**1. Introduction**

Animals have been used for research applications since the early centuries after Christ [1]. This practice has always been controversial. However, only in 1870, after discovering that animals feel pain, the theme began to be reconsidered [2]. Yet, the first significant milestone involving ethical issues in using animals for research occurred in 1954, when Charles Hume and William Russel proposed the concept of the 3Rs. They advocated "Replacing, Reducing and Refining" the use of animals to minimize pain or stress whenever possible [3]. This conception was further extended to the 6Rs to include "Read across", referring to the critical analysis of new results, "Relevance", which concerns ethical and educational visions, including good laboratory practices, and finally "Roadmaps," which evolves planning, communication, conference and technical implementation policies [4]. Recently, the American

#### *Monoclonal Antibodies*

Anti-Vivisection Society (AAVS) banned the production of ascites in animals, launching the "antibodies without animals" campaign [5, 6].

In view of this, the conventionally used strategies to generate and produce monoclonal antibodies (mAbs), initially proposed by Köhler and Milstein in 1975 [7] as the hybridoma technology, had to be reconsidered. This methodology is based on the immunization of animals with the antigen of interest, followed by the fusion of B lymphocytes with myeloma cells, resulting in the formation of hybridomas. After cloning and selection, the antibody-secreting stable monoclonal cell lines were used to produce mAb-enriched ascitic liquid [8], a now-banned practice. For many years, mAbs, which have wide applicability in research, diagnosis, and treatment, were generated and produced with this methodology. In this chapter, the main technologies that emerged as alternatives to the use of animals for the generation and production of mAbs are discussed.

#### **1.1 Overview of antibody structure and most common formats**

mAbs are immunoglobulin molecules with a molecular weight of ~150 kDa, made up of four polypeptide chains: one pair of identical light chains and another pair of identical heavy chains joined by disulfide and non-covalent bonds. Each chain contains a variable domain (VL and VH) at the N-terminal portion and one or three constant domains at the polypeptide's C-terminal portion. The antigenbinding fragment (Fab) has ~50 kDa and is composed of variable and constant regions of heavy and light chains. The variable fragment (Fv) includes only the two variable domains of both chains (**Figure 1a**) [9]. Each variable region is composed

#### **Figure 1.**

*Schematic representation of different antibody formats: (a) the classic IgG and its respective regions and chains; (b) monovalent antigen-binding formats: Fab and scFv; (c) bivalent antigen-binding formats: Fab'2, diabody and minibody. Different textures indicate different antigen specificities; (d) structures composed of scFv can form diabody, triabody and tetrabody; (e) other possible formats that can be constructed; (f) camelid Ig and shark IgNAR molecules.*

**25**

therapy [22].

*Alternative Methods to Animal Use for Monoclonal Antibody Generation and Production*

of six complementarity determining regions (CDRs), also known as hypervariable regions: three in the light chain (L1, L2, and L3) and three in the heavy chain (H1, H2, and H3), CDR H3 being the most variable in length, sequence, and structure. These regions promote the high specificity of functional binding of mAbs with the target antigen [10]. In spite of hypervariable regions, it should be noted that these regions assume conformations in the form of loops because of the presence of some

There are several formats of antibodies used for therapeutic, diagnostic, and basic research purposes. Smaller formats were initially generated by the removal of the constant domain (Fc) by proteolysis, with the enzymes papain and pepsin, and later by genetic manipulations, giving rise to monovalent antigen-binding formats: Fab and scFv (variable single-chain fragment) (**Figure 1b**) or bivalent formats such as Fab'2, diabody, minibody, among others (**Figure 1c**). These antibodies present characteristics in comparison to complete mAbs that may be useful depending on the application: besides retaining the antigen-binding affinity of the parental antibody, they have a reduced serum half-life and are less immunogenic [12, 13]. In addition to conventional shapes, camelids and sharks produce unusual antibodies composed only of heavy chains, with just a single domain in its variable antigenbinding site. In Camelidae, it is called variable domain of heavy chain antibodies (VhHs), while in some cartilaginous fish, like sharks, it is called new variable antigen receptor (V-NARs). The smaller sizes of VhHs and V-NARs allow them to be

