VHH characterization. Comparison of recombinant with chemically synthesized anti‐HER2 VHH

Abstract

In the continuous exploration of the VHH chemistry, biochemistry and therapeutic future use, we investigated two different production strategies of this small antibody‐like protein, using an anti‐HER2 VHH as a model. The total chemical synthesis of the 125 amino‐acid peptide was performed with reasonable yield, even if optimization will be necessary to upgrade this kind of production. In parallel, we expressed the same sequence in two different hosts: Escherichia coli and Pichia pastoris. Both productions were successful and led to a fair amount of VHHs. The integrity and conformation of the VHH were characterized by complementary mass spectrometry approaches, while surface plasmon resonance experiments were used to assess the VHH recognition capacity and affinity toward its “antigen.” Using this combination of orthogonal techniques, it was possible to show that the three VHHs—whether synthetic or recombinant ones—were properly and similarly folded and recognized the “antigen” HER2 with similar affinities, in the nanomolar range. This opens a route toward further exploration of modified VHH with unnatural amino acids and subsequently, VHH‐drug conjugates.

1. INTRODUCTION

In the therapeutic area, a paradigm shift had occurred these last years: from an almost‐all small molecule approach to drug, there has been a shift and an increasing interest for bigger objects united under the name of Biologics. Among them, and away from the nucleotide‐derived approaches, such as miR,1 antisense2 and siRNA,3 there have been essentially cells such as stem cells4 and proteins. Proteins have been used for more than a decade as substitutes of defective naturally produced ones such as insulin5 or EGF. It became increasingly obvious that antibody therapy6 might afford new tools to fight major diseases such as cancer.7 One of the main obstacles to the use of cytotoxic small molecules is the lack of specificity of those compounds toward cancer cells, leading to the well‐known side effects. A way to circumvent this lack of specificity is to address the compounds to the diseased cells, thanks to the use of antibodies raised specifically toward a target overexpressed in those cancer cells such as PD1, HER2, and so forth. The notion of antibody‐drug conjugates was born…

The next problems were thus (a) to attach cytotoxic compounds to the antibodies—that acts like cargos—in a reliable and reproductive way and (b) to produce this material to an industrial scale. The cost of this last item became a priority and a source of concern, as antibody‐based therapies might rapidly skyrocket, such as cell‐based ones. These facts nourished our immediate reflections on how to maintain the specificity and accessibility while gaining in terms of productivity and cost.

Camelids and sharks have surprisingly something in common: the way their antibody system is designed.8 Indeed, besides the classical double‐chain antibodies, shared by all living animals, they also have single‐chain ones. Even more interesting is the fact that the variable portion of their heavy chain bears the full antigen‐recognition site. This portion of the protein can be isolated as a single, rather short polypeptide while the antigen recognition potency is maintained. In other words, this approach can provide protein scaffolds far smaller than full antibodies (12 kDa vs. 150 kDa) or conventional antibody fragments (about 55 kDa for a Fab). They present immediate interest in term of production cost, but also in term of mastering their expression as well as their modification. The antibody fragments derived from camelids single‐chain antibodies were named VHHs or nanobodies.9 Many were raised versus several recognized target proteins, such as albumin, EGF receptor, PD1, HER2, and so forth. They obviously offer new possibilities in many therapeutic areas.10

Other types of proteins that have the possibility to specifically interact with a given target to form strong complexes have been described.11 À la façon of the antibody/antigen complex, they present a high level of specificity and fair levels of avidity. We saw in this area the possibility to develop a strategy by which the scaffold would be synthesized chemically, permitting the introduction in the sequence of one or several exotic amino acids. We applied those approaches to several proteins (ubiquitin12 or an enzyme: calstabin13) and were interested to extend this strategy to VHHs. Indeed, the size of those proteins permitted to anticipate two things: (a) the feasibility of their production by recombinant host(s) and (b) the possibility to reach those structures by a total chemical synthesis, even at an industrial scale. The possibility to access the introduction of exotic amino acids in a given sequence would open a new way to introduce new chemical functions that would be easily and specifically alkylated with cytotoxic compound(s) (or other, including fluorophore, for example) bearing the ad hoc chemical function (see review by Kent14).

The next step is to explore the possibility either to express a model VHH in two different hosts (Escherichia coli and Pichia pastoris) or to chemically produce it. Once the proteins were purified and characterized for their structure, we measured their affinities toward their antigen, HER2. The three proteins behaved in a similar way, opening an avenue of possibilities for future trends in this area, in particular with the possibility to include exotic amino acids that would permit to introduce new chemistry approaches, orthogonal to the classical peptide chemistry, leading to endless possibilities to build new entities such as armed VHHs for a more targeted delivery of cytotoxic molecules, for example.

2. RESULTS

2.1. Expression of anti‐HER2 VHH by Escherichia coli

The genetic construct was transformed in E. coli Shuffle T7 express strain for the anti‐HER2 VHH cytoplasmic expression. The protein was purified in a conventional manner, through a nickel‐affinity chromatography step, followed by a cation exchange chromatography. The anti‐HER2 VHH was produced in high quantity—in the 100 mg/L range—and in the correct oxidation state. As can be seen from Figure 1 (lane Ec), the VHH runs to the right molecular weight and seems to be fairly pure. Of course, minute information on the homogeneity of the preparations cannot be extrapolated from such an SDS‐PAGE gel, and only further characterization will be able to detail the quality of the VHH polypeptide.

Figure 1.

SDS‐PAGE analysis of anti‐HER2 VHH batches. Migration of ∼2 μg purified anti‐HER2 VHH in each lane. MW, molecular weight ladder; Pp, Pichia pastoris expression system; Ec, Escherichia coli expression system; Synth, Chemical synthesis

2.2. Expression of a secreted anti‐HER2 VHH using Pichia pastoris yeast

The anti‐HER2 VHH construct was introduced into P. pastoris by electroporation. Since the resulting yeast transformants potentially bear a variable number of plasmid copies integrated into their genome, isolated clones were then screened in a 96‐wells plate format for their performance in producing the recombinant VHH. Dot‐blot and Western‐blot immunodetection assays performed on the culture supernatants thus led to the selection of the clone secreting the highest amounts of the anti‐HER2 VHH (not shown). This clone was further used for the production step, whereby the VHH was directly purified from the culture supernatant with a single IMAC purification. A final concentration step eventually led to the recovery of about 10 mg of anti‐HER2 VHH per liter of yeast culture, with a high degree of purity (Figure 1, lane Pp).

2.3. Chemical synthesis of anti‐HER2 VHH

The synthesis strategy for preparing VHH entailed three peptide segments (F1, F2, and F3, see Figure 2). Due to the lack of solubility of the peptide F1, this segment was modified on Gly9 with a removable backbone linker.15 An hexahistidine tag was attached to the modification group (RMBO) using Fmoc solid phase peptide synthesis (SPPS). This tag increased the solubility of the fragment and facilitated the purification process with nickel affinity column.16 Peptides F1 and F2 were prepared as peptide hydrazides for subsequent in situ activation/thiolysis for native chemical ligation (NCL),17 while the peptide F3 was prepared with a C‐terminal carboxylic acid. The first ligation reaction, F1 to F2 at the Ser21‐Cys22 junction, was completed in 16 hr and the ligation product F1‐2 was then purified by HPLC and affinity column. F1‐2 was then ligated to F3 to obtain the full length VHH. The His‐tag was then removed under TFA treatment and purified on affinity chromatography column. The peptide was then folded and purified by HPLC yielding the pure protein. They were individually purified and analyzed by ESI‐MS (Figure 3). The resuspension of the lyophilized powder in an aqueous Tris buffer containing 5% DMSO (Figure 1, lane synth) enabled to increase its solubility from 5% to 18% compared to a buffer without DMSO. The major part of the VHH remains though insoluble, which might be improved with adjustments of the last purification steps and lyophilization procedure.

