 mg/mL ascorbic acid was used. The resulting radio-conjugates were filtered on 0.22 μm after which RCP was determined by iTLC with citric acid as running buffer (iTLC-SG, Pall Corporation) and measured > 96%.

Binding affinity and degree of internalisation of ¹¹¹In-DTPA-2F8 were evaluated as previously described [23]. In case of assessing binding affinity, CD38⁺ RPMI 8226 cells were incubated for 1 h at 4 °C with a serial dilution of ¹¹¹In-DTPA-2F8 (0.1–300 nM) alone or in combination with a 100-fold molar excess of unlabelled 2F8 to assess non-specific binding. Next, unbound sdAb was washed away, after which retained radioactivity was measured in a γ-counter. Data were plotted using GraphPad Prism software. To study the level of receptor-mediated internalisation, RPMI 8226 cells were incubated for 1 h at 4 °C with 10 nM of ¹¹¹In-DTPA-2F8, alone or in combination with a 100-fold molar excess of unlabelled #2F8 to assess unspecific binding. After washing, the cells were incubated at 37 °C up to 24 h. At different time points during incubation, cells were processed to obtain the corresponding dissociated, membrane-bound and internalised fraction, as described before [23]. All fractions were measured for radioactivity in a γ-counter.

Mice bearing s.c. CD38⁺ RPMI 8226 tumours (n = 3) were i.v. injected with ¹¹¹In-DTPA-2F8 (about 18.8 ± 0.1 MBq) or R3B23 and 150 mg/kg gelofusin. SPECT images were taken at different time points up to 48 h p.i. Images were obtained using the same scan parameters used as described for ^99mTc acquisitions, except for the number of positions [6] and time per position for the full body imaging (150 s) and number of iterations (6 instead of 7). After the last image acquisition, mice were killed and processed as described above. In parallel, a few animals (n = 3) were i.v. injected (4.5 ± 0.2 MBq) with ¹¹¹In-DTPA-2F8 or ¹¹¹In-DTPA-R3B23, alone or as a co-injection with 150 mg/kg gelofusin. They were imaged 1 h post-injection in order to evaluate the effect of gelofusin on the renal retention of the tracer.

Mice bearing CD38⁺ RPMI 8226 tumours were i.v. injected with ¹⁷⁷Lu-DTPA-2F8 (about 4 μg and 4.5 ± 0.2 MBq) co-infused with 150 mg/kg gelofusin (n = 3). Next, mice were killed at different time points up to 48 h p.i., followed by the isolation of different organs and tissues. All samples were weighed and counted for radioactivity content against a standard of known radioactivity using a γ-counter and expressed as percentage of the injected activity per gram of tissue mass (% IA/g), corrected for decay. For dosimetry purposes, the biodistribution data were time-integrated to obtain the residence time per gram tissue [22]. Briefly, the area under the curve between 1 and 48 h was made using the trapezoid integration method. Next, the absorbed doses were calculated using S values for ¹⁷⁷Lu obtained from RADAR phantoms (Unit Density Spheres).

Day 23 post-cell inoculation, mice with palpable CD38⁺ RPMI 8226 tumours were randomly categorised into three treatment groups (n = 10). Mice received three consecutive i.v. administrations, once every two days, of either: (i) a high radioactive dose (about 18.5 ± 0.5 MBq), (ii) a low radioactive dose (9.3 ± 0.3 MBq) ¹⁷⁷Lu-DTPA-2F8 + 150 mg/kg gelofusin, or (iii) an equal volume of vehicle solution (PBS). Animals were followed up over time through body condition scoring (weight, physical appearance, mobility and behaviour), and tumour volume was assessed via calliper measurements and bioluminescence imaging as described above.

Flow cytometry experiments revealed the specific binding of the different anti-CD38 sdAbs. Indeed, all of them are capable of binding the CD38 antigen on MM cell lines (RPMI 8226 and LP1) and non-Hodgkin’s lymphoma cells (LB5871-LYMP), while no specific binding was observed on the CD38⁻ cells (K562 cell line) and CD38^KO cell lines (Fig. 1). Non-targeting cAb-BcII10 did not show specific binding to CD38. 2F8 binds CD38 with an affinity in the low (K_D 1.5 nM) nanomolar range as illustrated in Fig. 2. Apparent midpoint of the heat-induced unfolding (Tm*), ranged from 69 to 89 °C, were obtained for the studied sdAbs with the #2F8 protein showing the highest one (88.7 ± 0.3 °C). To determine potential competition for CD38 targeting between daratumumab and the different sdAbs, competitive biolayer interferometry (BLI) experiments were performed. Different levels of competition with daratumumab were defined: total, partial or null. No binding of sdAbs #1053 or #375 was detected on CD38 that was pre-incubated with daratumumab, and vice versa. In contrast, sdAb #2F8 still bound in full capacity, and sdAb #551 in part, to the extracellular domain of CD38 in the presence of daratumumab (Fig. 2). The set of data acquired for the other three sdAbs are summarised in Table 1.

