LY6S, a New IFN-Inducible Human Member of the Ly6a Subfamily Expressed by Spleen Cells and Associated with Inflammation and Viral Resistance

The human genome contains at least 48 Lymphocyte Ag 6 (LY6) genes and the mouse genome at least 61 Ly6 genes, all coding for three-fingered proteins (TFPs) that contain a signature pattern of 10 cysteine residues, resulting in a distinctive disulfide bonding pattern. They contain a domain found in the urokinase-type plasminogen activator receptor protein (uPAR, encoded by PLAUR) and are also referred to as LY6/uPAR or LU proteins. These proteins locate extracellularly either linked to the cell membrane via a GPI anchor, as transmembrane proteins, or as fully secreted proteins as seen, for example, in the PATE proteins (1). The transmembrane proteins include the receptors for proteins belonging to the TGF-β family, such as BMP2 and activin (2). Only a few of the TFP/LY6/uPAR proteins have known functions, with the most well characterized being CD59 (3, 4), uPAR/PLAUR (5), and GPIHBP1 (6). A common theme for these proteins is that the LY6 domain mediates interactions with other proteins. The variety of functions both known and postulated for the LY6 family proteins suggests that nature has usurped the skeleton scaffold of the TFP fold to execute a large number of diverse activities (7–10).

Mouse chromosome 15 (mChr15) harbors a locus replete with 23 Ly6 family genes (Fig. 1a), most of which code for Ly6 proteins linked to the cell surface by a GPI anchor. One of the most widely investigated Ly6 genes is Ly6a, which is located within the chromosome 15 locus (11) and codes for an IFN-inducible GPI-linked protein, also known as T cell–Activating Protein and stem cell Ag-1. Ly6a is upregulated by IFNs and is implicated in the activation of T cells, as well as in inflammatory and immune-related processes. Similarly, the murine genes Ly6c and Ly6g, which genomically cocluster with Ly6a, are used extensively to analyze cells of the myeloid lineage, especially of the mouse spleen (e.g., Refs. 12–14), and Ly6c expression is used as a marker for murine blood monocytes (15, 16). Furthermore, the murine Ly6 genes Ly6a, Ly6c, and Ly6g are inducible by IFNs and code for proteins that are markedly upregulated in inflammatory processes and immune-related diseases (17–19). Despite the hundreds of publications on each of these three murine genes, many focusing on their role in inflammation, their precise functions have yet to be elucidated. It is all the more so surprising that no human homolog(s) to this murine Ly6a, Ly6c, and Ly6g gene subfamily has been previously identified.

We report in this article the identification of a new human LY6 gene designated LY6S by the HUGO Gene Nomenclature Committee (20) that codes for a LY6 protein that phylogenetic analyses indicate has an orthologous relationship with the murine Ly6a subfamily cluster. Human LY6S codes for a membrane-linked protein that is most highly expressed in the human spleen by a nonclassical human spleen cell and is associated with the regulation of cell growth, inflammatory and immune-related responses, and resistance to infection by certain viruses.

All chemicals and reagents were obtained from Sigma (St. Louis, MO), unless otherwise specified. Secondary Abs used in cell counterstaining or in immunohistochemical development were obtained from Jackson ImmunoResearch Laboratories (Bar Harbor, ME).

Mice were immunized with five consecutive intradermal immunizations spaced at 21-d intervals, with peptide E, derived from the amino acid sequence of LY6S-iso1 (see Fig. 6a), covalently conjugated to Keyhole Limpet Hemocyanin. The mice were purchased from Harlan Laboratories Limited (Israel) and were housed and maintained in laminar flow cabinets under specific pathogen-free conditions in the animal quarters of Tel Aviv University. All work was performed in accordance with current regulations and standards of the Tel Aviv University Institutional Animal Care and Use Committee. The initial immunization was done with Freund's adjuvant, and the following four peptide E covalently conjugated to Keyhole Limpet Hemocyanin boost immunizations together with IFA. Hybridomas were prepared by fusion of nonsecreting myeloma cells with immune splenocytes and screened by ELISA assay (see later).

ELISA Immunoassay plates (CoStar) were coated with recombinant hFc-LY6-iso1 protein, produced and secreted by HEK 293 cells stably transfected with and secreting hFc-LY6S-iso1 protein (see later Expression of hFc-LY6S fusion protein section). Spent culture media from the initial hybridomas were then applied to the wells. After incubation, samples were removed and the wells were washed with PBS/Tween. Detection of bound Abs was done with HRP-conjugated anti-mouse Ab.

The hFc-LY6S fusion protein was generated by inserting DNA coding for this protein into the pcDNA3.1 expression vector using standard molecular biology techniques. The makeup of this insert, the amino acid sequence of the resultant fusion protein and the DNA sequence of the insert, are shown in Supplemental Fig. 1.

RNA was isolated from the SCA5E12 hybridoma (anti–LY6-iso1 mAb) with TRIzol Reagent l, according to a technical manual for the reagent (Ambion, Foster City, CA). The sequence of the SCA5E12 Ab was determined as follows: cDNA was generated by reverse transcription of total RNA with the use of universal or isotype-specific antisense primers, according to the technical manual for PrimeScript First Strand cDNA Synthesis Kit (Takara Bio, Mountain View, CA). Amplification of V_H and V_L Ab fragments was carried out according to the standard operating procedure (GenScript), which involves rapid amplification of cDNA ends, followed by separate cloning into a standard cloning vector. Clones with inserts of the correct sizes were sequenced by colony PCR, and at least five colonies with such inserts were sequenced for each fragment, with the consensus sequence derived by alignment of the different clones.

