Different dysregulations of CYFIP1 and CYFIP2 in distinct types of dementia

In humans, the cytoplasmic FMR1 interacting protein (CYFIP) family consists of two members, namely CYFIP1 and CYFIP2. Both CYFIP1 and CYFIP2 function in the WAVE regulatory complex (WRC), which regulates actin polymerization. Additionally, these two proteins form a posttranscriptional regulatory complex with the fragile X mental retardation protein (FMRP), which suppresses mRNA translation. Thus, CYFIP1 and CYFIP2 are important signalling regulators at synapses, and mutations in their genes are associated with neurodevelopmental and neuropsychiatric disorders, including intellectual disabilities. Moreover, dysregulation of the CYFIP protein family is involved in Alzheimer ’ s disease (AD). However, the relevance of the CYFIP family in other dementias is largely unknown. Here, we compared CYFIP1/2 protein levels in the post-mortem hippocampus from patients with AD, dementia with Lewy bodies (DLB), vascular dementia (VaD) and frontotemporal lobar degeneration with TDP-43 pathology (FTLD-TDP). Consistent with previous findings, CYFIP2 was reduced in AD hippocampus. In DLB and VaD hippocampus, the protein level of CYFIP2 and CYFIP1 was unaltered. Finally, an increase in the protein level of both CYFIP1 and CYFIP2 was noted in FTLD-TDP hippocampus. These findings reveal that the protein levels of the CYFIP family is distinct in different types of dementia, suggesting that the pathogenesis of these neurodegenerative disorders has divergent impacts on hippocampal synaptic function.


Introduction
The human CYFIP (cytoplasmic FMR1 interacting protein) family consists of two paralogous proteins: CYFIP1 and CYFIP2.The primary function of CYFIP1 is to interact with RAC1, a member of the Rho small GTPases (Kobayashi et al., 1998).Subsequent studies revealed that CYFIP1 is also component of the WAVE regulatory complex, which is a ~400-kDa heteropentameric complex also containing WAVE1/2/3, abl interactor-1/2 (ABI1/2), Nck-associated protein 1 (NCKAP1), and haematopoietic stem/progenitor cell protein 300 (HPSC300) components (Takenawa and Suetsugu, 2007).The role of CYFIP1 is to maintain the WAVE complex in an inactivated state.Upon CYFIP1 dissociation, the WAVE complex facilitates actin cytoskeleton remodelling via the Arp2/3 complex (Chen et al., 2010;Eden et al., 2002;Kim et al., 2006), promoting dendritic spine remodelling and excitability (Yan et al., 2016).CYFIP1 also interacts with fragile X mental retardation protein (FMRP), forming a posttranscriptional regulatory complex at the synapse (Napoli et al., 2008).FMRP-bound CYFIP1 acts as a non-canonical eIF4E-binding protein, preventing the recruitment of translation initiation complex (eIF4E-eIF4G) to the eIF4E (Marcotrigiano et al., 1999;Richter and Sonenberg, 2005).Consequently, eIF4E-CYFIP1-FMRP complex represses the translation of target mRNA at dendritic and synaptic sites.Upon synaptic activation via tropomyosin receptor kinase B or group I mGluRs, eIF4E is released from CYFIP1-FMRP and permits the translation of target mRNAs (Napoli et al., 2008).Additionally, phosphorylation of CYFIP1 via the mitogen-activated protein (MAP)-kinase-interacting kinase-dependent pathway resulted in the translation in the early phase of long-term potentiation (LTP) (Panja et al., 2014).
Alzheimer's disease (AD) is a neurodegenerative condition characterised by progressive memory impairment, and synapse and neuronal loss (Darnell et al., 2011).The histopathological hallmarks of AD are amyloid β (Aβ) senile plaques and neurofibrillary tangles (NFTs) composed of hyperphosphorylated tau protein (Guo et al., 2017).Previous findings suggest that CYFIPs dysregulation contributes to the synapse loss in AD.CYFIP2 protein expression is significantly reduced in post-mortem hippocampus from patients with severe AD, and also reduced in Tg2576 mice which model a familial form of AD (Tiwari et al., 2016).Exposure of primary neurons to Aβ activates MAPK-interacting kinase 1 (Mnk1), ultimately resulting in reduced CYFIP2-eIF4E interaction and decreased CYFIP2 level, and a significant increase in protein synthesis (Ghosh et al., 2020).Cyfip2 +/-heterozygous knockout mice have increased synaptic APP protein expression and a subsequent increase in Aβ42 in the hippocampus in young-adult age (Tiwari et al., 2016).Amyloid and tau pathology, gliosis, spine loss and more severe memory impairment are developed in aged Cyfip +/-mice (Ghosh et al., 2020).
Here, we investigated CYFIP1 and CYFIP2 protein expression in postmortem hippocampus from patients with different dementia types, including AD, vascular dementia (VaD) and frontotemporal lobar degeneration (FTLD-TDP).We found that CYFIP2, but not CYFIP1 was significantly reduced in severe AD hippocampus when the expression was normalized to a synaptic marker.In DLB and VaD CYFIP1 and CYFIP2 expression was unaltered.Unexpectedly, we found a significant increase of both CYFIP1 and CYFIP2 in FTLD-TDP.