Figure 2.

Schematic representation of the strategy for the chemical synthesis of anti‐HER2 VHH. (a) Anti‐HER2 VHH amino acid sequence. (b) Amino acid sequences of the various peptide segments; X indicates position of the linker RMBO incorporation at the Gly9 residue. (c) Synthetic strategy

Figure 3.

Analytical data for the characterization of the various peptides of the anti‐HER2 VHH chemical synthesis. (a) HPLC traces and deconvoluted MS for initial peptide segments F1 (blue), F2 (red), and F3 (green). (b) HPLC trace and deconvoluted MS of first purified ligation product F1‐2. (c) HPLC trace and deconvoluted MS of purified ligation product F1‐2‐3. (d) HPLC trace and deconvoluted MS of final product full‐length VHH anti‐HER2

2.4. Mass spectrometry characterization of VHH constructs

We have used mass spectrometry based‐methods in order to compare the anti‐HER2 VHH obtained from the three production systems.

2.4.1. Purity and homogeneity/heterogeneity assessment of anti‐HER2 VHH products by LC‐UV‐MS and native MS

We first assessed the purity and homogeneity/heterogeneity of each anti‐HER2 VHH (detailed protein sequences are provided in Table 1) product using reversed phase HPLC–MS (Figure 4). LC‐UV‐MS analysis of the synthetic VHH sample revealed a highly homogeneous sample with only one main peak detected corresponding to the fully oxidized S‐S form of the VHH (14,022.7 ± 0.3 Da, expected mass 14,022.6 Da, see Table 2). LC–MS analysis of P. pastoris VHH revealed a slightly more heterogeneous sample with only two main peaks observed by LC‐UV corresponding to oxidized S‐S (retention time 13.5 min, 14,166.9 ± 0.5 Da, theoretical mass 14,166.7 Da) and the reduced SH forms (rt 14.9 min, 14,168.9 ± 0.4 Da, theoretical mass 14,168.7 Da) with a ratio of 62/38, respectively. Conversely, LC–MS analysis of E. coli VHH revealed a highly heterogeneous sample with two main clusters of peaks detected. The first main cluster (~15.3 min) could be attributed to E. coli anti‐HER2 VHH oxidized S‐S forms (as a mixture of intact—14,153.9 ± 0.3 Da—and C‐terminal truncated—13,605.4 ± 0.2 Da forms—while expected masses were 14,153.8 Da and 13,605.2 Da, respectively). The second cluster (~16.6 min) corresponds to reduced SH forms (14,155.9 ± 0.5 Da and 13,607.6 ± 0.4 Da, expected masses 14,155.8 Da and 13,607.2 Da, respectively) of a disulfide bridge formed between Cys23 and Cys96 (only two cysteines are in the VHH sequence). Interestingly, this S‐S bridge is not universal among the described VHHs, some of them do not have this feature, rendering it not completely canonical in this context. Of note, a third minor cluster of peaks could be detected at ~17.4 min with a mass increase of +28 Da on both intact and C‐terminal truncated anti‐HER2 VHH forms (14,183.9 ± 0.2 Da and 13,635.5 ± 0.4 Da, expected masses 14,183.9 Da and 13,635.2 Da, respectively), corresponding to N‐terminal methionine formylation (fMet), a well‐known posttranslational modification occurring in bacterial protein synthesis.18 From LC‐UV data, it could be deduced that the ratio of oxidized/reduced anti‐HER2VHH is 32/68. These results highlight that the expression of anti‐HER2 VHH is more heterogeneous (C‐terminal truncation and formylation) in E. coli than in P. pastoris, the more homogeneous sample originating from synthetic chemistry with 100% of S‐S bridge formed.

Table 1.

Detailed protein sequences

Production source	Amino acid sequence	Size	Theoretical mass (Da)
Escherichia coli	MEVQLVESGGGLVQAGGSLRLSCATSGITFMRYALGWYRQSPGKQREMVASINSGGTTNYADSVKGRFTISRDNAKNTVYLQMNSLKPEDTAVYYCNARWVKPQFIDNNYWGQGTQVTVSSHHHHHH*	127 AA	14,155.8 Da
Pichia pastoris	MRFPSIFTAVLFAASSALAAPVNTTTEDETAQIPAEAVIGYSDLEGDFDVAVLPFSNSTNNGLLFINTTIASIAAKEEGVSLEKRGSEVQLVESGGGLVQAGGSLRLSCATSGITFMRYALGWYRQSPGKQREMVASINSGGTTNYADSVKGRFTISRDNAKNTVYLQMNSLKPEDTAVYYCNARWVKPQFIDNNYWGQGTQVTVSSHHHHHH*	128 AA	14,168.7 Da
Chemical synthesis	EVQLVESGGGLVQAGGSLRLSCATSGITFMRYALGWYRQSPGKQREMVASINSGGTTNYADSVKGRFTISRDNAKNTVYLQMNSLKPEDTAVYYCNARWVKPQFIDNNYWGQGTQVTVSSHHHHHH*	126 AA	14,024.6 Da

Note: Underlined sequence: α‐factor secretion signal from Saccharomyces cerevisiae. It is endogenously cleaved by the yeast KEX2 protease during protein maturation in the Golgi apparatus. Protein size and molecular weight are given without this secretion signal.

Figure 4.

Heterogeneity/homogeneity assessment of anti‐HER2 VHH batches by LC‐UV‐MS and native MS. HPLC‐MS analysis in denaturing conditions of anti‐HER2 VHH batches: chemical synthesis batch (a), Pichia pastoris batch (b), and Escherichia coli batch (c) UV chromatograms

Table 2.