Chromogenic immunolabeling for SCA5E12 was performed on formalin‐fixed, paraffin-embedded human tissue sections. Briefly, after dewaxing and rehydration, slides were immersed in 1% Tween 20, then heat‐induced Ag retrieval was performed in a preheated steamer using Ag Unmasking Buffer (catalog #H-3300; Vector Labs) for 25 min. Slides were rinsed in PBST and endogenous peroxidase, phosphatase was blocked (catalog #S2003; Dako), and sections were then incubated with primary SCA5E12 mouse mAb (1:50 dilution) for 45 min at room temperature. For staining in the presence of competing peptides, primary Ab was diluted 1:10 in the peptide solution directly. The primary Abs were detected by 30-min incubation with HRP-labeled anti-mouse secondary Ab (catalog #PV6114; Leica Microsystems) followed by detection with 3,3′‐diaminobenzidine (catalog #D4293; Sigma‐Aldrich), counterstaining with Mayer’s hematoxylin, dehydration, and mounting.

Dual OPAL immunofluorescent labeling with SCA5E12 and anti-CD45 Abs (catalog #145M-94; Cell Marque/Sigma-Aldrich) was performed on formalin‐fixed, paraffin-embedded tissue sections following the manufacturer’s instructions (NEL810001KT; Akoya Biosciences, Menlo Park, CA). Briefly, after standard dewaxing and rehydration, slides were immersed in Ag Unmasking Buffer (H-3300 Vector Laboratories; Burlingame, CA) in a plastic chamber and retrieved in a microwave, which is also the subsequent Ab stripping step for sequential multiplex staining. Endogenous peroxidase and phosphatase were blocked (catalog #S2003; Dako), and sections were then incubated sequentially with each primary Ab (1:80 dilution for SCA5E12 and specific dilutions for the other primaries as listed later) for 45 min at room temperature with Ab stripping step performed in between. Slides were incubated with differently labeled anti-Mouse IgG (PV6114; Leica, Wetzlar, Germany) or anti-rabbit IgG (PV6119; Leica) secondary Abs, as appropriate, for 30 min. Fluorescent dyes OPAL 520 (for each of the Abs listed later) and OPAL 690 (for SCA5E12) were diluted in OPAL amplification buffer, and slides were stained for 10 min. Slides were counterstained with DAPI working solution for 10 min, washed, and mounted with ProLong Gold.

The specific primary Abs used for dual immunofluorescence with the SCA5E12 Ab were anti-CD4 (ab133616, 1:200, rabbit; Abcam), anti-CD8 (m0814, 1:100, mouse; Dako), anti-CD11b (ab133357, 1:8000, rabbit; Abcam), anti-CD11c (ab52632, 1:100, rabbit; Abcam), anti-CD19 (catalog #119M-14, 1:50, mouse; Cell Marque/Sigma-Aldrich), anti-CD20 (0755, 1:100, mouse; Dako), anti-CD31 (RB-10333-P, 1:50, rabbit; Thermo Fisher Scientific), anti-CD34 (catalog #134M-14, 1:75, mouse; Cell Marque/Sigma-Aldrich), anti-CD45 (catalog #145M-94, 1:100, mouse; Cell Marque/Sigma-Aldrich), anti-CD68 (m0814, 1:250, mouse; Dako), anti-CD117 (c-Kit; A450229-2, 1:400, rabbit; Dako), Pan-cytokeratin AE1/AE3 (ab961, 1:10, mouse; Abcam), and anti-FOXP3 (12653S, 1:50, rabbit; Cell Signaling Technologies).

Samples used for immunohistochemical stainings were procured from the Biomax tissue bank, with all required ethical approvals therein, as stated at http://biomax.us: “All tissue is collected under the highest ethical standards with the donor being informed completely and with their consent. We make sure we follow standard medical care and protect the donors' privacy. All human tissues are collected under HIPPA approved protocols.”

Total RNA was extracted from human melanoma M12-CB2 cells that were treated with IFN-γ for 48 h using the EZ-RNA Total RNA Isolation Kit (Biological Industries, Kibbutz Beit-Haemek, Israel), followed by cDNA synthesis with qScript cDNA Synthesis Kit (Quantabio). The overexpression construct of human LY6 was created by PCR amplification of cDNA by Phusion High-Fidelity DNA polymerase (Thermo Fisher Scientific), using the following primers LY-6S-Pac1-S-5′-GGAACGTTAATTAACGCTGAAGTTTGTCTGTGCACTA-3′, LY6S-EcoR1-AS-5′-AGAGAGGAATTCCCAAGCAGCTGGTGACGCACAG-3′. The generated fragment was digested with Pac1 and Ecor1 and ligated into the corresponding sites of pQCXIP vector (Clontech Laboratories, Mountain View, CA). PCR products of LY6S were sequenced. Cells were infected to stably express LY6S-pQCXIP retroviral vector (Clontech Laboratories) as described previously (21).

M12, YDFR, or U87 cells were infected with vesicular stomatitis virus (VSV)-New Jersey (a gift of Prof. M. Kotler, Hebrew University) at multiplicities of infection of 0.5, 0.15, or 0.05 for 24 or 48 h. Titer of inoculum or of spent medium of infected cells was measured by plaque assay on Vero cells: sequential 10-fold dilutions in DMEM, and 125,000 cells per well of a 12-well plate. Overlay was with 0.6% methylcellulose. Detection was by crystal violet staining.

Ag expression was determined using Flow cytometer S100EXi (Stratedigm, San Jose, CA) with CellCapTure software and FlowJo v10. Dead cells were gated out from the analysis.

Sequences were aligned by using the Muscle Multiple Sequence alignment software (PMID: 30976793) available through an online server at the EBI (https://www.ebi.ac.uk/Tools/msa/muscle/). AliView (PMID: 25095880) was used to view and edit the initial alignment of 303 columns. The alignment was edited to remove indel regions of ambiguous alignment, and columns with <20% gaps were retained, providing a final alignment of 113 aa.

