WO2021084104A1 - Tetravalent antibody molecules - Google Patents

Abstract

Disclosed herein are tetravalent immunoglobulins and F(ab')₂ fragments containing two Fv regions each consisting of two variable heavy (VH) domains, two constant light (CL) domains, and either two heavy chain constant regions, or two constant heavy 1 (CH1) domains and a hinge. Also disclosed herein are methods for producing such tetravalent immunoglobulins and F(ab')₂ fragments.

Description

TETRAVALENT ANTIBODY MOLECULES

Work leading to this invention has received funding from the European Communi ty’s Seventh Framework Programme (FP7/2007-2013) under grant agreement n°

602262.

FIELD OF THE INVENTION

The present invention relates to tetravalent immunoglobulins or F(ab’)2 fragments comprising four variable heavy (VH) domains and no variable light (VL) domain. These tetravalent immunoglobulins or F(ab’)2 fragments may be either bispecific or monospecif ic. The present invention also relates to a method of producing such tetravalent immuno globulins or F(ab’)2 fragments.

BACKGROUND OF THE INVENTION

Bispecific monoclonal antibodies have during the last decades emerged as a new growing class of antibody therapeutics (Sedykh et al 2018; Runcie et al 2018, Garbert et al 2014). A normal wild-type IgG is monospecific and typically binds one antigen, a bispecific antibody (BsAb), by design, binds two antigens simultaneously, either two dif ferent antigens, or two different epitopes on the same antigen. Bispecific antibodies can for example be used to recruit immune cells to tumors, block two different pathways sim ultaneously or increase antibody specificity, resulting in a more effective therapy (Runcie et al 2018) compared to conventional treatment with a monospecific antibody.

Numerous formats have been suggested for bispecific antibodies (Brinkmann et al 2017; Strohl 2018) and new ones are constantly being developed. The formats differ in many aspects including: how they are created, if they have an Fc-region, their size, how many antigens they can bind simultaneously, if they are symmetric etc. Formats con structed as whole IgGs share advantages with wild-type IgGs. This includes possibility to bind Fc-receptors on effector cells to mediate antibody-dependent cellular cytotoxicity (ADCC) and antibody-dependent cellular phagocytosis (ADCP), or induction of comple ment dependent cytotoxicity (CDC). Typically, antibodies in an IgG format also show long half-lives due their size and binding to neonatal Fc receptor (FcRn).

One particularly interesting bispecific format, due to its wild-type IgG structure, is the two-in-one antibody also called a dual action Fab (DAF), in which one variable do main of an antibody binds two different antigens. This was first reported by Bostrom et al who constructed an antibody binding to both HER2 and VEGF through mutagenesis of the HER2 specific antibody Herceptin (Bostrom et al 2009). This approach has since then been used to generate antibodies binding to additional antigen combinations such as HER3-EGFR (Schaefer et al 2011), VEGF-Angiopotenin 2 (Koenig et al 2015) or IL4- IL5 (Lee et al 2014). It is, however, challenging to find an antibody that binds two anti gens with high affinity. Another limitation is that each arm of the IgG molecule can only bind one antigen at the time. This has partly been overcome in the DutaMab format where the binding site of the antibody is restricted to 3 of the 6 complementary- determining regions (CDRs), making CDR H1, H3 and L2 binding one antigen and CDR L1, L3 and H2 binding a second antigen (Brinkmann et al 2017). These DutaMabs can thus bind two different antigens, if their size and special orientation allows, on one of the IgG arms simultaneously (WO 2012/163520 A1). Another approach to find a two-in-one type of antibody is to combine a variable heavy chain (VH) binding to one antigen with a variable light chain (VL) binding a second antigen (WO 03/002609 A2). All these strate gies are however both time-consuming and costly and methods to efficiently create the two-in-one antibody type of bispecific antibodies are currently lacking.

The smallest antibody fragment with antigen binding capacity is the VH domain (Ward et al 1989; Jirholt et al 1998; Rouet et al 2015), and in animals of the camelidae family (Hamers-Casterman et al 1993) and in sharks (Greenberg et al 1995) antibodies consisting of only heavy chains have been found. These findings have been used for discovery and development of single domain antibodies, called nanobodies, recently re viewed by Mir et al 2019.

WO 96/27011 A1 discloses a method for making heteromultimeric polypeptides, such as bispecific antibodies. This method is the so called “protuberance-into-cavity” strategy or “knob-in-hole” technique.

WO 2005/061547 A2 describes a bispecific antibody comprising two variable do mains on a single polypeptide chain. One of the variable domains binds to an effector antigen on a human immune effector cell and the other variable domain binds to another antigen on another cell than the human immune effector cell. Such bi-specific T-cell en gagers (BiTEs) are investigated for the use as anti-cancer drugs.

WO 2013/170168 A1 discloses a method of producing another type of bispecific or multispecific monoclonal antibodies wherein two different immunoglobulin heavy chain variable antibody regions, cloned in frame with an IgG constant domain modified such that they can no longer assemble to form immunoglobulin heavy chain homo-dimers, are co-expressed with a single immunoglobulin light chain variable antibody region that func tionally complements the two different immunoglobulin heavy chain variable regions.

An overview of bispecific antibodies is provided by R.E. Kontermann et al. in Drug Discovery Today, Volume 20, Number 7, July 2015, pages 838-847. Furthermore, in R. Kontermann (2012) Dual targeting strategies with bispecific antibodies, mAbs, 4:2, 182- 197, different formats of bispecific antibodies are presented. Recombinant approaches to IgG-like bispecific antibodies are described by J.S. Marvin and Z Zhu in Acta Pharmacologica Sinica 2005 Jun; 26 (6): 649-658. The differ ent approaches discussed therein are “Knobs-into-holes” BsAb IgG , common light chain BsAb IgG, single chain Fv-Fc “knobs-into-holes” BsAb, IgG-scFv fusions, C-terminal scFv fusion, N-terminal scFv fusion, diabody-Fc fusions, single chain diabody-Fc fusion, and di-diabody. They conclude that these approaches use some form of Fv, comprised of a VL and a VH, to bind antigen. They then discuss future perspectives, saying, for ex ample, that a novel approach in which BsAb can be constructed is to use the VL and VH as independent binding units, and mention that preferably, one single domain would be derived from a VL, and the other from a VH, to provide increased stability using a VL-VH interface. In a purely speculative way they suggest other imaginable combinations of sin gle domains.

BRIEF SUMMARY OF THE INVENTION The present inventors have found an approach to create a new type of bispecific (or monospecific; as will be explained below) antibodies by combining two variable heavy chain (VH) domains, from e.g. wild-type antibodies, in each Fab arm of an immunoglobu lin (Ig). For example, using an IgG, it is possible to obtain a tetra-VH IgG. The VHs are used as building blocks, where one VH is placed at its standard position, and the second VH replaces the variable light chain (VL) in a wild-type Ig.

The present invention relates to bispecific or monospecific tetravalent immuno globulins or F(ab’)2 fragments that comprise four VH domains; two in each Fv region. These four VH domains constitute four non-overlapping antigen binding sites, two on each Fab arm. Each Fv region in these tetravalent, tetra-VH, immunoglobulins can bind to two different or two identical antigens on each Fab arm, meaning that they are either bispecific or monospecific.

Thus, disclosed herein are tetravalent immunoglobulins or F(ab’)2 fragments consisting of: two Fv regions each consisting of two variable heavy (VH) domains; two constant light (CL) domains; and either two heavy chain constant regions, or two constant heavy 1 (CH1) domains and a hinge region, wherein two VH domains are linked to the two CL domains and the other two VH domains are linked to either the two heavy chain constant regions or the two CH1 do mains.

Disclosed are also a method of producing a tetravalent immunoglobulin or a tet ravalent F(ab’)2 fragment of an immunoglobulin, comprising the following steps: a) selecting a first variable heavy (VH) domain for its ability to bind to a first anti gen in absence of a variable light (VL) domain; b) selecting a second VH domain for its ability to bind to a second antigen in ab sence of a VL domain, wherein the first and second VH domains may be identical or dif ferent and wherein the first and second antigens may be identical or different; c) combining the first and second VH domains with two constant light (CL) do mains and either two heavy chain constant regions or two constant heavy 1 (CH1) do mains and a hinge region into a tetravalent antibody molecule in an Ig or F(ab’)₂ frag ment format, wherein the first VH domain is introduced in the position usually occupied by a VH domain in each heavy chain of a wild-type immunoglobulin and the second VH domain is introduced in the position usually occupied by a VL domain in each light chain of a wild-type immunoglobulin, or wherein the first VH domain is introduced in the posi tion usually occupied by a VL domain in each light chain of a wild-type immunoglobulin and the second VH domain is introduced in the position usually occupied by a VH do main in each heavy chain of a wild-type immunoglobulin.

Also disclosed are tetravalent immunoglobulins or F(ab’)₂ fragments obtainable or obtained by the above method.

Also disclosed are nucleic acids encoding a tetravalent immunoglobulin or F(ab’)₂ fragment mentioned above.

Also disclosed are vectors comprising nucleic acids mentioned above.

Also disclosed are host cells transfected with a vector mentioned above.

DETAILED DESCRIPTION OF THE INVENTION

The inventors have shown that it is possible to generate a new type of bispecific or monospecific immunoglobulin, a tetra-VH immunoglobulin, through the combination of two VH domains in each arm of an immunoglobulin molecule, one linked to CL (replacing the VL domain) and the other linked to CH1 (the normal position of a VH). Different VH domains can be combined in any combination, offering a convenient “plug and play” for mat to generate a mono- or bispecific immunoglobulin with the specificity of interest.

These new type of bispecific or monospecific tetravalent immunoglobulins may be used as previously known bispecific or monospecific immunoglobulins, for example in medicine as therapeutic antibodies, including immune-oncologic or immunomodulating antibodies.