Identification of anti‐HER2 VHH species by liquid chromatography coupled to mass spectrometry

Anti‐HER2 VHH batch	Peak annotation	Time (min)	Mass (Da)	Species
Chemical synthesis	I	15.1	14,022.7 ± 0.3	S‐S form
Pichia pastoris	I	13.5	14,166.9 ± 0.5	S‐S form
	II	14.9	14,168.9 ± 0.4	SH form
Escherichia coli	I	15.3	14,153.9 ± 0.3	S‐S form WT
	II	15.5	13,605.4 ± 0.2	S‐S form truncated
	III	16.6	14,155.9 ± 0.5	SH form WT
	IV	16.8	13,607.6 ± 0.4	SH form truncated
	V	17.4	14,183.9 ± 0.2	SH form WT + fMet
	VI	17.5	13,635.5 ± 0.4	SH form truncated + fMet

2.4.2. Native ion mobility‐mass spectrometry of anti‐HER2 VHH batches

To gain insight into the VHH construct conformations, we next used native MS hyphenated to ion mobility spectroscopy (IM‐MS), an emergent MS‐based technique to assess protein conformational heterogeneities/homogeneities.19, 20, 21, 22 Of note, no LC separation of VHH oxidized and reduced form is used upstream from the mass spectrometer, leading to overlapping charge state distributions of both species in native MS (see Figure 5); in these conditions, only the main form of the S‐S bridge was observable. As expected, no significant difference was observed on charge state distributions obtained in native conditions for the different VHH constructs (Figure 5). We then turned on ion mobility (IM) and compared arrival time distributions (ATDs) of VHH constructs (Figure 6). For the chemical synthetic VHH sample, a very homogeneous Gaussian peak is detected for the 7⁺ charge state, with a FWHM of 1.3 ms. Interestingly, increased ATD peak broadening is observed for P. pastoris (FWHM = 1.7 ms) and E. coli (FWHM = 2.0 ms) samples. Altogether, IM‐MS results suggest very close conformational properties for the three batches. However, as observed in LC–MS, the synthetic VHH seems to be more homogeneous in terms of conformations, with thinner ATD peaks than the other constructs, in agreement with synthetic VHH containing only the S‐S form whereas the two other samples are mixtures of oxidized and reduced VHH forms.

Figure 5.

Native mass spectrometry analysis of anti‐HER2 VHH batches. ESI mass spectra on the m/z range (1,000–3,500) of chemical synthesis (a), Pichia pastoris (b), and Escherichia coli (c) batches in native conditions obtained on a Q‐TOF instrument. (a) (A) 14,022.4 ± 0.4 Da: VHH WT (7+/2,004.1 m/z); (*)14,085.1 ± 0.4 Da (+63 Da): VHH WT (+noncovalent adduct). (b) (B) 14,166.4 ± 0.5 Da: VHH WT (7+/2,024.9); (*) 14,229.2 ± 0.4 Da (+63 Da): VHH WT (+noncovalent adduct). (c) (C) 14,155.1 ± 0.4 Da: VHH WT (7+/2,023.1 m/z); (#) 14,183.2 ± 0.5 Da (+28 Da): VHH WT (+fMet); (d) 13,607.7 ± 0.5 Da: C‐ter truncated VHH (1–123) (7+/1,944.7 m/z); (#)13,635.7 ± 0.5 Da (+28 Da): C‐ter truncated VHH (+fMet). (*) and (#) represent a noncovalent adduct and a formylation, respectively

Figure 6.

Native IM‐MS experiments for conformational characterization of anti‐HER2 VHH batches. ATDs of the 7+ charge state of the three VHHs at low collision energy (4 V)

2.4.3. Gas‐phase conformational stability of anti‐HER2 VHH batches by collision induced unfolding experiments

Finally, collision induced unfolding (CIU) experiments (Figure 7) were performed to compare the gas‐phase unfolding behavior of the three VHHs. This new approach has already been reported for the rapid characterization of intact monoclonal antibodies,23, 24, 25, 26 fragments27 and antibody drug conjugates.28, 29 Briefly, in CIU experiments, ions are progressively accelerated in the trap T‐wave of the mass spectrometer before IM separation in the IM cell. Drift times of the ions are reported as a function of collision energies. The highly homogeneous chemical synthesis sample, containing only the S‐S form, presents a single conformational species (t _D = 7.2 ms) across the 0–120 V collision energy range (Figure 7), suggesting a very stable construct resistant to gas phase unfolding across the collision energy range. Conversely, the two others VHHs, containing a mixture of oxidized and reduced forms, present a transition to a supplementary conformation in CIU experiments (see Figure 7d). This additional more extended conformation is observed at t _D = 9.0 ms from 30 to 120 V for VHHs from P. pastoris and E. coli. The proportion of this second conformation is more abundant in the E. coli VHH (39%) compare to the P. pastoris one (17%). By converting t _D into averaged collisional cross sections (CCS) of E. coli VHH, a variation from 13.2 nm² for the S‐S form to 14.9 nm² for the reduced form is observed, resulting in a total increasing of 11.4% in CCS between SS initial state and the reduced form (Table 3).

Figure 7.

Gas‐phase conformational stability of anti‐HER2 VHH batches by collision induced unfolding experiments. CIU plots of anti‐HER2 VHH batches. CIU fingerprints of the 7+ charge state for the VHH from chemical synthesis (a), Pichia pastoris (b), and Escherichia coli (c) batches from 0 to 120 V collision energies. ATD extraction of the 7+ charge state of the three VHHs at 35 V collision energy (d)

Table 3.

Table summarizing all the experimental ^TWCCS_N2 measurements

	Trap CV = 4 V (initial state)			Trap CV = 35 V (extended conformation)
Z	Chemical synthesis	Pichia pastoris	Escherichia coli	Chemical synthesis	P. pastoris	E. coli
	TWCCSN2 (nm2)	TWCCSN2 (nm2)	TWCCSN2 (nm2)	TWCCSN2 (nm2)	TWCCSN2 (nm2)	TWCCSN2 (nm2)
6	11.8 ± 0.1	11.9 ± 0.1	12.0 ± 0.1	–	–	–
7	13.0 ± 0.1	13.1 ± 0.1	13.2 ± 0.1	–	15.0 ± 0.1	14.9 ± 0.1
8	14.5 ± 0.1	14.5 ± 0.1	14.5 ± 0.1	–	–	–
–	14.2*	14.3*	14.3*	–	–	–

Note: ^TWCCS_N2 measurements measured in native IM + MS conditions (trap CV = 4 V) are denoted with the standard deviation for each individual charge state (standard deviation has been calculated from three different analysis) and the theoretical ^TWCCS_N2 values calculated from CCS =2.435 × (M)^(2/3) (asterisk).39 ^TWCCS_N2 values ranging from 11.8 to 14.5 nm² were obtained for the 6⁺ from 8⁺ charge states. Measured ^TWCCS_N2 values of the 8+ charge state (14.5 nm²) were in good agreement with both predicted collision cross‐section from VHHs considered as spherical proteins (14.2–14.3 nm²).

This different gas phase unfolding behavior can be attributed to the presence of reduced VHH forms in both samples. To unambiguously attribute the more extended conformation to the reduced form, we compared the CIU fingerprint of the fully reduced VHH from P. pastoris to the original mixture (Figure 8 ) and, as expected, we observed a stronger intensity of the extended conformation. Interestingly, the ratios of oxidized and reduced forms coming either from IM‐MS or from LC–MS were following the same trend. This difference in absolute values between IM–MS and LC–MS could be explained by an insufficient resolution of the IM separation in CIU experiment for liable relative quantification but enough for the detection of a second conformation more extended in this case.

Figure 8.

Gas‐phase conformational stability of nonreduced and reduced anti‐HER2 VHH produced by Pichia pastoris. CIU plots of the 7+ charge state for the nonreduced (a) and reduced (b) anti‐HER2 VHH samples

Altogether, CIU results reflect higher heterogeneity of P. pastoris and E. coli samples compared to the synthetic VHH. Oxidized and reduced VHH forms present different CIU behavior, suggesting a specific CIU signature for both forms. CIU results highlight that disulfide bridge formation could be characterized on a VHH and that it seems to stabilize, in our case, the anti‐HER2 VHH structure. Taken together, mass‐spectrometry‐based results have shown that E. coli does not permit the expression of homogeneous VHH batch (formylation, truncation) or a total formation of the disulfide bridges. P. pastoris allows obtaining more homogeneous VHH batches while synthetic VHH is the most homogenous form, with disulfide bridge being quantitatively formed.