ProtTest-3.4.2 (22) was used to determine the best fit evolutionary model for this maximum likelihood analysis. The RAxML-NG (PMID: 31070718) blackbox server (https://raxml-ng.vital-it.ch/#/) was used to perform a maximum likelihood analysis (JTT+G4m substitution model, stationary base frequencies taken from the model, with the proportion of invariant sites box selected, and four γ rate categories to model the among-site rate heterogeneity), with automatic bootstopping of the bootstrapping procedure, as implemented in RAxML. Trees were visualized and edited with FigTree v1.4.3 software (http://tree.bio.ed.ac.uk/software/figtree/).

Libraries were prepared by the Crown Institute for Genomics (G-INCPM, Weizmann Institute of Science) using an in-house poly(A)-based RNA sequencing (RNA-seq) protocol (INCPM-mRNA-seq). Briefly, the poly(A) fraction (mRNA) was purified from 500 ng of total input RNA followed by fragmentation and the generation of double-stranded cDNA. After Agencourt Ampure XP beads cleanup (Beckman Coulter), end repair, A base addition, adapter ligation, and PCR amplification steps were performed. Libraries were quantified by Qubit (Thermo Fisher Scientific) and TapeStation (Agilent). Sequencing was done on a Nextseq 75 cycles high-output kit, allocating 20M reads per sample (Illumina; single-read sequencing). The RNA-seq data, GEO accession number GSE188924, have been uploaded to the GEO database. This SuperSeries record provides access to all of the data relating to both the control human cells (U87 glioblastoma, M12 melanoma, and YDFR melanoma cells) and their counterparts expressing the LY6S-iso1 protein. To review, it can be accessed at: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE188924. The Paired End (PE) True Stranded RNA-seq data, GEO accession number GSE159456, have also been uploaded to the GEO database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE159456).

Poly-A/T stretches and Illumina adapters were trimmed from the reads by using cutadapt; reads <30 bp were discarded. Reads for each sample were aligned independently to human reference genome GRCh38 with STAR (23) and supplied with gene annotations downloaded from Ensembl (and with the EndToEnd option). Counting proceeded over genes annotated in Ensembl release 92 using the htseq-count (with stranded = “reverse” option for the PE) (24). Only uniquely mapped reads were used to determine the number of reads falling into each gene (intersection-strict mode). Differential analysis was performed with the DESeq2 package (25), with the betaPrior, cooksCutoff, and independentFiltering parameters set to False. Raw p values were adjusted for multiple testing by using the Benjamini and Hochberg procedure. Differentially expressed genes were determined by an adjusted p value of <0.05, absolute fold changes > 2, and max raw counts > 30. Functional analysis of DE genes was performed with Ingenuity Pathway Analysis (IPA; QIAGEN; https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis), together with Martin’s (26) method and also methods detailed elsewhere (23–25, 27).

Murine Ly6a resides on mChr15 in a 900-kb genomic locus replete with other genes coding for Ly6 family proteins (Fig. 1a). The syntenic human locus on chromosome 8 is also rich in LY6 genes, yet at ∼540 kb is considerably smaller than the corresponding mouse locus and contains far fewer LY6 family genes (Fig. 1b). The murine subregion containing Ly6a is demarcated by the Ly6e and Ly6l genes, spans ∼500 kb, and contains eight Ly6 genes. In contrast, the human LY6E and LY6L genes (Fig. 1b) span a region of ∼60 kb within which no annotated LY6 genes reside.

A phylogenetic tree analysis of the human chromosome 8 (hChr8) and mChr5 LY6 loci (Fig. 1e) shows that the eight Ly6 family genes located within the 500-kb subregion group together in a clade with high bootstrap support and are more distantly related to the Ly6 genes located outside of the Ly6e to Ly6l locus. Cumulatively, this indicates that the eight genes within the Ly6e to Ly6l locus form a Ly6 family subgroup, referred to in this study as the Ly6a subfamily. In contrast with the murine Ly6a subfamily locus, the corresponding syntenic human region also flanked by LY6E and LY6L is only ∼60 kb (Fig. 1b) and contains no annotated human LY6 genes.

Because the mouse Ly6a gene lies downstream of Ly6e, within the genomic neighborhood demarcated by the Ly6e and Ly6l genes, and because LY6L is predicted to be the human ortholog of Ly6l by multiple orthology resources (see the HUGO Gene Nomenclature Committee HCOP tool: https://www.genenames.org/data/gene-symbol-report/#!/hgnc_id/HGNC:52284), we expected that if a human ortholog of Ly6a were to exist, then it too should reside within the human syntenic segment bordered by LY6E and LY6L, which spans only ∼60 kb (see (Fig. 1b). No annotated human LY6 genes appear in this region. Yet, GenScan predicts a gene (chr8_73.103, here designated as LY6S; (Fig. 1c, and see later) that codes for an LY6 family protein that shows a high level of sequence homology to the mouse Ly6a protein. Furthermore, the TransMap cross-species alignment algorithm (28, 29) that maps genes and related annotations in one species versus another by using synteny-filtered pairwise genome alignments to determine the most likely ortholog, maps Ly6a at the same position as the GenScan predicted human LY6 gene.

