Inactivation of the EGF-TM7 receptor EMR4 after the Pan-Homo divergence

2.1 cDNA cloning, sequence analysis and chromosomal localization of EMR4

Evidence for the existence of a novel EGF-TM7 receptor gene on human chromosome 19 was received during a survey of the public databases. In line with the designation of other EGF-TM7 family members, we called this novel molecule EMR4. The complete cDNA sequence of the low-copy gene EMR4, comprising 2,732 nucleotides (nt), was determined (Fig. 1). Alignment of the cDNA sequence with the sequence of human chromosome 19 indicated that EMR4 consists of 16 exons. Unexpectedly, the single open-reading frame was found to shift in exon 8 to a different reading frame that terminates ten codons downstream. Comparison of the cDNA sequence with the recently identified murine homologue of EMR4, also designated F4/80-like receptor (FIRE) 12, 13, showed that this frameshift is caused by a one-nt deletion in codon 223. Due to this deletion, translation would result in a truncated 232-amino acid polypeptide, whereas a 663-amino acid polypeptide would be synthesized with no frame shift (Fig. 1).

Analysis of the amino acid sequence identified a hydrophobic signal sequence (residues 1–19). A full-length mature polypeptide would consist of 644 amino acids, with a predicted molecular mass of 72 kDa. Close to the N-terminus, EMR4 possesses two EGF domains (residues 31–69 and 70–120), the second one containing the consensus sequence for a class I calcium-binding site 21. The EGF domain region is followed, subsequently, by a stalk region (residues 121–339), a class B TM7 region (residues 340–586) and a cytoplasmic region (residues 587–663). The presence of a GPS immediately upstream to the TM7 region indicates that EMR4, like other EGF-TM7 receptors 4, 13, 14, 17, would be proteolytically processed into an extracellular α chain and a TM7/cytoplasmic β chain, which reassociate prior to expression at the cell surface. The predicted overall structure of EMR4 (Fig. 2A) is similar to other members of the EGF-TM7 family 1, 2. Pairwise comparison revealed an amino acid sequence identity of 45% with EMR3 11, 38% with EMR2 10, 34% with EMR1 7, 29% with CD97 3, 4 and 26% with ETL 13. The amino acid sequence identity between human EMR4 and its murine homologue 12, 13 is 59%.

To verify the effect of the frameshift mutation on the translation of EMR4, several fusion proteins were generated in which the sequence encoding the extracellular region of EMR4 was ligated in-frame upstream to the CH3-CH2-hinge region sequence of mouse IgG2b. Analysis of the resulting translation products demonstrated that synthesis of EMR4-mouse Fc (mFc) fusion proteins was abrogated in those constructs that possessed the one-nt deletion, but not in a construct that only contained the amino acid sequence upstream from the deletion (Fig. 2B, C).

By integration of genomic sequence with the physical map of chromosome 19, we assigned EMR4 to 19p13.3 just proximal to EMR1 (Fig. 3). The two genes are transcribed from opposite strands in a tail-to-tail arrangement, with the 3′-termini about 12 kb apart. The organization of the EMR4 gene, consisting of 16 exons (Fig. 1), is highly similar to that of other EGF-TM7 receptor genes 10, 22.

2.2 Transcription of EMR4 in dendritic cells

Characterization of the murine homologue of EMR4 revealed expression on monocytes, macrophages and dendritic cells 12, 13. On DC, mouse EMR4, like the murine homologue of EMR1, F4/80, is mainly expressed by the CD8^– cells, which are considered to represent myeloid DC 23. To investigate whether transcriptional regulation has been conserved in human EMR4, we examined transcription of EGF-TM7 receptors in human blood DC subsets, defined by the expression of blood DC antigen (BDCA) markers 24. Whereas BDCA-1 and BDCA-3 are expressed on myeloid DC, BDCA-4 is expressed on plasmacytoid DC that are likely of lymphoid origin. As delineated in Fig. 4, three transcription patterns were found. First, EMR3 and ETL are not detectable in DC. Second, CD97 and EMR2 are present in all three subsets. Third, EMR1 and EMR4 are transcribed in BDCA-1⁺ cells and at very low levels in BDCA-3⁺ cells, but not in BDCA-4⁺ cells. Although comparison of human and murine DC is hampered by the different expression of phenotypic markers, like CD8 and CD4 25, these data suggest that expression of EMR4 and EMR1, in both species, is restricted to myeloid DC. Possibly, the remarkable similarity in the presence of EMR4 and EMR1 in DC is due to coordinated transcriptional regulation of the very closely located genes.