Post-mortem human brain samples
Post-mortem human hippocampal tissue samples were obtained from the London Neurodegenerative Diseases Brain Bank, Institute of Psychiatry, Psychology and Neuroscience, King's College London.The sample set contained tissue from control subjects, and subjects with either AD, DLB, VaD or FTLD (n = 14 for controls, n = 14 for AD, and n = 8 for DLB and FTLD and n = 7 for VaD, Table 1).All human tissue was handled according to Human Tissue Authority regulations and guidelines set out by the London Neurodegenerative Diseases Brain Bank.

Isolation of human brain protein
Proteins were isolated from frozen hippocampal tissue using RIPA lysis buffer (Santa Cruz Biotechnology).Phenylmethane sulfonyl fluoride (PMSF), protease inhibitor cocktail and sodium orthovanadate were added to the buffer in a 1:100 ratio (Santa Cruz Biotechnology), and 0.25% sodium dodecyl sulphate (SDS) was added to the final volume.Tissue was homogenised on ice using a mechanical homogeniser (15 strokes at 700 rpm).Homogenates were centrifuged at 855 g for 10 min, and the supernatants were collected and stored at − 20 • C until use.

5 ×FAD mouse model
5 ×FAD mice were obtained from Paul Francis' lab at the Wolfson CARD, King's College London.Mice were bred and offspring maintained to 3 months-old.The work with wild-type mice and mutant mice was conducted under a Project Licence which was approved by the Home Office after local ethical reviewing.Mice were killed, either by terminal anaesthesia or by a Schedule 1 method as described in the Project Licence.Thus, all mouse procedures were conducted in accordance with the UK Animal Scientific Procedures Act 1986.

Preparation of genomic DNA and genotyping of 5 ×FAD mice
Mouse genomic DNA was extracted using the Extract-N-Amp™ Tissue PCR Kit (XNAT2, Sigma) according to the manufacture's instruction.Approximately a 5 mm piece of mouse tail was placed in a microcentrifuge tube and put into the mixture of 100 µl Extraction solution and 25 µl Tissue Preparation Solution.The tail sample was then incubated at room temperature for 10 min, followed by an incubation at 95 • C for 3 min for a complete digestion.The digestion was terminated by adding 100 µl Neutralization Solution B. Neutralized tissue extracts were stored at 4 • C for subsequent polymerase chain reaction (PCR).
Mice were genotyped blind, using a combination of separate PCR reactions using primer pairs that detect the specific mutant APP and PSEN1 allele.A pair of primers for the internal positive control (interleukin 2) is also included.Primers were designed using sequences provided based on previous literature and purchased from Integrated DNA Technologies (Table 2).

Western Blot Analysis
Thirty micrograms of total protein from human or mouse tissues were diluted in 5x Laemmli sample buffer and heated at 95 • C for 5 min.

Data analysis
Density of protein bands were quantified using NIH ImageJ software.All statistical analyses were conducted in SPSS (v23).Shapiro-Wilk tests of normality indicated that most of the data were normally distributed, and so independent samples t-tests were conducted.When data were not normally distributed, the non-parametric Kruskal Wallis H test was conducted.The ratio of protein to neuronal (NSE) and synaptic (synaptophysin) markers were normalised to control values.Samples were excluded from the analysis if they met the criteria of mean ± 2x standard deviation.Quantification values of the protein bands from all western blot films generated by image J were summarised in Excel spreadsheets, which are available in Supplementary Material.