Neither expressed sequence tags nor any other experimental evidence supports the GenScan chr8_73.103 (LY6S) gene. Notably, transcription of the LY6S gene (Fig. 1c) is in the opposite direction to that of a known gene, C8orf31, that is transcribed from the complementary strand and has extensive exon overlap with exons from the human LY6S gene (Fig. 2a, 2b). Oligonucleotide primers were identified that are specific solely for LY6S and do not overlap with C8orf31 (see (Fig. 2c and Supplemental Table I). RT-PCR analyses with these primers revealed a particularly prominent RT-PCR product in the human spleen (Fig. 2di, 2dii), which showed the highest expression (Fig. 2di, 2dii, 2e). Of all fetal tissues, spleen was also the highest LY6S expresser (inset, (Fig. 2e). In contrast with its high expression in the lymphoid-rich spleen, LY6S expression was not detected in peripheral blood leukocytes or in bone marrow (Fig. 2di, compare lanes 1 and 8 with lane 18 [bone marrow, leukocytes and spleen, respectively], (Fig. 2e). In addition to spleen, LY6S expression was detected in brain tissue (Fig. 2di, lane 2, and at higher contrast in (Fig. 2di′, lane 2). In line with these results, increasing the number of PCR cycles to 40 showed, as expected, highest expression in spleen followed by testis (Fig. 2h, lanes 14 and 16, respectively). Significant expression was also detected in the parietal lobe of the brain and in the spinal cord, and lower levels in the thalamus and temporal lobe (Fig. 2h, lanes 8, 13, 17, and 15, respectively). Use of a different set of primers, the F8 and R12 forward and reverse primers (Supplemental Table I), confirmed these results, and highest expression was again observed in the spleen, with significant expression observed in the parietal lobe and corpus callosum, and with lower levels in the cerebral cortex and temporal lobe (Fig. 2i, lanes 1, 8, 9, 4, and 7, respectively). With both sets of primers, the F7 and R10 and F8 and R12 pairs, the sizes of the observed RT-PCR products correlated precisely to those expected for expression of LY6S (Fig. 2di, 2di′, 2dii, 2h for the F7 and R10 primer pair; (Fig. 2f, 2g [see later], 2h–j for the F8 and R12 primer pair).

Whereas RT-PCR analyses revealed significant LY6S expression in human spleen tissue, similar analyses performed on a number of different human cell lines failed to detect LY6S expression. Ly6a, as well as other mouse Ly6 genes, such as Ly6c1, are highly induced by IFNs, and human LY6E located upstream to LY6S is induced by both type I (α/β) and type II (γ) IFNs. We postulated that LY6S may also be IFN inducible. A perfect IFN-stimulated response element positioned ∼500 bp upstream to the LY6S gene appears as TTCCTGTGAAATGGAAATTCAGGA (underlined sequence is identical to the IFN-stimulated response element conferring IFN inducibility on one of the human IFN-α genes) and conforms to the tandem GAAANNGAAA element that appears in most type I IFN-stimulated genes (ISGs). This stretch of 24 nt also contains the palindromic consensus sequence TTCN_2–4GAA (TTCCTGTGAA) that conforms to the consensus IFN-γ activation site. We thus assessed LY6S expression in different human cell lines, either untreated or exposed to IFN-α, IFN-β, or IFN-γ, and compared LY6S expression with that of LY6E. Because of the number of PCR cycles used in this experiment (36 cycles), LY6E expression was observed at high levels in all cell lines irrespective of IFN treatment (Fig. 2f, upper panel). In the absence of cytokine treatment, none of the human cell lines expressed significant levels of LY6S, as shown for the M12CB3 and YCB3 melanoma cells (Fig. 2g, lane 1, and (Fig. 2ji, 2jii, lane 1) and for the MCF7 breast cancer cells (Fig. 2jiii, lane 1). Treatment with IFN-β induced marked LY6S expression in several cell lines, including U87 glioblastoma cells (Fig. 2f, lanes 8 and 12), MCF7 breast cancer cells (Fig. 2f, lane 18, and (Fig. 2jiii, lane 2), and HEY ovarian cancer cells (Fig. 2f, lane 17). IFN treatment induced LY6S expression in M12CB3 melanoma cells (Fig. 2g, lanes 2–4, for IFN-α, IFN-β, and IFN-γ, respectively, and (Fig. 2ji, lane 2) and in YCB3 melanoma cells (Fig. 2jii, lane 2). Notably, LY6S isoforms 1 and 2 (see later for description of isoforms 1 and 2) were induced to varying extents by the different IFNs (Fig. 2g, compare lanes 2–4).

TOPO cloning and nucleotide sequencing of the LY6S RT-PCR product(s) obtained with human spleen cDNA (see (Fig. 2di, lane 18) revealed three inferred protein isoforms (see (Fig. 3 and Supplemental Fig. 2 for detailed information on LY6S-iso1 and iso2) corresponding to (1) a LY6S protein, which received a perfect GPI score from the GPI anchor predictor tool PredGPI, indicating that it is a GPI-linked membrane-bound protein (isoform 1, designated LY6S-iso1, in three of eight sequenced TOPO clones; (Fig. 3a, isoform 1); (2) a C-terminally truncated LY6S protein, likely to be secreted from the cell that retains both exon 1 (signal peptide [SP]) and exon 2 coding sequences but uses an alternative splice donor site toward the 3′ end of exon 2 (isoform 2, designated LY6S-iso2, for four of eight sequenced TOPO clones; (Fig. 3b, isoform 2); and (3) a protein that retains the exon 1 SP, but by use of an alternative splice acceptor site produces a frameshifted protein C-terminal to the SP, which is also likely to be secreted from the cell (isoform 3, for one of eight sequenced TOPO clones; (Fig. 3c). All the isoforms conformed to consensus splice donor and splice acceptor sites (“gt” and “ag” and their flanking sequences; (Fig. 3a). Providing additional confirmation, TOPO cloning and sequencing of the RT-PCR products obtained from the IFN-treated human melanoma cells (see (Fig. 2g) revealed the same LY6S isoforms to those observed in spleen.

Although belonging to the LY6/uPAR family of proteins, which normally contain 10 consensus cysteine residues, the LY6S-iso1 protein is special in that it contains a noncanonical serine residue in place of the sixth cysteine residue present in all other LY6-like proteins, hence the “S” in the LY6S designation. The serine codon was observed in cDNAs generated from both spleen and IFN-treated melanoma cells, and all genomic databases investigated show a codon at this position in the LY6S gene, in the genomic DNA, coding for a serine residue.

