# Critical Evaluation of Hypotheses for Complement-Mediated Synaptic Selectivity
## Overview
These hypotheses address a legitimate gap in understanding how C1q distinguishes between synapses for elimination during prolonged anesthesia. However, they vary substantially in mechanistic coherence, evidential support, and translational potential. I evaluate each systematically.
---
## Hypothesis 1: CREB-BDNF-TrkB Activity-Dependent Protection
### Weak Links
1. **Unproven activity sparing in vivo**: The mechanism assumes hippocampal-cortical and prefrontal circuits remain partially active during prolonged anesthesia, but sevoflurane and isoflurane broadly suppress arousal circuits including prefrontal networks. No direct measurement of circuit-specific neural activity during 6-hour anesthesia exposure supports this.
2. **Missing causal link**: The assumed chain CREB→BDNF→TrkB→CD46/CD55 upregulation is speculative. While BDNF can regulate complement genes in some contexts, direct evidence that TrkB signaling controls membrane complement regulators (CD46/CD55) at synapses is absent.
3. **Confounding interpretation of TrkB deletion**: The proposed TrkB flox deletion in cfos+ neurons will cause widespread neuronal dysfunction unrelated to complement regulation—developmental TrkB loss causes substantial neuronal death. Effects on synapse loss may be secondary to impaired neuronal health.
### Counter-Evidence
- Sevoflurane anesthesia suppresses hippocampal BDNF expression (PMID: 30735622)
- The "differential circuit sparing" model conflicts with neuroimaging evidence showing global hippocampal and cortical suppression during prolonged volatile anesthesia
- CD46/CD55 expression may be more constitutive in neurons than activity-dependent
### Falsifying Experiments
1. **Direct activity measurement**: Use fiber photometry with genetically encoded calcium indicators targeting CA1 pyramidal neurons AND parvalbumin interneurons during 6-hour sevoflurane exposure. If both show equivalent suppression, the selectivity mechanism cannot involve activity-dependent protection.
2. **Bidirectional modulation**: Test whether direct pharmacologic activation of TrkB (with BT13 or similar) during anesthesia actually preserves synapses AND upregulates CD46/CD55. Currently only correlative evidence exists.
3. **Synapse-specific complement regulator measurement**: Isolate synaptoneurosomes from behaviorally "spared" vs. vulnerable circuits and perform quantitative mass spectrometry for CD46/CD55. If complement regulators don't differ, this hypothesis fails.
### Revised Confidence: **0.58** (↓0.20)
The core assumption—that spared circuits have sufficient activity to maintain CREB-BDNF signaling—is likely false for prolonged anesthesia. The hypothesis conflates developmental activity-dependent synaptic protection with anesthesia-induced selective loss without establishing parallel mechanisms.
---
## Hypothesis 2: Astrocyte MFGE8/NPTX2 Heterogeneity
### Weak Links
1. **MFGE8 role is bidirectional**: MFGE8 bridges synapses to microglia via αvβ5, **promoting** phagocytosis. The mechanism claims reactive astrocytes downregulate MFGE8 to create "don't-eat-me" signals—but decreased MFGE8 would reduce synaptic tagging, not create active protection signals.
2. **NPTX2 function misapplied**: NPTX2 promotes excitatory synapse formation onto parvalbumin interneurons during stress (PMID: 29230024). This is a presynaptic organizer, not a "eat-me" signal for C1q binding. No evidence links NPTX2 to complement-mediated elimination.
3. **No mechanistic coupling**: The hypothesis proposes simultaneous MFGE8 downregulation AND NPTX2 upregulation but provides no molecular pathway connecting these events. They appear arbitrarily paired.
4. **Spatial transcriptomics resolution**: 10x Visium cannot resolve synapse-level events—RNA from a spot captures multiple astrocyte subtypes and their surrounding neuropil. Correlation between astrocyte cluster and synaptic C1q density at this resolution would be ecologically valid but mechanistically uninterpretable.
### Counter-Evidence
- MFGE8 is generally **protective** for synapses under inflammatory conditions; its loss correlates with excessive pruning in development
- Reactive astrocytes in different brain regions show heterogeneous responses, but no evidence supports a specific "suppressive MFGE8 + inductive NPTX2" pattern
- NPTX2 knockout mice show behavioral phenotypes but no complement pathway involvement has been demonstrated
### Falsifying Experiments
1. **Test the MFGE8 direction**: Conditional deletion of Mfge8 in astrocytes should **reduce** synaptic C1q deposition if MFGE8 promotes tagging. If it increases synaptic loss, then MFGE8 may normally suppress complement—but this contradicts existing literature.
2. **Measure NPTX2 directly**: Does anesthesia actually increase NPTX2 in vulnerable circuits? ELISPOT or microdialysis measurements should be performed before including this as a mechanism.