Bispecific tetra-VH IgGs produced this way contain an Fc part, which distin guishes them from several previously described bispecific formats in which building blocks of binding domains, such as scFv or Fabs, are combined and linked together, and which lack an Fc-part, and as a result cannot mediate ADCC, ADCP or CDC. The inventors have shown that VH domains binding different types of antigens can be easily isolated after initial selection using a scFv-display library and used as build ing blocks for construction of tetra-VH antibodies. Other approaches to create two-in-one type of antibodies require either mutagenesis of an already discovered antibody (Bostrom et al 2009; Schaefer et al 2011; Koenig et al 2015; Lee et al 1989; Eigenbrot et al 2013) to enable a second antigen to bind with high affinity, the use of two different binding sites within one IgG molecule (WO 2012/163520 A1) or the combination of a VH binding to one antigen with a VL binding a second antigen (W02003/002609 A2). The latter share the advantage with our strategy, that building blocks are generated, which can be reused in different combinations. It is however often challenging to find binding VL domains as the VL of an IgG is, at least in nature, much less variable than the VH, due to how antibody variability is created during B-cell development (Collins et al 2008). This limitation is avoided by the use of only VH domains in the creation of tetra-VH IgGs.

Standard methods, used for production of wild-type IgGs, were applied also on the tetra-VH IgGs, with the only exception that a preparative SEC was run to remove un bound VH-CL fragments after purification on protein A. This can presumably be avoided by using a different resin in the purification step (Seldon et al 2011), possibly in combina tion with the introduction of a mutation in CDR H2 (Bach et al 2015). The stability of the tetra-VH IgGs, over 3-years’ time, with respect to aggregation, binding affinity and speci ficity, was very good, which further demonstrates their wild-type like properties. The wild-type IgG architecture of the molecules most likely reduces the need for separate method-development relating to down-steam processing of these antibodies, which is often required for the development of other types of bispecific antibodies (Von Kreudenstein et al 2013; Spiess et al 2015).

In summary, the method of producing tetra-VH immunoglobulins described herein offers a novel, fast, and easy strategy for development of tetravalent mono- or bi-specific antibodies. As both VH domains retain their binding properties in the tetra-VH format, a tetravalent format where four antigen molecules can bind simultaneously to one tetra-VH IgG, is created. This offers possibilities beyond the traditional bispecific antibody formats, as it also enables construction of a new type of monospecific, tetravalent antibodies with expected enhanced binding and functional properties compared to conventional mono- specific IgGs.

Thus, as mentioned above, the present invention relates to tetravalent antibodies and to methods for producing such tetravalent antibodies. These are tetravalent in the sense that they comprise four antigen-binding domains. What is unique to these tetrava lent antibodies is that these four antigen-binding domains consist of four VH domains. These four VH domains may bind either to two different antigens, in which case they are bispecific, or to two identical antigens, in which case they are monospecific. Herein, such tetravalent antibodies are also referred to as tetra-VH antibodies.

Bispecific antibodies (bsAbs) may also be denoted dual-specific antibodies or immunoglobulins, dual-antigen-specific antibodies or dual-specificity antibodies. An ex ample of dual-specific antibodies is so called “two-in-one” antibodies, wherein each Fab can bind two different targets although only one target at each point of time. As men tioned above, several additional dual-specific antibodies have previously been described. However, as will be clear from below, the present tetra-VH antibodies are closely related to wild-type, naturally occurring antibodies in the sense that most of their structure may originate from a naturally occurring or wild-type immunoglobulins and only a small part is altered compared to the wild-type structure.

A normal or wild-type monomeric IgG, IgD or IgE antibody typically has a Y- shape and consists of two heavy and two light chains. For monomeric IgG and IgD, each heavy chain consists of a variable heavy (VH or VH) domain and a heavy chain constant region constituted by a constant heavy 1 (CH1 or CH1) domain, a hinge region, a con stant heavy 2 (CH2 or CH2) domain and a constant heavy 3 (CH3 or CH3) domain. Each light chain consists of a variable light (VL or VL) domain and a constant light (CL or CL) domain. The region on the immunoglobulin that binds to antigens is the antigen-binding fragment (Fab). There are two Fab regions (each constituting an “arm” of the Y) and one Fc (crystallizable fragment) region (the base or “leg” of the Y) in each immunoglobulin. Each Fab region is composed of one variable and one constant domain of each of the heavy and the light chain, i.e. VH, CH1, VL and CL. Two Fab fragments held together through a hinge region constitutes a F(ab’)2 fragment. The variable part of a Fab region, i.e. VH and VL, is referred to as the Fv or F_v region. Each normal or wild-type immuno globulin or F(ab’)2 fragment thus contains two Fv regions, each consisting of one VH do main and one VL domain. It is the Fv region that forms the antigen-binding site of an im munoglobulin or F(ab’)2 fragment.

Contrary to a normal or wild-type immunoglobulin or F(ab’)2 fragment, the tetrava lent antibodies described herein or produced with the method described herein contain no VL domains. Instead each of the two Fv regions consists of two VH domains, one oc cupying the place normally occupied by a VH domain in a wild-type immunoglobulin and one occupying the place normally occupied by a VL domain in a wild-type immunoglobu lin. Thus, two of the four VH domains are linked to the two CL domains and the other two VH domains are linked to either the two heavy chain constant regions or the two CH1 domains. In this context “linked” means that they are covalently bound.

In some embodiments, the tetravalent antibody is an immunoglobulin. In addition to the two Fv regions each consisting of two VH domains as described above, such a tetravalent immunoglobulin also contains two constant light (CL) domains, and two heavy chain constant regions. As explained above, this tetravalent immunoglobulin does not contain any VL domain. Such a tetravalent immunoglobulin is illustrated at the bottom of Fig. 1 A; both as a dualspecific tetravalent immunoglobulin and as a monospecific tetra valent immunoglobulin.

In some embodiments the tetravalent antibody is an immunoglobulin G (IgG). In some such embodiments it is of an lgG1, lgG2 or lgG4 subtype.

In other embodiments, the tetravalent antibody is an IgD or an IgE.

IgGs, IgDs and IgEs are all monomers. There are also multimeric antibody iso types, such as dimeric IgA and pentameric IgM. These consist of two and five, respec tively, Y shaped units. In an immunoglobulin or fragment described herein, each such Y shaped unit would then comprise two VH domains in each of the two Fv regions, as ex plained above, instead of one VH domain and one VL domain.

In some embodiments, the tetravalent antibody is instead a F(ab’)2 fragment. In addition to the two Fv regions consisting of two VH domains as described above, such a tetravalent F(ab’)2 fragment also contains two constant light (CL) domains, and two con stant heavy 1 (CH1) domains and a hinge region. As explained above, this tetravalent F(ab’)2 fragment does not contain any VL domain. Such a tetravalent F(ab’)2 fragment is similar to the immunoglobulin illustrated at the bottom of Fig. 1 A; with the exception that the CH2 and CH3 domains are not included while the two Fab regions still are held to gether by the hinge region.

A method of producing a tetravalent immunoglobulin or F(ab’)2 fragment as dis cussed above, comprises the following steps: a) selecting a first variable heavy (VH) domain for its ability to bind to a first anti gen in absence of a variable light (VL) domain; b) selecting a second VH domain for its ability to bind to a second antigen in ab sence of a VL domain; c) combining the first and second VH domains with constant heavy (CH) and constant light (CL) domains into a tetravalent antibody molecule in an Ig or F(ab’)2 for mat.

More precisely, in step c) one of the VH domains is combined with a constant light (CL) domain to form a light chain and the other VH domain is combined either with a heavy chain constant region or with a constant heavy 1 (CH1) domain and a hinge region to form a heavy chain. Two such light chains and two such heavy chains form together the tetravalent antibody molecule in an Ig or F(ab’)2 fragment format.

One of the VH domains selected in steps a) and b) is introduced in the position usually occupied by a VH domain in each heavy chain of a normal or wild-type immuno- globulin while the other VH domain selected in steps a) and b) is introduced in the posi tion usually occupied by a VL domain in each light chain of a normal or wild-type immu noglobulin. Thus, the first VH domain selected in step a) may be introduced in the posi tion usually occupied by a VH domain in each heavy chain of a normal immunoglobulin while the second VH domain selected in step b) is introduced in the position usually oc cupied by a VL domain in each light chain of a normal immunoglobulin. Alternatively, the first VH domain may be introduced in the position usually occupied by a VL domain in each light chain of a normal immunoglobulin while the second VH domain is introduced in the position usually occupied by a VH domain in each heavy chain of a normal immu noglobulin.

That a VH domain is able to bind to an antigen in absence of a VL domain means in this context that the VH domain binds to the antigen independently of the presence of a VL domain. This may be achieved for example by using a VH library which is screened for molecules binding to a specific antigen. A VH domain may then be selected from the molecules binding to the antigen; since no VL domains are present in such a library, it is clear that such a VH domain can bind to the antigen in the absence of a VL domain. Al ternatively, it is possible to select a VH-domain which binds to an antigen when com bined with a dummy-VL (VLD) domain, as explained further in the Examples below., i.e. a VL domain that ideally does not affect the VH chain and that is not necessary for the binding of the Fab, Ig or or F(ab’)₂ fragment to an antigen,. However, in some cases such VH domains may utilize the VLD for binding, so further testing will be required to ascer tain that such VH domains are suitable for tetra-VH Ig format. A further alternative is to combine a VH domain from an antibody known to bind to a specific antigen with a VH domain from another antibody known to bind to another specific antigen, followed by testing if the combination still binds both antigens. This can be tested in conventional immunochemical assays, such as ELISA, where the binding to recombinant antigens is analysed, or flow cytometry analysis, where the binding to both antigens is analysed. In this latter option, no dummy VL is used or needed. Instead the VH domains are directly tested in the final format as a tetra-VH IgG; if the tetra-VH IgG retains its binding to both antigens it is known that the tested VH domains are capable of binding their respectively antigen on their own. The included VHs can thereafter be used as building blocks and used in additional tetra-VH IgGs.

The VH domains selected shall be full-length VH domains, i.e. no fragments of or truncated VH domains.