2.5. Functional characterization of VHH interaction with HER2 protein using surface plasmon resonance (SPR)

The binding capacity of VHH samples to the immobilized C‐terminal part of HER2 protein was assessed using SPR experiments. We first determined the active concentration of each VHH batch, by measuring its ability to bind HER2 epitope, as it is not necessarily identical to the total concentration reflected by the absorbance of the sample at 280 nm. Furthermore, a correct VHH concentration is essential to determine affinity and the kinetic association rate constant. The percentage of active VHH was determined by Calibration‐Free Concentration Analysis (CFCA) (Figure 9a–c); it exceeded 80% of the soluble fraction for the three VHH sources (Table 4).

Figure 9.

Functional characterization of VHH‐antigen interaction by SPR. Panels A to C: Calibration‐free concentration analysis of VHH active concentration. CFCA fits for the binding between highly immobilized HER2 antigen (1,200 RU) and a VHH dilution (~1 nM) tested in duplicates at two different flow rates (10 and 80 μL.min⁻¹). The VHH samples were prepared from (a) Escherichia coli periplasmic expression, (b) chemical synthesis, and (c) Pichia pastoris secretion. The figures show overlay plots of experimental (black) and calculated (red) data. Calculated concentration in nM, closeness of the fit (chi²) and degree of mass transport limitation (QC ratio) for each sample: (a) 0.85 nM, chi² = 0.022, QC ratio = 0.691; (b) 0.99 nM, chi² = 0.047, QC ratio = 0.662; (c) 0.96 nM, chi² = 0.011, QC ratio = 0,718. Panels (d)‐(g) Sensorgrams of the interaction between immobilized HER2 antigen (100 RU) and increasing concentrations of VHH (0.33, 1, 3.33, 10, 30 nM, duplicates). (d) Escherichia coli periplasmic expression, (e) chemical synthesis, (f) Pichia pastoris secretion, and (g) anti‐ADORA2A VHH, negative control. For (d), (e), and (g), experimental curves (black) were fitted with a 1:1 Langmuir binding model (red) using the scrubber 2.0c software, residual plots obtained in global analysis of each binding curves are shown and are randomly distributed

Table 4.

Binding parameters of anti‐HER2 VHH to its antigen HER2 determined by SPR (n = 2)

Expression organism	k on (M−1.s−1)	k off (s−1)	K D (nM)	R max
Escherichia coli	2.0E6 ± 0.4E6	5.6E‐3 ± 0.2E‐3	2.8 ± 0.5	13.2 ± 1.1
Pichia pastoris	2.1E6 ± 0.3E6	4.4E‐3 ± 0.2E‐3	2.1 ± 0.4	13.2 ± 2.1
Chemical synthesis	2.5E6 ± 0.9E6	2.8E‐3 ± 0.4E‐3	1.3 ± 0.6	12.0 ± 1.7

To measure the binding parameters of VHH samples, a low level of HER2 protein was immobilized (100 RU) on a sensor surface. The binding rate was not affected by the flow rate (data not shown) indicating that mass transport did not limit the interaction which is suitable for kinetic measurements. The sensorgrams are presented in Figure 9d–g. As reported in Table 5, each anti‐HER2 VHH batch presents similar binding kinetics toward its antigen, which leads to comparable binding affinity and kinetic constants regardless of the expression system.

Table 5.

Percentage of active anti‐HER2 VHH in each batch determined by SPR

Expression organism	Soluble fraction (%)	Total protein concentration (M)	Active protein concentration (M)	Ratio of active VHH (%)
Escherichia coli	100%	1.20E‐04	1.02E‐04	85%
Pichia pastoris	100%	1.90E‐05	1.88E‐05	99%
Chemical synthesis	18%	7.70E‐06	7.39E‐06	96%

3. DISCUSSION

A boom in the area of antibody drug conjugates happened these last years with more than 300 clinical trials. Furthermore, the possible use of smaller protein(s), with similar specificity than antibodies is certainly a way to render this area more reachable to industrial scales. Indeed, a series of such compounds are entering clinical trials and market. Thus, a growing interest occurred these last years toward simplified antibodies, namely VHHs or nanobodies that comprise only a 100 amino acids.

Due to recent progress in the domain of peptide and protein synthesis (see review by Kent14), a realistic option arises for the chemical synthesis of such small proteins. To us, it is clearly a great advantage to move from recombinant to chemically‐synthesized objects, and this for at least two reasons: (a) the industrial scalability of such compounds, intimately associated with the possibility to achieve modern analytical characterization and (b), more importantly, the possibility to include at ad hoc places exotic amino acids, the side chain of which could bear a chemical function unique to a given chemistry step, as opposed to cysteine‐ or lysine‐based chemistry. Indeed, our recent work on either ubiquitin12 or an enzyme (calstabin, an isomerase)13 that were both chemically synthesized and biologically active, as many other reported proteins (see Bacchi et al13 for listed examples and Kent14 for review) opens avenues on the possibility to access to proteins of small molecular weight (between 100 and 150 amino acids) and possibly to incorporate new exotic amino acids susceptible with alternative, specific chemistry.30, 31 This will provide new possibility of arming antibody‐like proteins such as VHHs, with cytotoxic compounds, potentially delivered to the right target‐bearing cells. The first steps to take, though, are the systematic study of such molecules, to characterize their folding and their “antigen” recognition.

In order to be able to embark in such a task, the first step consisted to express various model VHHs in various hosts, and to document thoroughly their characteristics, particularly regarding their structural features. In parallel, we expressed the anti‐HER2 VHH in two hosts and we synthesized the same VHH for comparison purposes. Again, their respective biophysical characteristics were studied as well as their capacity to recognize the antigen, HER2. The positive results we obtained paved the way toward the possibility to synthesize VHHs in general while keeping their capacity to act as small antibodies. Indeed, we did not record any major differences between the recombinant and the synthetic protein. The main results in this Stage 2 process were certainly the facts (a) that a decent amount of chemical VHH was obtained at a high purity level even if not at an industrial‐compatible quantity, and (b) that this product was not distinguishable from the recombinant one, despite a thorough characterization, as we suggested as a result of the Stage 1 process.

Gaining access to such a synthetic protein is of major importance, because it opens new avenues according to two main axes: (a) the possibility to modify the primary sequence of the VHH without altering its affinity for its “antigen” and (b) the possibility to integrate almost any amino acid bearing an alternative chemical function. The latter can, thanks to additional chemical steps, lead to VHH armed with cytotoxic compounds, while maintaining their specificity toward therapeutic targets, chosen because they are preferentially expressed in cancer cells, for example. Overall, those new compounds will basically retain the specificity of their corresponding antibodies with an incredible gain in size, as we would have proteins of ~140 aminoacids instead of the initial ~900 to ~1,300 aminoacids respectively composing camelid and other mammals antibodies. Not only the gain in absolute mass is considerable, but also the structure of the VHH is far simpler than antibody's. Before embarking further in such studies, as suggested below, the first step will be to rationalize the yield of such synthesis. Indeed, it is only once the proof of concept has been done as in the present work that we will be able to work out the various steps of the synthetic schemes, in order to render the process scalable to a more industrial‐friendly approach.

