Target gene: Cd48 (Mus musculus, NCBITaxon:10090) — UniProt P18181 (CD48_MOUSE)
Focus type: function_assignment
Hypothesis slug: cd48-nk-activation-direction
Term under evaluation: natural killer cell activation (GO:0030101)
Seed hypothesis: Mouse CD48 contributes to natural killer cell activation (GO:0030101), i.e. its engagement of 2B4 (Cd244) and/or CD2 promotes rather than inhibits NK cell activation in mouse.
Verdict: REFUTED as stated (over-annotated for direction).
The seed hypothesis asserts a direction — that mouse CD48's engagement of 2B4 (CD244) and/or CD2 promotes rather than inhibits NK cell activation. The preponderance of direct mouse genetic and functional evidence contradicts this direction. In mouse, the CD48–2B4 axis is predominantly inhibitory, functioning as an MHC-class-I-independent NK-cell self-tolerance checkpoint. This is the opposite of the activating role the same molecular pair plays in human NK cells — a species difference stated verbatim in one of the seed's own reference papers (PMID:16002700).
Three convergent lines of mouse evidence drive this verdict. First, antibody-blocking and knockout experiments show that ligation of 2B4 by CD48 inhibits NK cytotoxicity and IFN-γ production (PMID:15123744, PMID:15356144), and the 2B4–CD48 pair constitutes a second, MHC-I-independent system for NK self-tolerance (PMID:15870174). Second, deleting the entire mouse Slam locus — which removes CD48 along with the other SLAM-family receptors — enhances NK activation toward hematopoietic targets, with the net inhibitory function mapping solely to 2B4 (PMID:27573813). Third, mouse SAP-family adaptors EAT-2 and ERT act as inhibitors (not activators) of NK function at 2B4 (PMID:16127454), and the dominant mouse 2B4-long isoform correlates with inhibition (PMID:15356144).
Two important caveats keep this from being a flat "wrong in all contexts" finding. (1) The single strongest positive mouse NK genetic result — that 2B4/CD48 interaction is required for generation of NK effector functions (PMID:15905190) — has been mechanistically reinterpreted by the same laboratory as avoidance of perforin-dependent fratricide, not as a genuine 2B4-dependent activation signal (PMID:17537992). (2) A genuine, context-restricted activating contribution of CD48 does exist in trans / missing-self settings, where CD48 (with Ly9) on MHC-I-deficient tumor targets significantly contributes to NK activation (PMID:27054584), and anti-CD48 crosslinking can deliver a bona fide activating signal that is not merely relief of inhibition (PMID:20881194). Crucially, CD48 itself is a GPI-anchored counter-receptor/ligand with no cytoplasmic signaling tail; the activation or inhibition signal is transduced by the partner receptor (2B4 or CD2), not by CD48. The existing GO:0030101 annotation on mouse Cd48 is an ISO (orthology-projected) carry-over that inherits the human "activating" directionality and is not supported by direct mouse experiment.