Because the LY6S gene and LY6S mRNAs described in this article are novel, it was important to find independent supportive evidence for their presence from freely available databases. Analyses of public domain human spleen RNA-seq data showed transcripts in the human spleen that are derived from the LY6S gene (Fig. 4, Supplemental Fig. 3), providing unambiguous validation for the existence of mRNAs coding for the LY6S-iso1 and LY6S-iso2 proteins.

In contrast with the sole human LY6S gene present in the region bordered by LY6E and LY6L, the syntenic mouse locus within which Ly6a resides contains eight Ly6 genes, all of which are highly similar to each other (Figs. 1a, 1e, 3d). These murine Ly6 genes demonstrate coordinated transcriptional responses to certain inflammatory cytokines and to those relating to IFN-mediated immune sensing (17, 18).

The eight Ly6 family genes are defined as belonging to the Ly6a subfamily (Figs. 1e, 3d). Both phylogenetic analyses (Fig. 1e) and protein similarity analyses (Fig. 3d) indicate that of all the mChr15-Ly6 proteins, the human LY6S-iso1 protein is most similar to the proteins encoded by the Ly6a subfamily. All the murine Ly6s-subfamily genes code for proteins with the amino acid sequence “ERAQGL” at the C-terminal regions of their SP, a sequence also conserved in the human LY6S protein (boxed red rectangles in Figs. 3, 5a), yet not present in any of the other hChr8 LY6 proteins.

The constructed phylogenetic maximum likelihood tree (Fig. 1e) shows LY6S branching at the base of the Ly6a subfamily clade containing the eight Ly6 mouse genes located within the 500-kb subregion. A bootstrap value of 89% shows that the phylogenetic signal from the whole alignment lends strong support to this grouping. This tree supports the notion that LY6S has evolved from the same ancestral gene as the Ly6a subfamily clade, and that there have been several duplication events in the mouse. The similarity of LY6S to the eight Ly6a subfamily genes, as well as their syntenic colocalization (both flanked by LY6E/Ly6e and LY6L/Ly6l), indicates that the eight Ly6a subfamily genes are homologous to the solitary human LY6S gene, and this is a 1:8 orthology relationship (Figs. 1a, 1e, 3d). Comparison with other well-annotated vertebrate genomes suggests that the most parsimonious explanation is that there has been a gene expansion of Ly6 genes in mouse, rather than one or more deletion events in other species.

A protein BLAST search with LY6S-iso1 yielded as the best hits the lymphocyte Ag Ly6a (Ly6A-2/Ly6E-1) proteins from Nannospalax galili (Upper Galilee Mountain blind mole rat), Microcebus murinus (gray mouse lemur), Merionus unguiculatus (Mongolian gerbil), and Peromyscus maniculatus bairdii (North American deer mouse) with e values of 1e−38, 3e−30, 2e−28, and 7e−28, respectively (see (Fig. 5a for a comparison of LY6S-iso1 with the Nannospalax and Peromyscus Ly6a [Ly6A-2/Ly6E-1] proteins). Subsequent pairwise comparison of each human LY6 protein present on chromosome 8 with the deer mouse Ly6A-2/6E-1–like protein showed that LY6S gave the best match by far (Fig. 5c–e).

To facilitate investigation into LY6S expression at the protein level, we generated LY6S-iso1–specific mAbs designated 5E12. The specificity of 5E12 was confirmed by the following results. First, 5E12 bound to recombinant hFc-LY6S fusion protein (Fig. 6b, lane 6), and binding was blocked by the immunizing LY6S peptide E and a peptide contained within this sequence (Fig. 6b, lanes 7 and 9, respectively; see (Fig. 6a for peptide sequences within the LY6S-iso1 protein), but not by peptide C, which is upstream of the immunizing peptide (Fig. 6b, lane 8), Second, 5E12 bound to a protein expected for the size of the LY6S-iso1 protein in Western blot analyses of HK293 cells infected with lentiviruses coding for the LY6S-iso1 protein (Fig. 6c, left panel, lanes 1 and 2), but not from cells infected with control lentiviruses (Fig. 6c, left panel, lane 3). Furthermore, LY6S-iso1 protein observed in cells infected with retroviral particles coding for the native LY6S-iso1 protein migrated slightly faster than the corresponding LY6S protein that contained a Flag-epitope at its N terminus, conforming to their difference in molecular mass (compare in (Fig. 6c, lane 1 [with the Flag epitope] and lane 2 [LY6S protein without the Flag epitope]). Third, by flow cytometry, 5E12 detected cell-surface LY6S-iso1 protein, albeit at low levels, on intact HK293 cells infected with retroviral particles coding for LY6S-iso1, as expected for a GPI-linked protein (Fig. 6dii, 6diii). Fourth, anti-LY6S mAb 5E12 immunofluorescent analysis of HK cells transiently transfected with plasmids coding for the LY6S protein showed positively staining cells (Fig. 6eii, 6diii), which were not seen with cells transfected with control empty plasmid (Fig. 6ei). The reproducible detection of expression at the cell surface at low levels (Fig. 6dii, 6diii) suggests that cells do not readily express the LY6S protein at the cell surface (and see later).

Human M12 melanoma cells treated with IL1-β (Fig. 6fi′), as well as with IFN-γ and IL-6 (Fig. 6fi′′, i′′′), demonstrated cell-surface LY6S protein expression. Similarly, U87 glioma cells treated with IFN-β showed cell-surface LY6S protein (Fig. 6fii′), as did cells treated with IL1-β (Fig. 6fii′′). Thus, cytokine treatment of cells reproducibly induces the expression of cell-surface LY6S in both M12 melanoma and U87 glioma cells. Such expression is not detected in the absence of cytokine treatment (Fig. 6fi, 6fii).