The Stage 3 will consist of modifying this anti‐HER2 VHH sequence with exotic amino acids at position without negative influence on the HER2 recognition. In doing so, we will be able to introduce chemical functions, orthogonal of cysteine and lysine side chain ones, leading to the possibility to maintain a high specificity of the cargo—the VHH—while maintaining a cost‐effective production, as opposed to the monoclonal antibody option.

4. MATERIALS AND METHODS

4.1. Cloning, expression, and purification of the anti‐HER2 VHH in E. coli

In brief, the coding DNA fragment for the anti‐HER2 VHH has been E. coli codon optimized, synthetized and cloned into high copy expression vector with a poly‐His‐tag at the C‐terminus end. The genetic construct was transformed into the E. coli strain Shuffle T7 express. Bacteria were cultured in a LB medium with ampicillin (final concentration of 100 μg/mL) and the expression was induced when the OD at 600 nm reached 0.6 with 0.1 mM IPTG. The temperature was decreased from 37°C to 22°C for the expression and the expression time was over night. Cells were firstly harvested by centrifugation (5,000 g, 20 min, 4°C), and then lysed by lysozyme membrane digestion (1 mg/mL) into a 20 mM Tris–HCl pH 8, 300 mM NaCl buffer supplemented with 10 μg/mL DNAse1 and protease inhibitor cocktail during 30 min at room temperature. Cell fragments were harvested by ultracentrifugation (100,000g, 30 min, 4°C). The supernatant was purified in three steps. A first step using affinity properties of the His‐tag for the Ni‐NTA agarose resin (Qiagen) was performed. The supernatant was injected onto an equilibrated Ni‐NTA agarose resin with a 20 mM Tris–HCl pH 8, 300 mM NaCl, 10 mM imidazole buffer. The column was washed with the equilibrium buffer supplemented with 20 mM imidazole, then the proteins were eluted with the equilibrium buffer supplemented with 250 mM imidazole. The second purification step was performed by cation exchange chromatography (IEX, SourceS, GE healthcare) after dialysis of the sample against a buffer composed of 50 mM Tris–HCl pH 7.5, 50 mM NaCl. The sample was injected onto the cation exchange column beforehand equilibrated with the dialysis buffer previously used. The proteins were eluted using a gradient of a buffer composed of 50 mM Tris–HCl pH 7.5 and 1 M NaCl. The eluted proteins were then concentrated by centrifugation using an Amicon (Millipore) with a cutoff of 3 kDa. The concentrated sample was finally purified by sized exclusion chromatography (SEC) using a HiLoad 26/600 Superdex 200 column (GE Healthcare) equilibrated with a 50 mM Tris–HCl pH 7.5, 150 mM NaCl buffer.

4.2. Cloning, expression, and purification of the anti‐HER2 VHH in P. pastoris

Anti‐HER2 VHH DNA sequence was fused in frame to the α‐factor secretion signal from Saccharomyces cerevisiae on its 5′ side and to a 6‐His affinity tag sequence on its 3′ side. It was cloned in a modified expression vector from the pPIC9K series. The protease‐deficient P. pastoris strain SMD1163 was transformed with the linearized plasmid, and recombinant clones presenting a stable VHH expression were selected based on both antibiotic resistance and expression level criteria, as extensively described in Hartmann et al.32 For protein expression, cells were grown overnight in BMGY (10 g/L yeast extract, 10 g/L peptone, 13.4 g/L yeast nitrogen base, 1% [v/v] glycerol, 0.4 mg/L biotin, 100 mM potassium phosphate pH 6.0). On the next day, the cells were diluted in BMGY and grown to an OD₆₀₀ of 5. The culture was centrifuged and the pellet resuspended in the same volume of BMMY (10 g/L yeast extract; 10 g/L peptone, 13.4 g/L yeast nitrogen base, 0.5% [v/v] methanol, 0.4 mg/L biotin, 100 mM potassium phosphate pH 6.0) for induction of protein expression. After 18 h at 22°C, the cells were spun down and the VHH‐containing supernatant was either used immediately for purification or frozen at −80°C. VHH‐containing supernatant was incubated overnight at 4°C with Ni‐NTA agarose batch resin (Qiagen) previously equilibrated with purification buffer (25 mM Tris HCl at pH 7.4, 200 mM NaCl, 10 mM imidazole), at a ratio of 4 mL resin for 100 mL supernatant. The resin was then poured into a disposable column and sequentially washed with (a) purification buffer containing 25 mM imidazole, (b) buffer containing 50 mM Tris HCl at pH 7.4, 1 M NaCl. Purified VHH was then eluted with 5 mL elution buffer (25 mM Tris HCl at pH 7.4, 200 mM NaCl, 250 mM imidazole). Imidazole was removed from the eluted fraction through dialysis in 50 mM Tris HCl at pH 7.4, 150 mM NaCl. The final sample was concentrated using Vivaspin® Centrifugal filter (membrane cut‐off of 3 kDa) (VivaProducts, Gottigen, Germany). Except for the initial resin incubation, all purification steps were performed at room temperature.