Fab favors the crystallization of several proteins facilitating the determination of their three-dimensional structure [15]. Therapeutic Fabs have been available since 1994, starting with the chimeric abciximab Fab (ReoPro), used as an antiplatelet agent. Ranibizumab (Lucentis) was approved by Food and Drug Administration (FDA) in 2006 for treatment of age-related macular degeneration, certolizumab pegol (Cimzia) was approved in 2009 for rheumatoid arthritis, and several other Fabs are now in clinical and pre-clinical trials. This format is also useful for diagnostic imaging, like arcitumomab (CEA-scan) approved in 1996 for colorectal

Another variant is scFv antibodies, with a molecular mass of ~30 kDa, composed of VH and VL domains joined with a peptide ligand. These structures can be presented as dimers, trimers, tetramers (**Figure 1d**) or other formats (**Figure 1e**) through genetic or chemical manipulations. Diabodies can present two identical antigen-binding sites, when it is called a bivalent diabody, or have two different antigen-binding sites, a bispecific diabody [19, 20]. Numerous scFvs have been constructed against haptens, proteins, carbohydrates, receptors, tumor antigens, and viruses for applications in therapy and diagnosis [21]. In 2019, the FDA approved the humanized scFv brolucizumab (Beovu) to treat neovascular age-related macular degeneration. Other scFvs are already in a pre-clinical study for targeted cancer

**2. Alternative methods for monoclonal antibody generation**

Antibody display libraries are powerful tools to isolate high-affinity antibodies for therapeutic and/or diagnostics applications. They can be divided into two groups: cell surface and cell-free display libraries. In the first case, the antibody is expressed on the surface of bacteria, yeasts, or mammalian cells, using own internal machinery [22] and, in the second, extracts of prokaryotic or eukaryotic

good candidates as biotechnological tools (**Figure 1f**) [14–16].

*DOI: http://dx.doi.org/10.5772/intechopen.95485*

conserved residues [11].

cancer screening [12, 17, 18].

**2.1 Antibody display libraries**

#### *Alternative Methods to Animal Use for Monoclonal Antibody Generation and Production DOI: http://dx.doi.org/10.5772/intechopen.95485*

of six complementarity determining regions (CDRs), also known as hypervariable regions: three in the light chain (L1, L2, and L3) and three in the heavy chain (H1, H2, and H3), CDR H3 being the most variable in length, sequence, and structure. These regions promote the high specificity of functional binding of mAbs with the target antigen [10]. In spite of hypervariable regions, it should be noted that these regions assume conformations in the form of loops because of the presence of some conserved residues [11].

There are several formats of antibodies used for therapeutic, diagnostic, and basic research purposes. Smaller formats were initially generated by the removal of the constant domain (Fc) by proteolysis, with the enzymes papain and pepsin, and later by genetic manipulations, giving rise to monovalent antigen-binding formats: Fab and scFv (variable single-chain fragment) (**Figure 1b**) or bivalent formats such as Fab'2, diabody, minibody, among others (**Figure 1c**). These antibodies present characteristics in comparison to complete mAbs that may be useful depending on the application: besides retaining the antigen-binding affinity of the parental antibody, they have a reduced serum half-life and are less immunogenic [12, 13]. In addition to conventional shapes, camelids and sharks produce unusual antibodies composed only of heavy chains, with just a single domain in its variable antigenbinding site. In Camelidae, it is called variable domain of heavy chain antibodies (VhHs), while in some cartilaginous fish, like sharks, it is called new variable antigen receptor (V-NARs). The smaller sizes of VhHs and V-NARs allow them to be good candidates as biotechnological tools (**Figure 1f**) [14–16].