| Citation (PMID) | Evidence type | Supports / Refutes / Qualifies / Competing | Claim tested | Key finding | Organism / context | Confidence & limitations |
|---|---|---|---|---|---|---|
| 15123744 | Mutant phenotype + blocking Ab | Refutes seed direction | Does CD48–2B4 ligation activate or inhibit mouse NK? | "NK lysis of CD48(+) tumor and allogeneic targets is inhibited by 2B4 ligation. Interferon gamma production by NK cells is also inhibited." 2B4−/− mice show increased clearance of CD48+ tumor; SAP-independent. | Mouse NK, in vitro + in vivo | High; direct genetic + Ab evidence |
| 15356144 | Direct assay / isoform | Refutes seed direction | Which 2B4 isoform dominates and what is the signal? | "Engagement of 2B4 by its counterreceptor, CD48... leads to an inhibition in NK cytotoxicity"; dominant 2B4-long isoform; SAP-independent | Mouse NK | High |
| 15870174 | In vivo + in vitro | Refutes seed direction | Is 2B4–CD48 a self-tolerance system? | 2B4–CD48 is a second, MHC-I-independent NK self-recognition/self-tolerance system; inhibits lysis of syngeneic/β2m-deficient targets | Mouse NK, BM rejection assay | High |
| 27573813 | Mutant phenotype (whole-locus KO) | Refutes seed direction | Does removing CD48 (+SFRs) raise or lower NK activation? | "Enhanced NK cell activation... inhibitory function of the Slam locus was due solely to 2B4"; SFR inhibition suppresses LFA-1 | Mouse NK, ~400-kb Slam-locus deletion | High; CD48 deletion included |
| 16127454 | Adaptor genetics | Refutes/qualifies | Are mouse SAP-family adaptors activating? | "Unlike SAP, EAT-2 was an inhibitor of NK cell function" (EAT-2/ERT) | Mouse NK | High |
| 16002700 | Review-level synthesis (seed ref) | Refutes seed direction | Human vs mouse signal direction | "In human NK cells, 2B4/CD48 interaction induces activation signals, whereas in murine NK cells it sends inhibitory signals." | Human vs mouse comparison | High relevance; review-level |
| 15634901 | Mutant phenotype | Refutes/qualifies | In vivo tumor rejection direction | WT reject CD48+ melanoma poorly vs CD48− (2B4 ligation inhibitory); male 2B4−/− reject CD48+ better; gender-specific CD48-independent defect | Mouse, B16 melanoma | Moderate-high; gender confound |
| 17537992 | Mechanistic reinterpretation | Qualifies/competing (undercuts main pro-seed result) | Why do 2B4−/− NK underperform? | "In the absence of 2B4 signaling, activated NK cells have defective cytotoxicity and proliferation because of fratricide and not due to the absence of a 2B4-dependent activation signal" | Mouse NK | High; same lab as PMID:15905190 |
| 15905190 | Mutant phenotype | Supports seed (contested) | Is 2B4/CD48 required for NK effector generation? | "2B4/CD48, but not CD2/CD48, interaction is essential for IL-2-driven expansion and activation of murine NK cells" | Mouse NK, homotypic NK–NK | Moderate; reinterpreted by PMID:17537992 |
| 27054584 | Direct assay (missing-self) | Supports (context-restricted) | Does CD48 contribute to NK activation in trans? | "CD48 and Ly9 (CD229) by MHC-I-deficient tumor cells significantly contributes to NK cell activation"; NK education can render 2B4 non-functional | Mouse NK, MHC-I-deficient targets | Moderate-high; context-specific |
| 20881194 | Direct assay | Supports (context-restricted) | Is anti-CD48 activation genuine or disinhibition? | "Induction of NK cell activation by anti-CD48 or by B cells is not due to the release of inhibitory effects of 2B4"; CD2/2B4 double-KO used | Mouse NK | Moderate; crosslinking artifact possible |
| 14666660 | In vivo Ab crosslinking | Supports (artifact-prone) | Does 2B4/CD48 activation reduce metastasis? | Anti-2B4/anti-CD48 crosslinking reduced B16F10 metastasis ~5-fold; IFN-γ-dependent | Mouse, B16F10 | Low-moderate; agonist Ab, not physiologic |
| 19586919 | Mechanistic (hybridoma) | Qualifies | How does CD244 inhibit? | "Inhibitory effects of mouse CD244 are accounted for by competition with CD2 at the cell surface for CD48"; CD244 recruits PLCγ1 via EAT-2 | Mouse T-cell hybridoma | Moderate; non-NK model |
