Assessment of a western blot signal for the Bcnt/Cfdp1, a tentative component of Srcap chromatin remodeling complex; trial to overcome off-target problems

The BCNT (Bucentaur) protein family is characterized by a conserved amino acid sequence at the C-terminus (BCNT-C domain) and plays an essential role in gene expression and chromosomal maintenance in fungi, fly, and chicken. The mammalian Bucentaur/Craniofacial developmental protein 1 (Bcnt/Cfdp1) is also a tentative component of the Srcap (SNF2-Related CBP Activator Protein) chromatin remodeling complex, but little is known about its properties, partly because there are few suitable antibodies to detect the endogenous protein. We used multiple anti-Bcnt/Cfdp1 antibodies against unrelated immunogens derived from BCNT-C domain and mouse-specific N-terminal peptide. To assign western blot signals and evaluate these antibodies, we utilized a stem cell line from mutant embryos of mouse Bcnt/Cfdp1, whose mRNA expression levels were reduced to 75% of the parental cells. In western blotting of these mutant and parental cell extracts with the anti-Bcnt/Cfdp1 antibodies, mouse Bcnt/Cfdp1 was detected as a doublet of approximately 45 kDa. LC-MS/MS analysis of the corresponding doublet for the Flag-tagged mouse Bcnt/Cfdp1 constitutively expressed in T-REx 293 cell (a HEK293 derivative) exhibited that the upper band was much more phosphorylated than the lower band and that there was preferential Ser phosphorylation in the WESF motif in the BCNT-C domain. Western blot with these validated antibodies indicated a preferential expression of Bcnt/Cfdp1 in the early stages of brain development in mouse and rat, which is consistent with the expression of Bcnt/Cfdp1 mRNA. This article describes the evaluation of anti-Bcnt/Cfdp1 antibodies, including a scheme to prepare a potential negative control for western blot, and discusses immune-cross reactions with off-target proteins, particularly immunoreaction probabilities.


2 3
Therefore, we used Flag-tagging to expect the lower tag specific background, though 1 2 4 having negatively charged sequence (DYKDDDDK). expression was significantly stronger than that of the constitutive expression (Fig 1).

4 0
These features are similar to those of His-tagged hBCNT/CFDP1, as previously 1 4 1 anti-His tag positive proteins [15]. Indeed, Nono (p54rnb) [20], which contains the 3 8 1 6 the other hand, the exogenous hBCNT/CFDP1 in the extract of G11 clone were well 3 9 7 recognized by both Abs. Therefore, we utilized a Cfdp1-K1 cell line that has been knockout cells [21], but the expected cells were not obtained so far (W. Kobayashi, 4 1 3 personal communication). Finally, we were able to assign the signal of endogenous 4 1 4 mBcnt/Cfdp1 detected by two Abs raised against unrelated immunogens. One of the 4 1 5 antigens is a mouse-specific N-terminal peptide, and the other is a peptide of BCNT-C 4 1 6 domain, which is highly conserved in mammalian Bcnt/Cfdp1.

1 7
The following is evidences that the ~45-kDa band is the signal of endogenous

2 2
Ab and A305-624A-M), each of which was generated using mutually unrelated

2 4
We confirmed the specificity of the anti-mBcnt-N Ab concerning the ~45-kDa signal by  The target band at ~45-kDa appears significantly smaller than signals reported by 4 2 8 many available anti-Bcnt/Cfdp1 Abs, including our custom-made anti-BCNT-C Ab.

3 0
Abs was significantly different from that of endogenous Bcnt/Cfdp1 (S1 and 4C); therefore, the 50-kDa signal is probably a common non-specific band(s). Since

0
It was evident that the anti-BCNT-C Ab detected a weak signal to the target 4 4 1 molecule (Fig 4C). A similarly difficult situation must occur with many Abs, which is 4 4 2 considered to be problematic (e.g., [24], [25] analysis of the molecule.

7 2
Moreover, a strategy for Ab validation has been proposed [28]. However, the focus of 4 7 3 this discussion appears to be blurred, at least regarding western blot analysis.

