Enhanced Effector Functions Due to Antibody Defucosylation Depend on the Effector Cell Fcγ Receptor Profile

Abs of the IgG isotype are glycosylated in their Fc domain at a conserved asparagine at position 297. Removal of the core fucose of this glycan greatly increases the affinity for FcγRIII, resulting in enhanced FcγRIII-mediated effector functions. Normal plasma IgG contains ∼94% fucosylated Abs, but alloantibodies against, for example, Rhesus D (RhD) and platelet Ags frequently have reduced fucosylation that enhances their pathogenicity. The increased FcγRIII-mediated effector functions have been put to use in various afucosylated therapeutic Abs in anticancer treatment. To test the functional consequences of Ab fucosylation, we produced V-gene–matched recombinant anti-RhD IgG Abs of the four different subclasses (IgG1–4) with and without core fucose (i.e., 20% fucose remaining). Binding to all human FcγR types and their functional isoforms was assessed with surface plasmon resonance. All hypofucosylated anti-RhD IgGs of all IgG subclasses indeed showed enhanced binding affinity for isolated FcγRIII isoforms, without affecting binding affinity to other FcγRs. In contrast, when testing hypofucosylated anti-RhD Abs with FcγRIIIa-expressing NK cells, a 12- and 7-fold increased erythrocyte lysis was observed with the IgG1 and IgG3, respectively, but no increase with IgG2 and IgG4 anti-RhD Abs. Notably, none of the hypofucosylated IgGs enhanced effector function of macrophages, which, in contrast to NK cells, express a complex set of FcγRs, including FcγRIIIa. Our data suggest that the beneficial effects of afucosylated biologicals for clinical use can be particularly anticipated when there is a substantial involvement of FcγRIIIa-expressing cells, such as NK cells.

I mmunoglobulin G is the most abundant class of Abs in human plasma, consisting of four subclasses: IgG1, IgG2, IgG3, and IgG4. Because the four subclasses differ in the structure of their constant regions (Fc domain), recognized by FcgRs and complement component C1q, they have different effector functions. This includes Ab-dependent cellular cytotoxicity (ADCC) and complement-dependent cytotoxicity (1,2).
The family of FcgRs consists of the high-affinity receptor FcgRI and the low-to-medium affinity receptors FcgRII and FcgRIII. All FcgRs, except for FcgRIIb and FcgRIIIb, are activating receptors through an ITAM. This motif is found either within the cytoplasmic tail of FcgRIIa/c or in the cytoplasmic tail of the common g-chain associated with the other activating FcgRs. Alternatively, FcgRIIb contains an ITIM and FcgRIIIb lacks an intracellular signaling motif (3,4). FcgRs bind IgGs of the four subclasses with different affinities. Moreover, polymorphisms in FcgRs influence the binding affinities, most notably for FcgRIIa (FCGR2A-His 131 Arg, with higher affinity for the 131 His variant) and for FcgRIIIa (FCGR3A-Val 158 Phe, with higher affinity for the 158 Val variant). Polymorphisms in FcgRIIIb (e.g., FCGR3B-NA1NA2) do not affect affinity for IgG (3).
All four IgG subclasses contain a conserved asparagine at position 297 to which a glycan is attached. This biantennary glycan consists of a core structure of N-acetylglucosamine and mannose residues and can be variably extended with galactose, sialic acid, core fucose, and a bisecting N-acetylglucosamine (5). The composition of the 297 Asn glycan can influence the quaternary structure of the IgG and thereby the interactions with FcgRs (2,(6)(7)(8). Furthermore, there is a direct carbohydrate-carbohydrate interaction between the Fc glycan at 297 Asn and the FcgR glycan on FcgRIII at 162 Asn, which affects the binding affinity (6,9,10). The glycoform of FcgRIII may change the interaction with IgGs, but this is beyond the scope of this study.
Hypofucosylation of IgG1 has been reported to result in a significantly increased binding to FcgRIIIa and FcgRIIIb (4,6,9,(11)(12)(13). This results in enhanced effector functions, most notably increased ADCC by freshly isolated PBMCs, purified peripheral blood monocytes, or NK cells through FcgRIIIa (11,(14)(15)(16). The increased affinity to the GPI-linked FcgRIIIb seems to adversely affect the phagocytosis function of neutrophils (14,17). In humans, hypofucosylated Abs can arise, as they have been described in alloimmune responses against Rhesus D (RhD) and platelet Ags, as well as in elite controllers of HIV infection (18). The degree of fucosylation correlates with FcgRIIIa-mediated ADCC such that Ab afucosylation enhances pathogenicity (14,19).
With this in mind, it is not surprising that several nonfucosylated therapeutic mAbs have been put to therapeutic use, especially focusing on cancer treatment (4,(20)(21)(22)(23)(24). Furthermore, there are indications that glycan composition plays a role in the working mechanisms of other Ig therapies, such as IVIg, a product of IgG pooled from the plasma of thousands of donors. IVIg is used as replacement therapy in primary immunodeficiencies and as an immune-modulating agent in various autoimmune and immune diseases, among which immune thrombocytopenia (ITP) and autoimmune hemolytic anemia (AIHA) (25)(26)(27)(28)(29)(30). Although the modes of action of IVIg as an immune-modulating agent are still largely unresolved (27), one of the proposed working mechanisms is that IVIg saturates FcgRs on splenic macrophages and thereby inhibits the phagocytosis of platelets or RBCs opsonized by autoantibodies (in case of ITP and AIHA, respectively) (27,29,31,32). Because the degree of Ab fucosylation alters the binding affinity to FcgRIII, it may also affect the working mechanisms of IVIg.
To study the functional extent and significance of changes in core fucosylation of human IgG, we generated a series of anti-RhD mAbs of all four IgG subclasses with and without high levels of fucose. Binding to all human recombinant FcgRs isoforms was assessed using surface plasmon resonance (SPR), and the impact was assessed using primary human NK cells and monocyte-derived macrophages.

