Was the First Atomic Resolution Structure of an Antibody Fragment Published in 1973?

Over a year-and-a-half ago, I did a deep dive into the history of antibodies by following multiple timelines which highlighted the important dates, studies, and discoveries throughout the century which led to the current prevalent antibody theory. Most of the research I did back then has already been added to this site while a few remain to be updated. In the case of this particular post, I was in a heated debate with myself over whether to even add the information I had gathered here on to the site. However, I ultimately felt that it can give valuable insight into how investigations and research can sometimes lead you down paths you may not need to go down yet can still provide some rather interesting revelations.

In 1973, the first structure of the antibody Fv fragment was said to be published. This was considered as a milestone in the history of antibodies as it apparently helped in the creation of monoclonal antibodies. However, you’d be hard-pressed to know as there isn’t much in the way of background information on this event. One source claimed that this was the first atomic resolution structure of an antibody and linked to a 1972 paper which I will highlight a little later:

“The first atomic resolution structure of an antibody fragment was published in 1973.”

https://www.genscript.com/history-antibodies.html

However, in the only other mention of this event that I could find, which was in Wikipedia’s antibody history, we see that there is no mention of this being an atomic resolution at all:

“The Fv fragment was prepared and characterized by David Givol.”

https://en.m.wikipedia.org/wiki/Antibody

Was this the first atomic resolution structure of an antibody fragment or was this simply yet another general characterization of a random particle assumed to belong to the invisible antibody? What does atomic resolution even mean and why would it matter in order to determine a never before identified and characterized fragment? The best source that I could find as to what atomic resolution is was a 2017 article which stated that the term “atomic resolution” itself is badly abused and seemingly can refer to many things. However, it gives us an approximate definition of what an atomic resolution image is supposed to be which is an electron-density map where the individual atoms can be seen:

‘Atomic resolution’: a badly abused term in
structural biology

“The term ‘atomic resolution’ was defined a long time ago and it is generally accepted to correspond to an electron-density map (or a map calculated using another type of data, for example nuclear density) in which individual atoms can be distinguished. It is usually assumed to correspond to a resolution dmin of 1.2 A˚ of the diffraction data, which is also known as the ‘Sheldrick criterion’ (Sheldrick, 1990; Morris & Bricogne, 2003). This limit is not arbitrary, since it reflects the ability to visualize separated atoms and serves as the basis for the determination of crystal structures by direct methods, but it might of course be adjusted if valid scientific reasons could be presented. The meaning of ‘near-atomic resolution’ is much more diffuse, but in our opinion it should be restricted to dmin <2A,˚ the resolution at which the backbone atoms of a protein chain can be assigned with a high degree of confidence. However, these terms are very often abused in order for the published structures to appear to be more accurate than they are in reality. Thus, many X-ray and neutron crystal structures claim ‘atomic resolution’ although they were determined on the basis of data extending to only -2 A˚ or less (Henderson et al., 2011; Taylor et al., 2011; Lyu et al., 2014; Miwa et al., 2016; Hong et al., 2016). This term is also sometimes used for structures determined by NMR, in which case the meaning of ‘resolution’ is even more uncertain (Wa¨lti et al., 2016). More recently, the term ‘near-atomic resolution’ has been used to describe cryo-EM structures determined at resolutions as low as 3.2–4.2 A˚ (Worrall et al., 2016; Galkin et al., 2015; Bartesaghi et al., 2014; Chua et al., 2016; von der Ecken et al., 2016) or an XFEL structure at 3.3 A˚ resolution
(Zhou et al., 2016). On the other hand, the term ‘near-atomic resolution’ is sometimes used to describe structures at the resolution as high as 1.0 A˚ (Romir et al., 2007). Since these terms are currently being used by scientists practicing different techniques for structure determination, an agreement on their exact meanings might be very helpful. In our opinion, the term ‘a structure at atomic resolution’ should not mean ‘a structure represented by individual atoms’, which can be constructed even at low data and map resolution from the known building blocks consisting of separate atoms. We would like to postulate that maybe all structural communities, including traditional macromolecular X-ray and neutron crystallography, XFEL and cryo-EM, among others, should adopt, or indeed respect, standard definitions of what these terms are supposed to mean. Such an agreement would help the readers of structural papers to obtain a realistic impression of the likely accuracy of the structures based on the resolution of the primary experimental data.”