In some embodiments, at least one of the VH domains is derived from a human antibody, i.e. of human origin. In some embodiments, both VH domains are of human origin. The selection of the VH domains in step a) and b) may be done using an antibody library, such as a library of antibody fragments. When a library of antibody fragments is used, the fragments should be large enough to encompass at least full VH domains.

In some embodiments a library of VH domains (or VH library) is used. VH librar ies are known to the skilled person (Ward et al 1989; Jirholt et al 1998; Rouet et al 2015). It is possible to use an existing VH library (such as a commercially available VH library) or to create a VH library specifically for the purpose of producing tetra VH-lgs as de scribed herein.

Alternatively, the selection of the VH domains in step a) and b) may be done us ing an scFv antibody library, such as n-CoDeR^® (Biolnvent, Soderlind E, et al NatBio- technol. 2000; 18(8): 852-6). Another alternative to make the selection of the VH domains in step a) and b) is to test VH domains from individual antibodies.

The VH domain selected in step a) may be identical to or different from the VH domain selected in step b).

Furthermore, the first antigen may be identical to or different from the second an tigen.

When the first antigen is different from the second antigen, the first VH domain is different from the second VH domain. The tetravalent immunoglobulin or F(ab’)2 fragment obtained when combining the two VH domains will be a bispecific tetravalent immuno globulin or F(ab’)2 fragment, i.e. an immunoglobulin or F(ab’)2 fragment with four binding sites binding two different antigens.

When the first and the second antigens are the same, the tetravalent immuno globulin or F(ab’)2 fragment obtained when combining the two VH domains will be a monospecific tetravalent immunoglobulin or F(ab’)2 fragment, i.e. an immunoglobulin or F(ab’)2 fragment with four binding sites all binding the same antigen. In a monospecific tetravalent immunoglobulin or F(ab’)2 fragment, the first and second VH domains may be either identical or different. If they are identical, they will bind the same epitope in the same antigen, while if they are different, they will bind the same or different epitopes in the same antigen.

In the examples below, it is shown that monospecific tetravalent antibodies as described herein demonstrate stronger agonistic activity compared to a conventional monospecific bivalent antibody (also shown in Fig. 10 C).

The first and second VH domain may bind any type of antigen. For example, ei ther one of or both of the first VH domain and the second VH domain may bind an anti gen selected from the TNFR superfamily (TNFRS). Such an antigen from the TNFRS may be selected from the group consisting of CD40 (cluster of differentiation 40), 0X40 (also known as CD134 and tumor necrosis factor receptor superfamily member 4, TNFRSF4) and 4-1 BB (or 41BB, also known as CD137 and tumor necrosis factor recep tor superfamily member 9, TNFRSF9). As another example, either one of or both of the first VH domain and the second VH domain may bind MCP-1 (monocyte chemoattractant protein-1 , also known as CCL2). As yet another example, either one of or both of the first VH domain and the second VH domain may bind CD72 (cluster of differentiation 72), which is a target belonging to proteins that contain C-type lectin (CLEC) domains.

In step c) of the above method, the first and second VH domains are combined with the constant heavy (CH) and constant light (CL) domains into a tetravalent antibody molecule in IgG or F(ab’)2 fragment format. Two independent VHs (binding the same or different antigens) are combined in the same human immunoglobulin molecule, such as an IgG. One VH is bound to CH (normal position) whereas the other VH is linked to CL replacing VL of a wild-type IgG molecule.

The constant heavy (CH) and constant light (CL) domains may be the constant regions from any immunoglobulin in a suitable format. For example, the CH and CL do mains may be from the antibody from which the first VH is obtained, or the CH and CL domains may be from the antibody from which the second VH is obtained. In some em bodiments, the CH and CL domains are derived from a human antibody or of human origin. In some embodiments, the CH and CL domains are derived from a murine anti body or of murine origin.

After the combination of the first and second VH domains with constant heavy (CH) and constant light (CL) domains into the tetravalent antibody molecule in IgG or F(ab’)2 fragment format, the binding should preferably be confirmed. This may be done using any form of immunochemical binding analysis, such as for example ELISA, flow cytometry and/or Biacore analysis.

In some embodiments, it is also preferable to control the stability of the generated tetravalent antibody molecule in IgG or F(ab’)2 fragment format. This may be done by an analysis for binding in ELISA and for lack of aggregation in size exclusion chromatog raphy at the time the tetravalent antibody molecule is generated and again a certain time thereafter for comparison. For example, the analysis is run immediately after production and purification and after long-term storage, such as 3 years storage at +4°C. Stable tet ravalent antibody molecules show no aggregation after the long-term storage and have similar EC50 values in ELISA compared to directly after generation.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 : A schematic outline of how tetravalent VH-lgGs can be constructed is shown in Fig. 1 A. Binding VHs are isolated from binding scFvs, VHs, Fabs or IgGs and combined in a mono- or bi-specific tetra-VH IgG molecule. A Schematic outline of how bispecific, tetravalent, tetra-VH-lgGs were produced in the examples below is shown in Fig. 1 B. Monospecific scFv were isolated from our phage display library n-CoDeR®, fol lowed by transfer of VHs from binding clones to a Fab vector containing a dummy VL (VLD). Generated VH-VLD Fab clones were screened for binding and after Sanger se quencing, for rediscovery of binding clones, unique binding VHs were combined in a tet- ra-VH IgG format.

Figure 2: Binding-analysis of monospecific scFv. Fig. 2A shows screening in FMAT of individual soluble scFv showing the binding to target expressing cells versus binding to mock transfected cells. Fig. 2B shows binding of unique clones to target ex pressing cells in the first screening versus their binding after re-expression. Fig. 2C shows binding of clones in ELISA to coated extra cellular domains (ECDs) of target pro tein and a non-related protein carrying the same tag as the target (non-target). Binding was detected using a tag-detection, mouse-anti-His together with anti-mouse-APC for FMAT (Fig. 2A, Fig 2B) or an AP-labeled anti-FLAG antibody together with a luminescent substrate in ELISA (Fig. 2C).

Figure 3: Screening of VH-VLD Fab clones (CD40 n=576, 0X40 n=576, 41 BB n=768) in ELISA for binding to target versus non-target protein. Clones marked in black were selected as binders and used for Sanger sequencing to identify unique sequences. Binding was detected using an AP-labeled anti-FLAG antibody together with a lumines cent substrate.

Figure 4: Dose response ELISA showing the binding of VH-VLD Fab clones, compared to their corresponding parental clones, in Fab format, to the target protein and a non-target protein (NOT) which is an irrelevant protein carrying the same tag as the target protein. Clones were serially diluted 1:3 starting at 167nM and binding was detect ed using an AP-labeled anti-FLAG antibody together with a luminescent substrate. Fig. 4 shows the results for four different anti-CD40 clones (4A-D), four different anti-OX40 clones (4E-H) and four different anti-41 BB clones (4I-L). (The VH comes from 4A a-005- A04 [SEQ. ID. NO. 76], 4B a-001-A04 [SEQ. ID. NO. 24], 4C a-004-B03 [SEQ. ID. NO. 4] 4D a-001-A10 [SEQ. ID. NO. 28], 4E b-001-B03 [SEQ. ID. NO. 80], 4F b-002-F02 [SEQ. ID. NO. 40], 4G b-001-A11 [SEQ. ID. NO. 32] 4H b-002-F07 [SEQ. ID. NO. 36], 4I c-001- B10 [SEQ. ID. NO. 84], 4J C-001-D05 [SEQ. ID. NO. 44], 4K C-005-G03 [SEQ. ID. NO. 60] and 4L C-006-F05 [SEQ. ID. NO. 56], respectively).

Figure 5: Binding-analysis of control antibodies (anti-CD40: mlgG2b anti hCD40- APC [R&D Systems Cat. No. FAB6321A], anti-OX40: rat lgG2a anti hOX40-APC [R&D Systems Cat. No. FAB3388A] and anti-41 BB: mlgG2b anti h4-1BB-APC [BioLegend Cat No 309809]) to transfected cells (in Fig. 5A) and endogenously expressing cells (in Fig. 5B), respectively, in flow cytometry. The endogenously expressing cells are a B-cell line (Raji) expressing CD40 and in vitro activated (IVA) CD4 positive T-cells expressing 0X40 and 4-1 BB. Isotype antibodies were included as negative control (negative control for CD40: mouse lgG2B-APC [BD Cat. No. 555745]; negative control for 0X40: rat lgG2A-APC [BD Cat. No. 560720]; negative control for 4-1 BB: mouse lgG2B-APC [BD Cat. No. 555745]). The results from using these control antibodies demonstrate that CD40, 0X40 and 4-1 BB are expressed on transfected cells and endogenously express ing cells, as expected.

Figure 6: Binding-analysis of generated bispecific tetra-VH IgGs. Antibodies were evaluated in ELISA on recombinant proteins (i.e. by analysis of the binding of the anti body to the different target proteins compared to a non-target protein) and flow cytometry on transfected cells (expressing either the different target proteins or only mock trans fected, i.e. transfected with an empty expression vector lacking gene encoding a protein) or endogenously expressing cells (expressing the different proteins). Binding was de tected using a HRP-labeled anti-human antibody and a luminescent substrate in ELISA and an APC-labeled anti-human antibody in flow cytometry. Fig. 6A shows the results for a tetra-VH IgG binding OX40-CD40; tetra VH IgG #2 wherein VH-CH 1=0X40 #2 (SEC. ID. NO. 36) and VH-CL=CD40 #5 (SEC. ID. NO. 16) (see Table 4 below). Fig. 6B shows the results for a tetra-VH IgG binding 41 BB-CD40; tetra VH IgG #4 wherein VH- CH1=41BB #4 (SEC. ID. NO. 52), VH-CL=CD40 #3 (SEQ. ID. NO. 12) (see Table 4 be low). Fig. 6C shows the results for a tetra-VH IgG binding 41BB-OX40-; tetra VH IgG #20 wherein VH-CH1=41BB #4 (SEQ. ID. NO. 52), VH-CL= OX40#2 (SEQ. ID. NO. 36) (see Table 4 below). The results demonstrate that the bispecific tetra- VH-lgGs bind trans fected cells, proteins and endogenously expressing cells as expected.