7 7 5
by the reaction probability. Epitopes are conventionally divided into two categories: linear or sequential and discontinuous or conformational epitopes [29]. However, Abs 4 7 7 do not recognize even linear epitopes as a series of amino acid residues but rather the 4 7 8 physicochemical and stereochemical states that they constitute, and these properties as a 4 7 9 whole constitute epitope [30,31]. For example, the following two cases may reflect   with 0.75 mL accutase at room temperature. After 5 min, the cell layer was   After washing with chilled HBS (10 mL/100-mm dish) followed by removing the 6 1 1 buffer completely using a piece of filter paper, the cells were homogenized with a cell  The frozen supernatant of E3 colony was thawed, and Nonidet P-40 (NP-40) was 6 2 9 added to a final concentration of 0.05%. After sonication for 30 s in the ice-water bath 6 3 0 (3 x 10-s pulses at 10-s intervals) followed by centrifuging at 10, 000 x g for 1 min at    Protein extracts from mouse and rat brain 6 6 0 Cerebrum and cerebral cortices were dissected from mice (P0 of C57BL/6J, male)

2
These isolated samples were snap-frozen in liquid N 2 and stored at -80 until use.

3
Frozen samples were crushed with a hammer on dry ice and immediately transferred to tissues were homogenized at 600 rpm using a digital homogenizer and boiled for 5 min.

7 0
A tandem gene duplication followed by recruitment of a retrotransposon       Caoili SE, Expressing Redundancy among Linear-Epitope Sequence      quality Abs for research use as well as Ab drugs and vaccines, it is critical not only to design immunogens that generate Abs with quite specific immunoreactivity for on-target molecules but also to predict more accurately cross-reactivities of Abs with "off-targets", which potentially induce side effects in therapies using Abs and vaccines (6). The amino acid sequences of epitopes have been considered to be a primary basis for specific immunoreactions of Abs with target molecules. However, in addition to Abs against proteins, those against nonprotein molecules, such as anti-DNA and anti-phospholipid Abs, are also well known. On the other hand, recent reports demonstrated that a multispecific monoclonal Ab recognizes completely different linear epitope sequences with high affinity (7), and progress is proceeding toward development of therapies by intentionally using such a dual-specific Ab (8). These examples are difficult to understand from a simple grasp of epitopes based on only their amino acid sequences. The epitopes are conventionally classified into two groups, linear or sequential and discontinuous or conformational (9, 10).
Such a conventional classification based on amino acid sequences is useful to simply explain epitopes of proteins but insufficient to more accurately understand epitopes in antigen-Ab interactions such as the above-described examples because these examples indicate that Ab recognition of target molecules is based on the physicochemical environment of epitopes and does not necessarily depend on their linear amino acid sequences. Accordingly, it is necessary to have a more comprehensive understanding of Ab specificities on the basis of physicochemical features of epitopes. In this report, we present a didactic example to consider these points.
We identified the Bcnt (Bucentaur) / Cfdp1 (craniofacial developmental protein 1) protein family, which probably function as a component of chromatin remodeling complex by the study of yeast ortholog Swc5 (11). Previously, we raised an Ab against a peptide of a C-terminal region of BCNT (called BCNT-C), EELAIHNRGKEGYIERKA, which is evolutionarily conserved from yeast to humans (12). This anti-BCNT-C Ab detected doublet bands of 43/45 kDa by Western blot analysis of proteins extracted from bovine and rat brains. We tried to isolate them by immunoprecipitation and identify them by mass spectrometry. Surprisingly, we obtained the unexpected result that the 43 kDa protein is glutamine synthetase (GS: NP_001035564.1); EC 6.3.1.2, also known as γ-glutamate: ammonia ligase (13,14), an enzyme entirely unrelated to Bcnt, which catalyzes the ATP-dependent condensation of glutamate and ammonia to form glutamine. The anti-BCNT-C Ab can specifically detect not only recombinant GS but also endogenous GS in rat brain. This quite specific reactivity of the Ab with an off-target molecule prompted us to seek a common epitope between GS and Bcnt to elucidate a chemical basis for the cross-reactivity of the anti-BCNT-C Ab with GS. We report here that the core amino acids of the common epitope causing the cross-reaction were identified as GYFE in GS and GYIE in Bcnt. Also, we discuss here the significance of physico-or stereochemical environments in epitope-paratope (10) or protein-protein interactions (15) and findings on the proteome-wide analyses of epitopes (9).