doi: 10.1107/S205979831700225X

A 2019 article in Nature stated that atomic resolution electron microscopy can heavily alter and/or destroy the structure attempted to be studied:

“Atomic-resolution electron microscopes utilize high-power magnetic lenses to produce magnified images of the atomic details of matter. Doing so involves placing samples inside the magnetic objective lens, where magnetic fields of up to a few tesla are always exerted. This can largely alter, or even destroy, the magnetic and physical structures of interest.”

https://www.nature.com/articles/s41467-019-10281-2

From what I can gather, if this 1972 study is the first atomic resolution structure of an unknown antibody fragment, this means that it should consist of images of individual atoms of the fragment being imaged and mapped out. However, it appears that obtaining such images can not only heavily alter the fragment being studied, it can destroy it entirely. Thus, it would seem contradictory to use this process to discover and characterize an unknown antibody fragment. Any information gleaned from such imaging would not tell much of anything about the structure of the fragment intended to be studied.

Regardless of whether atomic resolution was utilized or not, the linked study was either important in the context of antibody history or I was led down a quick and useless detour in my antibody research. In an attempt to shed more light on this matter, I decided to look at David Givol, the man who lead the study, to see if I could squeeze out any further information. What I learned was that his work uncovering the Fv fragment, considered the tiniest part of the antibody entity, was considered pivotal to the genetic engineering and creation of synthetic antibodies:

“Prof. David Givol spent a significant part of his career investigating the structure-function relationship of antibodies. He identified the smallest fragment of antibody containing all its binding properties, i.e. the part of the molecule responsible for recognizing the foreign substance attacking the organism.

Application

Today, this fragment – called Fv – is used in genetic engineering techniques for producing synthetic antibodies for treating various diseases.”

https://wis-wander.weizmann.ac.il/life-sciences/antibody-fragment-used-genetic-engineering-techniques

As Givol’s work led directly into the genetic engineering of synthetic antibody treatments, it does seem important to look at this work irregardless of whether or not it involved atomic resolution imaging. What you will see is that his study is a nice showcase of the ridiculous methods utilized by antibody researchers to try and establish a hypothetical framework of an invisible entity. It is also a good example of the very vague and non-committal language used by researchers when interpreting the results from their chemistry experiments. Highlights below:

Localization of Antibody-Combining Sites within the Variable Portions
of Heavy and Light Chains

“ABSTRACT The Fab’-fragment of a mouse IgA-myeloma (protein-315) was split by pepsin to yield a smaller fragment that retained the anti-2,4-dinitrophenyl activity
of the intact protein. This fragment, which we call Fv, has a molecular weight of about 30,000 (half that of Fab’), and is composed of two polypeptide chains (molecular weight 14,000) held together by noncovalent bonds. The N-terminal sequence of Fv suggests that it is composed of the N-terminal half of Fab’, and consists of the variable portions of the heavy and light chains. Since Fv has about one binding site with the same association constant as Fab’, this experiment provides direct evidence that the antibody site in this protein is contained entirely in the variable portion, and is independent of the constant portion, of the molecule.

An antibody-combining site is contained within an Fab-fragment (1), which is composed of a light chain (L) and the N-terminal half (Fd) of a heavy chain (2). It is generally agreed that antibody specificity results from different amino-acid sequences; this suggestion is strongly supported by experiments on refolding of completely unfolded Fab or peptide chains (3-6). However, only the N-terminal half of L-chain (105-109 residues) or of Fd (115-120 residues) contains sequences that vary from one immunoglobulin to another, whereas the C-terminal half of Fab has the same sequence for fragments derived from molecules of the same subclass. Theoretically, it is possible for a binding site in the constant region to be modulated by different sequences in the variable region. It is also possible that the constant region of the Fab is necessary for maintenance of the combining site.