Human samples
Peripheral blood from healthy volunteers was obtained in heparinized tubes. For NK cell isolation we only used PBMCs from FCGR-genotyped donors that have two copies of FCGR3-158F and do not have a FCGR2C-ORF allele to exclude FcgRIIc as a confounding factor (33). Genotyping was performed as described before (34,35).
The study was approved by the Medical Ethics Committee of the Academic Medical Center and was performed in accordance with the Declaration of Helsinki.
All IgGs were produced by transient transfection of HEK-freestyle cells (Thermo Fisher Scientific) (37). To block fucosylation, 400 mM 2-deoxy-2fluoro-L-fucose (2FF) was added to the cell suspension (14,38). After 5 d, IgG-containing supernatant from these cells was harvested by spinning twice at maximum speed and subsequent filtration with a 0.45-mm filter.

Ab isolation
IgG was isolated from cell supernatant with affinity chromatography columns HiTrap protein A HP (GE Healthcare) for IgG1, 2, and 4 and HiTrap protein G HP (GE Healthcare) for IgG3 on Ä KTA prime (GE Healthcare) according to standard procedures. Ab fractions were concentrated using protein concentrators, 9-kDa molecular mass cut-off (Pierce; Thermo Fisher Scientific) and dialyzed against PBS using Slide-A-Lyzer dialysis cassettes, 10-kDa molecular mass cut-off (Thermo Fisher Scientific).

Mass spectrometry analysis of Ab glycan
IgG was analyzed by nanoscale liquid chromatography coupled to electrospray ionization quadrupole time-of-flight mass spectrometry on an UltiMate 3000 RSLCnano (Thermo Fisher Scientific) coupled to a maXis Impact micrOTOF (Bruker Daltonics, Bremen, Germany), as described previously (38). The IgG1 glycopeptides (peptide sequence EEQYN-STYR) were eluted first, followed by IgG4 (EEQFNSTYR) and lastly IgG2 (EEQFNSTFR). Tryptic digestion of IgG3 resulted in a glycopeptide bearing a peptide sequence identical to that of IgG4, as well as a miscleaved glycopeptide (LREEQFNSTYR).
Using the three-dimensional Max Xtractor software, intensity values were extracted for each peak within a manually specified m/z window and retention time window. The background-subtracted peak intensity of the first three isotopic peaks in both 2+ and 3+ charge state was summed. For the two types of IgG3 glycopeptides, the intensity of both was summed. The values were subsequently normalized by dividing by the total intensity of all glycopeptides, yielding percentage data for each IgG subclass. These data were used to calculate (a)fucosylation levels, that is, the percentage of N-glycans carrying a core fucose.