| 27249817 | Direct assay (cis/trans) | Refutes/qualifies | Cis vs trans CD48–2B4 effect | "Interfering with the cis interaction... enhanced the lysis of CD48-expressing tumour cells"; cis dampens 2B4 phosphorylation | Mouse/human NK | Moderate-high |
| 20363967 | Mutant phenotype | Refutes | Is 2B4 inhibitory in NK homeostasis? | "2B4 is a dominant inhibitory receptor in SHIP-deficient NK cells" | Mouse NK | Moderate-high |
| 9841922 | Interaction/biophysics (seed ref) | Qualifies (molecular framing) | Is CD48 a ligand with two receptors? | "mCD48 bound m2B4 with a six- to ninefold higher affinity (Kd ≈ 16 µM at 37°C) than its other ligand, CD2" | Mouse recombinant proteins | High; establishes CD48 as counter-receptor |
| 17950006 | Structural (seed ref) | Qualifies (molecular framing) | Structure of 2B4–CD48 complex | Structure of mouse 2B4–CD48 complex; CD48 is the heterophilic ligand engaged by the 2B4 receptor | Mouse, crystal structure | High |
| 20813844 | Direct assay | Competing (human, activating) | Human NK requirement for 2B4/CD48 + CD2/CD58 | Both 2B4/CD48 and CD2/CD58 needed for human NK effector function | Human NK | High for human; species-divergent |
{{figure:cd48_nk_direction_tally_v2.png|caption=Tally of mouse NK direction evidence for the CD48–2B4 axis, including the contested positive result (PMID:15905190) and its fratricide reinterpretation (PMID:17537992). Inhibitory/refuting evidence dominates; the positive signal is context-restricted (trans/missing-self) or mechanistically reframed.}}
The core of the seed hypothesis is a directional claim, and the directional claim fails in mouse. Multiple independent groups using complementary approaches converge on inhibition. Lee et al. 2004 (PMID:15123744(https://pubmed.ncbi.nlm.nih.gov/15123744/)) showed with both antibody blocking and 2B4−/− NK cells that NK lysis of CD48+ tumor and allogeneic targets, as well as IFN-γ production, is inhibited by 2B4 ligation; consistent with this, 2B4−/− mice have increased peritoneal clearance of CD48+ tumor. Importantly, this inhibition was SAP (SH2D1A)-independent, distinguishing the mouse inhibitory mode from the SAP-dependent human activating mode. Mooney et al. 2004 (PMID:15356144(https://pubmed.ncbi.nlm.nih.gov/15356144/)) independently found that CD48 on target cells inhibits NK cytotoxicity, correlating inhibition with the dominant 2B4-long (2B4L) isoform in mouse, again SAP-independent. McNerney et al. 2005 (PMID:15870174(https://pubmed.ncbi.nlm.nih.gov/15870174/)) elevated this to a framework: 2B4–CD48 constitutes a second, MHC-class-I-independent system for NK-cell self-tolerance, with in vivo evidence from a bone-marrow rejection assay. Finally, Roncagalli et al. 2005 (PMID:16127454(https://pubmed.ncbi.nlm.nih.gov/16127454/)) showed the mouse SAP-family adaptors EAT-2 and ERT are inhibitors of NK function via 2B4 — the opposite of what an activating axis would predict. A seed reference, Mathew et al. 2005 (PMID:16002700(https://pubmed.ncbi.nlm.nih.gov/16002700/)), states the species divergence in plain language: "In human NK cells, 2B4/CD48 interaction induces activation signals, whereas in murine NK cells it sends inhibitory signals."
The hypothesis is not uniformly wrong in all contexts — it is wrong as a default directional assignment. There is a genuine activating contribution of CD48 in trans / missing-self settings. Alari-Pahissa et al. 2016 (PMID:27054584(https://pubmed.ncbi.nlm.nih.gov/27054584/)) showed that CD48 (together with Ly9/CD229) on MHC-I-deficient tumor targets significantly contributes to NK activation via 2B4 — but with a twist: NK education against CD48+ MHC-I-deficient cells can render 2B4 non-functional (tolerance), so even the activating contribution is conditional on NK education state. Sinha et al. 2010 (PMID:20881194(https://pubmed.ncbi.nlm.nih.gov/20881194/)) used CD2/2B4 double-KO mice to show that anti-CD48 or B-cell-driven NK activation "is not due to the release of inhibitory effects of 2B4," i.e., a genuine activating signal exists in a defined context. Johnson et al. 2003 (PMID:14666660(https://pubmed.ncbi.nlm.nih.gov/14666660/)) found in vivo anti-2B4/anti-CD48 crosslinking reduced B16F10 metastasis — but this uses agonist antibodies rather than physiologic ligand engagement.
