Identification of Liver Epithelial Cell-derived Ig Expression in μ chain-deficient mice

Growing evidence indicates that B cells are not the only source of immunoglobulin (Ig). To investigate this discovery further, we used μMT mice, which have a disruption of the first transmembrane exon of the μ heavy chain and do not express the membrane form of IgM. These mice lack mature B cells and thus serve as a good model to explore Ig expression by liver epithelial cells. We found that Ig heavy chains (μ, δ, γ and α) and light chains (κ and λ) were expressed in sorted liver epithelial cells of μMT mice. Surprisingly, each heavy chain class showed its respective variable region sequence characteristics in their variable region, instead of sharing the same VDJ usage, which suggests that class switching does not occur in liver epithelial cells. Moreover, the γ and α chains, but not the μ and δ chains, showed mutations in the variable region, thus indicating that different classes of Ig have different activities. Our findings support the concept that non-B cells, liver epithelial cells here, can produce different classes of Ig.

Different classes of Ig were found in livers of μMT mice. To further explore the expression of Ig, we analyzed different classes of Ig in livers of μMT mice using a Rapid Mouse Ig Isotyping Array. First, blood was drawn from the eye canthus to obtain serum, and then the mice were perfused to remove blood cells and serum Ig. Protein was extracted from the liver, and the Ig in the liver lysate and the serum were quantitated by fluorescence with Rapid Mouse Ig Isotyping Array. As expected, there were low levels of Ig in the serum of μMT mice compared with WT mice. However, the levels of IgM, IgG1, IgG2b and Igκ in the μMT liver lysate were the same as those in the WT liver lysate. Moreover, the levels of IgA, IgD, Igλ and IgG2a were much higher in the μMT liver lysate than in the WT liver lysate (Fig. 3A). We also confirmed this characteristic of IgM and IgG in the liver lysate by ELISA (Fig. 3B).
Subsequently, to confirm that liver cells of μMT mice could produce Ig, we analyzed the μ and κ chains in the μMT liver lysate by mass spectrometry (MS). An SDS-PAGE band of liver lysate of approximately 75 kDa, which was stained with a goat anti-mouse IgM antibody by Western blotting, contained 11 peptides that correspond to the mice Ig μ heavy chain (Fig. 3C). Similarly, a band of approximately 26 kDa, which was detected by an anti-mouse Ig κ antibody by Western blotting, yielded 1 peptide that corresponded to a κ chain variable region and 1 peptide that corresponded to a κ chain constant region (Fig. 3D). In addition, 1 peptide corresponding to the λ chain constant region was found in the 26 kDa band (Fig. 3E).

IgM and IgG were secreted by liver epithelial cells in μMT mice.
To verify that the Ig found in the liver was produced by epithelial cells, we performed flow cytometric sorting of liver epithelial cells using the epithelial marker CK18 (Fig. 4A). We cultured the sorted cells and used ELISPOT to determine whether IgM or IgG was being secreted. In fact, sorted liver epithelial cells from both μMT and WT mice could secret IgM and IgG (Fig. 4B). Moreover, the levels of IgM and IgG were comparable between μMT and WT mice (Fig. 4B).

Ig heavy and light chain transcripts in sorted liver epithelial cells of μMT mice.
To confirm the presence of μ and κ chain transcripts in the liver cells of μMT mice, we performed Northern blot analysis used the DNA fragments of Igμ or Igκ constant region as probes, the Igμ and κ chain transcripts were significantly found in liver tissues of μMT mice (Fig. 5A). Subsequently, the Igμ and κ chain transcripts were also determined in the liver epithelial cells of μMT mice with digoxigenin (DIG)-Labeled RNA fragments of Igμ CH3-CH4, or Jk-Ck as probes by in situ hybridization (Fig. 5B).
We used RT-PCR to analyze the expression of Ig in the sorted liver epithelial cells. The variable and constant regions of the different classes of heavy chains and light chains were amplified using a series of primers (Table S1). We detected rearranged transcripts of heavy chains (μ, δ , γ , and α ) and light chains (κ and λ ) in cDNA prepared from the sorted liver epithelial cells of μMT mice (Fig. 5C). However, we did not detect transcripts of the Ig ε chain. We also did not detect transcripts of CD20, a B cell marker, which confirmed that there was no B cell contamination in the cDNA libraries. The identities of all of the PCR products were confirmed by DNA sequencing, and these sequences have been submitted to the GenBank database (GenBank Accession Numbers: KT726173-KT726219).
