Small compounds such as synthetic drugs and natural compounds are insufficiently complex to induce an immune response and antibody production. For antibody production against small compounds, a hapten must be chemically conjugated to carrier proteins such as bovine serum albumin (BSA). The first step for production of such mAbs is the preparation of a hapten–carrier protein conjugate, which directly combines an immune antigen with the carrier protein. Thus, G-Rb₁ was oxidized with NaIO₄ solution to generate aldehyde groups on the sugar moiety, allowing the aldehyde group to react with BSA (Figure 1a) [4]. Subsequently, it was necessary to characterize the synthesized antigen G-Rb₁-BSA conjugate. However, direct and appropriate methods to determine the hapten-conjugated carrier proteins without differential UV analysis, radiochemical, or chemical methods were not reported. To determine the hapten density of immunoconjugates, Wengatez et al. used matrix-assisted UV laser desorption/ionization (MALDI) mass spectrometry [29]. We also performed direct analysis of hapten–carrier protein conjugates by MALDI-TOF mass spectrometry [1,2,3,4,5,6,7,8,9,10,11,12,13,14,30,31,32]. Figure 1b indicates the MALDI-TOF mass spectra of the G-Rb₁-BSA conjugate. A broad peak coinciding with the G-Rb₁-BSA conjugate appeared between 70,000 and 90,000 m/z and centered at approximately 79,469 m/z. The molecular weight of BSA is 66,433, and the calculated molecular mass of the G-Rb₁ moiety is 1,109. These data suggest that most BSA was effectively conjugated to G-Rb₁. The experimental results and molecular weight of BSA revealed that the calculated molecular mass of the conjugated G-Rb₁ is between 3,327 and 23,289, suggesting that 3–21 (average of 12) molecules of G-Rb₁ are conjugated to BSA [4]. This method is therefore suitable for the characterization of hapten–carrier protein conjugates.

The synthesized G-Rb₁-BSA conjugate was injected into mice [4]. Briefly, BALB/c female mice were injected intraperitoneally with G-Rb₁-BSA dissolved in phosphate buffer saline (PBS) four times. The first immunization (50 μg protein) was injected as a 1:1 emulsion in Freund’s complete adjuvant. The second and third immunization (50 μg protein in each injection) were injected as a 1:1 emulsion in Freund’s incomplete adjuvant. On the third day after the final immunization (100 μg protein), splenocytes were isolated and fused with a HAT-sensitive mouse myeloma cell line, P3-X63-Ag8-653, by the polyethylene glycol (PEG) method. After cell fusion, we assessed the binding activity against carrier protein BSA, and the hybridomas secreting antibody against BSA were removed. Hybridomas producing mAb reactive to G-Rb₁ were cloned by the limited dilution method [33]. Established hybridomas were cultured in eRDF medium supplemented with 10 μg/mL insulin, 35 μg/mL of transferrin, 20 μM ethanolamine and 25 nM selenium (ITES) [34]. The mAb was purified using a Protein G FF column [4]. The cultured medium containing the IgG was adjusted to pH 7 with 1 M Tris solution and subjected to the column, and washed the column with 10 mM phosphate buffer (pH 7). Absorbed IgG was eluted with 100 mM citrate buffer (pH 3). The eluted IgG was neutralized with 1 M Tris solution, then dialyzed against PBS (pH 7.4) three times, and finally lyophilized. To further examine the binding specificity of this mAb against G-Rb₁, but not the carrier protein BSA, we immobilized G-Rb₁ to human serum albumin (HSA) and adsorbed the G-Rb₁-HSA conjugate onto a polystyrene microtiter plate. Subsequently, the specificity of IgG-type mAb, 9G7, for G-Rb₁ was examined by competitive ELISA using a microtiter plate coated with the G-Rb₁-HSA conjugate. Under these conditions, the free mAb bound to free G-Rb₁ or the G-Rb₁-HSA conjugate and the full measurement range of the assay using this anti-G-Rb₁ mAb extended from 20 to 400 ng/mL. Since cross-reactivity is the most important factor in determining the utility of antibodies, assay specificity was tested by determining cross-reactivity of mAb with various related compounds. Cross-reactivity against G-Rc and G-Rd, which possess diglucose moieties attached to C-3 hydroxy groups, was weak compared with that against G-Rb₁ (0.024% and 0.020%, respectively), and G-Re and G-Rg₁ showed negligible cross-reactivity (<0.005%). Therefore, it is clear that this mAb reacts predominantly with G-Rb₁ and not with its structurally related compounds or the carrier protein BSA.

To couple mAb to the Affi-Gel Hz gel, the sugar moiety of the purified anti-G-Rb₁ mAb was oxidized by NaIO₄ to give a dialdehyde group on the sugar moiety. The oxidized anti-G-Rb₁ mAb was then coupled to the Affi-Gel Hz hydrazide gel to form a hydrozone-type immunoaffinity gel [28]. The immunoaffinity gel was poured into a plastic mini-column (Figure 2).