Identification of the 43 kDa protein
We found that we could immunoprecipitate the 43 kDa protein, but not the 45 kDa protein, from the extracts of bovine brain and rat olfactory bulb (16) when the extracts were boiled once in SDS, whereas we failed using a conventional Radioimmunoprecipitation assay (RIPA) buffer (see Supplementary Fig. S3 online). Using this method, we isolated the 43 kDa protein from the extracts by ammonium sulfate precipitation, phenyl-sepharose chromatography (see Supplementary Fig. S4A online), and immunoprecipitation using agarose beads coupled with the anti-BCNT-C Ab (see Supplementary Fig. S4B online).  Table S2 online). Detection of GS in these three distinct bands was probably caused by a mobility shift due to its modification by ubiquitination (17). These data suggest that GS, a protein not homologous to Bcnt, was the most promising candidate for the 43 kDa protein recognized by the polyclonal anti-BCNT-C Ab. It should be noted that the anti-BCNT-C Ab with cross-reactivity to GS was generated in seven of eight guinea pigs immunized with the 18-mer BCNT-C peptide in two independent immunizations, making it unlikely that generation of the anti-BCNT-C Ab with such a cross-reactivity was accidental.
that some characteristics of the immunoreactant recognized by the anti-BCNT-C Ab are also consistent with those previously reported for GS (see Supplementary Fig. S8 online, ref. 20,21). Because we confirmed that the anti-BCNT-C Ab really recognizes GS, a protein not homologous to Bcnt, we conducted epitope analysis of GS to investigate the mechanisms for such a peculiar cross-reactivity.

Epitope analysis of GS for cross-reactivity of the anti-BCNT-C Ab
First, we prepared several deletion mutants of GS protein in E. coli to identify the region  (Fig. 4B, lower panel). These results strongly suggest that the epitope must be included in the P1 peptide sequence. Furthermore, to identify key amino acids involved in the cross-reaction, we prepared graded deletion mutants of GS and examined their reactivity with the anti-BCNT-C Ab by Western blotting. As shown in Fig. 4C, the cross-reactivity was clearly observed for the Del-338 mutant but was conspicuously absent for the Del-336 with equivalent immunoreactivity (Fig. 5A, right), whereas the flow-through fraction did not react to F-mGS (Fig. 5A, middle). These results suggest that the subpopulation of the polyclonal anti-BCNT-C Ab cross-reacting with GS, derives from only Ab molecules recognizing epitopes including the core amino acids, GYFE. On the other hand, the immunoreactivity of the GYFE fraction was inhibited with a peptide EELAIHNRGKEGYIERKA-NH 2 , BCNT-C (GYIE) in Fig. 5B, but not with its mutant replacing GYIE with Alanine, BCNT-C (All-Ala) in Fig. 5B. This result shows that the subpopulation of the polyclonal anti-BCNT-C Ab cross-reacting with GYFE of GS specifically reacts to GYIE of Bcnt. We also isolated the subpopulation from the anti-BCNT-C Ab by affinity purification on the membrane using full-length recombinant proteins of wild-type GS or All-Ala GS mutant expressed in E. coli (see Supplementary Fig.   S10 online). These results indicate that a certain subpopulation exists in the polyclonal anti-BCNT-C Ab that reacts to both GS and Bcnt with similar affinities, and that GYFE of GS composes a core sequence of the epitope required for the cross-reactivity of the anti-BCNT-C Ab. It should be noted that the Ab recognized both GS and Bcnt with similar immunoreactivities in spite of the difference between Phe (GYFE) for GS and Ile (GYIE) for Bcnt.