Figure 7: Binding-analysis of bispecific tetra-VH IgGs to CD40 and 0X40 on overexpressing or endogenously expressing cells in flow cytometry and recombinant pro teins in ELISA. The two VHs, binding CD40 and 0X40 respectively, were either linked to CH1 or CL in the antibody. The two tetra-VH-lgGs were tetra VH lgG#2 wherein VH- CH 1=OX40#2 (SEQ. ID. NO. 36), VH-CL=CD40#5 (SEQ. ID. NO. 16) and tetra VH lgG#5 wherein VH-CH1=CD40#5 (SEQ. ID. NO. 16), VH-CL=OX40#2 (SEQ. ID. NO.

36). In the legend to Fig. 7, CD40 IgG #2 is short for binding to CD40 of lgG#2, which has CD40 VH#5 in VL position, CD40 of IgG #5 is short for binding to CD40 of lgG#5, which has CD40 VH#5 in VH position, 0X40 IgG #2 is short for binding to 0X40 of lgG#2, which has 0X40 VH#2 in VH position, and 0X40 IgG #5 is short for binding to 0X40 of IgG #5, which has 0X40 VH#2 in VL position. Binding was detected using an APC-labeled anti-human antibody in flow cytometry and a HRP-conjugated anti-human antibody in ELISA. This figure demonstrates that the binding of the antibody is not de- pendent on the position of each VH; i.e. binding is essentially the same regardless of if a specific VH is in the light or the heavy chain.

Figure 8: Binding-analysis of bispecific tetra-VH IgGs binding two different anti gens simultaneously. Fig. 8 A: ELISA analysis of an anti-CD40/OX40 tetra-VH IgG (tetra- VH IgG #2 wherein VH-CH1=OX40#2 [SEQ. ID. NO. 36], VH-CL=CD40#5 [SEQ. ID. NO. 16]; see Table 4) for binding to biotinylated CD40 or 0X40 after blocking with non- biotinylated CD40 or 0X40. Fig. 8 B: ELISA analysis of an anti-41 BB/CD40 tetra-VH IgG (tetra VH lgG#4 wherein VH-CH1=41 BB #4 [SEQ. ID. NO. 52], VH-CL=CD40 #3 [SEQ. ID. NO. 12]; see Table 4) for binding to biotinylated CD40 or 41BB after blocking with non-biotinylated CD40 or 41 BB. Fig. 8 C: ELISA analysis of an anti-OX40/41 BB tetra-VH IgG (tetra VH lgG#3 wherein VH-CH1=41 BB #2 [SEQ. ID. NO. 48], VH-CL=OX40 #2 [SEQ. ID. NO. 36]; see Table 4) for binding to biotinylated 0X40 or 41 BB after blocking with non-biotinylated 0X40 or 41 BB.

Figure 9: Binding-analysis of 0X40 and CD40 to an anti OX40/CD40 tetra-VH IgG, CD40 and 41 BB to an anti-41 BB/CD40 tetra-VH IgG and 0X40 and 41 BB to an anti 0X40/41 BB tetra-VH IgG in Biacore. Fig. 9 A and B: tetra-VH lgG#2 wherein VH- CH1=OX40#2 (SEQ. ID. NO. 36), VH-CL=CD40#5 (SEQ. ID. NO. 16) (see Table 4); Fig.

9 C and D: tetra-VH lgG#19 wherein VH-CH1=41 BB#4 (SEQ. ID. NO. 52), VH- CL=CD40#5 (SEQ. ID. NO. 16) (see Table 4); and Fig. 9 E and F: tetra-VH IgG #20 wherein VH-CH1=41BB#4 (SEQ. ID. NO. 52), VH-CL=OX40#2 (SEQ. ID. NO. 36) (see Table 4). Antibodies were captured on an immobilized catcher antibody followed by addi tion of 800 nM of the first antigen to achieve binding saturation. The second antigen was then added at 800 nM, diluted in 800 nM of the first antigen, to avoid signal loss due to dissociation of the first antigen. This Biacore analysis demonstrates bispecific, tetravalent binding.

Figure 10: Evaluation of anti-CD40 antibodies effect on B-cell proliferation. Fig.

10 A: Wild-type parental IgGs, tetra-VH IgGs and control antibodies, were analyzed for induction of B-cell proliferation on B-cells from healthy donors. The control antibodies were two agonistic anti-CD40 antibodies, a human lgG2, CP-870,893 (Pfizer/VLST), and a humanized lgG1, SGN-40 (also called Dacetuzumab or huS2C6 from Seattle Genet ics). No IgG as well as a human lgG1 isotype control antibody (in-house developed, con taining the same framework and constant regions as the evaluated anti-CD40 antibodies) were used as negative controls. B-cells were isolated from PBMC and incubated with 50nM of cross-linked antibodies (using anti-human-F(ab’)2 at a 1.5:1 molar ratio lgG:F(ab’)2) in media containing 10 ng/ml of IL4. After 5 days of incubation at +37°C, 8% CO2, cells were stained with anti-CD86-APC, anti-CD19-PE and a live dead marker and analyzed by flow cytometry. Proliferation was measured as the percent of CD19+ cells that were CD86+. Samples were run as duplicates and data from 2-4 different donors were plotted as mean with SEM using GraphPad Prism. Fig 10 B Evaluation of anti- CD40 antibodies effect on B-cell proliferation without cross-linking. The experiment was run as described for 10 A except that antibodies were analyzed without cross-linking. Fig. 10 C: Dose response analysis of antibodies effect on B-cell proliferation showing the mean, after subtraction of values for the isotype control, from 2 different experiments us ing B-cells from 4 different donors. The experiments were run as described for Fig. 10 A, except that antibody concentrations between 50 and 0.2 nM were used.

Figure 11 : Dose response binding-analysis of purified VH domains alone com pared to scFv. Fig. 11 A: binding to 41 BB of 41 BB VH# 4 compared to scFv composed of VH 41BB#4 (SEQ. ID. NO. 52) and VH OX40#2 (SEQ. ID. NO. 36). Fig. 11 B: binding to 0X40 of 0X40 VH #2 compared to scFv composed of VH 41BB#4 (SEQ. ID. NO. 52) and VH 0X40 #2 (SEQ. ID. NO. 36). Fig. 11 C: binding to CD40 of CD40 VH#5 com pared to scFv composed of VH OX40#2 (SEQ. ID. NO. 36) and VH CD40 #5 (SEQ. ID. NO. 16). Binding was detected using an anti-FLAG-AP antibody followed by a lumines cent substrate.

Figure 12: Analysis of binding to recombinant MCP-1 protein and a non-target protein of tetra-VH IgGs constructed from VH domains binding to MCP1 and VH domains binding CD40 (Fig. 12 A: tetra-VH-lgG #24 wherein VH-CH1=MCP1#6 [SEQ. ID No. 72], VH-CL =OX40#2 [SEQ. ID No. 36]), 0X40 (Fig. 12 B: tetra-VH-lgG #22 wherein VH- CH1=MCP1#6 [SEQ. ID No. 72], VH-CL =41BB#4 [SEQ. ID No. 52]) or 41 BB (Fig. 12 C: tetra-VH-lgG #23 wherein VH-CH1=MCP1#6 [SEQ. ID No. 72], VH-CL =CD40#5 [SEQ. ID No. 16]). Binding was detected using a HRP-labeled anti-human antibody followed by a luminescent substrate.

Figure 13: Binding-analysis of an anti-CD72 VH domain to recombinant CD72 (target protein) and a non-target protein (Fig. 13A: VH-CD72-002-F04 [SEQ. ID No. 64], Fig 13B: VH-CD72-001-C07 [SEQ. ID No. 68]). Binding was detected using an HRP- labeled anti-human antibody followed by a luminescent substrate.

Figure 14: Size exclusion chromatography (SEC) analysis of tetra-VH IgGs di rectly after purification (time = 0) and after 3 years storage (time = 3 years). The ana lyzed tetra-VH s are in Fig. 14 A tetra-VH lgG#2 wherein VH-CH1=OX40#2 (SEQ. ID.

NO. 36), VH-CL=CD40#5 (SEQ. ID. NO. 16); in Fig. 14 B tetra-VH lgG#5 wherein VH- CH1=CD40#5 (SEQ. ID. NO. 16), VH-CL=OX40#2 (SEQ. ID. NO. 36); in Fig 14 C tetra- VH lgG#7 wherein VH-CH1=CD40#3 (SEQ. ID. NO. 12), VH-CL=CD40#3 (SEQ. ID. NO. 12); in Fig 14 D tetra-VH lgG#8 wherein VH-CH1=CD40#5 (SEQ. ID. NO. 16), VH- CL=CD40#5 (SEQ. ID. NO. 16); in Fig 14 E tetra-VH lgG#13 wherein VH-CH1=CD40#3 (SEQ. ID. NO. 12), VH-CL=CD40#5 (SEQ. ID. NO. 16); in Fig 14 F tetra-VH lgG#14 wherein VH-CH1=CD40#5 (SEQ. ID. NO. 16), VH-CL=CD40#3 (SEQ. ID. NO. 12); in Fig 14 G tetra-VH lgG#17 wherein VH-CH1=OX40#2 (SEQ. ID. NO. 36), VH-CL=CD40#3 (SEQ. ID. NO. 12); and in Fig 14 H tetra-VH lgG#19 wherein VH-CH1=41BB#4 (SEQ.

ID. NO. 52), VH-CL=CD40#5 (SEQ. ID. NO. 16). No aggregation is seen after long term storage.