Ig variable region in sorted liver epithelial cells of μMT mice displayed distinct sequence characteristics. We cloned 86 VDJ rearrangements (from the heavy chain variable region) and 35 Vκ Jκ rearrangements (from the κ chain variable region) that were expressed by the sorted liver epithelial cells from 5 μMT mice. Sequence analysis showed that 82 of the 86 VDJ rearrangements and all 35 Vκ Jκ rearrangements were products of functional Ig rearrangements ( Table 1). The analysis of Ig repertoires showed that similar to our previous findings 6 , in the same sorted epithelial cell population, the V, D, and J usage and VDJ sequences were dissimilar among the μ, γ , α and δ heavy chains. In fact, each class of heavy chain showed its own unique VDJ pattern ( Table 1). The results further suggested that the liver epithelial cells did not possess a class switching mechanism. Interestingly, the μ, γ and κ transcripts showed a restricted clonal VDJ pattern, whereas the α and δ transcripts displayed VDJ diversity patterns similar to those exhibited by B cell-derived Ig and distinguished from Igμ and Igγ in liver epithelial cells (Table 1). To exclude an influence of the primers, we also amplified variable regions of the μ and κ chains from WT mice spleen cells and found the rearrangement patterns to be diverse (Table 2). In addition, we analyzed mutations in the representative VDJ and Vκ Jκ rearrangements (Fig. 6A). The primers for the λ light chain were designed specific to V λ 1/J λ 1, and the products were confirmed to be V λ 1/J λ 1 by DNA sequencing (data not shown). N/P addition in the V-D, D-J and V-J junctions is a classical feature of B cell-derived Ig, so we analyzed N/P addition in Ig transcripts derived from the liver epithelial cells. Markedly, 93% of the VDJ rearrangements had classical N/P addition, whereas only 5% of the Vκ Jκ rearrangements had N/P addition (Fig. 6B). Interestingly, we found that 86% of the Vκ Jκ rearrangements used Jκ 2 and that VDJ rearrangements frequently used JH4 and JH2 in Ig expressed by the liver epithelial cells of μMT mice (Fig. 6C).

Discussion
In this study, by using B cell-deficient μMT mice, we further provided evidence that Ig can be expressed by non-B cells under physiological conditions. We found that liver epithelial cells produced multiple classes of Ig, including IgG, IgM, IgA and IgD. These liver epithelial cell-derived Ig had distinctive characteristics in their variable regions.
In past 10 years, many studies have shown that non-B cell cancer cells, including epithelial cancer cells, and other types of malignant cells can produce Ig, especially IgG, which is involved in cell survival, proliferation, tumorigenesis and metastasis [35][36][37][38][39] . Ig expression in non-B cells was thought to be unique to malignant cells due to the genomic instability in these cells. However, in recent years, growing evidence has indicated that normal non-B cells, such as epithelial cells, endothelial cells, neurons, and even germ cells, can also express Ig. The expression of Ig in these non-B cells is of unknown significance, and its function remains poorly understood. Because these findings challenge the important classical theory that B cells are the only source of Ig, some immunologists are still questioning the possibility that non-B cells produce Ig. To address this issue, we avoided the contamination of B cells by using μMT mice (which have a deletion in the transmembrane domain of the Ig μ chain, thus blocking B cell development after the pro-B stage) as a model to determine whether Ig can be produced in non-B cells.
First, we found that mature B cells are not present in peripheral blood and lymphoid organs of μMT mice; furthermore, these mice have very low levels (~100 times lower than WT mice) of Ig in the bloodstream. Subsequently, using a series of techniques, including immunohistochemistry, western blotting, ELISPOT, MS, RT-PCR and DNA sequencing, we found that different classes of Ig could be expressed and secreted from non-B cells (such as liver epithelial cells) at both the mRNA and protein level. Moreover, the levels of Ig in non-immune organs in μMT mice was similar to those in WT mice, which suggests that non-B cells can spontaneously express Ig that is retained locally, whereas the Ig in the bloodstream may be derived mainly from B cells.