Immunofluorescent staining of human spleen tissue showed a discrete subpopulation of LY6S-iso1–positive cells that accounted for ∼5% of all spleen cells (Fig. 6gi). The immunizing peptide E abrogated staining (Fig. 6giii) confirming Ab specificity, whereas peptide C, which was not used for immunization (Fig. 6gii), had no effect. LY6S-iso1 localized to the cell surface (Fig. 6giv′) but was also observed within the cell (Fig. 6giv). LY6S-iso1 protein expression was barely detectable in other lymphoid tissues, such as thymus and tonsils, as assessed by immunohistochemical staining (Fig. 6hii, 6hiii), yet was seen clearly in spleen cells (Fig. 6hi) stained under identical conditions. Spleen tissue costained for LY6S-iso1 together with Abs directed against additional cell-surface proteins that define particular cell types detected a prominent subpopulation of LY6S-positive cells constituting ∼5% of all the human spleen cells (Figs. 6gi, 6hi, 7b′, 7b′′, 7e–o). Staining of thymus, tonsil, and spleen with anti-CD45, a pan-leukocyte marker (leukocyte common Ag), showed many CD45⁺ cells in all three lymphoid-rich tissues (Fig. 7a–c). Costaining with anti-LY6S Abs revealed that only the spleen contained LY6S-positive cells (Fig. 7b′), distinguishing it from the LY6S-negative thymic and tonsillar tissues (Fig. 7a′, 7c′). Abs against CD8 (Fig. 7e), CD4 (Fig. 7f), CD34 (Fig. 7g), CD19 (Fig. 7h), CD11b (Fig. 7i), FoxP3 (Fig. 7j), CD11c (Fig. 7k), CD20 (Fig. 7l), CD117-cKit (Fig. 7m), CD68 (Fig. 7n), and CD31 (Fig. 7o) showed no colocalization of any of these markers with LY6S-iso1–positive cells, suggesting that the LY6S-positive cells do not belong to a classical lineage of lymphoid cells. In contrast with the other tested leukocyte markers, a discrete subpopulation representing ∼10% of the LY6S-positive splenic cells also expressed the leukocyte common Ag marker CD45 (Fig. 7b′′, 7d), indicating that a subset of LY6S-iso1^+ve and CD45^+ve cells express both proteins.

That LY6S expression was observed only after cytokine treatment of cells (Figs. 2f, 2g, 2j, 6f, 6fi′′) suggested that LY6S expression may be related to an activated cell phenotype. To test this supposition, we infected human cell lines with retroviral particles coding for the GPI-linked cell-surface LY6S-iso1 protein or for the secreted LY6S-iso2 protein. Cells expressing the secreted LY6S-iso2 protein showed no differences in gross morphology compared with control cells stably infected with control pQCXIP plasmid (Fig. 8ai, compare right inset [LY6-iso2] with left inset [control vector]). However, cells infected with particles coding for the cell-surface-linked LY6S-iso1 protein showed changes in cell morphology (compare cells expressing LY6-iso1 in (Fig. 8aii, 8aii′, 8bii, 8bii′, 8cii, 8cii′ with cells expressing control vector in (Fig. 8ai, 8ai′, 8bi, 8bi′, 8ci, 8ci′), where cells expressing LY6S-iso1 protein were large and many contained prominent cytoplasmic vacuoles (see, e.g., (Fig. 8aii′). Cells displaying this phenotype included U87 (glioma; (Fig. 8aii, 8aii′), MCF7 (breast carcinoma; (Fig. 8bii, 8bii′), and YDFR (melanoma; (Fig. 8cii, 8cii′). The LY6S-iso1–expressing puromycin-resistant cells initially grew slowly (as shown for YDFR melanoma cells expressing LY6S-iso1; (Fig. 8d). However, after three to four passages, the LY6S-iso1 cells resumed growth, reached plate confluency, and could then be subcultured (Fig. 8e).

Because transcription of the LY6S gene is IFN inducible and LY6S-iso1 expression is associated with changes in cell morphology and growth, we investigated whether LY6S-iso1 protein expression impacts cell resistance to viral infection. In all three human cell lines investigated, M12, YDFR, and U87, LY6S-iso1 protein expression was associated with a marked diminishment in VSV viral replication (Fig. 9). In M12 melanoma cells, LY6-iso1 expression led to a decrease in viral yield that exceeded two orders of magnitude (Fig. 9c, 9d), and a marked decline of viral yields was also seen in human YDFR melanoma and human U87 glioblastoma cells that express the LY6-iso1 protein (Fig. 9a, 9b, 9d).

The changes observed in the phenotype of LY6S-iso1–expressing cells suggested an altered gene expression profile. To test for this, we performed next generation sequencing (both Illumina single-read sequencing and PE True Stranded) on RNA isolated from triplicate U87 glioblastoma cell cultures infected with control pQCXIP plasmid (U87-control) or from cells stably expressing cell-surface-linked LY6S-iso1 protein (designated “U87-LY6S-iso1”). Genes coding for chemokines, chemokine receptors, cytokines, and other proteins associated with an inflammatory cell phenotype were markedly upregulated in the LY6S-iso1–expressing cells (Fig. 10a, 10b, shown as volcano and heatmap plots, respectively; see Supplemental Table II). These included the following: cytokines (IL-1A, IL-1B, IL-6, IL-11, IL-24, IL-36B, CD274 [PDL1], CSF2, CSF3, EREG, FGF2, GDNF, HBEGF, LIF, and TNFSF8), chemokines (CXCL1, CXCL2, CXCL3, CXCL5, CXCL8, CX3CL1, and CCL20), chemokine receptors (ACKR3, CMKLR1, and IL-36RN), genes directly related to inflammation and immune responses (BCL2A1, BGN, LCN2, LCP1, NFkB1, NLRP12, RCAN1, PTGS2, RGS16, and SDC3; downregulated: CXADR_, DEPTOR, DKK1_, HMOX1), and genes related to IFN and cytokine signaling (IRF5, NFKBIZ, and SOCS1; downregulated: ID1 and SNAI1). Ingenuity analyses (Qiagen IPA as detailed in Materials and Methods) of the RNA-seq data identified the following biological pathways with very high p values, canonical pathways [−log(p value)], and annotations of “diseases and functions”(p value): (1) granulocyte adhesion and diapedesis [12.8], (2) role of cytokines in mediating communication between cells [9.86], (3) differential regulation of cytokine production in macrophages and Th cells by IL-17A and IL17-F [6.89], (4) role of pattern recognition receptors in recognition of bacteria and viruses [6.82], (5) acute inflammatory response (2.99E−13), (6) accumulation of myeloid cells (1.26E−17), (7) accumulation of leukocytes (6.75E−17), (8) cytokine- and chemokine-mediated signaling pathways (3.29E−16), and (9) accumulation and recruitment of granulocytes (2.2E−14) and phagocytes (1.76E−14). These pathways reflected the alterations in gene expression patterns elicited by LY6S-iso1 expression, particularly of genes coding for cytokines, chemokines, chemokine receptors, IFN, and immune- and inflammatory-related proteins (Fig. 10a, 10b).