Figure 15: Dose response binding analysis to recombinant proteins in ELISA af ter incubation in 50%human serum at 37°C after different timepoints (t=0, t=1 day, t=3 days, t=4 days or t=7 days. A control with a freshly prepared antibody is included. The analyzed tetra-VHs are in Fig. 15 A and B tetra-VH lgG#2 wherein VH-CH1=OX40#2 (SEQ. ID. NO. 36), VH-CL=CD40#5 (SEQ. ID. NO. 16); binding to CD40 (A) and 0X40 (B), in Fig 15 C and D tetra-VH lgG#19 wherein VH-CH1=41BB#4 (SEQ. ID. NO. 52), VH-CL=CD40#5 (SEQ. ID. NO. 16) binding to CD40 (C) and 4-1 BB (D) in Fig 15 E and F tetra-VH IgG #20 wherein VH-CH1=41BB#4 (SEQ. ID. NO. 52), VH-CL=OX40#2 (SEQ. ID. NO. 36) binding to 0X40 (E) and 4-1 BB (F). The antibodies show similar binding at t=0 and after incubation in human serum at 37°C for 7 days. Binding was detected using an HRP-labeled anti-human antibody followed by a luminescent substrate.

Figure 16: Thermal stability of tetra-VH IgGs in nano differential scanning fluo- rometry. The analyzed tetra-VHs are in Fig. 16 Tetra VH IgG #1 wherein VH-CH1=41BB #2 (SEQ. ID. NO. 48), VH-CL=CD40 #3 (SEQ. ID. NO. 12); Tetra VH IgG #15 wherein VH-CH1=41BB #2 (SEQ. ID. NO. 48), VH-CL=41BB #4 (SEQ. ID. NO. 52); hlgGI L- 41 BB-2-VLD wherein VH-CH 1=41 BB #2 (SEQ. ID. NO. 48), VH-CL=VLD (SEQ. ID. NO. 85); Tetra VH IgG #2 wherein VH-CH 1=0X40 #2 (SEQ. ID. NO. 36), VH-CL=CD40 #5 (SEQ. ID. NO. 16); Tetra VH IgG # 9 wherein VH-CH1=OX40 #2 (SEQ. ID. NO. 36), VH- CL=OX40 #2 (SEQ. ID. NO. 36); Tetra VH IgG #17 wherein VH-CH 1=0X40 #2 (SEQ. ID. NO. 36), VH-CL=CD40 #3 (SEQ. ID. NO. 12);Tetra VH IgG #21 wherein VH-CH1=OX40 #2 (SEQ. ID. NO. 36), VH-CL=41 BB #4 (SEQ. ID. NO. 52); hlgG1 L-OX40-2-VLD where in VH-CH 1=0X40 #2 (SEQ. ID. NO. 36), VH-CL=VLD (SEQ. ID. NO. 85); Tetra VH IgG #5 wherein VH-CH1=CD40 #5 (SEQ. ID. NO. 16), VH-CL=OX40 #2 (SEQ. ID. NO. 36); Tetra VH IgG #8 wherein VH-CH1=CD40 #5 (SEQ. ID. NO. 16), VH-CL=CD40 #5 (SEQ. ID. NO. 16); Tetra VH IgG #14 wherein VH-CH1=CD40 #5 (SEQ. ID. NO. 16), VH- CL=CD40 #3 (SEQ. ID. NO. 12); hlgG1 L-CD40-5-VLD wherein VH-CH 1=CD40 #5 (SEQ. ID. NO. 16), VH-CL=VLD (SEQ. ID. NO. 85); Tetra VH IgG #7 wherein VH-CH1=CD40 #3 (SEQ. ID. NO. 12), VH-CL=CD40 #3 (SEQ. ID. NO. 12); Tetra VH IgG #13 wherein VH-CH 1=CD40 #3 (SEQ. ID. NO. 12), VH-CL=CD40 #5 (SEQ. ID. NO. 16); hlgGIL- CD40-3-VLD wherein VH-CH1=CD40 #3 (SEQ. ID. NO. 12), VH-CL=VLD (SEQ. ID. NO. 85); Tetra VH IgG #11 wherein VH-CH 1 =41 BB #4 (SEQ. ID. NO. 52), VH-CL=41BB #4 (SEQ. ID. NO. 52); Tetra VH IgG #16 wherein VH-CH1=41 BB #4 (SEQ. ID. NO. 52), VH- CL=41BB #2 (SEQ. ID. NO. 48); Tetra VH IgG #19 wherein VH-CH 1=41 BB #4 (SEQ. ID. NO. 52), VH-CL=CD40 #5 (SEQ. ID. NO. 16); Tetra VH IgG #20 wherein VH-CH1=41BB #4 (SEQ. ID. NO. 52), VH-CL=OX40 #2 (SEQ. ID. NO. 36); hlgG1L-41BB-4-VLD where in VH-CH1=41 BB #4 (SEQ. ID. NO. 52), VH-CL=VLD (SEQ. ID. NO. 85).

EXAMPLES

Specific, non-limiting examples which embody certain aspects of the invention will now be described.

In these examples several tetra-VH-lgGs are exemplified, and for some of these VH sequences presented in Table 1 were used. Table 1 contains additional sequences suitable for tetra-VH-lgGs. Additional VH, presented in Table 2, were also tested, but these did not have the ability to bind to a first antigen when in a VH-VLD Fab format.

Table 1 : Examples of VH sequences binding as VH-VLD Fab or as VH

Table 2: Examples of VH sequences not suitable for tetra-VH-lgG

Table 3: VL dummy (VLD) sequence used in the examples and sequences of constant chains suitable for tetra-VH-lgGs (amino acids marked in bold in the VL dummy se quences represent the CDRs)

Table 4: Summary of VH sequences of tetra-VH-lgGs used in the examples

Preparation of antigens

CHO-S cells (Thermo Fisher) were transfected with full-length CD40, 41 BB and 0X40 (pCMV/hygro vectors from Sino Biological; CD40: Cat. No. HG10774-M-N, 0X40: Cat. No. HG10481-G-N and 4-1 BB: Cat. No. HG10041-M-N) using Freestyle™ MAX Re agent (Thermo Fisher) and used either 48h after transfection (41 BB and 0X40) or for stable cell line development (CD40). Stable cell lines were developed by addition of se lection pressure, 600 pg/mL Hygromycin, (Thermo Fisher) and limiting dilution of surviv ing transfected cells. The extracellular domain (ECD), PCR amplified from full length constructs above, of CD40 (the extracellular domain corresponding to amino acids 1-193), 0X40 (the ex tracellular domain corresponding to amino acids 1-216) and 41 BB (the extracellular do main corresponding to amino acids 1-186), respectively, was ligated into an expression vector (in-house developed; containing a CMV-promoter, a BGH poly A signal and an ampicillin resistance gene) containing a 8x His-tag and produced in HEK293 EBNA cells (ATCC-CRL- 10852, ATCC) adapted to serum-free conditions through sequential reduc tion of the serum concentration. The proteins were purified with Ni-NTA Affinity Chroma tography (AKTA purifier, GE Healthcare). Isolation of monospecific antibody fragments against CD40, 0X40 and 4-1 BB

The phage display n-CoDeR^® single chain Fv (scFv) library (Biolnvent, Soderlind E, et al Nat Biotechnol. 2000; 18(8) :852-6) was used to generate CD40, 0X40 and 41 BB specific scFv, respectively. A combination of protein coated on polystyrene beads (Pol- ysciences, Cat. No. 17175), biotinylated proteins loaded on Streptavidine beads (Dyna- beads™ M-280 Streptavidin, Thermo Fisher) and target expressing cells was used. Be fore each selection, phages binding beads with a non-related protein carrying the same tag as the target protein (non target) or mock transfected cells were removed. The de pleted phage stock was then left to bind the target coated beads or target expressing cells before extensive washing to remove unbound phages. Bound phages were recov ered by trypsin elution (0.5 mg/ml, Sigma-Aldrich) followed by inactivation with Aprotenin (0.2 mg/ml Sigma-Aldrich). Eluted phages were used to infect exponentially growing E. coli HB101F’ (in-house constructed from E. coli HB101 (Thermo Fisher Scientific) modi fied so that it expresses F-pili required for phage infection)). The bacteria were then spread on selective agar plates and incubated overnight at +30°C before colonies were pooled and cultivated to produce a phage stock using R408 (Agilent Technologies, Cat. No. 200252) as helper phage. This phage stock was used for a second selection round, in total three consecutive selections were done.

Genes encoding scFv fragments were digested from the phagemide vector (puri fied according to the manufacturer’s instruction (QIAprep Spin Miniprep Kit, Qiagen, Cat. No. 27104), ligated into a protein expression vector (in-house developed suitable for E. coli) and used for transformation of chemically competent E. coli TOP10 (Thermo Fisher Scientific Cat. No. C404010). Individual transformants were picked (Qbot, Molecular De vices) and used for production of soluble scFv in microtiter plates.

Screening of monospecific antibody fragments

Individual scFv clones were analysed for binding to target expressing CHO cells and mock transfected CHO cells in FMAT (Flourometric Microvolume Assay Technolo gy). 2500 cells/well in 40 pi DM EM media (Thermo Fisher, Cat. No. 11330032) were added to FMAT plates (384 well, Thermo Fisher) followed by 10 mI/well of scFv superna tant and 20 mI/well of anti-His (0.6 pg/ml, R&D Systems Cat. No. MAB050) and anti- mouse-APC (1.5 pg/ml, Jackson ImmunoResearch, Cat. No. 109-136-098). After 10 h incubation at room temperature (approximately 20-25 °C), the plates were read in a 8200 Cellular detection system (Thermo Fisher Scientific). Unique clones binding specifically to the target cells (defined as no detected binding to mock transfected cells and a signal above 2000 on target cells) were identified with Sanger sequencing (the sequencing was performed by Eurofins Genomics).

Unique scFv clones were re-expressed and binding to cells confirmed in FMAT as described above. Unique clones were further analyzed for binding to CD40, 0X40 and 41 BB, respectively, in ELISA. ECD of the target proteins were coated to ELISA plates overnight. After washing scFv supernatant, 10 mI/well, diluted in block buffer 40 pl/well, (PBS (Invitrogen) with 0.05 % Tween 20 and 0.45 % fish gelatin (both from Sigma- Aldrich)) 40pl/well, were left to bind for 1 h at room temperature (approximately 20-25°C). Bound scFv was detected with anti-FLAG-AP (Sigma Aldrich, Cat. No. A9469) and a luminescent substrate (CDP Star Emerald II, Thermo Fisher, Cat. No. T2216), and plates were read in a plate reader (Tecan Ultra).