In AD hippocampus CYFIP2, but not CYFIP1 protein expression is reduced when normalized to a synaptic marker
We first analysed the expression level of CYFIP1 and CYFIP2 protein in AD hippocampus.Hippocampal homogenates of severe AD brain (Braak stage VI) were analysed through western blot using antibodies against CYFIP1 or CYFIP2.Meanwhile, antibodies against synaptophysin or neuron-specific enolase (NSE) were employed as reference markers for synapse and neuron numbers, respectively.Hippocampal samples of age-matched subjects that were not affected by AD or any neurodegenerative disease were employed as control group.In line with previous findings, significant changes in the protein level of CYFIP1, CYFIP2, and synaptophysin were noted in the AD group in comparison to the control group (Fig. 1 and Fig. S1).When normalized to NSE, the protein level of CYFIP2 was reduced by 43% in comparison to the control group (p = 0.0003, unpaired t-test).A 40% decrease in the protein level of CYFIP1 was also observed (p = 0.001, unpaired t-test).These results evidently showed AD pathology caused drastic reduction of CYFIP1 and CYFIP2.In parallel, the level of synaptophysin was 25% lower than in the control group (p = 0.0001, unpaired t-test), indicating the emergence of synapse loss in AD brains (Arendt, 2009).
Notably, the protein level of CYFIP2 was reduced by 36% when normalized to synaptophysin (p = 0.0164, unpaired t-test) whereas such decrease was not seen for CYFIP1 expression (Fig. 1 and Fig. S1).The divergent changing patterns between CYFIP1 and CYFIP2 suggest that CYFIP1 level remains unchanged in the surviving synapses in AD hippocampus, and the reduction of total CYFIP1 is most likely to due to synapse loss.In contrast, depletion of CYFIP2 occurred prior to synapse loss.

Reduced CYFIP2 expression in hippocampus of 5 ×FAD mice
The 5 ×FAD mice are widely used as a model of familial AD featured by rapid development of amyloid pathology.In this AD mouse model,   high level accumulation of intraneuronal Aβ1-42 and extracellular amyloid deposition begins as early as 2 months whereas synaptophysin levels begin to decline by 4 months of age (Oakley et al., 2006), suggesting that synapse loss is a later event than development of Aβ pathology.Hence, we used 3-month-old 5 ×FAD mice to investigate whether changes in the protein levels of CYFIP1 and/or CYFIP2 occur before loss of synapses takes place.In line with the results obtained in human brain, we also discovered that CYFIP2, but not CYFIP1 was reduced in the brains of 5 ×FAD mice.The genotype of the animals are determined by the presence of specific transgene products (Supplement Fig. 3).As shown in Fig. 2, Western blot results of hippocampal homogenates showed that in 3-month-old 5 ×FAD mice there was a significant decrease in the protein level of CYFIP2 by about 73% in comparison to the wild-type mice (p = 0.0034, unpaired t-test).In contrast, no change in the protein level of CYFIP1 was detected.These results were in agreement with those in Fig. 1, indicating depletion of CYFIP2 emerges earlier than synapse loss in AD pathological cascade.

CYFIP2 and CYFIP1 protein expression is unaltered in DLB hippocampus
Next, we investigated the protein levels of CYFIP1 and CYFIP2 in post-mortem DLB hippocampus (Fig. 3).In line with previous findings (Lippa, 2004), synaptophysin protein expression was unchanged in DLB hippocampus when compared to the control group (Fig. 3B, p = 0.412, unpaired t-test).Western blot analysis revealed that CYFIP2 protein expression was unaltered in DLB hippocampus, when normalised to NSE and synaptophysin (p = 0.2407 and 0.4661, respectively, unpaired t-test).Similar results were observed for CYFIP1.These results indicate that CYFIP2 and CYFIP1 expression are not altered in the absence of synapse loss in the hippocampus.

CYFIP1 and CYFIP2 protein expression is unaltered in VaD hippocampus
Similar to the cases of DLB hippocampus, the protein levels of CYFIP1 and CYFIP2 are not compromised in the hippocampus of VaD patients (Fig. 4).CYFIP2 protein level was found to be unchanged in VaD hippocampus when normalised to both NSE and synaptophysin (p = 0.7635 and 0.9909, respectively, unpaired t-test).Meanwhile, synaptophysin level was not altered in VaD hippocampus compared to control (p = 0.6575, unpaired t-test).There was no significant difference in the CYFIP1 level between VaD and control hippocampus when normalised to NSE or synaptophysin (p = 0.3840 and 0.1723 respectively, unpaired t-test).These results suggests that both CYFIP1 and CYIPF2 protein are not affected by VaD pathology.