Fab favors the crystallization of several proteins facilitating the determination of their three-dimensional structure [15]. Therapeutic Fabs have been available since 1994, starting with the chimeric abciximab Fab (ReoPro), used as an antiplatelet agent. Ranibizumab (Lucentis) was approved by Food and Drug Administration (FDA) in 2006 for treatment of age-related macular degeneration, certolizumab pegol (Cimzia) was approved in 2009 for rheumatoid arthritis, and several other Fabs are now in clinical and pre-clinical trials. This format is also useful for diagnostic imaging, like arcitumomab (CEA-scan) approved in 1996 for colorectal cancer screening [12, 17, 18].

Another variant is scFv antibodies, with a molecular mass of ~30 kDa, composed of VH and VL domains joined with a peptide ligand. These structures can be presented as dimers, trimers, tetramers (**Figure 1d**) or other formats (**Figure 1e**) through genetic or chemical manipulations. Diabodies can present two identical antigen-binding sites, when it is called a bivalent diabody, or have two different antigen-binding sites, a bispecific diabody [19, 20]. Numerous scFvs have been constructed against haptens, proteins, carbohydrates, receptors, tumor antigens, and viruses for applications in therapy and diagnosis [21]. In 2019, the FDA approved the humanized scFv brolucizumab (Beovu) to treat neovascular age-related macular degeneration. Other scFvs are already in a pre-clinical study for targeted cancer therapy [22].

### **2. Alternative methods for monoclonal antibody generation**

#### **2.1 Antibody display libraries**

Antibody display libraries are powerful tools to isolate high-affinity antibodies for therapeutic and/or diagnostics applications. They can be divided into two groups: cell surface and cell-free display libraries. In the first case, the antibody is expressed on the surface of bacteria, yeasts, or mammalian cells, using own internal machinery [22] and, in the second, extracts of prokaryotic or eukaryotic

*Monoclonal Antibodies*

Anti-Vivisection Society (AAVS) banned the production of ascites in animals,

In view of this, the conventionally used strategies to generate and produce monoclonal antibodies (mAbs), initially proposed by Köhler and Milstein in 1975 [7] as the hybridoma technology, had to be reconsidered. This methodology is based on the immunization of animals with the antigen of interest, followed by the fusion of B lymphocytes with myeloma cells, resulting in the formation of hybridomas. After cloning and selection, the antibody-secreting stable monoclonal cell lines were used to produce mAb-enriched ascitic liquid [8], a now-banned practice. For many years, mAbs, which have wide applicability in research, diagnosis, and treatment, were generated and produced with this methodology. In this chapter, the main technologies that emerged as alternatives to the use of animals for the genera-

mAbs are immunoglobulin molecules with a molecular weight of ~150 kDa, made up of four polypeptide chains: one pair of identical light chains and another pair of identical heavy chains joined by disulfide and non-covalent bonds. Each chain contains a variable domain (VL and VH) at the N-terminal portion and one or three constant domains at the polypeptide's C-terminal portion. The antigenbinding fragment (Fab) has ~50 kDa and is composed of variable and constant regions of heavy and light chains. The variable fragment (Fv) includes only the two variable domains of both chains (**Figure 1a**) [9]. Each variable region is composed

*Schematic representation of different antibody formats: (a) the classic IgG and its respective regions and chains; (b) monovalent antigen-binding formats: Fab and scFv; (c) bivalent antigen-binding formats: Fab'2, diabody and minibody. Different textures indicate different antigen specificities; (d) structures composed of scFv can form diabody, triabody and tetrabody; (e) other possible formats that can be constructed; (f) camelid* 

launching the "antibodies without animals" campaign [5, 6].

**1.1 Overview of antibody structure and most common formats**

tion and production of mAbs are discussed.