Fab-fragment (molecular weight 50,000) can be obtained by various methods (7). An early report (8) on the production of a smaller fragment (molecular weight about 13,000) by digestion of Fab with papain in 6 M urea was not followed up, and other attempts to produce an active fragment with molecular weight smaller than 46,000 have failed (9).

We present direct evidence for the localization of the antibody combining site in the variable region of both H and L chains by obtaining a fragment, denoted Fv, that consists of the N-terminal half of the Fab, and retains the complete antibody-binding site. Our approach was based partly on the
observation that under acidic conditions, light chain can be
split with pepsin into variable and constant fragments (10). It seems possible, therefore, that the Fd-piece could also be split in the middle under similar conditions. Such a cleavage of Fab would yield a fragment half the size of Fab. The mouse myeloma protein-315, an IgA with anti-2,4-dinitrophenyl (Dnp) activity (11), was chosen for this purpose.

MATERIALS AND METHODS

The myeloma protein-315 and its pepsin-produced Fab’-fragment were prepared (12) from the serum of tumor-bearing mice. Further digestion of the Fab’-fragment with pepsin (Worthington) was performed in 0.1 MI sodium acetate buffer (pH 3.6) at 370 for 4 hr. The digest was fractionated by affinity chromatography on e-Dnp-lysine Sepharose (12), and by gel filtration on Sephadex G-75 columns in 0.01 M sodium phosphate-0.14 M NaCl (pH 7.4).

Measurements of sedimentation velocity were made in a Spinco model E ultracentrifuge at 56,100 rpm, at a constant temperature of 20.30. Diffusion measurements were made in the same centrifuge with a synthetic boundary cell at 5000 rpm. All samples were equilibrated by dialysis against 0.14 M NaCl-0.01 1\I sodium phosphate (pH 7.4). Molecular weight was calculated according to the Svedberg equation: M = RT s20,w/D20,w(1-ip). The partial specific volume (v) of 0.723 was calculated for Fv from its aminoacid composition (13).

Disc gel electrophoresis was performed in neutral sodium
dodecyl sulfate on 10% polyacrylamide (14). Protein markers were heavy and light chains of rabbit IgG, ovalbumin, trypsin, lysozyme, and cytochrome c. Gels were stained with Coomassie brilliant blue, and the RF of each band was plotted against
the logarithm of the molecular weight.

Aminoacid analyses were performed on a Beckman 120B aminoacid analyzer. Protein samples were hydrolyzed in evacuated tubes with 6 N HCl at 110° for 24 hr. The value for half-cystine was obtained from aminoacid analysis of samples oxidized with performic acid. Tryptophan content was calculated from the absorbance of the protein in 6 M guanidine hydrochloride at 288 and 280 nm (15). The sequence of N-terminal residues was determined (16) with a Beckman protein sequencer, model 890.

The binding activity of the proteins was assessed by equilibrium dialysis at room temperature in 0.01 ‘M sodium phosphate-0.15 M NaCl (pH 7.4) buffer in 0.5-ml chambers with [14CJe-N-2,4 Dnp-lysine as a ligand, and by quenching of protein fluorescence with Dnp-lysine (17). Fluorescence measurements were made on a Turner model 210 recording spectrofluorimeter, and absorbance was determined with a Zeiss PAIQ II spectrophotometer.”

DISCUSSION

“Our results show clearly that peptic cleavage of protein-315 yields a small fragment (Fv) that retains the binding activity of the original protein. This fragment is half the size of the Fab’-fragment, and its aminoacid composition and limited N-terminal aminoacid sequence are consistent with the view that it contains the N-terminal half of L chain and Fd, i.e., the variable segments designated VH and VL. The similar binding constant of Fv and Fab’ indicates that the antibody site in this protein is contained entirely in the variable domains and is independent of the constant portion of the molecule. Fv contains 263 residues (Table 2), about 33 residues more than expected for the sum of VH and VL. It is very likely, therefore, that each of the chains in Fv has 15-16 residues from the constant region. This is not surprising, since the C-terminal parts of VL and VH are in hypervariable regions that are involved in antibody-combining sites (21) and, therefore, must be left intact if antibody site is to be preserved. It seems likely that these additional 15 residues from the constant parts are in the interdomains area and, although not proven, we assume that they have no effect on the binding site. Currently, we are trying to digest Fv with carboxypeptidases to see how many of its C-terminal residues can be removed without diminution of its binding properties.