CYFIP2 and CYFIP1 protein expression is significantly increased in FTLD-TDP hippocampus
Finally, we investigated CYFIP2 and CYFIP1 protein levels in FTLD hippocampus (Fig. 5).Specifically, we analysed FTLD-TDP post-mortem tissue as all cases had TDP-43 pathology.Synaptophysin levels were found to be similar between FTLD-TDP and control hippocampus (p = 0.2825, unpaired t-test).In contrast, we found that CYFIP2 protein expression was significantly increased when normalised to both NSE (p = 0.0161 unpaired t-test) and synaptophysin (p = 0.0225, unpaired t-test).Similarly, we found a significant, two-fold increase in CYFIP1 protein levels in FTLD-TDP hippocampus compared to control, when normalised to NSE (p = 0.0033, unpaired t-test) and synaptophysin (p = 0.0102, unpaired t-test).

Discussion
CYFIP1 and CYFIP2 regulate protein synthesis and dendritic spine morphology (Davenport et al., 2019;Napoli et al., 2008;Tiwari et al., 2016), functions that are important for hippocampal physiology and memory (Sala and Segal, 2014).In the present study, we firstly report that different disease manifestations of dementia have different impact on the protein levels of CYFIP1 and CYFIP2 in post-mortem hippocampus among AD, DLB, VaD and FTLD-TDP.Secondly, even within the same disease condition, CYFIP1 and CYFIP2 displayed divergent changing patterns.In summary, the loss of both synaptophysin and CYFIP1/2 is noted in the hippocampus of AD patients.In DLB patients, the level of all the three proteins remained unchanged when comparing the DLB patients with healthy controls.Similarly, no significant changes in the protein level of the four proteins was observed in VaD patients in comparison to healthy controls.In contrast, elevated protein level of CYFIP1 and CYFIP2 were found in FTLD-TDP cases.
Decreased level of synaptophysin indicated the loss of synapse in late-stage AD patients (Braak stage VI).Decreased synaptophysin levels were also observed, indicating loss of neurons in the AD brains.Depletion of both CYFIP1 and CYFIP2 was exemplified when the protein levels of CYFIP1 and CYFIP2 were normalized to the total neuronal marker, NSE.There is a concomitant reduction in CYFIP1 and synaptophysin, indicating depletion of CYFIP1 occurs with synapse loss synchronously.In contrast, CYFIP2 declines more drastically than synaptophysin, suggesting that the reduction of CYFIP2 is not only a result of synapse loss, but also due to other mechanisms which occur before synapse loss.This notion is supported by previous findings.In the early stages of Alzheimer's disease (Braak stages I-II), CYFIP2 level is significantly decreased whereas CYFIP1 level remained unchanged (Tiwari et al., 2016).Accumulating results pointed out that CYFIP1 and CYFIP2 have different cell-type-expression patterns in mouse brain.While CYFIP1 was detected in both neurons and astrocytes, CYFIP2 displayed a neuron-specific localization (Ma et al., 2022;Zhang et al., 2019).In line with these results, CYFIP2 levels are more susceptible to synapse loss and neuronal death in comparison to CYFIP1, which is supported by our current results.Consequently, only 8 out of 131 proteins in the CYFIP1 interactome overlaps with the interactome of CYFIP2 (Ma et al., 2022).These results suggest CYFIP1 and CYFIP2 can have distinct functions that may be mediated by their different subcellular localizations and/or interactors.Our data provided further evidence by showing both shared and distinct changing pattern CYFIP1 and CYFIP2 in human brains affected by different type if dementias, contributing to improved understanding of the pathophysiology mediated by CYFIP1 and CYFIP2.
It is evident that Aβ exposure is a driving factor of CYFIP2 reduction (Tiwari et al., 2016).In the context of AD, Aβ is suggested to reduce CYFIP2 expression within the neurons in the absence of synapse loss or cell death.Such CYFIP2 reduction will lead to not only changes in synapses morphology, but also translation of mRNA at synapses.Indeed it has been shown that CYFIP2 reduction leads to synapse loss, and causes other key features of AD pathology, including elevated Amyloid-beta precursor protein (APP) expression, amyloid accumulation, hyperphosphorylation of tau and tau accumulation, as well as astrocytosis and elevation of microglia, and including memory impairment (Ghosh et al., 2020).Given the dual role of CYFIP2 in the WAVE and FMRP complex, CYFIP2 depletion can contribute to synapse loss directly through inducing abnormal cytoskeleton organization, and/or indirectly by facilitating Aβ production and Aβ-mediated toxicity.Subsequently, synapse loss resulted in the decrease of other synaptic proteins including CYFIP1 and synaptophysin.
Unlike AD, in FTLD-TDP hippocampus protein levels of CYFIP1 and CYFIP2 are elevated in the absence of significant change in the number of synapses.Previous reports suggested a functional/physical partnership between FMRP and TDP-43 in regulating the translation of a subclass of neuronal proteins.This partnership potentially explains elevated CYFIP1 expression is a consequence of deregulated protein level of TDP-43 (Majumder et al., 2016).Enhanced expression of CYFIP may also be detrimental to neuron function, as excessive amount of CYFIP would overly inhibit the translation of certain mRNAs (Darnell et al., 2011).
Both CYFIP1 and CYFIP2 are unaffected in DLB and VaD.Meanwhile, the level of synaptophysin remained unchanged.Our data from in the current study has exhibited a simultaneous reduction of synaptophysin and CYFIP1 in AD cases.Hence, it is not surprising to observe unaffected CYFIP1 level in the absence of synapse loss.The absence of CYFIP2 reduction further indicates that in DLB and VaD, neuronal function is undermined by different mechanisms from AD and FTLD-TDP.Data from DLB and VaD cases highlighted that the protein level of CYFIP2 is specifically responds to AD pathology.
Together, our study showed a correlation between the protein level of CYFIP1/2 and AD.Reduction of CYFIP1/2 is a pathological feature specific to AD. Particularly, whilst CYFIP1 reduction appears to be a consequence of synapse loss, CYFIP2 is suggested to represent early synaptic dysfunction in response to Aß action.As a conclusion, CYFIP2 holds significant potential to be a specific biomarker for the AD.