**24**

**Figure 1.**

*Ig and shark IgNAR molecules.*

cells, mainly from rabbit or wheat germ, are used to transcribe and translate genetic information contained in the library [23, 24]. Although they use different paths, these methodologies have a common property: they are potential alternative methods to the use of animals for antibody generation.

These methodologies involve two main steps: the construction of the library and the selection of the antibody of interest. According to the origin of the genetic material used in their construction, the libraries are classified as naïve, immune, synthetic, or semi-synthetic. The immune library is obtained from animals or humans that have been immunized and developed antibodies against a particular antigen [25, 26]. The other types are known as universal libraries: naïve is cloned from non-immunized donors; the semisynthetic is created using both naturally and synthetically (*in silico*) randomized CDRs, which increases the diversity of the library without requiring a large number of donors; and synthetic, based only in *in silico* design and gene synthesis to optimize individual amino acids, hence expanding its diversity, expression, and stability [27].

#### *2.1.1 Cell surface display*

Among the display libraries, the phage display stands out for being the first described and currently more used in the generation of antibodies [28]. The technique was developed by George Smith in 1985 and it uses the bacteriophage's ability to infect bacteria. In this way, a foreign DNA sequence is inserted into the genes III or VIII, which encode the pIII and pVIII coat proteins, respectively. The recombinant protein is displayed on the outer surface of the phage as a fusion protein in an immunologically accessible form [29]. The displayed antibodies can be of either Fab or scFv formats. The phage display library is generated by assembling DNA sequences that encode antibody fragments in the phage or phagemid vectors. The phage vector has a complete phage genome, but it is not effective for large proteins. A phagemid vector is a plasmid that contains phage coat gene (gIII or gVIII), and phage and plasmid's origin of replication. The vectors are used to transform *E. coli*. The phage vector has all the ability to produce phage particles and display the fusion antibody, while the phagemid needs to infect the bacterium with a helper phage to enable the recombinant DNA package, as a single-strand DNA into virion particles, and to display the antibody fragments [25, 28, 30]. The screening of displayed antibodies is performed by biopanning, a process in which the phages are incubated with the immobilized antigen, and the non-binding phages are removed by extensive washing. The bound phages are then eluted and enriched by reinfection of *E. coli* and thus successive rounds of selection can be carried out as many times as needed [31].

Bacterial display, an alternative to phage display, allows libraries of greater diversity. In this system, the expression of recombinant proteins is easier and the transformation by DNA is more efficient than phage display. The methodology is fast, easy to handle, and eliminates the stage of infection by the phage. The library can be displayed on the membrane in the periplasmic space or fused to the filament flagellar or fimbrial adhesin proteins [32, 33]. To generate the bacterial library, the sequences of DNA, encoding scFv or Fab fragments, are inserted into the appropriate display vector used to transform *E. coli*, the most common bacterial strain used in this technique [34]. The target antigen, adsorbed on magnetic beads or fluorophore-labeled, is used for the screening of antibody libraries by cell sorting [35].

In the cases of yeast and mammalian displays, eukaryotic systems, folding and post-translational modifications, which are relevant to the function and stability of the antibody, are more effective when compared to what occurs in the prokaryotic system [22]. Briefly, the yeast library is created by linking the antibody

**27**

*Alternative Methods to Animal Use for Monoclonal Antibody Generation and Production*

gene sequence into suitable yeast display vectors. The transformed yeast by the plasmid generated is induced to express the recombinant library. The system can display scFv, Fab, or full-length antibody formats [36], that are expressed in fusion with anchor proteins of the glycosylphosphatidylinositol (GPI) family, such as α-agglutinin and a-agglutinin. The screening is performed by cell sorting [37]. When using the mammalian system, the post-translational modifications are still more effective than those possible in the yeast system. The antibody library can be usually displayed on the surface of Human Embryonic Kidney 293 T (HEK 293 T) and Chinese Hamster Ovary (CHO) cells after transient or stable transformation [38]. The antibody expressed is fused to the transmembrane domain of human platelet-derived growth factor receptor (PDGFR), which anchors the antibody on the outer surface of the cell membrane [39]. The library screening is performed as already described for other cell-surface display systems [40].

