Synthesis of Heterobifunctional Protein Fusions Using Copper-Free Click Chemistry and the Aldehyde Tag

Heterobifunctional protein fusions are gaining interest as next-generation biopharmaceuticals.1–5 Combining proteins with disparate functions can enable multidrug therapy with a single chemical entity,6, 7 add a targeting element to an otherwise nonspecific therapeutic,8, 9 or improve the pharmacokinetic profile of a rapidly cleared molecule.10, 11 Indeed, heterobifunctional proteins, such as immunoglobulin G (IgG) Fc domain fusions, are among the top-selling biotherapeutics on the market today.12 These biomolecules are primarily generated as genetic fusions. The DNA sequences that encode the individual protein components are fused in tandem to direct the expression of a single polypeptide that comprises the two proteins joined together at their N and C termini, respectively. However, this limited topology is not ideal for every protein combination, as some polypeptides require unmodified termini for optimal bioactivity13 or can suffer from expression difficulties as a result of folding and processing issues.3, 14, 15

. Linkers functionalize aldehyde-tagged hIgG with azide for subsequent Cu-free click chemistry. SDS-PAGE analysis of 5 µM hIgG reacted with AO-AF488 or AO linkers 1-3 (pH 4.5, RT, 18h) followed by DIBAC-488 (37 ºC, 1h). Top, fluorescent scan; bottom, coomassie stain. Figure S4. SDS-PAGE and MALDI-TOF MS analysis of reactions with DIBAC-488. 30 µM hGH was reacted with DIBAC-488 after conjugation with increasing concentrations of linkers 1-3. Lanes 1-6, protein was reacted overnight with aminooxy linker in NaH2PO4/NaCit buffer pH 4.5 at 32 ºC followed by the addition of excess DIBAC-488 and reacted for an additional 16 h at 4º C. Lane 7, negative control reaction with only DIBAC-488 without linker conjugation. Lanes 8-10, conjugated protein was buffer exchanged to remove linker and then 15 µM of the azide tagged protein was reacted with only 2 equiv. of DIBAC-488 again for 16 h at 4º C. Aliquots were taken and run on SDS-PAGE or diluted with matrix solution for MALDI-TOF MS analysis. Figure S5. SDS-PAGE of aldehyde-tagged human serum albumin (HSA) reacted with DIBAC-488 with and without azide linker 1 conjugation. HSA was reacted at 25 µM with 1 mM linker 1 (pH 4.5, 35 ºC, 16h) and buffer exchanged into PBS. 5 µM of aldehydeor azide(1)-tagged HSA was reacted with DIBAC-488 (pH 7.4, 4 ºC, 20h). While robust labeling was seen with the linker 1 conjugated protein (lanes 3,4), little to no reactivity was found for the HSA containing no azide but still a free thiol (lanes 1,2). Top, fluorescent scan; bottom, coomassie stain.   Figure S4.    Figure S10. Western blot analysis of Cu-free click conjugations of hGH-Az and MBP-Az (linker 1) with hIgG-DIBAC (linker 4). Lanes 1-4, samples were reduced with βME to show specific labeling of the IgG heavy chain (IgG-HC); lane 4 includes reverse conjugate where linkers 1 and 4 were pre-click reacted and incubated with aldehyde tagged proteins to demonstarte lack of conjugation based on oxime formation alone; lanes 5-8, illustrate full-length conjugates containing one or two hGH/MBP molecules per hIgG dimer. Top, ponceau stain; middle, blot probed with α-hIgG 647 and imaged by fluorescent scan; bottom, same blot probed with α-hGH or α-MBP and imaged via α-mIgG-FITC to highlight higher MW conjugates containing the hGH or MBP conjugates. Figure S11. Representative histograms of fluorescence (x-axis) vs % of total cell counts (y-axis) for SKOV3 cells incubated with aldehyde tagged α-HER2 hIgG without and with conjugation to hGH/MBP. Cells were treated with a 1/500 dilution of 5 µM click protein-protein conjugation reactions followed by α-hIgG-649 to label the aldehyde-tagged α-HER2 hIgG and mouse α-hGH or α-MBP followed by α-mIgG to label the presence of hGH/MBP.  Figure S11. Representative dot-plots of forward scatter (x-axis) vs side scatter (y-axis) for SKOV3 cells incubated (A) without aldehyde tagged α-HER2 hIgG conjugates and with (B) hIgG-hGH and (C) hIgG-MBP conjugates. No apparent abnormalities or gross cell death was observed throughout the sample set.