Mechanistically, all of this must be read through the molecular identity of CD48. Brown et al. 1998 (PMID:9841922(https://pubmed.ncbi.nlm.nih.gov/9841922/), a seed reference) established that CD48 binds 2B4 with 6–9× higher affinity (Kd ≈ 16 µM at 37 °C) than its other counter-receptor CD2, and Velikovsky et al. 2007 (PMID:17950006(https://pubmed.ncbi.nlm.nih.gov/17950006/), a seed reference) solved the mouse 2B4–CD48 complex structure. Both confirm CD48 is a GPI-anchored heterophilic ligand/counter-receptor with no cytoplasmic signaling domain. Whatever activation or inhibition occurs is transduced by the partner receptor (2B4 or CD2), not by CD48 itself. This is central to the curation decision: assigning a "natural killer cell activation" BP term to CD48 implies CD48 drives an activating process, whereas CD48 is the passive ligand whose engagement, in mouse, net-inhibits the partner.
UniProt P18181 cross-references show GO:0030101 "natural killer cell activation" annotated with evidence code ISO:GO_Central (Inferred from Sequence Orthology) — i.e., projected from the ortholog rather than supported by a direct mouse experiment. Because the human CD48/2B4 axis is activating, orthology projection inherits the human directionality and applies it to mouse, where the direction is reversed. The same UniProt record also carries the more informative MF term GO:0048018 "receptor ligand activity" (ISO) and CC terms (external side of plasma membrane IDA:MGI; membrane raft; GPI-anchor) that correctly describe CD48 as a GPI-anchored surface ligand.
The decisive loss-of-function test is Guo et al. 2016 (PMID:27573813(https://pubmed.ncbi.nlm.nih.gov/27573813/)), which deleted the entire ~400-kb mouse Slam locus — removing CD48 along with the six SLAM-family receptors. These mice showed enhanced NK activation toward hematopoietic targets, with the inhibitory function mapping solely to 2B4, and SFR-mediated inhibition acting by suppressing LFA-1 activation. Supporting evidence: Fortenbery et al. 2010 (PMID:20363967(https://pubmed.ncbi.nlm.nih.gov/20363967/)) found 2B4 is a dominant inhibitory receptor whose deficiency restores cytolysis in SHIP-deficient NK cells, and Claus, Wingert & Watzl 2016 (PMID:27249817(https://pubmed.ncbi.nlm.nih.gov/27249817/)) addressed the seed's explicit cis-vs-trans question: cis CD48–2B4 interaction (on the same cell) reduces trans binding and dampens 2B4 phosphorylation, and interfering with the cis interaction enhanced lysis of CD48+ tumor cells.
The strongest positive mouse genetic evidence for the seed is Lee et al. 2006 (PMID:15905190(https://pubmed.ncbi.nlm.nih.gov/15905190/)): 2B4/CD48 (but not CD2/CD48) interaction is essential for IL-2-driven expansion and activation of mouse NK cells, with defective calcium signaling and impaired cytotoxicity/IFN-γ in its absence, via homotypic NK–NK interactions. However, Taniguchi et al. 2007 (PMID:17537992(https://pubmed.ncbi.nlm.nih.gov/17537992/)), from the same laboratory, reinterpreted this: in the absence of 2B4–CD48, activated mouse NK cells undergo perforin-dependent fratricide, so the apparent "defective cytotoxicity and proliferation" arises from self-killing, not from loss of a 2B4-dependent activation signal. This substantially undercuts the one result that most directly supports the seed. Two further points narrow the seed: Clarkson & Brown 2009 (PMID:19586919(https://pubmed.ncbi.nlm.nih.gov/19586919/)) showed mouse CD244's inhibitory effect is "accounted for by competition with CD2 at the cell surface for CD48," and Kim et al. 2010 (PMID:20813844(https://pubmed.ncbi.nlm.nih.gov/20813844/)) confirmed the activating 2B4/CD48 + CD2/CD58 requirement is a human NK phenomenon — reinforcing that the seed's directional claim is imported from human biology.