The biodistribution and tumour-targeting potential of different ^99mTc-labelled sdAbs were assessed via pinhole μSPECT/CT imaging and dissection. The resulting uptake values in tumours and non-target organs are summarised in Table 1. Administration of ^99mTc-H₆-2F8 revealed a tumour uptake of 2.22 ± 0.47% IA/g compared to only 0.79 ± 0.11% IA/g for ^99mTc-H₆-cAb-BCII10, with good contrasts observed early after tracer administration (Fig. 3). In addition, ^99mTc-H₆-2F8 was found concentrated in kidneys and bladder, in line with the fast blood clearance of sdAbs [24]. Only low amounts of radioactivity were measured in blood (0.39 ± 0.09% IA/g) and non-targeted organs and tissues (from 0.02 ± 0.01 to 0.53 ± 0.12% IA/g) (Additional file 1: Table 1).

Based on the superior in vitro and in vivo characteristics, sdAb #2F8 was selected as lead CD38-targeting compound (Table 1). Untagged #2F8 was conjugated to p-SCN-Bn-CHX-A”-DTPA and was shown to maintain its binding characteristics as determined with radioactivity and BLI experiments (Figs. 2 and 4). After radiolabelling with ¹¹¹In, both the maximal effective concentration (EC₅₀) and the level of internalisation were assessed on RPMI 8226 cells. EC₅₀ for ¹¹¹In-DTPA-2F8 measured 16.7 ± 3.2 nM. ¹¹¹In-DPTA-2F8 bound well to CD38 + RPMI 8226 cells but did not induce receptor-mediated internalisation, as this fraction remained low at all time points until 24 h (Fig. 4, panel Ci). Flow cytometry confirmed the minor internalisation of sdAb #2F8 (Fig. 4, panel Cii).

Mice bearing CD38⁺ RPMI 8226 tumours were i.v. injected with either ¹¹¹In-DPTA-2F8 alone, ¹¹¹In-DPTA-2F8 in combination with 150 mg/kg gelofusin or ¹¹¹In-DTPA-R3B23. ¹¹¹In-DPTA-2F8 showed high and specific uptake in tumour 1 h after tracer injection. High levels of radioactivity were observed in kidneys (23.00 ± 2.62% IA/g) after 1 h, which was reduced by co-injection with 150 mg/kg gelofusin to only 13.00 ± 0.02% IA/g Fig. 4B and Additional file 1: Table 4. ¹¹¹In-DTPA-R3B23 revealed no relevant uptake in tumour (0.54 ± 0.14% IA/g), while the amount in kidneys was similar. Uptake of CD38-specific sdAbs in all other organs and tissues was low and considered as background (Fig. 4). A head-to-head comparison of uptake in tumour and kidneys between ^99mTc-H₆-tagged-2F8, ¹¹¹In- and ¹⁷⁷Lu-labelled untagged-2F8 is depicted in Fig. 5B. Removal of the hexahistidine tag and co-infusion with 150 mg/kg gelofusin reduced kidney retention by 86 and 57%, respectively. Uptake in tumour of ¹⁷⁷Lu-DTPA-2F8 co-injected with 150 mg/kg gelofusin measured about 4.5% IA/g and remained constant over time up to 48 h p.i. (Fig. 5A). The renal uptake peaked at 18% IA/g 1 h after administration and decreased to 2% IA/g at 24 h p.i. and about 1.2% IA/g at 48 h p.i. The uptake values over time for different organs and tissues were used for dosimetry calculations. Both tumour and kidneys received the highest absorbed doses per unit of activity, with values of 0.1 and 0.22 Gy per MBq of ¹⁷⁷Lu-DTPA-2F8 (Additional file 1: Table 7). Absorbed doses to additional organs and tissues measured below 0.006 Gy per MBq.

Therapeutic efficacy of ¹⁷⁷Lu-DTPA-2F8 was assessed in mice bearing CD38⁺ RPMI 8226 tumours. Mice received three i.v. injections of either high radioactive dose (18.5 ± 0.5 MBq) or low radioactive dose (9.3 ± 0.3 MBq) regimen of ¹⁷⁷Lu-DTPA-2F8, or an equal volume of vehicle solution, which led to cumulative absorbed doses to kidneys and tumour of 6.21 and 2.77 Gy, and 12.31 and 5.50 Gy for the low- and high-dose groups, respectively. Figure 6A illustrates tumour volume evolution for all three treatment groups. The direct comparison of tumour volumes at day 43 after tumour inoculation revealed a dose-dependent tumour reduction for both dose regimens compared to vehicle solution (Fig. 6C). These changes were confirmed by similar reduction in tumour burden, quantified by bioluminescence (not shown). The median survival (Fig. 6B) of vehicle-treated mice was 43 days, compared to 62 days for the low radioactive dose regimen (p = 0.027) and 65 days for the high radioactive dose regimen (p = 0.0007).