2.3 Analysis of the one-nucleotide deletion in human and nonhuman primates

The completely preserved exon-intron organization and regulated transcription in DC suggest that the frameshift mutation in exon 8 of EMR4 manifested during recent evolution. To investigate this, we examined the DNA sequence of this region from various human and nonhuman primates. No polymorphism at the one-nt deletion site was detected in either, Caucasians or sub-Saharan Africans (data not shown, for details see Sect. 3). However, as shown in Fig. 5, an additional nt was found in all primates studied, including chimpanzees. Although the presence of inactivating mutations elsewhere in the primate EMR4 genes cannot be excluded, this indicates that EMR4 lost its function only after human speciation that occurred approximately 5 million years ago 26. It makes EMR4 one of the very few genes 27–29 and, to our knowledge, the only one encoding an individual molecule involved in immunity that, on current evidence, are nonfunctional in humans but still might be active in chimpanzees. The only biochemically characterized gene that became inactive after divergence from chimpanzee is the cytidine monophosphate-N-acetylneuraminic acid (CMP-Neu5Ac) hydroxylase gene (CMAH), which contains a 92-nt deletion in humans 30. Consequently, the great apes, but not humans, express high levels of the sialic acid derivative N-glycolylneuraminic acid (Neu5Gc). Another example, the C-type lectin gene Ly49L, became nonfunctional due to a splice donor site mutation after divergence of the common ancestor of chimpanzee and humans from gorilla, approximately 8 million years ago 31. The function of Ly49L is still elusive.

2.4 Concluding remarks

Taken together, we here provide evidence that the EGF-TM7 receptor gene EMR4 became inactive in humans due to a frameshift mutation. Whether a truncated, soluble form of EMR4 is secreted by human leukocytes remains to be demonstrated with specific antibodies. The most interesting questions, actually, deal with the identification of the ligand and the function of EMR4. Binding of multivalent, soluble mouse EMR4 probes to the B lymphoma cell line A20 suggests that EMR4, like other EGF-TM7 receptors, mediates cellular interactions 13. Unravelling the physiological role of EMR4 might provide the unique opportunity to study a genetic difference between our species and our evolutionary closest relatives, a challenge that is even more intriguing since humans and chimpanzees differ substantially in their susceptibility to diseases like malaria, epithelial cancers, Alzheimer's disease and AIDS 27, 29.

3.1 cDNA cloning and sequence analysis

A survey of the DDBJ/EMBL/GenBank databases for novel EGF-TM7 receptor genes identified several exons of the thus far unknown family member EMR4 in a working draft sequence of human chromosome 19 (accession no. AC020895). Based on this sequence, primers were designed and used to derive the complete EMR4 cDNA sequence by a combination of RT-PCR from human PBMC first-strand cDNA and 5′- and 3′-rapid amplification of cDNA ends (RACE) from human spleen Marathon-Ready cDNA (Clontech, Palo Alto, CA). Sequences were determined with the BigDye terminator cycle sequencing kit (Applied Biosystems, Warrington, GB). Amino acid sequences were compared with the ClustalW software. Protein architecture was analyzed using the SMART program 32.