These results were confirmed at the protein level with cytokine/chemokine arrays, which showed increased secretion by the LY6S-iso1–expressing U87 cells of CXCL1, IL-1ra, CSF2, IL-6, and MIP-1a (Fig. 10c, 10d).

That LY6S-iso1 elicits expression of genes associated with an inflammatory phenotype was substantiated in the additional human cell lines, M12 and YDFR. In each of the three cell lines investigated (M12, YDFR, and U87), LY6S-iso1 elicited increased expression (using the stringent criteria of adjusted p < 0.05, counts > 30, and a fold change > 2) of IL-1B, CXCL8, LIF, PCDH1, ADAMTS4, and STC1, all genes associated with an inflammatory cell phenotype (Fig. 11). Less stringent criteria showed that in at least two of the three cell lines investigated, LY6S-iso1 elicited increased expression of the chemokines CXCL1, CXCL2, CXCL3, and CCL20 and of the ILs IL-6, IL-11, IL-12A, IL-23A, IL-33, and IL-36B (Fig. 11).

The following inflammation-associated genes were also upregulated in all three LY6S-iso1–expressing cell lines: NRG1 (7.7-, 42-, and 2.7-fold for M12, YDFR, and U87 cells, respectively), PTGS2 (1.6, 11, and 3.9), RCAN1 (1.5, 1.5, and 2.1), and SOCS1 (1.2, 1.9, and 2.8). For all three (LY6S-iso1–expressing) cell lines, the following canonical functional networks appeared in the top 14 pathways identified using the IPA (Supplemental Table II): dendritic cell maturation (2.236, 2.449, and 3.545, Z scores for M12, U87, and YDFR, respectively), IL-6 signaling (2, 1.89, and 2.121), acute-phase response signaling (2, 1.663, and 1.941), IL-8 signaling (2.236, 1.342, and 1.291), and neuroinflammation signaling (1.633, 1.342, and 1.512). The IPA-derived upstream regulator analysis provided further support for involvement of the LY6S-iso1 protein with an inflammatory cell phenotype. The list of upstream regulators (Supplemental Table II) included the following top Z-scoring upstream regulators for all three cell lines expressing LY6S-iso1: TNF (#1 in the list, with Z scores of 3.3, 5.0, and 5.3 for M12, U87, and YDFR cells, respectively), tetradecanoylphorbol acetate (#2; 3.6, 4.4, and 4.3), IL-1B (#3; 2.4, 5.1, and 3.6), LPS (#4; 2.9, 5.3, and 2.8), IL-17A (#5; 2.2, 4.7, and 4.0), NF-κB (complex) (#7; 3.0, 4.4, and 2.9), ERK (#10; 3.1, 3.2, and 3.4), IKBKB (#12; 2.7, 4.1, and 2.6), IL-1A (#13; 3.2, 3.7, and 2.7), and IL-1 (#16; 3.2, 3.7, and 2.4). Additional upstream regulators included poly(rI:rC)-RNA (#20; 2.4, 3.8, and 2.9), TGF-B1 (#23; 2.4, 2.2, and 4.1), TLR3 (#25; 2.5, 3.6, and 2.5), and TLR9 (#26; 2.4, 3.8, and 2.5).

This study documents a previously unannotated human gene now designated LY6S. Analyses both at the RNA and protein levels, as well as TransMap and public domain RNA-seq data, provide evidence for the existence of the LY6S gene. At the RNA level, cloning and sequencing of the cDNA products generated by RT-PCR of human spleen samples showed mRNAs that code for three LY6S isoforms: LY6S-iso1 codes for a cell-surface protein, whereas LY6S-iso2 and LY6-iso3 code for secreted proteins. LY6S-iso1 has the “fingerprint” features characteristic of the large LY6 protein family, including (1) spacing of cysteine residues, (2) exon/intron makeup, (3) the cysteine-cysteine doublet, followed by (4) the cysteine-asparagine doublet. What differentiates it from other LY6 family members is the consensus sixth cysteine residue replaced in LY6S-iso1 by a serine residue. We are unaware of any other member of the LY6 protein family that shows this deviation from the LY6 consensus. Because the unpaired, nondisulfide cysteine residue is now free to form disulfide bonds with unbonded cysteine residues in other proteins, the odd number (9) of cysteine residues in LY6S-iso1 likely affects LY6S-iso1 protein interactions. Yet 9 of the 10 consensus LY6S cysteine residues are retained in the LY6-iso1 protein, indicating that the four disulfide bridges formed between the LY6 consensus cysteine (C) residues C#1 and C#5, C#2 and C#3, C#7 and C#8, and C#9 and C#10 (8) should all be intact. Snake toxins that contain eight cysteine residues, associated with four disulfide bridges, form the characteristic three-finger structure (7), as do the LY6 proteins that have five disulfide bridges (30). Furthermore, the N-terminal LU/uPAR domain of uPAR lacks a cysteine pair (31), yet still forms the LU/uPAR domain. It is reasonable to assume, therefore, that the LY6S-iso1 protein also adopts a TFP structure.