Hepatocytes are epithelial cells that are involved in synthesizing protein, cholesterol, bile salts, fibrinogen, phospholipids and glycoproteins. So far, there are no reports indicating that hepatocytes can express Ig or perform Ig-related functions. However, mouse liver tissue was used as a negative control by Tonegawa (1987 Nobel laureate in Physiology for discovery of Ig gene rearrangement) when he confirmed that Ig genes are rearranged in Ig-producing cells. In 1976, he used Southern blotting to show that a mouse myeloma cell line, but not liver tissue, was able to bind to a probe made from the mouse myeloma cell line's mRNA fragment of an Ig variable region [40][41][42] . It was concluded that Ig gene rearrangement occurred in the myeloma cells but not in liver tissue. This conclusion further strengthened the classical concept that only B cells can produce Ig. However, in the current study, we found that Ig gene rearrangement also occurs in μMT mouse liver epithelial cells sorted by FACS with the specific liver epithelial cell marker CK18. In light of the current knowledge of Ig gene rearrangement, it is likely that the conclusion made by Tonegawa did not take substantial differences in Ig gene rearrangement among Ig-producing cells into account. Different conclusion could have been reached if Tonegawa had used a liver cell-derived Ig variable region mRNA fragment as a probe in his experiments. In addition, our results reveal a novel Ig source in μMT mice because Ig can be produced from many non-B cells, such as liver epithelial cells, but is not limited to z "a small population of B cells" as described previously [32][33][34] . In fact, we did not find any B cells in peripheral blood or mature B cells in BM in the μMT mice we used. V(D)J gene rearrangement and transcription provide crucial evidence for Ig expression. We detected rearranged transcripts of Ig genes in sorted liver epithelial cells. To exclude the possible contamination of any other types of cells, we gated on the larger liver cells when CK18 + liver cells were sorted. As expected, rearranged functional heavy chain and light chain transcripts of IgG, IgM, IgD and IgA, but not IgE, have been found in the sorted liver epithelial cells. Sequencing analysis revealed that similar to B cells, the liver epithelial cells derived V(D)J or VJ showed classical Ig gene rearrangement patterns and had typical N/P additions among the V, D and J segments. Moreover, liver epithelial cell-derived IgA and IgG, but not IgM and IgD, also tended to show hypermutation. However, in the same liver epithelial cell population of μMT mice, each Ig heavy chain, including μ, δ , γ , α , showed distinctive variable region sequences, but not the same VDJ sequences. This finding suggests that class switching does not occur in liver epithelial cells, unlike the classical concept of class switching when IgM switches to IgA, IgG or IgE, only the constant region (not the variable region containing the VDJ sequences) is replaced. Indeed, we also found that non-B cells lack class switch mechanisms in our previous studies 6 . This finding suggests that the mechanism of Ig production in non-B cells is different from that in B cells, although the mechanism remains unknown.
Based on our current and previous results, non-B cell are able to express Ig; however, its function remains unclear. Our previous studies suggested that cervical cancer cell-produced IgM have natural IgM activity that could bind to ssDNA, dsDNA and bacteria 22 . Jiang et al. found that normal epidermal cells can express and secrete IgG and IgA, which can bind to different strains of bacteria 9 . We have preliminary results showing that IgM extracted from liver tissue of both WT and μMT mice possesses natural IgM activity (data not shown). This finding suggests that Ig from non-B cells, especially epithelial cells, may serve natural antibody functions and may be involved in natural immunity in local tissues. Our focus will be on the function of these Ig. Immunohistochemistry. At first, tissues were sectioned and fixed in 10% formalin, and then was embedded in paraffin. The paraffin sections were rehydrated by ethanol with the concentrations of 75%, 80%, 90%, 95% and 100% after deparaffinization. Antigen retrieval was performed in tris-EDTA buffer (pH 9.0) at 90 °C for 5-min in a microwave oven. To block the andogenous peroxidase activity, the slide was treated with 0.3% hydrogen peroxide for 5 min, and washed in PBS. Then, the sections were blocked with 10% normal goat serum for 10 min. Slides were incubated with primary antibody, goat anti-mouse IgM (VECTOR LAB, CA, USA) or goat anti-mouse Igκ (Southern Biotech, UBA, USA), for 60 min at 37 °C in a humidified chamber. After washed, the sections were incubated with the HRP-conjugated second antibodies (rabbit anti-goat IgG) at 37 °C for 40 min. After washing with PBS, the signal was detected using DAB (Dako, CA, USA). Sections stained without primary antibody were used as negative controls.