Biosensor affinity measurements
Using a Continuous Flow Microspotter (Wasatch Microfluidics, Salt Lake City, UT), biotinylated FcgRs were spotted onto a single SensEye Gstreptavidin sensor (Ssens, Enschede, the Netherlands) for binding affinity measurements of each Ab to any of the spotted FcgRs in a parallel manner on the IBIS MX96 (IBIS Technologies, Enschede, the Netherlands) (39). The biotinylated FcgRs were spotted in 3-fold dilutions, ranging from 100 to 3 nM for FcgRIIb and fusion FcgRIIIb-IgG2-Fc. All of the other FcgRs were spotted in 3-fold dilutions, ranging from 30 to 1 nM in PBS 0.0075% Tween 80 (Amresco, Solon, OH), pH 7.4. The IgGs were then injected over the IBIS at 23 dilution series starting at 0.98 nM up to 2000 nM in PBS 0.075% Tween 80.
For FcgRI affinity measurements, we used his-tagged FcgRI. Biotinylated anti-his-tagged Ab (GenScript, Piscataway, NJ) was spotted in 3-fold dilutions, ranging from 30 to 1 nM. Prior to the IgG injection, 50 nM his-tagged FcgR was injected. The IgGs were then injected over the IBIS at 3-fold dilution series starting at 0.41 nM up to 100 nM as the highest concentration because of the inherently higher affinity.
Regeneration of the IBIS MX96 was carried out after every sample with acid buffer (10 mM Gly-HCl, 0.075% Tween 80, pH 2.5). Calculation of K D was done as described previously (40). Analysis and calculation of all binding data were carried out with Scrubber software version 2 (Biologic Software, Campbell, ACT, Australia).

NK cell ADCC
NK cells were isolated from PBMCs by a CD56 MACS isolation kit (Miltenyi Biotec), according to the manufacturer's description.
Erythrocytes of RhD-positive donors were isolated and labeled with radioactive chromium ( 51 Cr). The cells were subsequently opsonized with anti-RhD mAb in different subclasses and glycovariants at 10 mg/ml for 30 min at 37˚C, after which excess Ab was washed away. Erythrocytes were incubated with NK cells for 2 h at 37˚C in a 1:1 ratio. Afterwards, cytotoxicity was quantified by counting the released chromium with a gamma counter.

Target cell staining and preparation
The amount of Ab deposition on the erythrocyte surface was determined by staining with goat anti-human IgG and analysis by flow cytometry.
To control for the amount of Ab deposition on the red cell surface, we varied Ag density by loading the erythrocytes with different concentrations of the TNP target trinitrobenzene sulfonic acid (TNBS) (0.05-5 mM) (Sigma-Aldrich), after which the erythrocytes were opsonized with anti-TNP with or without fucose at a concentration of 10 mg/ml. The amount of Ab deposition was subsequently determined by staining with goat anti-human IgG.

Culture of monocyte-derived macrophages
Monocytes were isolated and cultured into monocyte-derived macrophages, as described previously (41). In short, monocytes were isolated using a CD14 MACS isolation kit and cultured for 9 d in IMDM, supplemented with 10% FCS, glutamine, and antibiotics, containing either 10 ng/ml GM-CSF or 50 ng/ml M-CSF.

Phagocytosis by monocyte-derived macrophages
Phagocytosis assays with monocyte-derived macrophages were performed as described previously (41). RhD-positive erythrocytes were isolated and stained with CFSE (Life Technologies). The cells were opsonized at an optimal dose of human polyclonal anti-RhD Abs (1.56 IU/ml, RheDQuin; Sanquin, Amsterdam, the Netherlands) or anti-RhD mAbs of different subclasses with and without core fucose at 10 mg/ml for 30 min at 37˚C, or left unopsonized. In other experiments, the level of Ab deposition was controlled by labeling the RBCs with TNBS and subsequently opsonizing with anti-TNP mAb, as described above. After washing excess Ab away, the cells were added to the monocyte-derived macrophages in a ratio of 10:1. After incubation, the nonphagocytized RBCs were lysed and the percentage of phagocytosis was determined by flow cytometry.
Phagocytosis of anti-RhD-opsonized erythrocytes was inhibited by adding anti-TNP Abs with or without fucose (as described above), 5 min prior to the addition of erythrocytes, at concentrations of 0.1, 1.0, or 10 mg/ml.