4.3. Chemical synthesis of anti‐HER2 VHH

A schematic representation of the synthesis strategy is shown in Figure 2. All peptides were synthesized at 100 μmole scale using Fmoc‐SPPS on a Symphony X instrument (Protein Technologies, Inc.). Synthesis‐grade Fmoc‐AA reagents were purchased from Iris Biotech. Specific L‐pseudoproline dipeptides and (Dmb)Gly that were used to facilitate the peptide syntheses are indicated in the peptide sequences in bold and underline font, respectively. HATU and 2‐chlorotrityl chloride resin (200–400 mesh, 1.2 mmol/g) were purchased from IRIS Biotech. DMF, CH₂Cl₂, methanol, diethyl ether, and HPLC‐grade acetonitrile were from purchased from Aldrich. The standard deprotection‐coupling cycle for each residue consisted of six steps: (a) Wash with 5 mL of DMF (3 × 30 s); (b) deprotect Fmoc group with 5 mL of 20% piperidine in DMF (3 × 3 min); (c) wash with 5 mL of DMF (3 × 30 s); (d) Couple Fmoc‐AA (2 × 60 min); (d1) 5 mL of Fmoc‐AA dissolved at 200 mM in DMF; (d2) 2 mL of HATU dissolved at 500 mM in NMP; (d3) 2 mL of iPr₂NEt dissolved at 1 M in NMP; (e) capping with 5 mL of 10% Ac₂O in DMF (7 min) and finally (f) wash with 5 mL of DMF (3 × 30 s). Upon completion of the synthesis, peptide resins were washed three times with DMF and three times with DCM. Peptide resins were then cleaved with one of three TFA cleavage cocktails (per 100 μmole scale synthesis): (a) Cocktail #1 for peptides containing neither Met nor Cys: 18 mL TFA, 500 μL TIS, 500 mg phenol, 1,000 μL water; (b) Cocktail #2 for peptides containing Cys, but not Met: 17.5 mL TFA, 500 μL TIS, 500 mg phenol, 1,000 μL water, 500 μL EDT; or (c) Cocktail #3 for peptides containing Met: 17.5 mL TFA, 500 μL TIS, 500 mg phenol, 1,000 μL water, 500 μL EDT and 500 μL tetrabutylammonium bromide. After 3 h of TFA cleavage, the cleavage solution was filtered from the resin, and peptides were precipitated by addition to 40 mL of ice‐cold ether. After >1 hr at −20°C, ether solutions were centrifuged at 3,500 RCF, and the supernatants were decanted. Pellets were washed twice more with ether, and then dried for >3 hr in vacuum desiccator prior to dissolution, analytical characterization, and purification. Peptides containing C‐terminal acids and hydrazides were used in this work. Peptide acids and hydrazides were custom loaded at low density (~0.2 mmol/g) starting with 2‐chlorotrityl chloride resin (1.5 mmol/g, IRIS Biotech, Germany). In the case of peptide acids, 0.35 mmol Fmoc‐AA was dissolved in 30 mL CH₂Cl₂, and then 1.5 mmol Pr2NEt was added to the Fmoc‐AA solution. This solution was next added to 1 g of resin and mixed for 2 hr at room temperature. Unreacted groups were capped with 17:2:1 CH₂Cl₂:MeOH:Pr2NEt for 30 min. Fmoc‐AA‐resin was then washed 3× with DMF, 3× with CH₂Cl₂ and vacuum dried for >20 min. The loading was determined by the titration of the Fmoc group deprotection (UV, 301 nm). In the case of peptide hydrazides, 0.35 mmol Fmoc‐NH‐NH₂ was dissolved in 30 mL CH₂Cl₂, and then 1.5 mmol Pr2NEt was added to the solution. This solution was next added to 1 g of resin and mixed for 2 hr at room temperature. Unreacted groups were capped with 17:2:1 CH₂Cl₂:MeOH:Pr2NEt for 30 min. Fmoc‐AA‐resin was then washed 3× with DMF, 3× with CH₂Cl₂ and vacuum dried for >20 min. The loading was determined by the titration of the Fmoc group deprotection (UV, 301 nm). Native chemical ligation (NCL) reactions were performed according to standard methods using MPAA as thiol catalyst and TCEP as reducing agent. Specifically, all ligation reactions employed peptide hydrazide method, whereby peptides were dissolved and activated (conversion of hydrazide to acyl azide) in “activation buffer” (6 M GuHCl, 100 mM phosphate, pH 3) for 20 min at −20°C by addition of freshly prepared 20 mM sodium nitrite. Following activation, a solution containing freshly prepared 200 mM MPAA pH 7 in “ligation buffer” (6 M GuHCl, 200 mM phosphate, pH 7) was added, and the final pH was adjusted to 6.0–6.5 to initiate thiolysis and ligation reaction. Upon completion (based on analytical HPLC and LC/MS), reactions were treated with freshly prepared 200 mM TCEP in 6 M GuHCl (pH > 6) for 10 min, spun at 5,000 RCF to remove any aggregates, and the supernatant was purified by preparative/semipreparative HPLC. The peptides were analyzed by UPLC and ESI‐MS mass spectrometry. The instruments were equipped with Jupiter 4 μm C12 Proteo 90 å (250 × 21.2 mm) (flow rate: 20 mL/min) or XBridge Protein BEH C4 4 μm 300 å (250 × 19 mm) (flow rate: 20 mL/min). Solvents A and B were 0.1% TFA in water and 0.1% TFA in acetonitrile, respectively.

4.4. Synthesis of peptide F1

The peptide F1 [EVQLVESGXGLVQAGGSLRLS‐NHNH₂, see Figure 3b, blue peptide], where X is G(H6‐RMBO), was prepared by standard Fmoc‐SPPS as described above, with a few special steps elaborated below. The peptide chain was synthesized using standard Fmoc building blocks and protecting groups, with the exception of the Fmoc‐Gly(RMBO)‐OH building block at the position 10. The N‐terminus of the peptide was protected Boc‐Glu(Otbu)‐OH building block in cases. Upon completion of the linear peptide synthesis, the aloc group was removed. PhSiH₃ (25 eq.) in 5 mL of CH₂Cl₂ was added to the resin followed by Pd(PPh₃)₄ (0.25 eq.) in CH₂Cl₂. After agitating the resin for 1 hr in the dark, the solution was drained and the reaction was repeated twice. The resin was washed with three volumes of CH₂Cl₂, four volumes of 1 M pyridine in DMF, and three volumes of DMF and standard Fmoc‐SPPS conditions were used to fuse the HisTag (His6) sequence to the VHH. Standard peptide cleavage conditions (described above) were used following the synthesis of the HisTag with the G(RMBO) part of the peptide. F1 crude peptide (350 mg) was dissolved in 20 mL of 25/75 acetonitrile/water (0.1% TFA) and purified on a Jupiter 4 μm C12 Proteo 90 å (250 × 21.2 mm) column using a gradient of 10–25% buffer B over 30 min. HPLC purification yielded 67 mg of pure material (yield 21%).

4.5. Synthesis of peptide F2

The peptide F2 [CAASGRTFSSYAMGWFRQAPGKEREFVVAINWSSGSTYYADSVKGRFTISRD NAKNTVYLQMNSLKPEDTAVYY‐NH‐NH2, see Figure 3b, red peptide] was prepared by standard Fmoc‐SPPS as described above, using standard Fmoc building blocks and protecting groups, with the exception of the Fmoc‐Asp(OEpe)‐OH building block at the position 51, Fmoc‐Asp(OtBu)‐Thr(psiMe,Mepro)‐OH building block at the position 69 and, Fmoc‐Asp(OtBu)‐Ser(psiMe,Mepro)‐OH building block at the position 41. After standard peptide cleavage conditions (described above), F2 crude peptide (900 mg) was dissolved in 20 mL of 25/75 acetonitrile/water (0.1% TFA) and purified on a XBridge Protein BEH C4 4 μm 300 å (250 × 19 mm) column using a gradient of 20–35% buffer B over 30 min. HPLC purification yielded 40 mg of pure material (yield 5%).

4.6. Synthesis of peptide F3

The peptide F3 [CNARWVKPQFIDNNYWGQGTQVTVSSHHHHHH‐OH, Figure 3b, green peptide] was prepared by standard Fmoc‐SPPS as described above, using standard Fmoc building blocks and protecting groups. After standard peptide cleavage conditions (described above), the F3 crude peptide (650 mg) was dissolved in 20 mL of 25/75 acetonitrile/water (0.1% TFA) and purified on a Jupiter 4 μm C12 Proteo 90 å (250 × 21.2 mm) column using a gradient of 20–35% buffer B over 30 min. HPLC purification yielded 80 mg of pure material (yield 21%).