To determine the best conditions for adsorption and elution, 400 μg of G-Rb₁ was dissolved in phosphate buffer (PB) and loaded into the immunoaffinity column. After washing with 20 mM PB containing 0.5 M NaCl, different elution buffers were loaded into the column and the recovery efficiency of G-Rb₁ was examined by ELISA. Recovery of G-Rb₁ was somewhat increased by elution with 20 mM PB containing 10% MeOH and 0.5 M potassium thiocyanate (KSCN). Elution was optimal with 100 mM AcOH buffer (pH 4), which was used as a substitute for 20 mM PB. Although 20% MeOH enhanced the elution of G-Rb₁, the MeOH concentration >20% was ineffective. Taken together, elution with 100 mM AcOH buffer containing 20% MeOH and 0.5 M KSCN gave the best recovery of G-Rb₁.

To examine the anti-G-Rb₁ mAb-coupled immunoaffinity column, a crude extract of P. ginseng roots was loaded into the column. After washing the column with washing buffer to remove the unbound compounds, the bound compounds were eluted with elution buffer. Figure 3 shows that fractions 1–8 contained the overloaded G-Rb₁, as determined by ELISA. The Eastern blotting procedure, a highly sensitive on-membrane quantitative analysis of anti-G-Rb₁ mAb, indicated that these fractions contained other ginsenosides such as G-Rg₁, Rc, Re, and Rd. After washing, a single sharp peak was detected in the eluted fractions 21–24, which contained G-Rb₁. However, these eluted fractions were contaminated with a small amount of malonyl-G-Rb₁, as detected by Eastern blotting, which indicated that anti-G-Rb₁ mAb has almost the same reactivity with malonyl-G-Rb₁ [28]. Therefore, to convert malonyl-G-Rb₁ to pure G-Rb₁, these fractions were treated with a mild alkaline solution (0.1% KOH in MeOH) at room temperature [17]. Overloaded G-Rb₁ in washing buffer was repeatedly loaded, and purified G-Rb₁ was finally obtained. The activity of the anti-G-Rb₁ mAb-coupled immunoaffinity column was stable during all procedures. The capacity of the immunoaffinity column was approximately 20 μg of G-Rb₁/mL of gel, which did not decrease after 10 repeated purification cycles under the same conditions, as reported in one-step purification of forskolin from a crude extract of Coleus forslohlii root [15].

This methodology made it possible to purify G-Rb₁ from the crude extract without using complicated procedures and may be applicable to related studies of other families of saponins for which an acceptable one-step purification method has not yet been developed. In addition, it may be possible to separate the total ginseng saponins using a widely cross-reactive mAb against ginsenosides, such as anti-G-Re mAb [35]. A combination of immunoaffinity column chromatography, ELISA, and Eastern blotting could be used for high-sensitivity detection of G-Rb₁ from plants and/or experimental animals and humans. Using this combination of methods, we detected G-Rb₁ from Kalopanax pictus Nakai, which was not previously reported to contain ginsenosides [36].

We separated G-Rb₁ from the crude ginseng extract using the anti-G-Rb₁ mAb-coupled immunoaffinity column. The capacity of the immunoaffinity column is 20 μg of G-Rb₁/mL of gel [28]. Upon loading the crude extract into the immunoaffinity column without exceeding the G-Rb1 binding capacity, the column specifically eliminated G-Rb₁ from the crude ginseng extract. After loading the ginseng extract into the immunoaffinity column, the washing and eluted fractions were collected. Figure 4 indicates the TLC profile of each fraction stained by H₂SO₄. A standard of ginsenosides was spotted on lanes 1 (G-Rd, G-Rc, and G-Rb₁) and 2 (G-Rg₁ and G-Re). The crude extract, washing fraction, and eluted fraction were spotted on lanes 3, 4, and 5, respectively. In the crude extract (lane 3), all spots of ginsenosides were clearly detected. In contrast, the washing fraction (lane 4) contained all ginsenosides, except G-Rb₁, and G-Rb₁ was detected in the eluted fraction (lane 5). These data clearly demonstrate that the G-Rb₁ molecule from the ginseng extract was captured by the anti-G-Rb₁ mAb-coupled immunoaffinity column. Thus, the washing fraction was referred to as the G-Rb₁-KO extract because it lacked G-Rb₁ [37,38]. This G-Rb₁-KO extract may be useful for pharmacological characterization of G-Rb₁ in ginseng extracts and traditional medicines such as Kampo medicines and TCMs that prescribe ginseng.