Statistical analysis
Statistical analysis was performed using GraphPad Prism 6.07. For comparison of IgG-mediated phagocytosis, a Mann-Whitney U test was used. For comparison of expression levels, cytotoxicity, or blocking studies, testing was performed with unpaired t tests.

Results
Abs with low Fc fucosylation were produced by adding 2FF to the culture medium To investigate the effects of Ab fucosylation, we produced IgG against the RhD blood group in the four different subclasses with and without core fucose. We added 2FF, a competitive inhibitor of fucosyl transferase, to the culture medium of the anti-RhD Abproducing HEK cells. This resulted in a decrease in fucosylation from an average of 90% for all IgG subclasses to 26, 13, 29, and 60% for IgG1, IgG2, IgG3, and IgG4, respectively (Fig. 1). For IgG4, the reduction in fucosylation was consistently less efficient than for the other subclasses.

Hypofucosylated Abs have a higher binding affinity to FcgRIII
We subsequently determined the binding affinity of the mAbs of different IgG subclasses with high and low fucose for all human FcgRs and their most relevant allotypic variants (FcgRI, FcgRIIa-131 His, FcgRIIa-131 Arg, FcgRIIb/c, FcgRIIIa-158 Val, FcgRIIIa-158 Phe, FcgRIIIb-NA1, and FcgRIIIa-NA2). We found similar binding affinities for normal (highly fucosylated) IgG as previously reported (Fig. 2) (3). IgG1 and IgG3 bound all FcgR variants, IgG2 only bound FcgRIIa, and IgG4 bound FcgRI, FcgRIIa, FcgRIIb/c, and FcgRIIIa, but not FcgRIIIb. IgG defucosylation increased the binding affinity to FcgRIIIa and FcgRIIIb, with the extent depending on the IgG subclass, but it did not change the binding to any of the other FcgRs (Fig. 2).
We confirmed that the affinity of low-fucosylated IgG1 to FcgRIIIa was 5-and 14-fold increased compared with highfucosylated IgG1 for the FcgRIIIa-158 Val and the FcgRIIIa-158 Phe variant (Fig. 2). Similarly for IgG3, an 11-and 22-fold increase in binding upon hypofucosylation of IgG3 was observed for FcgRIIIa-158 Val and FcgRIIIa-158 Phe, respectively. Hypofucosylation of IgG1 and IgG3 also increased the binding to FcgRIIIb 7-to 8-fold, irrespective of the NA1/NA2 allotype. The affinity of IgG2 to both allotypes of FcgRIIIa became significant and quantifiable upon hypofucosylation, but not to FcgRIIIb. Hypofucosylation of IgG4 also affected its binding to FcgRIIIa, which went from observable but not reliably quantifiable to true reproducible binding. Upon hypofucosylation, a significant and reliable binding of IgG4 was also observed to FcgRIIIb of both allotypes (Fig. 2).

Hypofucosylated IgG1 and IgG3 exert enhanced NK cell effector functions
Because afucosylated Abs have an increased affinity for FcgRIIIa (4,11,13), we performed NK cell ADCC assays to investigate the importance of Ab hypofucosylation in a biological FcgRIIIa system. We used primary NK cells from donors expressing FcgRIIIa as the only IgG receptor on their cell membrane as confirmed by flow cytometry (Fig. 3A).
Normally, FcgR-expressing cells carry out their function in the presence of competing irrelevant IgG, the concentration of which is increased in patients treated with IVIg, which most likely contributes to the effector mechanism of IVIg. We therefore tested whether hypofucosylation of aspecific IgG blocks NK cell-mediated ADCC more efficiently than does fucosylated IgG, using irrelevant recombinant anti-hapten (TNP) IgG1. Only hypofucosylated IgG1 Abs induced a significant inhibition of anti-RhD IgG1-mediated erythrocyte ADCC (Fig. 3H).