CD48 is a GPI-anchored ligand. Its NK-relevant signaling partner is the receptor 2B4 (CD244), which carries cytoplasmic ITSM motifs that recruit SAP-family adaptors. The directionality of the output depends on species, adaptor repertoire, isoform, and cis/trans geometry:
CD48 (GPI-anchored ligand; NO cytoplasmic tail)
| engages (Kd ~16 µM, 6-9x > CD2)
v
+--------- 2B4 / CD244 (ITSM receptor) ----------+
| |
HUMAN NK MOUSE NK
SAP-dependent SAP-independent inhibition;
ACTIVATING ---------------------------> EAT-2 / ERT = INHIBITORS
(PMID:20813844) dominant 2B4-LONG isoform
NET INHIBITORY (self-tolerance)
(PMID:15123744/15356144/15870174/16127454)
cis CD48-2B4 (same cell): dampens 2B4 phosphorylation -> less killing (PMID:27249817)
trans CD48 on MHC-I-deficient target: context-specific ACTIVATION via missing-self (PMID:27054584)
Whole Slam-locus deletion (removes CD48 + SFRs): ENHANCED NK activation;
inhibition maps SOLELY to 2B4 (PMID:27573813)
The central interpretive point for curation: the seed conflates the human and mouse directions of an identical receptor–ligand pair. In mouse, net physiologic output of CD48–2B4 is inhibitory (self-tolerance checkpoint), with activation emerging only in restricted trans/missing-self contexts. Additionally, CD48 is the ligand, not the signal transducer; a BP term implying CD48 drives NK activation attributes a receptor-side process to a ligand-side molecule.
| GO term | Aspect | Current status | Evidence-based lead (curator to verify) |
|---|---|---|---|
| GO:0030101 natural killer cell activation | BP | ISO carry-over (human-biased direction) | Do not treat as a positive mouse function. Direction refuted in mouse; if retained, needs qualifier/context; consider removal or NOT-qualification for positive activation |
| GO:0032815 / GO:0045953 negative regulation of NK cell activation / cytotoxicity | BP | Absent | Candidate add (lead) — better reflects net mouse inhibitory role; curator to confirm exact term and evidence |
| GO:0048018 receptor ligand activity | MF | ISO present | Retain — informative MF; avoid "protein binding" as final recommendation |
| GPI-anchor / external side of plasma membrane / membrane raft | CC | IDA:MGI / ISO present | Retain — correctly describes CD48 as GPI-anchored surface ligand |
{{figure:cd48_nk_direction_tally.png|caption=Initial tally of mouse-specific evidence on CD48–2B4 signal direction, built during iteration 2. Inhibitory/self-tolerance evidence outweighs activating evidence, which is confined to trans/missing-self or agonist-antibody contexts.}}
Human-vs-mouse species divergence (the central conflict). The same CD48–2B4 pair is activating in human NK (PMID:20813844) and inhibitory in mouse NK (PMID:16002700, PMID:15123744). The seed appears to import the human direction into mouse. This is the dominant alternative explanation for how GO:0030101 landed on mouse Cd48 — ISO orthology carry-over.
Ligand vs receptor conflation. CD48 is GPI-anchored with no signaling tail (PMID:9841922, PMID:17950006). Any "activation" or "inhibition" is a property of the partner receptor's signaling, not of CD48. A BP term implying CD48 performs NK activation mislocates the activity.
Fratricide artifact. The main pro-seed result (PMID:15905190) is best explained by NK self-killing in the absence of 2B4–CD48 (PMID:17537992), not by loss of an activation signal — a classic loss-of-function-phenotype-vs-direct-function distinction.