Spatial similarity of stereochemical environment between GS epitope and BCNT-C antigen
To understand the similar immunoreactivities of the anti-BCNT-C Ab against both GS and Bcnt in spite of the one amino acid difference of the core sequence, we considered the possibility that the cross-reactivity was caused by a spatial similarity of the stereochemical environment between the BCNT-C antigen peptide and a region of GS including GYFE. As described above, weak but significant immunoreactivity was detected for an alanine substitution mutant of GS (F337A) but not for other alanine-substitution mutants (Fig. 4D).
Taken together, we conjectured that F337 of GS is not directly involved in the interface of the antigen-Ab interaction between GS and the anti-BCNT-C Ab. To investigate how F337 actually participates in the antigen-Ab interaction, we prepared substitution mutants of F337 of GS with various amino acids having different steric bulks (the amount of space that side chain atoms of amino acids occupy) or polarities and tried to semi-quantitatively analyze the effect of those substitutions on stereochemical and physiochemical changes of GS by Western blotting. As shown in Fig. 6A, when four hydrophobic and branched-chain amino acids were substituted in the order of steric bulk from isoleucine to alanine, the cross-reactivity of the anti-BCNT-C Ab was attenuated with an accompanying decrease in the steric bulk of the substituted amino acid, but it was not completely abolished. This result further supports our conjecture that F337 of GS is not directly engaged in the contact site of the antigen-Ab interaction but importantly constitutes the stereochemical environment of the epitope. Fig. 6B shows a 3D layout of a region around GYFE of the common epitope of GS, which was drawn based on X-ray crystallography data of GS (PDB code 2OJW, ref. 22).
Note that Phe (F337) resides on the opposite side of neighboring Y336 and E338, occupying rather a space by its aromatic ring. This arrangement of F337 supports our speculation above that this amino acid is not directly involved in the contact site of the antigen-Ab interaction, but has a steric bulk effect on the stereochemical environment of the epitope.
On the other hand, the cross-reactivity was approximately equal to the F337Y mutant as compared with the wild-type GS (F337) (Fig. 6A). These data suggest that the polarity of tyrosine, a hydroxyl group adduct of phenylalanine, does not significantly affect the stereochemical environment of the epitope. In conclusion, the results obtained strongly suggest that the spatial similarity of stereochemical environments formed by the core amino acids of common epitopes of GS and Bcnt is critically important in the cross-reaction of the anti-BCNT-C Ab with GS.