Fig. 1 .
Fig. 1.Reduced CYFIP2, but not CYFIP1 expression in post-mortem Alzheimer's disease (AD) hippocampus when normalized to a synaptic marker.(A) Representative Western blots images showing protein levels of CYFIP2, CYFIP1, NSE, and synaptophysin in AD (n = 6) and control (n = 6) hippocampus.(B-F) Statistical analysis of the difference in the protein level of CYFIP1, CYFIP2 and synaptophysin in AD (n = 14) and control (n = 14).(B) Synaptophysin protein levels were significantly reduced, when normalised to NSE, in AD hippocampus compared to control.(C, E) CYFIP2 and CYFIP1 protein levels were significantly reduced in AD hippocampus when normalised to NSE. (D, F) CYFIP2, but not CYFIP1 levels were significantly reduced in AD hippocampus when normalised to synaptophysin.Mean ± standard error of the mean (SEM) is shown.**p < 0.01, ***p < 0.001, ns= not significant.

Fig. 2 .
Fig. 2. CYFIP2, but not CYFIP1 expression is reduced in hippocampus of 3 month-old 5 ×FAD mice.(A) Western blots analysis of crude mouse hippocampal synaptosomal fractions, using antibodies against CYFIP1, CYFIP2, and NSE.Molecular markers are shown on the left.(B) Quantification of the protein level of CYFIP1 normalized to NSE in the wild-type mice and in the 5 ×FAD mice.(C) Quantification of the protein level of CYFIP2 normalized to NSE in the wild-type mice and in the 5 ×FAD mice.Wild-type group is used as the reference for comparison (100%).n = 5 for wild-type animals, and n = 6 for 5 ×FAD mice; bars are mean ± S.E.M; unpaired t test, **p < 0.01, ns= not significant.

Fig. 3 .
Fig. 3. CYFIP2 and CYFIP1 protein expression is unaltered in post-mortem dementia with Lewy bodies (DLB) hippocampus.(A) Western blots showing protein expression of CYFIP2, CYFIP1, NSE and synaptophysin in DLB (n = 7) and control (n = 8) hippocampus.(B) Synaptophysin protein levels, normalised to NSE, were similar in control and DLB hippocampus.(C, E) CYFIP2 and CYFIP1 protein levels were unaltered when normalised to NSE. (D, F) CYFIP2 and CYFIP1 protein levels were also unaltered when normalised to synaptophysin.Mean ± SEM are shown, ns= not significant.

Fig. 4 .
Fig. 4. CYFIP2 and CYFIP1 protein expression is unaltered in post-mortem vascular dementia (VaD) hippocampus.(A) Western blots showing protein expression of CYFIP2, CYFIP1, NSE, and synaptophysin in VaD (n = 7) and control (n = 7) hippocampus.(B) Synaptophysin protein levels were unaltered in VaD hippocampus when normalised to NSE. (C, E) CYFIP2 and CYFIP1 protein levels were not significantly altered when normalised to NSE. (D, F) CYFIP1 and CYFIP2 were unaltered in VaD hippocampus when normalised to synaptophysin.Mean ± SEM are shown, ns= not significant.

Table 1
Brain Samples for Western Blot Analysis.
M31 † Excluded from data analysis using the criterion mean ± 2 standard deviations X. Peng et al.

Table 2
Primers Used for Genotyping.