3. **Dissociate astrocyte effects**: Use Aldh1l1-Cre;Mfge8-flox mice but measure synapse loss specifically in NPTX2-positive vs. -negative circuits. If astrocyte MFGE8 loss affects both equally, NPTX2 involvement is excluded.
### Revised Confidence: **0.45** (↓0.27)
This hypothesis has the lowest coherence—the MFGE8 mechanism is inverted, the NPTX2 mechanism is misapplied, and no causal pathway connects them. The spatial transcriptomics prediction, while technically feasible, cannot resolve synapse-level selectivity.
---
## Hypothesis 3: Neuronal MHC-I/LilrB2 Targeting
### Weak Links
1. **Adult neurons downregulate MHC-I**: Neuronal MHC-I expression is prominent during development and low/absent in healthy adult CNS. Whether anesthesia-induced ER stress is sufficient to reactivate surface MHC-I in adult mice is unestablished.
2. **Dissociation from complement pathway**: The MHC-I/LilrB2 mechanism is presented as facilitating C1q-opsonized synapse internalization, but LilrB2/PirB-mediated pruning is documented as **complement-independent**. This hypothesis conflates two distinct pruning pathways.
3. **Species mismatch**: H2-Kb/H2-Db are mouse MHC-I orthologs; human relevance would require demonstrating HLA-ABC expression on human neurons, which is rarely observed.
### Counter-Evidence
- Neuronal MHC-I is largely retained intracellularly in adult CNS, with surface expression limited to specific conditions (viral infection, autoimmune disease)
- The paper referenced (PMID: 20048153) describes developmental pruning, not pathological anesthesia-induced loss
- ER stress markers may be epiphenomena unrelated to MHC-I trafficking
### Falsifying Experiments
1. **Direct surface MHC-I measurement**: Use cell-surface biotinylation combined with flow cytometry to quantify surface (not total) H2-Kb on hippocampal neurons 24h post-anesthesia. If surface levels don't increase, the mechanism fails.
2. **Isolate LilrB2 requirement**: Test Lilrb4-/- mice. If anesthesia-induced synapse loss persists despite LilrB2 deletion, this pathway is not essential.
3. **Overexpression controls**: The CRISPR-dCas9 activation approach needs careful validation—H2-Kb overexpression alone may cause ER stress and indirect synaptic effects unrelated to microglial recognition.
### Revised Confidence: **0.62** (↓0.13)
This hypothesis is mechanistically plausible but conflates developmental and pathological pruning pathways. The critical test is whether adult neurons actually express surface MHC-I in response to anesthesia.
---
## Hypothesis 4: Complement Regulators CD55/CD46
### Weak Links
1. **C1q binding can be complement-independent**: C1q binds directly to synapses through its globular domains recognizing surface patterns, independent of complement cascade initiation. Synapses lacking CD55/CD46 could still be opsonized if they have alternative C1q recognition sites.
2. **Peptide delivery uncertainty**: Membrane-permeable peptides containing CD55 domains will incorporate into all neuronal membranes, not specifically into vulnerable synapses. Off-target effects in inhibitory neurons could cause circuit-level dysfunction.
3. **Unknown baseline distribution**: Whether excitatory synapses on CA1 distal dendrites actually express lower CD55/CD46 than inhibitory synapses has not been demonstrated.
### Counter-Evidence
- The reference (PMID: 31611251) demonstrates CD55 protection but in the context of excitotoxicity, not anesthesia-induced complement activation
- C1q binding studies show pattern recognition is primary; complement regulator protection may be secondary
- Global complement regulator enhancement could impair necessary synaptic remodeling and immune surveillance
### Falsifying Experiments
1. **Baseline comparison**: Perform quantitative immunofluorescence for CD55 and CD46 on CA1 excitatory vs. parvalbumin inhibitory synapses in adult mice, normalized to PSD-95 vs. gephyrin. If expression is equivalent, differential susceptibility cannot be explained by this mechanism.
2. **Peptide selectivity test**: The myristoylated CD55 peptide will incorporate into all membranes. To test selectivity, compare synaptic C1q density in CD55-high vs. CD55-low regions after peptide treatment. If both decrease equivalently, the mechanism is non-specific.
3. **C3a dependence**: Test whether the protective effect of CD55 enhancement requires C3aR1. If C3aR1-/- mice still show protection, the mechanism bypasses complement amplification.
### Revised Confidence: **0.70** (↓0.10)
Despite modest evidence, this hypothesis has the strongest translational potential because complement regulators can be targeted pharmacologically and it directly addresses the biochemical selectivity question. The main weakness is that C1q binding may occur independently of complement regulation