It has been demonstrated that in its reaction with Dnp ligands, protein-315 exhibits essentially all the characteristics of binding sites of conventionally prepared anti-Dnp antibodies (11). Hence, it is likely that an active Fv-fragment may be obtainable from other antibodies, although the procedure may vary for antibodies from different species, and classes. Indeed, recent work on the light chains of rabbit antibodies to pneumococcal polysaccharides demonstrated that the N-terminal half of these light chains (VL) can be obtained by cleavage with dilute acid (22). These cleaved chains, when recombined with homologous heavy chains, restored 78% of the original antibody activity.

Our results also suggest a modification of the “domain hypothesis” (23), according to which each homologous region in an immunoglobulin chain (110-120 residues) folds separately as a globular domain, with less-tightly folded stretches of the peptide chain between domains. The present findings indicate that the VH and VL domains are intercalated and folded as a single globular unit, resistant to enzymic digestion. This is particularly consistent with recent x-ray analyses of human Fab’ crystals (24) that clearly show the presence of two globular subunits (V and C) in Fab’, each comprised of two domains, one of which probably corresponds to Fv (VL + VH) and the other to intercalated CL + CH1.

The availability of Fv-fragment may also permit the chemical synthesis of antibody-combining sites.”

doi: 10.1073/pnas.69.9.2659.

When dissecting this study, which was linked to the initial source as the first atomic resolution structure of the Fv fragment of an antibody, what really stood out to me is that the images provided in this paper leave a lot to be desired:

Now, I’m no expert in atomic resolution images and I could be wrong (I’m not), but it seems fairly obvious to me that these are not only NOT atomic resolution images of an Fv fragment, there are no images of any fragment at all within the paper. In fact, the whole study was essentially a detailed explanation of the numerous processes used to break a supposed protein down into a smaller piece which could then be used to pluck into a hypothetical model of an entity never observed before. It would seem logical that prior to claiming a broken fragment belongs to this invisible entity, one must observe the entity in its entirety first. It is clear that this paper provided neither atomic resolution images nor any images of an Fv fragment, let alone any images of antibodies of any kind.

Since the first source seemingly led me down an incorrect path, I turned my attention to the 1973 Givol study linked in the Wikipedia article listed as the characterization of the Fv fragment. As it consists of much of the same methods, I am not going to do an in-depth recap of the processes used. However, I do want to highlight a few statements made as well as some of the images provided. What you will see is that the researchers performed the same deconstruction of the supposed protein said to come from the blood of tumor-bearing mice and then applied their findings to a hypothetical model of what they assumed an antibody looked like. In other words, they were attempting to reverse engineer an invisible fictional entity from random protein particles:

An Active Antibody Fragment (Fv) Composed of the Variable Portions of Heavy and Light Chains

“The amino acid sequences of immunoglobulins1 indicate some general features of the gross architecture of these molecules. (1) Each chain is comprised of homology regions, 110-120 residues long. The light chain has two such regions and the heavy chain four. The N-terminal region of each chain is the variable portion, whose sequence varies from one molecule to another, whereas the rest of the sequence is identical within subclass. (2) The intrachain disulfide bonds are linearly and periodically distributed in the structure such that each homology region includes one disulfide bond. This leads to the hypothesis that each homology region folds independently into a compact domain stabilized by one disulfide bond and linked to the neighboring domains by less tightly folded stretches of the peptide’s chain (Edelman, 1971). Another consequence of this arrangement is that noncovalent interactions take place between two identical or homologous domains on different peptide chains, to yield a globular unit which performs a specific function.