The results of the screening of individual clones is shown in Fig. 2 A. Unique clones, identified through Sanger sequencing as described above, were re-produced and the binding to transfected cells was confirmed for most of them as shown in Fig. 2B. The majority of the clones also bound to recombinant proteins in ELISA, as shown in Fig. 2C resulting in around 100 clones / target for downstream evaluation.

Isolation of monospecific VH binders

VH from all unique scFv were amplified by PCR, pooled and ligated into a Fab expression vector (in-house developed suitable for E.coli) containing a dummy VL gene, composed of the light chain variable gene fragment DPL3 (lgLV1-47, accession number Z22189) expressing the VL dummy sequence presented in Table 3 [SEQ. ID. NO. 85] rearranged to the light chain joining gene fragment JL2. Chemocompetent E. coli Top10 were transformed and individual clones picked and used for production of soluble Fab. Soluble Fab were analyzed for binding to coated proteins in ELISA as described above for scFv. Clones specifically binding to CD40, 0X40 and 41 BB, respectively, were sequenced with Sanger sequencing (Eurofins Genomics) for rediscovery of clones binding irrespectively of the VL chain.

Unique VH-VLD binding Fab clones were purified from periplasm of E. coli. A 10 ml TB culture was cultivated to exponential phase followed by 0.5 mM IPTG (Sigma- Aldrich) addition to induce Fab production. Bacteria were pelleted, supernatants discarded, and the periplasm prepared by addition of a lysozyme-sucrose solution (1 mg/ml lysozyme (Sigma-Aldrich) in 20% sucrose (BDH chemicals) and 1 h incubation, with rotation at +4°C. Cell debris were removed by centrifugation and the supernatants (containing the periplasm) transferred to, and left to bind, a washed Ni-NTA plate (His MultiTrap HP, 96 well, GE Healthcare, Cat. No. 28-4009-89). After multiple washing steps, Fabs were eluted using 250 mM imidazole (Sigma-Aldrich) and the concentration was measured as A280 (UV Star plate, Greiner, Cat. No. 655801), in a plate reader (Tecan Infinite F500, Tecan) before addition of protease inhibitors (NaN3_, BDH chemicals and benzami- dine, Sigma-Aldrich).

Purified Fab clones were titrated for binding their corresponding target antigen in a dose response ELISA. A target protein and a non-target protein carrying the same tag was coated to ELISA plates overnight at +4°C. Purified Fab was serially diluted 1:3, start- ing at 200 nM, and added to the coated and washed ELISA plates. Bound Fab clones were detected as described previously.

Results of screening of individual VH-VLD Fab for binding to recombinant pro teins in ELISA are shown in Fig. 3. It was seen that a fraction of the VH-VLD clones bound to their target and their corresponding genes were sequenced for rediscovery of input clones that were functional also in this context. In Fig. 4, the results of dose re sponse ELISA analyses of some purified unique VH-VLD Fabs are shown, confirming target binding for sub-sets of the VH-VLD Fabs. Binding of the parental clone, containing the original VL, was analyzed as control, as shown in Fig. 4. Different binding patterns were seen including: binding of parental Fab with no binding of the VH-VLD Fab (Fig. 4 A, E, I), stronger binding of parental Fab compared to VH-VLD Fab (Fig. 4 B, F, J), simi lar binding of VH-VLD Fab and parental Fab (Fig. 4 C, G, K), and stronger binding of VH- VLD Fab than the parental Fab (Fig. 4 D, H, L). The latter showed that the VL of some selected scFv may even inhibit its VH partner from binding, as the binding, in some cas es, was stronger after exchange of the parental VL for another VL. In total, around 7 per cent of the generated monospecific scFv, mean of all three targets, were binders also in the VH-VLD format (see Table 5 below), with similar or better affinities than the parental clones (Fig. 4)

Table 5:

* Produced and binding in tetra-VH IgG format

Construction of tetra-VH IgG antibodies

A schematic outline of how tetravalent VH-lgGs are produced is shown in Fig. 1 B. The VH regions of VH-VLD binding Fab clones were combined to create tetra-VH hu man IgG antibodies. Two independent VH (binding the same or different target antigens) were combined in the same human IgG molecule. One VH is bound to CH (normal posi- tion) whereas the other VH is linked to CL replacing VL of a wild-type IgG molecule. In this example, the constant human g1 heavy and light chains shown in Table 3 were used. The individual VH fragments were PCR amplified and ligated into two expression vectors (in-house developed; these vectors contain a CMV-promoter, a BGH poly A sig nal and an ampicillin resistance gene; in addition, the heavy chain vector contains the antibody heavy chain constant region and the light chain vector contains the antibody light chain constant region) one for heavy and one for light chain containing the constant parts for a human lgG1. The vectors were prepared from E. coli according to manufac turer’s instructions (Qiagen miniprep kit, Cat. No. 27104) and used for transient co transfection of suspension adapted HEK 293 EBNA (Thermo Fisher Scientific) with poly- ethylenimine (PEI) (Polyscience Inc, Cat. No. 23966). The cells were incubated for4h, +37°C, 8% CO2, 300 rpm before addition of feed, UltraPep™ Soy (Sheffield Bio-Science). After 6 days, cell supernatants were harvested and IgG purified on Mabselect (GE Healthcare) followed by preparative size-exclusion chromatography (SEC) to remove unpaired light chains.

Binding analysis of tetra-VH IgG antibodies

The binding of purified tetra-VH IgG antibodies was confirmed in ELISA and flow cytometry. For ELISA, target proteins were coated to ELISA plates over night at +4°C. The next day antibodies were serially diluted 1 :3 starting at 133 nM and left to bind the washed ELISA plate for 1 h at room temperature (approximately 20-25°C). Binding was detected using a horseradish peroxidase (HRP) conjugated anti-human-Fc antibody (Jackson Immunoresearch, Cat. No. 709-036-098,) followed by a lumincecent substrate (SuperSignal™ ELISA Pico Chemiluminescent Substrate, Thermo Fisher, Cat. No. 37070) and reading in a plate reader (Tecan Ultra, Tecan).

For cell binding analysis, transfected CHO cells, in vitro activated T cells, as de scribed below, or a B cell line (Raji ATCC, Cat. No. CCL-86) were added 50000 cells/well. IgGs were serially diluted 1:3 starting at 133 nM and left to bind the cells for 1 h at +4°C. After washing bound IgG was detected with allophycocyanin (APC) labelled anti-human F(ab)’2 (Jackson ImmunoResearch, Cat. No. 109-136-098) in flow cytometry (FACS Verse, BD Biosciences). For in vitro activation of T cells, CD4+ T-cells were puri fied from peripheral blood mononuclear cells (PBMC) by negative selection (human CD4+ T cell isolation kit, Miltenyi, Cat. No. 130-096-533) followed by activation with IL2, 50 ng/ml (R&D Systems) and CD3/CD28 beads (Dynabeads™, Thermo Fisher Scientific, Cat. No. 11161D) at a 1:1 cell:bead ratio. Cells were incubated 2-5 days at +37°C, 8% CO2 before usage. Target expression on transfected cells, the included B-cell line (CD40) and T-cells (0X40 and 41 BB) was confirmed with commercial control antibodies in flow cytometry before evaluation of tetra-VH antibodies and are shown in Fig 5. The results of the eval uation of functionality of the tetra-VH antibodies, with respect to antigen binding, wherein antibodies were titrated for binding to recombinant proteins in ELISA and transfected cells, primary T-cells and a B-cell line in flow cytometry, are shown in Fig 6. Importantly, several (40%) of the VHs, identified as VH-VLD binders, could be produced and retained binding as tetra-VH IgGs (Table 5). The tetra-VH antibodies were bispecific and bound two antigens, either CD40/OX40, 41BB/CD40 or 0X40/41 BB, as recombinant proteins or expressed on the cell surface (Fig. 6). This demonstrated that monospecific VHs could be combined in various combinations, offering a novel approach to generate bispecific antibodies. Importantly, the binding VH domains could be placed at both the CL- and CH1-position without effecting the binding affinity and specificity of the antibody (Fig. 7).

ELISA analysis of tetra-VH IgG antibodies forevaluation of valency

To analyze if the constructed bispecific tetra-VH IgGs were tetravalent and thus could bind two antigens simultaneously on each Fab arm, the antibodies were analyzed in ELISA and Biacore (see below). In ELISA, non-biotinylated antigen was added at dif ferent concentrations to captured tetra-VH IgGs followed by addition and detection of a fix amount of biotinylated antigen. More specifically, IgG was captured at 16 nM on a coated (7.5 pg/ml) goat anti-human-Fc antibody (Jackson ImmunoResearch, Cat. No. 109-005-008). After washing, non-biotinylated antigen was diluted to 400 nM, titrated 1:3 and added. Biotinylated antigen was then added at a fix concentration resulting in a high, but titratable, signal. Bound biotinylated antigen was detected using HRP-labelled strep- tavidin (Jackson ImmunoResearch, Cat. No. 016-030-084) followed by a luminescent substrate (SuperSignal™ ELISA Pico Chemiluminescent Substrate, Thermo Fisher, Cat. No. 37070) and reading in a plate reader (Tecan Ultra, Tecan).