General materials
Synthetic reagents were purchased from Sigma-Aldrich, Acros, and TCI and used without purification unless noted otherwise. DIBAC-488, DIBAC-NHS ester, and sulfo-DIBAC-NHS ester were purchased from Click Chemistry Tools (Scottsdale, AZ). Aminooxy-FLAG peptide containing an N-terminal aminooxy acetic acid [1] was custom synthesized by New England Peptide (Gardner, MA). Linkers 2 and 3 were custom synthesized by Tandem Sciences Inc. (Menlo Park, CA). Anhydrous DMF and MeOH were purchased from Acros in sealed bottles; all other anhydrous solvents were obtained from an alumina column solvent purification system. All reactions were carried out in flame-dried glassware under N 2 unless otherwise noted. In all cases, solvent was removed by reduced pressure with a Buchi Rotovapor R-114 equipped with a Welch self-cleaning dry vacuum. Products were further dried by reduced pressure with an Edwards RV3 high vacuum. Lyophilization was performed on a LABCONCO FreeZone instrument equipped with an Edwards RV2 pump. Thin layer chromatography was performed with Silicycle 60 Å silica gel plates and detected by UV lamp or charring with p-anisaldehyde in acidic EtOH. Flash chromatography was performed using Silicycle 60 Å 230-400 mesh silica. HPLC was performed on a Varian system attached to a absorption detector using a C18 reverse phase column (5 µm, 250 x 4.6 mm, Agilent(Varian); Carlsbad, CA) for analytical or a C18 reverse phase column (8 µm, 250 x 21.4 mm, Agilent(Varian); Carlsbad, CA) for preparative purifications. HPLC solvents were A: ddH 2 O with 0.1% TFA and B: MeCN with 0.1% TFA. All 1 H and 13 C NMR spectra are reported in ppm and referenced to solvent peaks (1H and 13C). NMR spectra were obtained on Bruker AVQ-400, DRX-500, AV-500, or AV-600 instruments. High resolution electrospray ionization (ESI) mass spectra were obtained from the UC Berkeley Mass Spectrometry Facility an LTQ Orbitrap (Thermo Fisher Scientific). MALDI-TOF (Matrixassisted laser desorption/ionization (time-of-flight)) analysis was carried out on an Applied Biosystems Voyager DE-Pro machine.

b. Expression and purification of aldehyde-tagged proteins
Recombinant proteins produced in E. coli (MBP and hGH) were expressed as previously described. [1] In short, BL21(DE3) E. coli harboring both a pET plasmid containing the aldehyde tagged-protein and a pBAD plasmid containing the M. tuberculosis FGE were grown in LB media supplemented with 100 µg/mL ampicillin and 50 µg/mL kanamycin at 37º C. When OD 600 reached 0.5, arabinose was added at 0.25% and the culture was shaken at 37º C for an additional 1 h. The temperature was then reduced to 18º C for 1 h and IPTG (0.1 mM) was added with further shaking for 14-18 h. Cells were lysed by homogenization, and the His 6 -tagged proteins were purified using Ni-NTA-agarose beads (Qiagen) under standard purification procedures. MBP was eluted in 50 mM NaH 2 PO 4 , 300 mM NaCl, 250 mM imidazole, pH 7.4 and dialyzed into PBS with 10% glycerol for storage while hGH was eluted in 50 mM Tris, 500 mM NaCl, 300 mM imidazole, 10% glycerol, pH 7.5 with 1 mM DTT, TCEP, and methionine to protect against oxidation and frozen directly for storage at -20 ºC. Protein concentrations were determined using the calculated absorbances at 280 nm.
When required, proteins were treated with additional FGE in vitro to fully convert Cys-to-fGly if intracellular conversion was sub-optimal. Briefly, aldehyde-tagged proteins (0.5 mg/mL) were incubated in 50 mM Tris, 100 mM NaCl, 0.1-1 mM β-Me, pH 9 with 0.05 equiv. M. tuberculosis FGE [6] for 20 h at rt. Proteins were then purified by size exclusion or buffer exchanged by spin concentration and stored in PBS.
Human serum albumin (HSA) and full-length human IgG aldehyde-tagged proteins were generated similar to previously described procedures. [7] Vectors bearing the heavy and light chain were constructed and nucleotides encoding the aldehyde tag were inserted at the Cterminus using standard molecular biology techniques. Protein production was performed by Redwood Bioscience using CHO-S cells overexpressing human FGE following standard protein expression protocols. HSA was purified from clarified media by Ni-NTA-agarose beads and eluted with high imidazole following typical His 6 -tag purification procedures. Full-length human IgG was purified from clarified media by affinity chromatography using Protein-A agarose resin. The eluent was dialyzed into PBS and stored at -20 ºC. Cys-to-fGly conversion was assessed by standard addition on LCMS/MS of trypsinized IgG and demonstrated high conversion rates (Table S3). Table S3. fGly conversion calculated on C-terminally aldehyde-tagged hIgG after in vitro conversion with additional FGE and analysis by LCMS/MS after trypsinization. Conversion rate was quantitated by standard addition of known amounts of aldehyde tag peptide fragments.