This general arrangement of the molecule (Figure 1) finds support in recent X-ray crystallographic studies of the Fab’ fragment (Poljak et d., 1972) and the light chain (Epp et al., 1972). Moreover, it is possible, under controlled conditions, to split the molecule between different globular units or domains.”

“We have demonstrated that the Fab’ of protein 315 (mouse IgA myeloma protein possessing anti-Dnp activity, Eisen et al., 1968) can be digested by pepsin to yield a fragment which is half the size of Fab’ and retains all the binding activity of the intact protein (Inbar et al., 1972). This fragment (denoted Fv) is the N-terminal fragment of Fab’ and contains the variable portions of both heavy and light chains. Its size may, therefore, be the minimal possible size for a fragment which possesses an antibody site. This finding also implies that the constant portions of Fab do not contribute to antigen binding.”

“It is obviously necessary to extend this finding to antibodies and myeloma proteins from other species. The possibility of obtaining such a small molecular weight fragment which contains all the structural requirements for the antibody binding site may prove to have been an important step in the fine structure analysis of antibody sites as well as in the preparation of synthetic antibodies.”

doi: 10.1021/bi00730a018

In Summary:

• The term ‘atomic resolution’ was defined a long time ago and it is generally accepted to correspond to an electron-density map (or a map calculated using another type of data, for example nuclear density) in which individual atoms can be distinguished
• However, these terms are very often abused in order for the published structures to appear to be more accurate than they are in reality
• Many X-ray and neutron crystal structures claim ‘atomic resolution’ although they were determined on the basis of data extending to only -2 A˚ or less
• This term is also sometimes used for structures determined by NMR, in which case the meaning of ‘resolution’ is even more uncertain
• The term ‘near-atomic resolution’ is sometimes used to describe structures at the resolution as high as 1.0 A˚
• Since these terms are currently being used by scientists practicing different techniques for
  structure determination, an agreement on their exact meanings might be very helpful
• The authors postulated that maybe all structural communities, including traditional macromolecular X-ray and neutron crystallography, XFEL and cryo-EM, among others, should adopt, or indeed respect, standard definitions of what these terms are supposed to mean
• Such an agreement would help the readers of structural papers to obtain a realistic impression of the likely accuracy of the structures based on the resolution of the primary experimental data
• Atomic-resolution electron microscopes utilize high-power magnetic lenses to produce magnified images of the atomic details of matter
• Doing so involves placing samples inside the magnetic objective lens, where magnetic fields of up to a few tesla are always exerted which can largely alter, or even destroy, the magnetic and physical structures of interest
• Prof. David Givol spent a significant part of his career investigating the structure-function relationship of antibodies where he identified the smallest fragment of antibody containing all its binding properties, i.e. the part of the molecule responsible for recognizing the foreign substance attacking the organism
• Today, this fragment – called Fv – is used in genetic engineering techniques for producing synthetic antibodies for treating various diseases