4.7. Synthesis of peptide F1‐2

Native chemical ligation of F1 to F2 to generate F1‐2 [EVQLVESGXGLVQAGG SLRLSCAASGRTFSSYAMGWFRQAPGKEREFVVAINWSSGSTYYADSVKGRFTISRDNAKNTVYLQMNSLKPEDTAVYY‐NHNH2 where X is G(H6‐RMBO, Figure 3c, purple peptide)], was performed using peptide hydrazide method. Here, 30 mg of F1 was dissolved in 3 mL of activation buffer, and 40 mg of F2 was dissolved in 3 mL of ligation buffer. F1 was then activated (conversion of hydrazide into azide) for 20 min at −20°C by addition of 200 μL of 200 mM NaNO2, pH 3.0. At the same time, 100 mg of MPAA (200 mM) was dissolved in the F2 solution and the pH was adjusted to 6.5. After activation, the two solutions were mixed, and the final pH was adjusted to 6.5 to initiate ligation. Reaction was complete after 6 hr. The reaction was then reduced for 10 min by treatment with 100 mM TCEP at pH 7.5, then diluted out to 3 M GuHCl and washed with 4 × 5 mL of Et₂O. The solution was adsorbed on affinity chromatography column (HiTrap™ Chelating HP, 5 mL) and the column was washed with 3 × 5 mL of binding buffer (6 M GuHCl, 20 mM Na₂HPO₄, pH 7.5). The product was eluted with 3 × 10 mL of elution buffer (6 M GuHCl, 20 mM Na₂HPO₄, 0.5 M Imidazole, pH 7.5). The fractions containing the desired product were purified by HPLC on a XBridge Protein BEH C4 4 μm 300 å (250 × 19 mm) column using gradient of 20–45% B over 50 min. HPLC purification yielded ~24 mg (45% yield, based on limiting F2 peptide) of pure material.

4.8. Native chemical ligation between peptides F1‐2 and F3

Native chemical ligation of F1‐2 to F3 to generate F1‐3 [EVQLVESGXGLVQAGGSLRLSCAASGRTFSSYAMGWFRQAPGKEREFVVAINWSSGSTYYADSVKGRFTISRDNAKNTVYLQMNSLKPEDTAVYYCNARWVKPQFIDNNYWGQGTQVTVSSHHHHHH‐OH where X is G(H6‐RMBO), Figure 3c, orange peptide)], was performed using the peptide hydrazide method. Here, 24 mg of F1‐2 was dissolved in 2 mL of activation buffer, and 17 mg of F3 was dissolved in 2 mL of ligation buffer. F1‐2 was then activated (conversion of hydrazide into azide) for 20 min at −20°C by addition of 200 μL of 200 mM NaNO₂, pH 3.0. At the same time, 65 mg of MPAA (200 mM) was dissolved in the F3 solution and the pH was adjusted to 6.5. After activation, the two solutions were mixed, and the final pH was adjusted to 6.5 to initiate ligation. Reaction was complete after 16 hr. The reaction was then reduced for 10 min by treatment with 100 mM TCEP at pH 6.5, then diluted out to 3 M GuHCl and washed with 4 × 5 mL of Et₂O. The solution was purified by HPLC on a XBridge Protein BEH C4 4 μm 300 å (250 × 19 mm) column using gradient of 20–45% B over 50 min. HPLC purification yielded ~14 mg (45% yield, based on limiting F2 peptide) of pure material.

4.9. Synthesis of the anti‐HER2 VHH

The final anti‐HER2 VHH [EVQLVESGGGLVQAGGSLRLSCAASGRTFSSYAMG WFRQAPGKEREFVVINSSGSTYYADSVGRFTISRDNAKNTVYLQMNSLKPEDTAVYYCNARWVKPQFIDNNYWGQGTQVTVSSHHHHHH‐OH, Figure 3c, black peptide] was obtained by dissolving 14 mg of the peptide F1‐3‐HisTag in 5 mL of TFA/DTT/iPr₃SiH (92.5:5:2.5) at room temperature. After 5 hr, the peptide was precipitated with Et₂O, washed two times with Et₂O and dissolved in 10 mL of 6 M GuHCl solution. The solution was adsorbed on affinity chromatography column (HiTrap™ Chelating HP, 5 mL) and the column was washed with 3 × 5 mL of binding buffer (6 M GuHCl, 20 mM Na₂HPO₄, pH 7.5). The product was eluted with 3 × 5 mL of elution buffer (6 M GuHCl, 20 mM Na₂HPO₄, 0.5 M Imidazole, pH 7.5). The peptide was desalting by solid phase extraction using Sep‐Pak C18 Cartridge (Waters, 2 g) and lyophilized yielding ~7 mg (53% yield) of material.

4.10. Folding of the synthetic VHH

The peptide (7 mg) was dissolved in 50 mL of folding buffer (6 M GuHCl, 100 mM TRIS, pH 8.5) and left 2 days at 4°C. The solution was purified by HPLC on a Jupiter 5 μm C4 300 å (250 × 4.6 mm) column using gradient of 20–40% B over 40 min at 70°C. HPLC purification yielded ~3.5 mg (50% yield) of pure material. The solution containing peptide was lyophilized on a CHRIST Gamma 2‐16 LSCplus for 24 hr at room temperature and 0.3 mbar.

4.11. SDS‐PAGE electrophoresis

Protein preparations were submitted to electrophoresis as follows. They were diluted twice in sample buffer (100 mM Tris HCl at pH 6.8, 25% glycerol, 8% SDS, 0.2 g/L Coomassie Brilliant Blue G250, 200 mM DTT) and loaded on a 10% polyacrylamide‐SDS tricine gel, in parallel with PageRuler™ Prestained Protein Ladder (Thermo Scientific™). The separating layer of the gel was composed of 10% acrylamide 19:1, 1 M Tris HCl at pH 8.45, 0.1% SDS, 0.1% (w/v) APS, 0.06% (v/v) TEMED. The stacking part was composed of 6% acrylamide 19:1, 0.8 M Tris HCl at pH 8.45, 0.08% SDS, 0.15% (w/v) APS, 0.01% (v/v) TEMED. Proteins were resolved for 1 hr at 100 V in tris‐tricine buffer (100 mM Tris HCl at pH 8.3, 100 mM tricine, 0.1% SDS) and stained with Coomassie blue (QuickCoomassie Stain, Generon, Slough, UK).

4.12. Dot‐blot

Dot‐blot assay was performed in a 96‐well plate format using a Bio‐dot microfiltration manifold (Biorad). Twenty‐five microliters of culture supernatant total was suspended in 25 μL sample buffer and absorbed onto a 0.45‐μm nitrocellulose membrane by gravity flow. The membrane was then washed three times with 200 μL of PBS per well. The immunodetection was performed using anti‐His antibody (Penta‐His Antibody, Qiagen) diluted at 0.1 μg/mL in PBST containing 5% (m/v) of BSA. Primary antibody were revealed using secondary anti‐mouse antibody coupled to a fluorophore (Goat anti‐Mouse antibody DyLight® 488 Conjugated, Bethyl Laboratories). The fluorescence signal was visualized on an Odyssey scanner (Li‐Cor).