The cell-free display libraries, unlike previously described ones, do not depend on the efficiency of transduction or transfection. Among them, the ribosome and mRNA systems have been the most described, offering around 1012–1014 variants, a

–1010) systems [41]. While the ribosome display connects nascent proteins to their encoding mRNA through the generation of stable protein–ribosome–mRNA complexes, the mRNA system uses an antibiotic, puromycin, that mimics the structure of an aminoacylated tRNA, to modify mRNA that also is linked to its respective nascent protein. In a brief description, a ribosome system display construct is designed to be used with cell extracts to allow the downstream mRNA synthesis. In the construct, it must be present a ribosome binding site to the start codon where protein synthesis begins, for recruitment and pairing of the ribosome. The open reading frame is followed by the library of binding proteins and a spacer. The spacer provides flexibility to the display library in order to fold outside of the ribosome tunnel. Another important point is that the ribosome is stalled at the 3′-end by deleting the stop codon to

), eukaryotic

wider diversity over other display techniques, like phage (around 109

couple the nascent polypeptide with its encoding mRNA [23, 42, 43].

The mRNA display stands out in relation to the ribosome display for offering a large degree of control over experimental conditions [41]. Briefly, the DNA antibody library is *in vitro* transcribed in mRNA. In a second step, a covalent interaction between mRNA and puromycin is produced, providing after translation the formation of the mRNA-puromycin-protein complex, which is reverse transcribed into cDNA to obtain the heteroduplex cDNA-mRNA, more stable than mRNA alone. After screening, the selected cDNA is amplified by PCR. The amplified constructs are subjected to a new cycle to obtain the mRNA-puromycin-protein complexes, and then the heteroduplexes obtained are ready for a new round of screening [41, 43]. Both methodologies enable the generation of different formats of the antibodies,

including the full-length ones. The selection of the antibodies of interest is performed by binding to an immobilized antigen. Ribosome and mRNA systems have gained relevance for allowing efficient and low-cost antibody production and for their advantageous ability to screen large libraries. Although promising, so far there

Advances in DNA and protein sequencing techniques associated with X-ray crystallography approaches to evaluate the antibody structure at an atomic level

is no commercially available antibody generated by cell-free technologies.

**2.2 Antibody design via** *in silico* **modeling**

*DOI: http://dx.doi.org/10.5772/intechopen.95485*

*2.1.2 Cell-free display*

), and prokaryotic (108

(106 –107

#### *Alternative Methods to Animal Use for Monoclonal Antibody Generation and Production DOI: http://dx.doi.org/10.5772/intechopen.95485*

gene sequence into suitable yeast display vectors. The transformed yeast by the plasmid generated is induced to express the recombinant library. The system can display scFv, Fab, or full-length antibody formats [36], that are expressed in fusion with anchor proteins of the glycosylphosphatidylinositol (GPI) family, such as α-agglutinin and a-agglutinin. The screening is performed by cell sorting [37].

When using the mammalian system, the post-translational modifications are still more effective than those possible in the yeast system. The antibody library can be usually displayed on the surface of Human Embryonic Kidney 293 T (HEK 293 T) and Chinese Hamster Ovary (CHO) cells after transient or stable transformation [38]. The antibody expressed is fused to the transmembrane domain of human platelet-derived growth factor receptor (PDGFR), which anchors the antibody on the outer surface of the cell membrane [39]. The library screening is performed as already described for other cell-surface display systems [40].