The anti-GC mAb was prepared using the same procedure as for G-Rb₁ [7]. Cross-reactivity of the anti-GC mAb against glycyrrhetic acid-3-O-glucuronide and glycyrrhetic acid was 0.585% and 1.865%, respectively. Cross-reactivity with other related compounds, including deoxycholic acid, ursolic acid, and oleanolic acid, was <0.005%. Thus, we established competitive ELISA and Eastern blotting using anti-GC mAb to specifically detect GC [7,45]. In addition, we used the anti-GC mAb-coupled immunoaffinity column to prepare the GC-KO extract. After purifying 60 mg of anti-GC mAb and coupling it to 25 mL of an Affi-Gel Hz gel, the immunoaffinity gel was packed into a plastic column [18]. To eliminate GC from LE, 12 mg of LE, which contained 1275.0 μg of GC in loading buffer (5% MeOH) was applied to the immunoaffinity column. After circulation of the loading buffer to enhance the binding efficiency of GC, the unbound fraction was collected. The column was washed with washing buffer (5% MeOH) and then eluted with elution buffer (20 mM PB containing 30% MeOH). After elution of bound compounds, each fraction was deionized, the solvent was lyophilized, and the GC concentration was determined by ELISA. The unbound fraction contained 3.50 μg of GC (0.27% of the loaded GC), while the bound fraction contained 1269.26 μg of GC (99.55% of the loaded GC), suggesting that the anti-GC mAb-coupled column eliminated 99.55% of the loaded GC. In the unbound fraction, 3.50 μg of GC (0.27% of the applied GC) was detected. In contrast, 1269.26 μg of GC (99.55% of the applied GC) was found in the bound fraction. These data suggest that the anti-GC mAb-coupled column eliminated 99.55% of the loaded GC and produced a GC-KO extract [46].

To characterize this GC-KO extract, we performed Eastern blotting and TLC analysis [46]. As shown in Figure 5a, Eastern blotting using anti-GC mAb demonstrated that GC was detected in LE (lane 2), but the spot corresponding to GC was not present in the GC-KO extract (lane 3). Furthermore, TLC analysis (Figure 5b) showed that many compounds, including GC, were present in LE (lane 2), whereas GC was completely absent in the GC-KO extract (lane 3). Taken together, these data clearly demonstrate that GC molecules in LE can be eliminated using the anti-GC mAb-coupled immunoaffinity column and that the unbound fraction was indeed a GC-KO extract.

The GC-KO extract is a useful tool for clarifying the interaction between GC and other compounds contained in LE. To investigate the potential function of GC in LE, we used macrophages to compare the effects of LE and the GC-KO extract on the production of nitric oxide (NO). The free radical NO has multiple effects on various systems, including host defense against pathogens, inhibition of tumor cells, and neurotransmission [47,48,49]. However, excess production of NO, which is synthesized by the inducible NOS (iNOS), can be harmful and can trigger rheumatoid arthritis, gastritis, bowel inflammation, neuronal cell death, and bronchitis [50,51]. Macrophages produce large quantities of NO after stimulation with bacterial lipopolysaccharide (LPS), and concomitantly produced inflammatory cytokines participate in the pathogenesis of inflammatory diseases [52]. Overproduction of NO by iNOS appears to be involved in the pathogenesis of various inflammatory diseases [53]. Therefore, inhibition of NO production by blocking iNOS expression may be useful in treating NO overexpression-mediated diseases.

To examine the inhibitory effect of LE on NO production, we used murine RAW264 macrophage cells, which produce NO upon stimulation with LPS. NO induction by LPS was blocked by treatment with LE in a dose-dependent manner [46]. Furthermore, Western blotting and RT-PCR showed that iNOS protein and mRNA expression was downregulated in a dose-dependent manner by LE and was completely suppressed in the presence of 100 μg/mL of LE. ELISA using anti-GC mAb demonstrated that 100 μg of LE contains 10.6 ± 0.618 μg of GC. However, significant inhibition of NO production and iNOS protein expression was not observed with GC at approximately 10.6 μg/mL.

Figure 6a shows that LE treatment (100 μg/mL) markedly inhibits NO production (inhibition ratio (IR) = 57.7%) compared with LPS treatment alone. In contrast, treatment with GC (10.6 μg/mL) did not suppress NO production. Subsequently, we treated cells with the GC-KO extract (89.4 μg/mL) or the combination of GC-KO extract (89.4 μg/mL) and GC (10.6 μg/mL). Interestingly, although GC alone did not block NO production, the inhibitory effect of the GC-KO extract (IR = 17.8%) was weaker than that of LE. Moreover, co-treatment with the GC-KO extract and GC significantly improved inhibition (IR = 33.5%). To determine whether these effects depended on iNOS expression, we performed Western blotting. As shown in Figure 6b, inhibition by the GC-KO extract was weaker than that by LE and addition of GC to the GC-KO extract rescued the inhibition. These data suggest that GC alone cannot suppress iNOS expression; however, in combination with the other constituents of LE, GC suppresses iNOS expression. Thus, in vitro and in vivo analyses using KO extracts can elucidate the functions of target natural compounds in crude extracts. Moreover, we have compared activity using GC contained in elution fraction and commercially available GC when the cells were treated with GC-KO extract. Since the activity was almost same level, the function of purified GC was not lost during the purification step. In future studies, we will investigate the detailed mechanisms including the differences between the original LE and GC-KO extract, the interactions between GC and GC-KO extract, and the action of GC in LE.