Although the data indicate that human LY6S is homologous to the Ly6a subfamily of LY6 genes, we do not know whether LY6S is the ortholog of a single gene of the murine Ly6a subfamily, or whether all eight murine genes and their protein products are an expansion of the solitary human LY6S gene. The phylogenetic analysis, as well as our additional similarity analyses, show that the eight murine Ly6 genes constitute a gene cluster, in which an ancestral gene likely underwent several duplication events, thereby forming the Ly6a subfamily cluster. Moreover, the TransMap algorithm of the UCSC (University of California, Santa Cruz) genome browser maps Ly6a to the human LY6S gene. These considerations led us to conclude that this is a one-to-many orthologous relationship, and LY6S is the human ortholog of the Ly6a subfamily genes.

Genes of the murine Ly6a subfamily play pivotal roles both in inflammation and in hematopoietic stem cells (e.g., Refs. 32–36), and many of the Ly6a subfamily genes are related both to an inflammatory cell phenotype and to IFN-related pathways (18, 19). Just like the genes of the Ly6a subfamily, it appears from our analyses that LY6S is also involved in inflammatory processes. RNA-seq analyses in three different human cell lines showed that expression of the LY6S-iso1 protein is associated with the expression of genes coding for chemokines, cytokines, and other proteins classically connected to an inflammatory cell phenotype. These findings lend further support, in addition to that provided by the phylogenetic and protein similarity analyses, to the notion that LY6S has a one-to-many orthologous relationship with the genes of the murine Ly6a subfamily gene cluster.

Like the genes of the Ly6a subfamily, the human LY6S gene is also an ISG. By inducing expression of hundreds of genes, IFNs and the protein products of the induced ISGs are critical players in the restriction of viral infections. In line with our findings that LY6S expression is both IFN inducible and associated with increased expression of genes related to inflammation and immune responses, we observed that LY6S-iso1 expression markedly inhibited viral replication, a phenomenon seen in each of the three different human cell lines investigated. This is likely indirectly mediated by LY6S-iso1, which, as noted earlier, leads to altered patterns of gene expression, which in turn affects virus replication. Thus, one of the functions of LY6S may be to elicit protection from viral infection, as already noted for LY6E (37).

LY6S expression at the RNA level was highest in spleen, whereas it was undetectable in bone marrow and peripheral blood leukocytes. At the protein level, immunostaining analyses identified the LY6S-iso1 protein in ∼5% of all spleen cells, yet it could not be detected in classical lymphoid-rich tissues, such as thymus and tonsil. These findings suggest that the LY6S⁺ cells are likely not of a known classical hematopoietic or lymphoid cell lineage. In this respect, the expression pattern of LY6S in cells and tissues is different from that of murine Ly6a, which, in contrast with LY6S, is expressed in peripheral blood leukocytes, on lymphoid precursor cells and hematopoietic stem cells in the mouse bone marrow (38, 39), and in many CD4⁺ T cells in the spleen.

In line with the notion that LY6S expression does not designate a classical cell of the human myeloid lineage, immunofluorescent analysis for expression of CD11b, a prototypical marker of human macrophages and possibly for many cells of human myeloid lineages, failed to show significant colocalization in the spleen with LY6S-expressing cells, just as classical T and B cell markers also failed to colocalize with LY6S expression. However, ∼10% of all CD45⁺ spleen cells, a pan-leukocyte marker, are also LY6S⁺ (see (Fig. 7d), and very few CD11b⁺ cells, amounting to <1% of these cells, are also LY6S⁺. Certain subsets of spleen macrophages, such as red pulp, marginal zone, and marginal metallophilic macrophages, do not express high levels of CD11b, if at all, and because of this, the LY6S⁺ cells may belong to such a subset.

In an attempt to learn more about the LY6S-iso1⁺ spleen cells, we queried publicly available human spleen single-cell RNA-seq data, but because of their limited gene coverage and low sensitivity, LY6S-iso1⁺ cells could not be identified in the available single-cell RNA-seq datasets. Sorting of human spleen cells with the anti–LY6S-iso mAbs that we have generated into LY6S-iso1^+ve cells and LY6S-iso1⁻ cells, followed by RNA-seq analyses, might assist in the future identification of the [LY6S-iso1⁺] cell type.

In summary, to our knowledge, we have identified the new IFN-inducible human LY6S gene that is likely the long-sought human ortholog of genes belonging to the Ly6a subfamily. LY6S is expressed in a discrete subset of human spleen cells of a nonclassical lineage, and its expression is associated with an inflammatory cell phenotype and with the restriction of viral replication. Of particular therapeutic clinical interest are the recent findings showing that Ly6e and Ly6a serve as receptors for viral entry into the cell. Ly6e is a receptor for HIV (40), and Ly6a, expressed on the surface of murine brain endothelial cells, is the cell receptor for a recombinant adeno-associated virus (AAV-PHP.B), via which AAV-PHP.B can penetrate the blood–brain barrier and transduce gene cargo into the brain (41–45). Our findings indicate that human LY6S is the ortholog to the genes of the murine Ly6a subfamily, and as we have observed LY6S expression in human brain tissue, the discovery of LY6S as reported in this article may open up new possibilities for introducing therapeutic genes into the human brain. Furthermore, the LY6S proteins may serve as receptors for additional pathogenic viruses, thus leading to novel antiviral therapeutic strategies.

The authors have no financial conflicts of interest.