3.2 Generation, expression and detection of mFc fusion proteins

To generate mFc fusion constructs 13, the N-terminal part of EMR4 was amplified by RT-PCR from human PBMC first-strand cDNA using the specific primers 5′-CTGACCACGTCAAGCTTCAGAGAG-3′ [nt 2–25 of the (+) strand, an introduced HindIII site is underlined], 5′-GAAATTGGATCCTTTCCTGGAGAAG-3′ [nt 551–575 of the (-) strand, an introduced BamHI site is underlined], 5′-GGATGCATTCAGAGGATCCCCAAGAG-3′ [nt 730–755 of the (-) strand, an introduced BamHI site is underlined], and 5′-CAGCACAGGATCCTCCTTGGGGGC-3′ [nt 1038–1061 of the (-) strand, an introduced BamHI site is underlined]. PCR products were ligated in-frame immediately upstream to the CH3-CH2-hinge region sequence of mouse IgG2b, linked to a C-terminal biotinylation sequence, in pcDNA3.1/ Neo(+) (kindly provided by M. Stacey, Sir William Dunn School of Pathology, Oxford). COS cells were transfected with the expression constructs and after 1 day, medium was replaced by serum-free Opti-MEM. Five days later, culture supernatant was collected and 5 μl of each transfection were analyzed by immunoblot for soluble EMR4-mFc protein using HRP-labeled anti-mouse Ig (DAKO, Glostrup, Denmark).

3.3 Chromosomal localization

EMR4 was identified in genomic sequence (accession number AC020895) as described. Coding sequences for previously identified EGF-TM7 receptor genes were localized in chromosome 19 genomic sequence by blast analysis with the corresponding cDNA sequences. The cosmids and BAC from which genomic sequence was derived had been previously localized on the chromosome 19 map by a combination of restriction mapping and high-resolution pronuclear FISH (33, 34; for updated map see http://bbrp.llnl.gov/ bbrp/genome/html/chrom_map.html). Mouse chromosomal locations for regions corresponding to the human EGF-TM7 receptor gene locations on chromosome 19 were derived from the mouse-human comparative map constructed by Kim et al. 35 and the location of homologous reference genes in mouse genomic sequence 36.

3.4 RT-PCR on human DC subsets

Blood DC of the myeloid and plasmacytoid lineage were isolated from human PBMC according to the protocol of Dzionek et al. 24. In short, after depletion of CD19⁺ B cells using anti-CD19 microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany), cells were labeled with PE-conjugated mAb directed against BDCA-1, BDCA-3 or BDCA-4 for 15 min at 4°C in FcR-blocking reagents (Miltenyi Biotec). Next, anti-PE microbeads were added and incubated for another 15 min at 4°C. BDCA-1⁺, BDCA-3⁺ and BDCA-4⁺ cells were subsequently isolated using immunomagnetic cell sorting (MACS, Miltenyi Biotec). To obtain >98% pure cell populations, the enriched fractions were further purified by FACS sorting using a FACSVantage (Becton Dickinson, San Jose, CA). From the sorted cells, first-strand cDNA was prepared and amplified by PCR (35–40 cycles, 30 s at 93°C, 30 s at 58°C, 40 s at 72°C) using primer combinations depicted in Table 1. PCR products were separated on a 1% agarose gel.

3.5 Sequencing of exon 8 in human and nonhuman primates

A 182-bp fragment from exon 8 was amplified by PCR (35 cycles, 30 s at 94°C, 30 s at 58°C, 30 s at 72°C) with the specific primers 5′-TTTCTTCAGAGAGCACTGCAG-3′ [nt 698–709 of the (+) strand, nt 1–9 belong to the preceding splice acceptor] and 5′-GGCGAAAAGTCAGGAACACAGG–3′ [nt 852–873 of the (-) strand]. As template, genomic DNA was used from Caucasians (n=8, donors recruited from our laboratory, kindly provided by E. Remmerswaal), sub-Saharan Africans (n=9, Coriell Institute for Medical Research, Camden, NJ) and different nonhuman primates including chimpanzee (Pan troglodytes, n=3), orangutan (Pongo pygmaeus, n=2), rhesus macaque (Macaca mulatta, n=3) and crab-eating macaque (Macaca fascicularis, n=3) (Biomedical Primate Research Center, Rijswijk, The Netherlands). PCR products were sequenced as described.