Isoform specificity. Mouse expresses a dominant 2B4-long (inhibitory) isoform (PMID:15356144); the activating 2B4-short isoform is relatively minor, biasing the mouse net output toward inhibition.
cis vs trans geometry. Cis CD48–2B4 dampens signaling (PMID:27249817); trans CD48 on MHC-I-deficient targets can activate (PMID:27054584). The net organismal output (whole-locus KO) is inhibition (PMID:27573813).
Agonist-antibody artifacts. Positive in vivo results (PMID:14666660) rely on crosslinking antibodies that may not reflect physiologic ligand engagement.
Action on GO:0030101 (natural killer cell activation, BP):
- Lead: Do not treat GO:0030101 as a positive, mouse-experimentally-supported function of Cd48. The term is ISO orthology carry-over inheriting the human activating direction. In mouse the net direction is inhibitory. Consider removal of the positive-activation annotation, or qualification/NOT-annotation for the activation direction, and/or restriction to a trans/missing-self context note.
- Candidate replacement/add (BP): GO:0032815 (negative regulation of natural killer cell activation) or GO:0045953 (negative regulation of NK cell mediated cytotoxicity) — better reflects the net mouse role as a 2B4 ligand in an inhibitory self-tolerance axis. Curator to select the precise term and supporting evidence code.
- Retain (MF): GO:0048018 receptor ligand activity as the informative molecular function; avoid "protein binding" as a final recommendation.
- Retain (CC): GPI-anchor / external side of plasma membrane (IDA:MGI) / membrane raft.
Candidate references with snippets to verify:
- PMID:16002700(https://pubmed.ncbi.nlm.nih.gov/16002700/): "In human NK cells, 2B4/CD48 interaction induces activation signals, whereas in murine NK cells it sends inhibitory signals." — supports species-specific direction flip.
- PMID:15123744(https://pubmed.ncbi.nlm.nih.gov/15123744/): "NK lysis of CD48(+) tumor and allogeneic targets is inhibited by 2B4 ligation. Interferon gamma production by NK cells is also inhibited." — direct mouse inhibition.
- PMID:27573813(https://pubmed.ncbi.nlm.nih.gov/27573813/): "enhanced NK cell activation responses... the inhibitory function of the Slam locus was due solely to 2B4." — loss-of-function direction.
- PMID:17537992(https://pubmed.ncbi.nlm.nih.gov/17537992/): "in the absence of 2B4 signaling, activated NK cells have defective cytotoxicity and proliferation because of fratricide and not due to the absence of a 2B4-dependent activation signal." — reinterprets the pro-seed result.
- PMID:27054584(https://pubmed.ncbi.nlm.nih.gov/27054584/): "CD48 and Ly9 (CD229) by MHC-I-deficient tumor cells significantly contributes to NK cell activation." — the one genuine context-restricted activating contribution.
Suggested curator questions:
- Is the GO:0030101 annotation on P18181 experimental or ISO? (Expected: ISO.)
- Should the review explicitly flag the human→mouse direction inheritance as the source of over-annotation?
- Does the review want a context-qualified activation annotation for trans/missing-self, or a cleaner negative-regulation annotation?
Suggested experiments: clean Cd48−/− NK assay; perforin-background fratricide control; 2B4 isoform swap (see Discriminating Tests).
The seed hypothesis — that mouse CD48 engagement of 2B4/CD2 promotes rather than inhibits NK cell activation — is refuted as a directional claim. Mouse CD48–2B4 is a predominantly inhibitory, MHC-I-independent NK self-tolerance axis; the existing GO:0030101 annotation is an orthology-projected import of the human activating direction; the single strongest positive mouse result is better explained by fratricide avoidance; and a genuine positive CD48 contribution exists only in restricted trans/missing-self contexts. Curators should avoid treating GO:0030101 as a positive mouse function, consider a negative-regulation BP term, and retain GO:0048018 (receptor ligand activity) as the informative molecular function.