#### *2.1.2 Cell-free display*

*Monoclonal Antibodies*

cells, mainly from rabbit or wheat germ, are used to transcribe and translate genetic information contained in the library [23, 24]. Although they use different paths, these methodologies have a common property: they are potential alternative

These methodologies involve two main steps: the construction of the library and the selection of the antibody of interest. According to the origin of the genetic material used in their construction, the libraries are classified as naïve, immune, synthetic, or semi-synthetic. The immune library is obtained from animals or humans that have been immunized and developed antibodies against a particular antigen [25, 26]. The other types are known as universal libraries: naïve is cloned from non-immunized donors; the semisynthetic is created using both naturally and synthetically (*in silico*) randomized CDRs, which increases the diversity of the library without requiring a large number of donors; and synthetic, based only in *in silico* design and gene synthesis to optimize individual amino acids, hence expand-

Among the display libraries, the phage display stands out for being the first described and currently more used in the generation of antibodies [28]. The technique was developed by George Smith in 1985 and it uses the bacteriophage's ability to infect bacteria. In this way, a foreign DNA sequence is inserted into the genes III or VIII, which encode the pIII and pVIII coat proteins, respectively. The recombinant protein is displayed on the outer surface of the phage as a fusion protein in an immunologically accessible form [29]. The displayed antibodies can be of either Fab or scFv formats. The phage display library is generated by assembling DNA sequences that encode antibody fragments in the phage or phagemid vectors. The phage vector has a complete phage genome, but it is not effective for large proteins. A phagemid vector is a plasmid that contains phage coat gene (gIII or gVIII), and phage and plasmid's origin of replication. The vectors are used to transform *E. coli*. The phage vector has all the ability to produce phage particles and display the fusion antibody, while the phagemid needs to infect the bacterium with a helper phage to enable the recombinant DNA package, as a single-strand DNA into virion particles, and to display the antibody fragments [25, 28, 30]. The screening of displayed antibodies is performed by biopanning, a process in which the phages are incubated with the immobilized antigen, and the non-binding phages are removed by extensive washing. The bound phages are then eluted and enriched by reinfection of *E. coli* and thus successive rounds of selection can be carried out as many times as

Bacterial display, an alternative to phage display, allows libraries of greater diversity. In this system, the expression of recombinant proteins is easier and the transformation by DNA is more efficient than phage display. The methodology is fast, easy to handle, and eliminates the stage of infection by the phage. The library can be displayed on the membrane in the periplasmic space or fused to the filament flagellar or fimbrial adhesin proteins [32, 33]. To generate the bacterial library, the sequences of DNA, encoding scFv or Fab fragments, are inserted into the appropriate display vector used to transform *E. coli*, the most common bacterial strain used in this technique [34]. The target antigen, adsorbed on magnetic beads or fluorophore-labeled, is used for the screening of antibody libraries by cell sorting [35]. In the cases of yeast and mammalian displays, eukaryotic systems, folding and post-translational modifications, which are relevant to the function and stability of the antibody, are more effective when compared to what occurs in the prokaryotic system [22]. Briefly, the yeast library is created by linking the antibody

methods to the use of animals for antibody generation.

ing its diversity, expression, and stability [27].

*2.1.1 Cell surface display*

**26**

needed [31].

The cell-free display libraries, unlike previously described ones, do not depend on the efficiency of transduction or transfection. Among them, the ribosome and mRNA systems have been the most described, offering around 1012–1014 variants, a wider diversity over other display techniques, like phage (around 109 ), eukaryotic (106 –107 ), and prokaryotic (108 –1010) systems [41].

While the ribosome display connects nascent proteins to their encoding mRNA through the generation of stable protein–ribosome–mRNA complexes, the mRNA system uses an antibiotic, puromycin, that mimics the structure of an aminoacylated tRNA, to modify mRNA that also is linked to its respective nascent protein. In a brief description, a ribosome system display construct is designed to be used with cell extracts to allow the downstream mRNA synthesis. In the construct, it must be present a ribosome binding site to the start codon where protein synthesis begins, for recruitment and pairing of the ribosome. The open reading frame is followed by the library of binding proteins and a spacer. The spacer provides flexibility to the display library in order to fold outside of the ribosome tunnel. Another important point is that the ribosome is stalled at the 3′-end by deleting the stop codon to couple the nascent polypeptide with its encoding mRNA [23, 42, 43].