4.13. Liquid chromatography coupled to mass spectrometry (LC–MS)

Liquid chromatography coupled to mass spectrometry (LC–MS) analysis were performed using an Alliance 2,695 (Waters, MA) coupled to a micrOTOF‐Q (Bruker, Billerica, MA). A volume equivalent to 15 μg of sample preparation was injected on a XBridge BEH300, C4, 3.5 μm, 2.1 × 150 mm column (Waters) set at 60°C. The gradient was generated at a flow rate of 250 μL/min using 0.1% trifluoroacetic acid (TFA) for mobile phase A and acetonitrile containing 0.08% TFA for mobile phase B. B was raised from 20 to 60% in 20 min followed by a 2 min washing step at 90% B and a 15 min reequilibration period. Signal acquisition was realized by UV absorbance measurement at 214 nm. The micrOTOF‐Q was operated in positive mode with a capillary voltage of 4,500 V. Acquisitions were performed on the mass range 500–5,000 m/z with a 1 s scan time. External mass calibration of the TOF was achieved before each set of analyses using Tuning Mix (Agilent Technologies, Paolo Alto, USA) in the mass range of 622–2,732 m/z. Data analysis was performed using the Compass DataAnalysis 4.2 software (Bruker).

4.14. Native mass spectrometry and ion mobility mass spectrometry (IM‐MS)

Prior to native MS and IM‐MS experiments, all anti‐HER2 VHHs were buffer exchanged against ammonium acetate (150 mM, pH 7.4) buffer (Sigma, St. Louis, MO), using Zeba Spin desalting columns (Thermo Fisher Scientific, Rockford, IL). Sample concentrations were determined by UV absorbance using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). Native mass spectrometry analysis were carried out on a hybrid electrospray quadrupole time‐of‐flight mass spectrometer (Synapt G2 HDMS, Waters, Manchester, UK) coupled to an automated chip‐based nanoelectrospray source (Triversa Nanomate, Advion Biosciences, Ithaca) operating in the positive ion mode. Mass spectrometer calibration was performed using singly charged ions produced by a 2 mg/mL solution of cesium iodide in 2‐propanol/water (1v/1v) over the m/z range 500–8,000. Instrumental parameters have been optimized to get optimal high m/z ion transmission and resolution by raising the backing pressure to 6 mbar and the cone voltage to 20 V. Data interpretation was performed using MassLynx 4.1 (Waters). Ion mobility experiments were performed on a TWIMS‐MS Synapt G2 instrument (Waters). MS and IMS parameters were carefully tuned in order to obtain good IM resolution without ion activation using low accelerating voltages. In the Z‐spray source, sampling cone was fixed to 20 V and the backing pressure to 6 mbar. For IM‐MS characterization, the IM cell parameters have been optimized as described below. The argon flow rate in the traveling‐wave based ion trap was 5 mL/min and the trap collision voltage has been set to 4 V. In the helium compartment, ions were thermalized with a flow rate of 120 mL/min and the ion mobility cell was filled with a constant N₂ flow rate of 30 mL/min for ion separation. The IM wave height and velocity were set 30 V and 1,000 m/s, respectively. After the IM cell, the transfer collision voltage was fixed to 2 V in order to transmit ions up to the TOF analyzer. Ion mobility calibration was performed using β‐lactoglobulin as described elsewhere33 in order to calculate collision cross sections for each anti‐HER2 VHH batch. IM‐MS measurements were performed in triplicate under identical conditions.

4.15. Collision induced unfolding experiments (CIU)

We also characterized our proteins by collision‐induced unfolding experiments. Ions were activated in the trap cell by increasing progressively the trap collision voltage in 5 V steps from 0 to 120 V prior to IM separation during 1 min run. Generation and data interpretation of CIU plots were performed using the open‐source CIU_Suite software, allowing the extraction of the arrival time distribution (ATD) of ions of interest at each trap collision voltage as described previously.22, 27

4.16. VHH‐“antigen” interaction measurements

SPR experiments were performed to measure the interactions between antibodies and HER2 protein were investigated using a BIAcore T200 at 25°C. Sensor surfaces CM5 sensorchips (BR100530), amine coupling kit and other Biacore consumables were purchased from GE Healthcare. HER2 protein was immobilized on the sensor surface and the VHH antibodies were injected over the surface in a continuous flow. To immobilize HER2 protein onto the ships, we used standard amine coupling procedures34 at a flow rate of 10 μL/in in a running buffer composed of 10 mM HEPES at pH 7.4, 150 mM NaCl, 0.05% P20 surfactant. Surfaces were activated by an injection of a 1:1 mix of 0.2 M N‐ethyl‐N′‐(3‐dimethylaminopropyl)‐carbodiimide hydrochloride (EDC) and 0.05 M N‐hydroxysuccinimide (NHS) for 10 min. The his‐tagged N‐terminal extracellular domain of human HER2 protein (Sino Biological, Cat. Number 10004‐H08H) was diluted at a concentration of 5 and 25 μg/mL in acidic sodium acetate buffer (pH 5.5) and injected on two chip surfaces, until immobilization reached the respective levels of 100 RU and 1,200 RU. The remaining active groups were then deactivated with an injection of ethanolamine hydrochloride (1 M pH 8.5). Reference surfaces were treated similarly except that HER2 injection was omitted. VHH samples were prepared in 50 mM Tris HCl at pH 7.5, 400 mM NaCl and 0.05% P20 surfactant, which constitute the running buffer. All the samples and blanks were injected in duplicate. The HER2 surfaces were regenerated with two successive injections (10 s) of 3 M MgCl₂. The concentration measurements, using CFCA, were performed as follows: the VHH samples (1 nM) were injected for 60 s on the high‐density immobilized surface, at flow rates of 10 μL/min and 80 μL/min. For each sample, the dilution was globally fitted using the CFCA functionality in Biacore T200 evaluation software v 1. For the analysis, a diffusion coefficient of 1.28 × 10⁻¹⁰ m²/s and a VHH molecular weight of 14,024 Da were used. Concentration measurements validation criteria were those recommended by Biacore, that is, a sufficiently high initial binding rate at low flow (about 0.2–0.3 RU.s⁻¹ at 5 μL/min) and limited by mass transport visible as binding curves spread out when binding rates are dependent of flow rate. The degree of mass transport limitation was estimated by the QC ratio which vary from 0 to 1 for kinetic or totally mass transport controlled interaction.35, 36 The kinetic and affinity were measured as follows: VHH samples at concentrations in the 0.25 nM to 30 nM ranges were serially injected for 300 s over reference and HER2 surfaces. Each VHH injection was followed by a 900 s buffer. The flow rate was 50 μL/min. SPR data were processed using the softwares BIAevaluation T200 evaluation version 1 or Scrubber 2.0c (BioLogic Software, Campbell, Australia). Sensorgrams were automatically XY‐zeroed before the injection start and corrected for signals recorded on the empty reference surface. The overlaid sensorgrams were further processed by subtracting the average of the three to five running buffer responses.37, 38 The protein concentration was determined from the binding data on the high‐density HER2 surface using the CFCA evaluation feature of the BIAevaluation T200 software. The association (k _on) and dissociation (k _off) rate constants, the dissociation equilibrium constant (K _D = k _off/k _on), the maximum response R _max were determined using a simple 1:1 L interaction model by regression analysis.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Hartmann L, Botzanowski T, Galibert M, et al. VHH characterization. Comparison of recombinant with chemically synthesized anti‐HER2 VHH. Protein Science. 2019;28:1865–1879. 10.1002/pro.3712