DISCUSSION
In this paper, we showed that the anti-BCNT-C Ab quite specifically cross-reacted with GS, which is a protein not homologous to Bcnt, in addition to the original target molecule. GS is present predominantly in the brain where it participates in the metabolic regulation of glutamate (14,23). The anti-BCNT-C Ab recognized exogenous GS expressed in HEK cells and also showed strong immunoreactivity in astrocyte of adult rat cerebellum (see Supplementary Fig. S7 online), consistent with the localization of GS in brain previously reported (20). It is most likely that strong immunoreactivities of the anti-BCNT-C Ab in Leydig cells of bovine testis (24) and astrocytes of rat brain, we had conjectured as Bcnt/Cfdp1 so far, are immunoreactivities for endogeneous GS (20,25). In addition to its strong immunoreactivity, this Ab detects GS under different denaturing conditions: thermally and reductively denatured GS in Western blotting, chemically fixed GS in situ in immunocytochemistry and immunohistochemistry, and a more naïve form of GS in aqueous solution in immunoprecipitation. These data strongly suggest that the epitope including GYFE, which is reactive to the anti-BCNT-C Ab, localizes on the surface of the native GS.
Although no significant homology of amino acid sequences was found between Bcnt and located in this site of GS and GYIE of Bcnt, constitute similar stereochemical environments.
It is also noted that the conjectured core amino acids of the common epitope are bracketed by charged amino acid residues such as lysine and arginine (Fig. 4C, lower left). The hydrophilicity of these charged residues may bring the core amino acids of each epitope to the surface of GS or BCNT. Elucidation of this characteristic of epitopes awaits further studies.
It is well known now that whereas cross-reactions of Abs with structurally unrelated peptides of off-targets frequently occur in binding assays of Abs using small peptides, such cross-reactions rarely occur in those using protein fragments and in Western blot analysis (9). It has been generally believed that Abs recognize linear epitopes of target molecules under the denatured condition of SDS-PAGE, but some Abs can also react to proteins via conformational epitopes, probably depending on the protein renaturation on transferred membranes after denaturation in SDS-PAGE (26,27). The anti-BCNT-C Ab reacted to each GS mutant for Phe of GYFE in a steric bulk-dependent manner (Fig. 6A), in contrast to the loss of immunoreactivities in alanine-substituted mutants for Gly, Tyr, and Glu of GYFE (Fig. 4D). These results reflect differences in whether these amino acids exist at the contact site for the Ab (Gly, Tyr, and Glu) or not (Phe). This result, which is consistent with the configuration of the core amino acids displayed in the 3D structure of GS (Fig. 6B), suggests that Bcnt and GS are partially renatured and that their conformations are partially maintained on transferred membranes in Western blot analysis. Such renaturation of Bcnt and GS is inferred to be one of the reasons why we were able to elucidate the similarity of stereochemical environments between GYIE of Bcnt and GYFE of GS by Western blot analysis.
Recent reports including this paper indicate that cross-reactions of Abs with off-target molecules cannot be avoided, even when choosing a very specific amino acid sequence as an immunogen (9,28). Therefore, we have to pay attention to the fact that Abs do not necessarily recognize only "on-targets" but potentially react with "off-targets" that comprise similar stereochemical or physicochemical environments as that of on-target molecules.
Recent advances in computational and bioinformatic analyses using extensively accumulated data of epitope-paratope interfaces make it possible to group T cell receptors of common specificity using grouping algorithms of lymphocyte interactions by paratope hotspots (29). Furthermore, these advances have also revealed the functional redundancy of epitopes based on physicochemical similarities at a level of amino acid residues involved in the antigenic cross-reaction (30). In addition to these accumulated data, further comprehensive understanding of the physicochemical environment of epitope-paratope interfaces including both "on-targets" and "off-targets" may make it possible to evaluate and predict more accurately antigenicity and immunogenicity, and these advances will bring about therapeutic Abs and vaccines with high specificity and without adverse effects.

Ethical approval
All the genetic recombination experiments and all the animal experiments in the present study were approved by the Genetic Recombination Experiment Safety Committee and the Animal Care and Use Committee, respectively, of Tokushima Bunri University.

Reagents
Generation of anti-BCNT-C Ab, anti-mBcnt-N Ab, and anti-Bcnt-Cter Ab is described in Supplementary Materials and Methods. Abs against Flag tag (DYKDDDDK tag; 014-22383) and GS (GTX109121) were obtained from Wako Pure Chemical Industries and GeneTex, respectively. Unstained or prestained SDS-PAGE molecular weight standard markers (Bio-Rad) were used differently for each experiment.

Construction of expression vectors for mGS and mBcnt
A plasmid carrying Myc-tagged mGS was a gift from Dr. T. Araki (Natl. Inst. Neurosci., Tokyo) (17). Using cloning primers (see Supplementary Table S3-1

Preparation of extracts of E. coli expressing GS and its mutants
Culture of E. coli and induction of coded molecules by cDNAs in pCold II vector were described previously (11). E. coli were harvested and washed with PBS by centrifugation at Extracts were sonicated by a Bioruptor (BM Equipment) in an ice-water bath (15 x 10 s pulses at 10 s intervals) and followed by centrifugation at 28,000 x g for 10 min.
Supernatants were subjected to Western blot analysis.

Immunoblotting
Procedures of SDS-PAGE and blotting onto membranes were essentially the same as previously described (11)

Mass spectroscopy analysis
The CBB stained protein bands in the gel were excised and de-stained. The protein bands were digested with trypsin (TPCK-treated, Worthington Biochemical) at 37°C for 12 h in 50 mM Tris-HCl (pH 8.0). The digests were analyzed by nano LC-MS/MS using a Q Exactive mass spectrometer exactly as described previously (11).