For an OX40/CD40 bispecific antibody it was seen that CD40 inhibits the binding of biotinylated CD40 and 0X40 inhibits the binding of biotinylated 0X40 in a dose de pendent manner, as expected. However, CD40 showed no inhibition of biotinylated 0X40 and vice versa, as shown in Fig. 8 A. The same pattern was seen for antibodies targeting 41BB/CD40 (Fig. 8 B) and for antibodies targeting 0X40/41 BB (Fig. 8 C). This indicated tetravalent binding as the first antigen was unable to affect the binding of the second antigen to the tetra-VH IgG and vice versa. Biacore analysis of tetra-VH IgG antibodies forevaluation of valency

Tetravalent binding was confirmed in Biacore. Tetra-VH IgGs were captured on an immobilized catcher antibody followed by sequential antigen additions. A CM5 chip (GE Healthcare) was immobilized with anti-human IgG, 25 pg/ml, (GE Healthcare, Cat. No. BR 1008-39). IgG was added at 10 nM, 10 mI/min for 3 min followed by the first anti gen addition, 800 nM, 30 mI/min, for 3 min followed by the second antigen addition (dilut ed in 800 nM of the first antigen to avoid signal loss due to dissociation of the first anti gen), 800 nM, 30 mI/min, for 3 min. The surface was regenerated with 10 mM glycine pH 1.5 between each cycle.

After binding saturation with the first antigen, addition of the second antigen re sulted in increased binding. The same amount of the second antigen bound regardless if the first antigen was added or not. No added signal was seen after addition of a non target protein or repetitive addition of the same antigen. All antigen combinations, OX40/CD40 (Fig. 9 A and B), 41 BB/CD40 (Fig. 9 C and D) or 41 BB/OX40 (Fig. 9 E and F), showed the same result. Importantly, this clearly demonstrates that the generated tetra-VH IgGs are tetravalent and can bind two antigens simultaneously on each Fab.

B-cell proliferation assay

In order to evaluate the functionality in regard to agonistic effect, tetra-VH IgGs were compared to known agonistic antibodies in a B-cell proliferation assay. PBMC from healthy donors were prepared from buffy coats by gradient density centrifugation (Ficoll Paque PLUS, GE Healthcare). B cells were purified by negative selection according to manufacturer’s instructions (pan B cell isolation kit, Miltenyi, order no: 130-091-151) and added to plates 100000/well. Antibodies were added with or without crosslinking with a F(ab)’2-anti-human-Fc (Jackson ImmunoResearch, cat no 109-006-098) at a molar ratio 1.5:1 between lgG:F(ab’)2 . After 5 days incubation at +37°C, 8% CO2, cells were stained with anti-CD19-BV421 (BD Biosiences, Cat. No. 562440), anti-CD86-APC (BD Biosienc- es, Cat. No. 555660) and live dead marker (propidium iodide, Thermo Fisher) and ana lyzed by flow cytometry (FACS Verse, BD Biosiences). B-cell proliferation was evaluated by the fraction of live CD19+, CD86 high cells.

Thus, the functionality of generated tetra-VH IgGs, with respect to agonistic activi ty, was analyzed in a B-cell proliferation assay. Anti-CD40 tetra-VH antibodies including two different anti-CD40 VH domains were combined and analyzed as: wildtype mono- specific IgGs, bispecific tetra-VH IgGs or monospecific tetra-VH IgGs. The results are shown in Fig.10. No effect of the tetra-VH IgGs or the positive control antibody SGN was seen without crosslinking (Fig. 10 B). After crosslinking, the tetra-VH IgGs containing VH No. 3, either in a monospecific or in a bispecific format, induced B-cell proliferation simi lar to the positive control antibody CP-870.839 and more than the positive control anti body SGN (Fig. 10A). Thus, all tetra-VH IgGs containing VH No. 3 were shown to have agonistic activity, in contrast to the tetra-VH IgGs containing VH No. 5 but not VH No. 3, which showed no agonistic activity. Importantly, this showed that the functional activity (here agonist activity) of the VH domain was maintained in the tetra-VH format regard less of the position and VH-partner. Dosing of a few of the most potent antibodies demonstrated that combining two agonistic VHs in one IgG induced about twice as much proliferation as an antibody containing only one agonistic VH (Fig. 10 C).

Binding analysis of VH domains

As most of the identified VHs, identified in the VH-VLD Fab screening, could be used to construct tetra-VH IgGs, the limiting step is to identify VH-binders. Instead of pairing VHs with a dummy VL during the screening, one alternative is to screen VH fragments alone for binding and this was tested for a few clones. VH genes from three VH-VLD binding clones were amplified by PCR, ligated into an expression vector (same as above) containing a 6xHis and 3xFLAG tag and used for production, purification and ELISA binding analysis as described for VH-VLD-Fabs previously.

Purified VH clones bound similar to the target antigen as an scFv containing the same VH, for 2 out of 3 tested clones (Fig. 11). The VH-VLD screening can thus be re placed with a “VH only” screening for identification of VHs to be used for construction of tetra-VH IgGs.

Generation of VHs against additional antigens

As CD40, 0X40 and 41 BB all belong to the tumor necrosis factor receptor super family, the broad applicability of this approach was demonstrated by combining the VH from a previously generated cytokine (MCP-1) binding antibody with VHs against CD40, 0X40 or 41 BB in tetra-VH IgGs.

VH genes from three MCP-1 binding clones were amplified by PCR and ligated into the Fab expression vector containing a dummy VL gene, described above. After production and purification, binding was analysed using ELISA as described for VH-VLD- Fabs previously. One VH binder was identified and combined with VHs binding to CD40, 0X40 and 41 BB respectively in a tetra VH IgG format and binding was confirmed as de scribed previously.

The data presented in Fig. 12 shows that the anti-MCP-1 specific VH could be combined with any of the tested VHs, i.e. VH binding CD40 (Fig. 12 A), 0X40 (Fig. 12 B) and 4-1 BB (Fig. 12 C) with retained binding. Retained binding demonstrates that the tet- ra-VH format works, since the binding otherwise would be lost.

In an additional example, VH domains from scFv against CD72, generated previ ously by Ljungars et al 2018, belonging to the c-type lectin family, were screened for tar- get binding. VH genes from CD72 binding scFv clones were amplified by PCR, ligated into an expression vector (same as above) containing a 6xHis and 3xFLAG tag, used for production, purification and ELISA binding analysis as described for VH-VLD-Fabs previ ously.

The data presented in Fig. 13 shows the results of binding analysis of two anti- CD72 VH domains. This confirms that VH binders can be isolated also against this anti gen, and thus that tetra-VH IgGs containing such VHs can be produced.

Stability of the tetra-VH IgGs

To examine the stability of tetra-VH IgGs, binding in ELISA and lack of aggregation in size exclusion chromatography (SEC) were compared shortly after purification and after 3 years storage at +4°C. The antibodies showed similar EC50 values in ELISA at both time points, as shown in Table 6, and no aggregation was observed after long term stor age (Fig. 14), demonstrating high stability. In a separate experiment tetra-VH IgGs were incubated in 50% human serum at+37°C for up to 7 days and then analyzed for binding in ELISA (Fig. 15) as described above in section “Binding analysis of tetra-VH IgG anti bodies”. The antibodies show similar binding also after incubation in human serum which further demonstrate their stability.

The thermal stability of tetra-VH IgGs was evaluated with nano differential scan ning fluorometry in Prometheus NT.48 from Nano Temper Technologies with a tempera- ture ramp of 1°C/min from 20 to 95°C. The tetra-VH IgGs show similar variation in ther mal stability as conventional IgGs.

Table 6: EC50 values on binding to coated CD40 in a dose response ELISA for tetra-VH IgG at time 0 and after 3 years of storage.

In summary VH binders against a broad range of antigens were isolated and could be used to create tetra-VH IgGs. These antibodies are tetravalent and bind four epitopes, two on each Fab arm, simultaneously and can be used for the construction of both mono and bi-specific antibodies.

REFERENCES

S. E. Sedykh, V. V. Prinz, V. N. Buneva and G. A. Nevinsky: Bispecific antibodies: design, therapy, perspectives. Drug Des Devel Ther, 12, 195-208 (2018) doi:10.2147/DDDT.S151282

K. Runcie, D. R. Budman, V. John and N. Seetharamu: Bi-specific and tri-specific antibodies- the next big thing in solid tumor therapeutics. Mol Med, 24(1), 50 (2018) doi: 10.1186/s10020-018-0051-4

K. Garber: Bispecific antibodies rise again. Nat Rev Drug Discov, 13(11), 799-801 (2014) doi: 10.1038/nrd4478

U. Brinkmann and R. E. Kontermann: The making of bispecific antibodies. MAbs, 9(2), 182-212 (2017) doi:10.1080/19420862.2016.1268307

W. R. Strohl: Current progress in innovative engineered antibodies. Protein Cell, 9(1), 86-120 (2018) doi: 10.1007/s13238-017-0457-8

J. Bostrom, S. F. Yu, D. Kan, B. A. Appleton, C. V. Lee, K. Billed, W. Man, F. Peale, S. Ross, C. Wiesmann and G. Fuh: Variants of the antibody herceptin that interact with HER2 and VEGF at the antigen binding site. Science, 323(5921), 1610-4 (2009) doi: 10.1126/science.1165480

G. Schaefer, L. Haber, L. M. Crocker, S. Shia, L. Shao, D. Dowbenko, K. Totpal, A. Wong, C. V. Lee, S. Stawicki, R. Clark, C. Fields, G. D. Lewis Phillips, R. A. Prell, D. M. Danilenko, Y. Franke, J. P. Stephan, J. Hwang, Y. Wu, J. Bostrom, M. X. Sliwkowski, G. Fuh and C. Eigenbrot: A two-in-one antibody against HER3 and EGFR has superior inhibitory activity compared with monospecific antibodies. Cancer Cell, 20(4), 472-86 (2011) doi:10.1016/j.ccr.2011.09.003

P. Koenig, C. V. Lee, S. Sanowar, P. Wu, J. Stinson, S. F. Harris and G. Fuh: Deep Sequencing-guided Design of a High Affinity Dual Specificity Antibody to Target Two Angiogenic Factors in Neovascular Age-related Macular Degeneration. J Biol Chem, 290(36), 21773-86 (2015) doi:10.1074/jbc.M115.662783

C. V. Lee, P. Koenig and G. Fuh: A two-in-one antibody engineered from a humanized interleukin 4 antibody through mutation in heavy chain complementarity-determining regions. MAbs, 6(3), 622-7 (2014) doi:10.4161/mabs.28483