Hypofucosylated IgGs do not influence macrophage effector functions
In vivo, FcgR-expressing effector cells other than NK cells are also targeted by IgG-mediated immune responses. We therefore investigated the effect of Ab fucosylation with human macrophages, which express a broader spectrum of FcgRs than NK cells and are considered relevant in immune-mediated clearance reactions by tissue-resident macrophages in liver and spleen.
We compared human GM-CSF-and M-CSF-cultured macrophages, representing the so-called M1 and M2 type of pro-and anti-inflammatory macrophages (41). These macrophages differ in their relative expression of FcgRI, with GM-CSF-cultured macrophages expressing more of this receptor. The anti-inflammatory M-CSF-cultured macrophages express significantly more FcgRIIa than the GM-CSF-cultured macrophages (Fig. 4A, 4B). Both cell types have considerably less FcgRIIIa expression than NK cells. GM-CSF-cultured macrophages phagocytize significantly more anti-RhD IgG3-opsonized RBCs than IgG1-opsonized cells; IgG4opsonized RBCs were phagocytized to an even lower extent, whereas IgG2-opsonized cells were not phagocytized at all. M-CSF-cultured macrophages phagocytized anti-RhD IgG1-, IgG3-, and IgG4-opsonized erythrocytes to a similar extent, whereas IgG2opsonized cells were again not phagocytized. Upon glycoengineering, all four anti-RhD IgG subclasses generated as hypofucosylated isoforms showed identical opsonizing capacity of the erythrocytes. No increase in phagocytosis was observed upon hypofucosylation of any IgG subclass (Fig. 4C).
To test whether the Ag density and corresponding level of opsonizing Ab binding would contribute to the difference between myeloid and NK cell effector cells, we made use of our mAb against TNP. Erythrocytes were TNP haptenized to different degrees and opsonized with either normally fucosylated or hypofucosylated anti-TNP IgG1 mAb (Fig. 5A). When performing phagocytosis assays with these erythrocytes now expressing a graded level of TNP, we found a strong positive correlation between the level of TNP-labeling, opsonization efficiency, and the percentage of uptake by macrophages (p , 0.0001). Still, we did not observe any difference in erythrocyte phagocytosis with respect to the fucosylation state of the opsonizing mAb (Fig. 5B).
We then tested whether hypofucosylated IgG1 was superior or not in blocking the phagocytosis of Ab-opsonized erythrocytes by macrophages. In this setting we used a clinically applied polyclonal anti-RhD product purified from plasma (RheDQuin; Sanquin) for Ab deposition (Fig. 5C). As expected from previous experiments (41), IgG Abs can block phagocytosis by competing in IgG binding to the FcgRs in a dose-dependent manner. Again under our conditions of macrophage phagocytosis, uptake of anti-RhDopsonized erythrocytes was blocked irrespective of the fucosylation status of the blocking IgG (Fig. 5D).
Collectively, our data demonstrate that the fucosylation state of IgG greatly impacts the binding affinity to FcgRIIIa, but whether this affinity change results in an enhanced effector function depends on the effector cell type and its level of FcgRIII expression.

