**Background:** Fear extinction in adult animals typically does not erase the original fear memory but forms a new inhibitory memory that can decay (spontaneous recovery) or re-emerge (renewal, reinstatement). In juvenile animals, extinction leads to permanent erasure. The transition from erasable to persistent fear memory is thought to involve maturation of brain circuits, particularly the condensation of chondroitin sulfate proteoglycans (CSPGs) into perineuronal nets (PNNs) around neurons. Enzymatic removal of CSPGs in the amygdala can reinstate juvenile-like erasure in adults. However, it was unknown whether the specific aggregation of CSPGs into PNNs (driven by the cartilage link protein Crtl1/HAPLN1) is necessary for this transition. This study used Crtl1-knockout (Crtl1-KO) mice, which have attenuated PNNs but normal total CSPG levels, to test whether preventing PNN formation is sufficient to make fear memories susceptible to erasure, and to examine the underlying neural activation patterns.
**Methods:** Adult Crtl1-KO and wild-type (Crtl1-WT) littermates (P75-P120) were subjected to auditory cued fear conditioning (5 pairings of CS: 10 s, 7.5 kHz, 80 dB; US: 1 s footshock, 0.6 mA) in context A. Extinction training occurred on days 2 and 3 in context B (12 unreinforced CS presentations per day). Freezing was scored manually. Spontaneous recovery and fear renewal were tested 7 and 42 days after late extinction using 4 CS presentations in context B and A, respectively. A separate cohort underwent extinction starting 9 days after conditioning to rule out memory consolidation deficits. For pupillometry, head-fixed mice underwent virtual fear conditioning (visual CS, tail shock US) in a custom apparatus; pupil diameter was recorded with an infrared camera and analyzed using MEYE deep learning. To assess neuronal activation, a separate group of conditioned and pseudo-conditioned (unpaired CS and US) mice were perfused 1 hour after early extinction (10 CS presentations) and brain sections were immunostained for Zif268 (immediate early gene). Zif268+ cells were counted in lateral amygdala (LA), basolateral amygdala (BLA), central amygdala medial (CeM) and lateral (CeL) divisions, and infralimbic cortex (IL). Statistical analyses used ANOVA, RM-ANOVA, and post-hoc Sidak tests; significance set at p<0.05.
**Key Results:** During early extinction, Crtl1-KO mice showed significantly accelerated freezing reduction compared to Crtl1-WT, with lower freezing from the 3rd block of 2 CS onward (genotype p<0.001; interaction p<0.001). Crtl1-KO significantly reduced freezing from block 1 to 6 (p<0.001), while Crtl1-WT did not (p=0.836). During late extinction, Crtl1-KO maintained lower freezing through block 4 (p<0.05 to p<0.001), and both genotypes reached comparable low levels by block 5. Pupillometry confirmed faster extinction in Crtl1-KO: pupillary response to CS was significantly lower in Crtl1-KO than Crtl1-WT during the second block of 5 CS on early extinction (p=0.010) and first block of late extinction (p<0.01). Spontaneous recovery and fear renewal at 7 and 42 days post-extinction showed persistently lower freezing in Crtl1-KO across all CS presentations (p<0.001 for all comparisons at 7 days; at 42 days, renewal was lower for CS1 and CS2, p<0.001). Context-dependent freezing in the conditioned context was also lower in Crtl1-KO (p<0.001 at both time points). No differences in freezing were observed in the unconditioned context. When extinction started 9 days after conditioning, Crtl1-KO again showed accelerated freezing reduction (genotype p<0.01; interaction p<0.001), with significantly lower freezing from block 3 onward (p<0.001) and significant reduction from block 1 to 5 (p<0.001), while Crtl1-WT did not (p=0.520). Zif268 analysis after early extinction revealed that conditioned Crtl1-KO mice had significantly fewer Zif268+ cells in LA (p<0.001 vs. conditioned Crtl1-WT), BLA (p<0.001), and CeM (p<0.001), and their activation levels were indistinguishable from pseudo-conditioned mice (LA: p=0.858 vs. pseudo-WT; CeM: p>0.99 vs. pseudo-WT). In contrast, conditioned Crtl1-WT showed significantly higher Zif268+ density in these regions compared to pseudo-conditioned controls. CeL activation was elevated in both conditioned groups vs. pseudo-conditioned (p<0.001 for both genotypes). IL activation did not differ significantly between conditioned Crtl1-KO and Crtl1-WT (p=0.124), though both showed higher activation than pseudo-conditioned (WT: p<0.05; KO: p=0.058, trend). After memory recall (one block of 2 CS), Crtl1-KO showed significantly higher Zif268+ cells in LA (p=0.001), CeM (p<0.0001), CeL (p<0.0001), BLA (p=0.006), and IL (p<0.0001) compared to Crtl1-WT.
**Clinical Implications:** This study demonstrates that selective disruption of PNNs via Crtl1 deletion is sufficient to render adult fear memories susceptible to erasure by extinction, mimicking juvenile plasticity. The accelerated and persistent reduction of fear responses, along with the absence of amygdala activation after extinction, suggests that PNNs normally protect fear memories from erasure by stabilizing amygdala circuits. These findings highlight the extracellular matrix, particularly PNNs, as a potential therapeutic target for anxiety disorders such as PTSD and phobias, where maladaptive fear memories are resistant to extinction. The use of pupillometry as a non-aversive physiological readout of fear extinction in mice also offers a translational tool for assessing treatment efficacy. Future research should explore whether pharmacological or genetic manipulation of PNN components can safely enhance extinction-based therapies in humans.