Buffer pH: Aldehyde-tagged MBP was reacted at 25 µM with 400 µM AO-AF488 in 20 µL of buffer at 37 ºC for 18 h. Aldehyde-tagged hIgG was reacted at 20 µM with 500 µM AO-AF488 also in 20 µL of buffer at 37 ºC for 18 h. Buffer conditions are listed in Table S1 and made to the indicated concentration by dilution from a 10x stock in ddH 2 O. Upon completion, the reactions were quenched with the addition of 4 µL 1M Tris, pH 7.5 and a 4 µL aliquot was taken for SDS-PAGE analysis. The remaining labeled protein was diluted into PBS, pH 7.4 and washed 4 times by concentration to remove unconjugated dye. The degree of labeling (DOL, % conjugate) was calculated by ratio of dye/protein concentration determined by UV-Vis absorbance (NanoDrop; AF488 at 494 nm; protein at 280 nm). An 11% correction was subtracted from the A 280 based on the A 494 reading to account for the absorbance overlap of Alexa Fluor 488 at 280 nm. Gels were scanned for fluorescence by a Typhoon 9410 imaging system (Amersham) before Coomassie stain to access protein loading.
Time and temperature: 5 µM of aldehyde-tagged hIgG was treated with 200 µM AO-CF488A (Biotium) in 40 µL of 100 mM KOAc, pH 4.6 and incubated at rt or 37 ºC. At 14, 24, 40, and 62 h, a 5 µL aliquot was taken and quenched by the addition of 1 µL 1 M Tris, pH 7.5. Aliquots were kept at -20 ºC until analysis at which 4x loading dye was added and loaded onto SDS-PAGE. Relative fluorescence was measured using ImageJ imaging software and given as fluorescence band density ratios after normalizing to coomassie stain density to control for loading.
Reagent concentration: 10 µM aldehyde-tagged hIgG was incubated at rt for 16 h in 10 µL of 5% MeCN, 0.02% FA with varying concentrations of AO-FLAG. Upon completion, the reactions were probed by western blot (5% milk/PBST; anti-FLAG M2, Sigma, 1/1000; anti-mIgG-AP, Jackson Immunolabs, 1/1000) and visualized by Western Blue (Promega). d. Linker and oxime vs. Cu-free click conjugations DIBAC-488 was purchased from Click Chemistry Tools and kept as a 10 mM stock in DMSO. Linkers 1-3 and DIBAC-FLAG were generated as described and kept at 10 mM in ddH 2 O. Protein loading was confirmed by Coomassie or Ponceau stain (0.2% ponceau in 3% AcOH). Blocking and antibody incubation conditions were conducted in 1x Dulbecco's phosphate buffered saline with 0.1% Tween-20 (PBST) with component concentrations as described. Membranes were developed by chemiluminescence using the SuperSignal West Pico kit (Thermo) or scanned for fluorescence by a Typhoon 9410 imaging system (Amersham). MALDI-TOF MS (Matrix-assisted laser desorption/ionization (time-of-flight) mass spectrometry) analysis was carried out on an Applied Biosystems Voyager DE-Pro machine. Typically, samples were diluted 8x in a saturated Sinapinic/Sinapic acid solution of 60% MeCN, 0.2% FA and 1 µL was deposited and dried under vacuum on a stainless steel sample plate (Applied Biosystems).
To compare yields among linkers at varying concentrations, 30 µM of aldehyde-tagged MBP or hGH was incubated with 300 or 750 µM of linker 1, 2, or 3 in NaH 2 PO 4 /NaCit buffer, pH 4.5 at 32 ºC for 18 h. Reactions were then supplemented with 500 µM or 1 mM DIBAC-488 and incubated at 4 ºC for 18 h before SDS-PAGE. Alternatively, the 750 µM linker reactions were diluted into PBS and concentrated 3 times to remove excess aminooxy linker. 15 µM of linker conjugated azide MBP or hGH was further reacted in PBS with 30 µM DIBAC-488 at 4 ºC for 18 h. Before SDS-PAGE, aliquots were taken for analysis by MALDI-TOF MS. Percent conjugation was determined by comparing peak heights of unreacted aldehyde-tagged protein to fluorophore conjugated peaks (+634 Da for DIBAC-488).
FLAG peptide conjugations: 50 µM MBP was reacted with 1 mM aminooxy linkers 1, 2, or 3 in 20 µL of 100 mM KOAc, pH 4.6 for 16 h at 37 ºC. The reactions were then diluted into PBS and concentrated 3 times to remove excess aminooxy linker. 10 µM aldehyde-tagged MBP was reacted with 100 or 500 µM AO-FLAG or DIBAC-FLAG in 100 mM KOAc, pH 4.6 at 37 ºC. The linker labeled MBP was also reacted at 10 µM with 100 or 500 µM DIBAC-FLAG in PBS, pH 7.4 at rt. Reactions were flash frozen at 2 h or run for 16 h before analysis by western blot. Transferred nitrocellulose was probed with anti-FLAG M2-HRP (Sigma, 1/20000) in 5% milk/PBST and developed by film exposure with chemiluminescence.
To compare FLAG conjugations by oxime versus Cu-free click reaction over time, MBP and hGH were first preconjugated with linker 1 as previously described. 20 µM aldehyde-tagged protein was incubated with 40 or 400 µM AO-FLAG in 100 mM MES, pH 4.6 at 37 ºC or azide linker labeled protein was incubated with 40 or 400 µM DIBAC-FLAG in 10 µL PBS at rt. Aliquots were taken at the specified times, diluted with sinapinic acid matrix solution and dried on the MALDI plate for mass spec analysis. Percent conjugation was determined by comparing