The mRNA display stands out in relation to the ribosome display for offering a large degree of control over experimental conditions [41]. Briefly, the DNA antibody library is *in vitro* transcribed in mRNA. In a second step, a covalent interaction between mRNA and puromycin is produced, providing after translation the formation of the mRNA-puromycin-protein complex, which is reverse transcribed into cDNA to obtain the heteroduplex cDNA-mRNA, more stable than mRNA alone. After screening, the selected cDNA is amplified by PCR. The amplified constructs are subjected to a new cycle to obtain the mRNA-puromycin-protein complexes, and then the heteroduplexes obtained are ready for a new round of screening [41, 43].

Both methodologies enable the generation of different formats of the antibodies, including the full-length ones. The selection of the antibodies of interest is performed by binding to an immobilized antigen. Ribosome and mRNA systems have gained relevance for allowing efficient and low-cost antibody production and for their advantageous ability to screen large libraries. Although promising, so far there is no commercially available antibody generated by cell-free technologies.

#### **2.2 Antibody design via** *in silico* **modeling**

Advances in DNA and protein sequencing techniques associated with X-ray crystallography approaches to evaluate the antibody structure at an atomic level and the increasing availability of the generated data in public domains provided a fundamental basis to the in-silico generation of mAbs.

Computer-assisted design of new mAbs consists of high-throughput algorithm analyses of antibody structures modeled from query residue sequences. These models are typically obtained by homology with precompiled antibody scaffold templates [44], which is possible because, despite the unique spatial identity of mAbs, the geometry of their variable regions is well conserved, with most CDR loops having a limited number of conformations, known as canonical classes [11]. In general, the established modeling tools coupled with refined protein–protein docking [45] and machine learning methods have been found useful for predicting the VH and VL domain arrangements and the potential antibody electrostatic complementary interface [46].

Examples of platforms available for antibody modeling are the "Prediction of ImmunoGlobulin Structure" (PIGS) [47], the Rosetta Antibody Modeling [48], and the "Web Antibody Modeling" (WAM) [49]. These servers comprise fully automated homology-based modules that predict with high accuracy the tridimensional antibody structure, including most of the hypervariable regions of the antigenbinding site [50]. An exception is the H3 loop. Unlike the other CDRs, the H3 structure has unique conformations that do not follow a canonical form and are also not found in any described protein, with ~75% of its fragments not having structural neighbors in the known non-immunoglobulin protein world [51]. Therefore, the H3 loop cannot be predicted by selecting templates from a database and this is an important obstacle for the *in silico* antibody design. Some alternative algorithms, based on candidate conformations obtained computationally and energy functions, have been developed, but they often fail to produce sub-angstrom structure models [50, 52] and the problem persists.

Other concerns also affect antibody modeling. The limited number of highquality X-ray crystal structures of mAbs in public protein databases may not be sufficient to allow a proper antibody shape prediction [50]. Regarding the docking protocols, it should be noted that, despite the great advances in the bioinformatic field, most of the antibody algorithms still need to be optimized to consider the molecular backbone flexibility and the transient conformational changes following protein–protein interactions [53, 54]. Another relevant point is the time needed for antibody modeling. High-throughput computational design of mAbs can still be as time-consuming as experimental cellular approaches, even when well-consolidated prediction systems are used. As an example, the Rosetta Antibody server was previously found to take 570,000 CPU hours to generate ~2,000 antibody models [55].