E. S. Ward, D. Gussow, A. D. Griffiths, P. T. Jones and G. Winter: Binding activities of a repertoire of single immunoglobulin variable domains secreted from Escherichia coli. Nature, 341(6242), 544-6 (1989) doi: 10.1038/341544a0

P. Jirholt, M. Ohlin, C. A. K. Borrebaeck and E. Soderlind: Exploiting sequence space: shuffling in vivo formed complementarity determining regions into a master framework. Gene, 215(2), 471-476 (1998) doi:https://doi.org/10.1016/S0378-1119(98)00317-5 R. Rouet, K. Dudgeon, M. Christie, D. Langley and D. Christ: Fully Human VH Single Domains That Rival the Stability and Cleft Recognition of Camelid Antibodies. J Biol Chem, 290(19), 11905-17 (2015) doi:10.1074/jbc.M114.614842

C. Hamers-Casterman, T. Atarhouch, S. Muyldermans, G. Robinson, C. Hammers, E. B. Songa, N. Bendahman and R. Hammers: Naturally occurring antibodies devoid of light chains. Nature, 363(6428), 446-448 (1993) doi:10.1038/363446a0 A. S. Greenberg, D. Avila, M. Hughes, A. Hughes, E. C. McKinney and M. F. Flajnik: A new antigen receptor gene family that undergoes rearrangement and extensive somatic diversification in sharks. Nature, 374(6518), 168-73 (1995) doi: 10.1038/374168a0 M. A. Mir, U. Mehraj, B. A. Sheikh and S. S. Hamdani: Nanobodies: The "magic bullets" in therapeutics, drug delivery and diagnostics. Hum Antibodies, 12(10), HAB-190390 (2019)

A. M. Collins, Y. Wang, V. Singh, P. Yu, K. J. Jackson and W. A. Sewell: The reported germline repertoire of human immunoglobulin kappa chain genes is relatively complete and accurate. Immunogenetics, 60(11), 669-76 (2008)

T. A. Seldon, K. E. Hughes, D. J. Munster, D. Y. Chin and M. L. Jones: Improved Protein-A separation of V(H)3 Fab from Fc after papain digestion of antibodies. J Biomol Tech, 22(2), 50-2 (2011)

J. Bach, N. Lewis, K. Maggiora, A. J. Gillespie and L. Connell-Crowley: Differential binding of heavy chain variable domain 3 antigen binding fragments to protein A chromatography resins. J Chromatogr A, 1409, 60-9 (2015) doi: 10.1016/j. chroma.2015.06.064

T. S. Von Kreudenstein, E. Escobar-Carbrera, P. I. Lario, I. D'Angelo, K. Brault, J. Kelly, Y. Durocher, J. Baardsnes, R. J. Woods, M. H. Xie, P. A. Girod, M. D. Suits, M. J. Boulanger, D. K. Poon, G. Y. Ng and S. B. Dixit: Improving biophysical properties of a bispecific antibody scaffold to aid developability: quality by molecular design. MAbs, 5(5), 646-54 (2013) doi:10.4161/mabs.25632

C. Spiess, Q. Zhai and P. J. Carter: Alternative molecular formats and therapeutic applications for bispecific antibodies. Mol Immunol, 67(2 Pt A), 95-106 (2015) doi:10.1016/j.molimm.2015.01.003

A. Ljungars, L. Martensson, J. Mattsson, M. Kovacek, A. Sundberg, U. C. Tornberg, B. Jansson, N. Persson, V. K. Emruli, S. Ek, M. Jerkeman, M. Hansson, G. Juliusson, M. Ohlin, B. Frendeus, I. Teige and M. Mattsson: A platform for phenotypic discovery of therapeutic antibodies and targets applied on Chronic Lymphocytic Leukemia. NPJ Precis Oncol, 2, 18 (2018) doi:10.1038/s41698-018-0061-2

Claims

1. A tetravalent immunoglobulin or F(ab’)2 fragment consisting of: two Fv regions each consisting of two variable heavy (VH) domains, wherein one VH domain (“the first VH domain”) in each Fv region binds a first antigen and the other VH domain (“the second VH domain”) in each Fv region binds a second antigen, wherein the two VH domains in each Fv region may be identical or different and wherein the first and second antigens may be identical or different; two constant light (CL) domains; and either two heavy chain constant regions, or two constant heavy 1 (CH1) domains and a hinge region, wherein two VH domains are linked to the two CL domains and the other two VH do mains are linked to either the two heavy chain constant regions or the two CH1 domains.

2. A method of producing a tetravalent immunoglobulin or F(ab’)2 fragment, com prising the following steps: a) selecting a first variable heavy (VH) domain for its ability to bind to a first anti gen in absence of a variable light (VL) domain; b) selecting a second VH domain for its ability to bind to a second antigen in ab sence of a VL domain, wherein the first and second VH domains may be identical or dif ferent and wherein the first and second antigens may be identical or different; c) combining the first and second VH domains with two constant light (CL) do mains and either two heavy chain constant regions or two constant heavy 1 (CH1) do mains and a hinge region into a tetravalent antibody molecule in an Ig or F(ab’)2 frag ment format, wherein the first VH domain is introduced in the position usually occupied by a VH domain in each heavy chain of a wild-type immunoglobulin and the second VH domain is introduced in the position usually occupied by a VL domain in each light chain of a wild-type immunoglobulin, or wherein the first VH domain is introduced in the posi tion usually occupied by a VL domain in each light chain of a wild-type immunoglobulin and the second VH domain is introduced in the position usually occupied by a VH do main in each heavy chain of a wild-type immunoglobulin.

3. A tetravalent immunoglobulin or F(ab’)2 fragment according to claim 1 or a method according to claim 2, wherein the tetravalent immunoglobulin or F(ab’)2 fragment is an IgG or F(ab’)2 fragment of an IgG.

4. A tetravalent immunoglobulin or F(ab’)2 fragment according to claim 1 or 3, or a method according to claim 2 or 3, wherein the first VH domain and/or the second VH domain is/are selected from a library of antibody fragments comprising VH domains.

5. A tetravalent immunoglobulin or F(ab’)2 fragment according to claim 4, or a method according to claim 4, wherein the library is a library consisting of variable heavy domains.

6. A tetravalent immunoglobulin or F(ab’)2 fragment according to any one of the claims 1 or 3-5, or a method according to any one of the claims 2-5, wherein the first VH domain, the second VH domain and/or the constant domains are of human origin.

7. A tetravalent immunoglobulin or F(ab’)2 fragment according to any one of the claims 1 or 3-7, or a method according to any one of the claims 2-6, wherein both the first VH domain and the second VH domain bind an antigen selected from the TNFR su perfamily (TNFRS) or wherein either the first VH domain or the second VH domain bind an antigen selected from the TNFR superfamily (TNFRS).

8. A tetravalent immunoglobulin or F(ab’)2 fragment according to any one of the claims 1, or 3-6, or a method according to any one of the claims 2-6, wherein the first antigen and the second antigen are different antigens, and wherein the tetravalent im munoglobulin or F(ab’)2 fragment is a bispecific tetravalent immunoglobulin or F(ab’)2 fragment.

9. A bispecific tetravalent immunoglobulin or F(ab’)2 fragment according to claim 8, or a method according to claim 8, wherein both the first VH domain and the second VH domain bind an antigen selected from the TNFR superfamily (TNFRS) and wherein: the first VH domain binds 0X40 and the second VH domain binds 4-1 BB, or the first VH domain binds 4-1 BB and the second VH domain binds 0X40, or the first VH domain binds 0X40 and the second VH domain binds CD40, or the first VH domain binds CD40 and the second VH domain binds 0X40, or the first VH domain binds CD40 and the second VH domain binds 4-1 BB, or the first VH domain binds 4-1 BB and the second VH domain binds CD40.

10. A tetravalent immunoglobulin or F(ab’)2 fragment according to any one of the claims 1 or 3-7, or a method according to any one of the claims 2-7, wherein the first and second antigens are identical and wherein the tetravalent immunoglobulin or F(ab’)2 fragment is a monospecific tetravalent immunoglobulin or F(ab’)2 fragment.

11. A monospecific tetravalent immunoglobulin or F(ab’)2 fragment according to claim 10, wherein the first and second VH domains are different domains binding differ ent epitopes within the antigen.

12. A monospecific tetravalent immunoglobulin or F(ab’)2 fragment according to claim 10, wherein the first and second VH domains are identical domains binding the same epitope within the antigen.

13. A monospecific tetravalent immunoglobulin or F(ab’)2 fragment according to any one of the claims 10-12, or a method according to any one of the claims 10-12, wherein both the first VH domain and the second VH domain bind CD40, or both the first VH domain and the second VH domain bind 0X40, or both the first VH domain and the second VH domain bind 4-1 BB.

14. A bispecific tetravalent immunoglobulin or F(ab’)2 fragment according to claim 8, or a method according to any one of the claims 2-8, wherein either the first VH domain or the second VH domain bind an antigen selected from the TNFR superfamily (TNFRS).

15. A bispecific tetravalent immunoglobulin or F(ab’)2 fragment according to claim 14, or a method according to claim 14, wherein: the first VH domain binds MCP1 and the second VH domain binds CD40, or the first VH domain binds CD40 and the second VH domain binds MCP1 , or the first VH domain binds MCP1 and the second VH domain binds 0X40, or the first VH domain binds 0X40 and the second VH domain binds MCP1 , or the first VH domain binds MCP1 and the second VH domain binds 4-1 BB, or the first VH domain binds 4-1 BB and the second VH domain binds MCP1.

16. A tetravalent immunoglobulin or F(ab’)2 fragment, obtainable by a method ac cording to any one of the claims 2-15.

17. A nucleic acid encoding tetravalent immunoglobulin or F(ab’)2 fragment to any one of claims 1 or 3-16.

18. A vector comprising a nucleic acid according to claim 17.

19. A host cell transfected with a vector according to claim 18.