• The Fab’-fragment of a mouse IgA-myeloma (protein-315) was split by pepsin to yield a smaller fragment that retained the anti-2,4-dinitrophenyl activity
  of the intact protein
• The N-terminal sequence of Fv suggested that it is composed of the N-terminal half of Fab’, and consists of the variable portions of the heavy and light chains
• It is stated that it is generally agreed that antibody specificity results from different amino-acid sequences; this suggestion is strongly supported (i.e. not confirmed) by experiments on refolding of completely unfolded Fab or peptide
  chains
• It is claimed that it is theoretically possible for a binding site in the constant region to be modulated by different sequences in the variable region
• It was also said that it was possible that the constant region of the Fab is necessary for maintenance of the combining site
• Other attempts to produce an active fragment with molecular weight smaller than 46,000 failed
• It seemed possible, therefore, that the Fd-piece could also be split in the middle under similar conditions
• The myeloma protein-315 and its pepsin-produced Fab’-fragment were prepared from the serum of tumor-bearing mice
• Ever wonder why they don’t attempt to get these antibody fragments from human blood rather than from the blood of mice with tumors?
• Givol felt that their results showed clearly that peptic cleavage of protein-315 yields a small fragment (Fv) that retains the binding activity of the original protein
• In other words, he was certain that if they tried to break the protein, they could obtain a smaller fragment…shocking…
• This fragment was half the size of the Fab’-fragment, and its aminoacid composition and limited N-terminal aminoacid sequence were consistent with the view that it contained the N-terminal half of L chain and Fd, i.e., the variable segments designated VH and VL
• He concluded that it was very likely that each of the chains in Fv has 15-16 residues from the constant region
• It also seemed likely that those additional 15 residues from the constant parts were in the interdomains area and, although not proven, he assumed that they have no effect on the binding site
• Givol felt that it had been demonstrated that in its reaction with Dnp ligands, protein-315 exhibits essentially all the characteristics of binding sites of conventionally prepared anti-Dnp antibodies
• Hence, it was likely that an active Fv-fragment may be obtainable from other antibodies, although the procedure may vary for antibodies from different species, and classes
• Givol claimed that his results also suggested a modification of the “domain hypothesis” where each homologous region in an immunoglobulin chain (110-120 residues) folds separately as a globular domain
• His findings indicated that the VH and VL domains are intercalated and folded as a single globular unit, resistant to enzymic digestion
• In other words, Givol felt his guess was better than the previous guess
• This was said to be particularly consistent with recent x-ray analyses of human Fab’ crystals which “clearly showed” the presence of two globular subunits (V and C) in Fab’, each comprised of two domains, one of which probably corresponds to Fv (VL + VH) and the other to intercalated CL + CH1
• Givol ended his paper by claiming that the availability of Fv-fragment may also permit the chemical synthesis of antibody-combining sites
• In other words, now that they could break proteins into smaller and smaller pieces, this somehow granted the means to create fake antibodies

• The amino acid sequences of immunoglobulins1 are said to indicate some general features of the gross architecture of these molecules (in other words, they are claiming that they can model the form of the entity based on sequence information)
• This led to the hypothesis that each homology region folds independently into a compact domain stabilized by one disulfide bond and linked to the neighboring domains by less tightly folded stretches of the peptide’s chain
• Givol claimed that the general arrangement of the molecule seen in the drawing in Figure 1 found support in X-ray crystallographic studies of the Fab’ fragment and the light chain
• He claimed to have demonstrated that the Fab’ of protein 315 (mouse IgA myeloma protein possessing anti-Dnp activity, Eisen et al., 1968) can be digested by pepsin to yield a fragment which is half the size of Fab’ and retains all the binding activity of the intact protein (i.e. they broke the particle into a smaller piece…magic!)
• It was stated that its size may, therefore, be the minimal possible size for a fragment which possesses an antibody site and that this finding also implied that the constant portions of Fab do not contribute to antigen binding
• Givol admitted that it was obviously necessary to extend their finding to antibodies and myeloma proteins from other species
• He claimed that this work showed the possibility of obtaining such a small molecular weight fragment which contained all the structural requirements for the antibody binding site (how would they know all of the structural requirements for a fragment of an antibody never seen in a purified and isolated state?)
• Givol concluded that this may prove to have been an important step in the fine structure analysis of antibody sites as well as in the preparation of synthetic antibodies

What are we to take away from this quick detour into the supposed first atomic resolution imaging and characterization of the antibody Fv fragment? It is clear to see that there was absolutely no atomic imaging done whatsoever. It is also clear that Givol broke down what he claimed was the the Fab fragment from the blood of a tumor-bearing mouse, split it into a smaller fragment, named the resulting smaller particle as the Fv fragment, and then designed theoretical functions and characteristics around it. Mind you, there is no actual image of either a Fab or Fv fragment in the entire 1972 study, just some nifty little graphs and charts. The 1973 article does provide images of particles claimed to be antibody fragments which were then inserted into the hypothetical model of what the researchers believe an antibody looks like as they had never observed one as a whole entity before. Ultimately, this foray into the Fv fragment serves as a further reminder of the unscientific and illogical steps these researchers will take to sell their science fiction stories to the unsuspecting public.