With many challenges ahead, currently there are few reports of functional antibodies completely designed by in-silico approaches. A successful attempt in this field is the mAb described by Nimrod and co-workers, which was based on robust predictions of specific residue-residue interactions rather than modeling the entire antigen–antibody complex [56]. On the other hand, computational protocols have been used with increasing frequency to improve the physicochemical properties of previously generated mAbs, as well as to engineer humanized versions of murine full-length immunoglobulins, making them like those found in humans [57]. Molecular structure-based iterative algorithms have been shown to optimize the generation of humanized antibody scaffolds without a significant drop in affinity and specificity toward the antigen, compared to the original murine one, and with reduced occurrence of structure failures, important drawbacks commonly found following conventional humanization techniques, which are mostly guided by linear antibody residue sequences [58].

The overall computational antibody discovery scenario is promising and, although the design of new biologically active mAbs is still deeply dependent on

**29**

*Alternative Methods to Animal Use for Monoclonal Antibody Generation and Production*

living animals, the advances in structure prediction methods set the scene for an ongoing technological evolution that should potentially lead the future generation

As previously mentioned, one of the major utilities of the mammalian expression

The primary technique for mAb obtention was already originally dependent on a mammalian cell: the hybridoma cell [7]. As the high specificity of the monoclonal antibodies was making these molecules increasingly useful for various applications, a hybridoma large-scale cultivation became a great demand in the industry. Therefore, the ascites method production was no longer enough to supply bulk production, nor feasible due to ethical matters. That way, most research and diagnostic proposed mAbs are now produced *in vitro,* through the harvest and following purification of mAb-enriched media obtained in dynamic or non-dynamic cell

In a therapeutic context, although the hybridoma cell lines are still responsible for the generation of more than 50% of the FDA-approved mAbs [63], these antibodies are bulk produced in other mammalian host systems [64]. This is due mainly to the highly immunogenic nature of murine mAbs for humans, demanding the antibodies to be genetically modified (humanization or generation of fragments) for human therapeutic use [65]. Besides, many of these mammalian cells had their expression machinery highly optimized for recombinant protein production [66]. In the mammalian expression system, cells are readily transfected or transduced to introduce foreign DNA that codes for the target protein and then, they are cultivated preferably in suspension in a chemically defined serum-free media [59]. The preferred mammalian cell lines for protein expression in research and industrial fields are CHO and HEK-293 cells [67]. CHO cells are dominant in heterologous protein production in industry, mainly because of advantages like the property to provide complex post-translational modifications similar to those of humans, their ease to scale-up, and for being easily adapted to grow in serum-free suspension cultures [65]. CHO cells are more suited for stable expression, for its transfection renders low yields of recombinant protein secretion in this lineage. Since establishing a stable cell line is time and labor-consuming, transient transfection is a suitable option to gather high amounts of proteins in a shorter period. In this case, HEK cells might represent an interesting option, since they are wellknown for being rather suitable for transient transfection. This cell line also has rapid doubling time and grows in high-density concentrations, just like CHO cells,

systems is to produce complex biomolecules such as antibodies that require posttranslational modifications like glycosylation [38]. Though other eukaryotic systems do provide this modification, their capability of doing so is limited and might result in the addition of glycans that are not common in human proteins [59]. This event might result in misfolding and biologically inactive immunoglobulins, undesirable features in human therapeutic and diagnostic monoclonal antibodies [60]. Also, expression in prokaryotic systems might lead to contamination with endotoxins, which increases downstream processes to clear these endotoxins from the final product. Thereby, the mammalian expression systems are valuable tools to produce monoclonal antibodies as well as other proteins with proper structure and activity. Indeed, there are numerous FDA-approved mAbs produced in mammalian expression systems in contrast to prokaryotic systems and other eukaryotic cells [61].

*DOI: http://dx.doi.org/10.5772/intechopen.95485*

**3.1 Mammalian production systems**

culture systems [62].

of these molecules using only in-silico approaches.

**3. Alternative methods for mAb production**

living animals, the advances in structure prediction methods set the scene for an ongoing technological evolution that should potentially lead the future generation of these molecules using only in-silico approaches.