Discussion
In the present study we investigated the functional effects of Ab hypofucosylation in different cellular systems. Whereas .90% of IgG normally contains fucose, we reduced this level to ,30% upon applying a well-characterized bioengineering method (14,38). Human NK cells, which express FcgRIIIa as their only IgG receptor, showed increased affinity of hypofucosylated IgG1 and IgG3 for FcgRIIIa, whereas no such increased ADCC was observed with IgG2 and IgG4. Even though a gain of function of   IgG2 and IgG4 has been previously reported upon fucose removal (1,42), and was observed in our study with isolated recombinant FcgRs tested in an SPR array, the enhanced binding affinity of nonfucosylated IgG2 and IgG4 must be considered insufficient to exert any biological effect under our conditions. In other cellular systems, hypofucosylated IgG2 and IgG4 may increase ADCC, as was shown by Niwa et al. (1), who reported enhanced ADCC of Daudi cells opsonized with anti-CD20.
Besides the increased binding affinity to FcgRIIIa and FcgRIIIb upon Ab defucosylation, no effect to other FcgRs is known. There is equivalent binding of defucosylated and fucosylated IgG1 to the neonatal Fc receptor (11), implying that defucosylated Abs have a half-life comparable to fucosylated Abs. Finally, fucose removal also does not change the binding to C1q and have equal complement-dependent cytotoxicity as fucosylated Abs (1, 11) (G. Dekkers et al., manuscript in preparation).
To compare the enhanced effector function of NK cells with hypofucosylated IgG, we also studied human macrophages. We previously reported that GM-CSF-cultured macrophages mainly phagocytize via FcgRI, but also, albeit less, through FcgRIIIa (41). Consistent with IgG1 and IgG3 being the major IgG subclasses recognized by both FcgRI and FcgRIIIa, we found that IgG1 and IgG3 showed significant phagocytosis capacity, whereas IgG2-opsonized erythrocytes were not phagocytized, in accordance with IgG2 not being recognized by FcgRI. IgG4 is recognized by FcgRI, resulting in less phagocytosis of IgG4-opsonized RBCs compared with IgG1-and IgG3-opsonized cells. In accordance with the major role of FcgRI in this system (41) and the main effect of hypofucosylation being restricted to FcgRIII, no role for IgG hypofucosylation was found using macrophages.
Ab hypofucosylation did not affect macrophage effector function. This could be explained by the fact that macrophages express a variety of FcgRs, including FcgRI and FcgRIIa (41). Apparently, the contribution of FcgRIIIa to the phagocytosis process is too small to be affected by Ab defucosylation in our phagocytosis assay.
We also tested the impact of IgG fucosylation on macrophage effector function in a different approach, as Ag-specific Abs can also compete for FcgR-binding with the overwhelming IgG concentration found in plasma (41,43). In this study, we showed that adding an irrelevant IgG mAb to macrophages prior to the addition of anti-RhD-opsonized erythrocytes inhibited phagocytosis in a dosedependent manner, similar to the addition of IVIg (41). Although the precise working mechanism of IVIg is still debated (32), one hypothesis is that IVIg saturates FcgRs on splenic and/or liver macrophages, hence inhibiting the phagocytosis of platelets or RBCs opsonized by autoantibodies (in the case of ITP and AIHA, respectively) (27,29,31,32). Our studies indicate that irrelevant anti-TNP mAbs inhibit phagocytosis of opsonized red cells by macrophages irrespective the level of fucosylation. If saturation of FcgRs were indeed the major working mechanism of IVIg, defucosylation alone would not alter the efficacy of IVIg treatment.
Recently, nonfucosylated mAbs have been generated for clinical use, because the increased affinity for FcgRIIIa is thought to improve effector functions against cancer (20,21,44,45). In this study, we show that this approach may represent an oversimplified After opsonization, the cells were stained with goat anti-human Ig and analyzed by flow cytometry to measure the amount of Ab deposition. The median fluorescence intensity is shown on the y-axis (n = 2-5). (B) Phagocytosis of TNPylated erythrocytes opsonized with anti-TNP with (black) and without fucose (red) by monocyte-derived macrophages cultured with GM-CSF (left) and M-CSF (right). The median fluorescence intensity as a measure of Ab deposition on erythrocytes is shown on the x-axis. The percentage of positive macrophages, normalized to the amount of macrophages that phagocytized anti-D RheDQuin-opsonized erythrocytes, is shown on the y-axis (n = 9-10). (C) Ab deposition on erythrocytes after the erythrocytes were opsonized with RheDQuin (polyclonal anti-RhD). The cells were stained with goat anti-human Ig and analyzed by flow cytometry. The median fluorescence intensity is shown on the y-axis (n = 54). (D) Phagocytosis of anti-RhD RheDQuin opsonized erythrocytes upon the addition of anti-TNP (black) and anti-TNP 12FF (gray) nonspecific blocking Abs, relative to the percentage of positive macrophages in unblocked macrophages that phagocytized anti-RhD RheDQuinopsonized erythrocytes. The concentration of blocking Ab is depicted on the x-axis (n = 4). **p , 0.01. ns, not significant. strategy and will definitely not work in all clinical approaches equally well. We suggest that the clinical efficacy of nonfucosylated Abs will also depend on the cell type predominating in a certain biological effect in vivo. In processes in which NK cells play a predominant role, it would be beneficial to defucosylate therapeutic Abs. However, in a more complex cellular system, where multiple FcgRs contribute to the cellular effector function, no difference may be expected between fucosylated and defucosylated Abs.
In sum, alteration of the 297 Asn glycan of IgG Abs may enhance the clinical response depending on the FcgR expression pattern of the effector cell type that predominates. Alternatively, altered levels of IgG fucosylation may be only one of the alterations that could be used to afford enhanced binding activity and effector function beyond the FcgRIII expression, processes that we are currently studying in much greater detail by further bioengineering steps.

Disclosures
The authors have no financial conflicts of interest.