Protein extraction. Heart perfusion was performed to the mice anaesthetized by chloralic hydras before sacrification. The tissues were cut into small pieces and lysed in TSD buffer (50 mM Tris-HCL, 0.5% sodium dodecyl sulfate, 5 mM DTT) followed by ultrasonication. Lysate were centrifuged at 13,000 rpm for 20 min at 4 °C, and the supernatants were collected for ELISA or Western blot.
Rapid Mouse Ig Isotyping Array. As mentioned above, tissues lysate was prepared. Serum from mice was      Laboratories, USA) for overnight at 4 °C. Following the blocking with 10% FBS, cells were incubate with plates for 24 h at 37 °C. After washed with PBS, HRP conjugated detection antibody were incubated for 1 h at room temperature. Then AEC coloration was performed and analyzed by ELISPOT reader.
Northern blot analysis. Total RNA was extracted from liver or spleen cells with TRIzol reagent. A total of 10 μg liver-derived RNA or 5 μg spleen-derived RNA was separated in 1% agarose gel with formaldehyde, and then transferred to a nylon membrane. The sequence of probe specific to Igμ constant region gene was: ttcatctctgcgacagctggaatgggcacat, and the probe specific to Igκ constant region was: cgccattttgtcgttcactgccatcaatcttc. The probes were conjugated with biotin. The membrane was hybridized with 60 ng/ml denatured probe in hybridization buffer (Thermo) after prehybrization at 56 °C for 12-16 h. And the signal was detected with North2South ® Chemiluminescent Hybridization and Detection Kit (Thermo) according to the manufacturer's instructions.
In situ hybridization. In situ hybridization was performed on 6 μm serial sections of paraffin-embedded tissue sections. Plasmids inserted with constant region fragments of the Igμ and Igκ , which were obtain by PCR with Igμ CH3-CH4 primers and Jk-Ck primers (Table S1), were linearized. The RNA probes were labeled with digoxigenin (DIG) and transcribed by T7 RNA polymerase (for the antisense probe) or SP6 RNA polymerase (for the sense probe) by DIG northern starter kit (Roche, Rotkreuz, Switzerland). Paraffin-embedded tissue sections were performed deparaffinized and dehydrate. Then sections were treated with Proteinase K, fixed with paraformaldehyde, prehybridized at 42 °C for 2 h, and hybridized with the DIG-labeled RNA probe (5 μg/mL) at 42 °C overnight. After hybridization, the sections were washed in 2 × SSC and 0.1 × SSC at 37 °C, respectively, and then treated with RNase A. The samples were incubated with alkaline phosphatase-conjugated antidigoxigenin antibody (1:250; Roche, Rotkreuz, Switzerland). BCIP/NBT (Sigma, Saint Louis, USA) was used for the color reaction. Corresponding sense probes were used as controls.
RNA extraction and RT-PCR. Total RNA in spleen was extracted using TRIzol reagent (Invitrogen, Carlsbad, USA). For the sorted liver epithelial cells, RNA was extracted with RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the standard protocol. Reverse transcription (RT) was carried out with RevertAid First Strand cDNA Synthesis Kit (Fermentas, Glen Burnie, USA) by following the manufacturer's instructions. And then the cDNA was prepared for PCR. The primers used in this study were shown in Table S1. The Ig constant region genes were amplified at the annealing temperature of 56 °C for 35 cycles. Semi-nested PCR was performed for Ig V region genes. cDNA derived from BALB/c mice spleen cells was used as positive control. PCR products were separated with 1% agarose gel by electrophoresis, and then were stained with ethidium bromide. The PCR products were cloned to pGEM-T Easy Vector (Promega) to be sequenced by ABI 3730XL Genetic Analyzer. Sequence analysis. In order to analyze the usage and junctions of Ig variable region genes, all sequence derived from PCR products were submitted to the IMGT V-QUEST program. Alignments were performed with Lasergene software (DNAStar) or with BLAST in NCBI to compare with published sequences. Statistical analysis. All statistical calculations were performed using the statistical software program (GraphPad Prism 5.0 software). Differences between different groups were evaluated by the Student's t test. Differences were considered to be statistically significance when P was < 0.05.