Cathepsin B Deficiency Improves Memory Deficits and Reduces Amyloid-β in hAβPP Mouse Models Representing the Major Sporadic Alzheimer’s Disease Condition

The lysosomal cysteine protease cathepsin B (CTSB) has been suggested as a biomarker for Alzheimer’s disease (AD) because elevated serum CTSB in AD patients has been found to correlate with cognitive dysfunction. Furthermore, CTSB gene knockout (KO) in non-transgenic and transgenic AD animal models showed that elimination of CTSB improved memory deficits. However, conflicting CTSB KO results on amyloid-β (Aβ) pathology in transgenic AD models have been reported. The conflict is resolved here as likely being due to the different hAβPP transgenes used in the different AD mouse models. CTSB gene KO reduced wild-type (Wt) β-secretase activity, brain Aβ, pyroglutamate-Aβ, amyloid plaque, and memory deficits in models that used cDNA transgenes expressing hAβPP isoform 695. But in models that used mutated mini transgenes expressing hAβPP isoforms 751 and 770, CTSB KO had no effect on Wt β-secretase activity and slightly increased brain Aβ. All models expressed the AβPP transgenes in neurons. These conflicting results in Wt β-secretase activity models can be explained by hAβPP isoform specific cellular expression, proteolysis, and subcellular processing. CTSB KO had no effect on Swedish mutant (Swe) β-secretase activity in hAβPP695 and hAβPP751/770 models. Different proteolytic sensitivities for hAβPP with Wt versus Swe β-secretase site sequences may explain the different CTSB β-secretase effects in hAβPP695 models. But since the vast majority of sporadic AD patients have Wt β-secretase activity, the CTSB effects on Swe β-secretase activity are of little importance to the general AD population. As neurons naturally produce and process hAβPP isoform 695 and not the 751 and 770 isoforms, only the hAβPP695 Wt models mimic the natural neuronal hAβPP processing and Aβ production occurring in most AD patients. Significantly, these CTSB KO findings in the hAβPP695 Wt models demonstrate that CTSB participates in memory deficits and production of pyroglutamate-Aβ (pyroglu-Aβ), which provide rationale for future investigation of CTSB inhibitors in AD therapeutics development.


INTRODUCTION
Evidence for participation of the lysosomal cysteine protease cathepsin B (CTSB) in Alzheimer's disease (AD) memory deficits has been provided by human clinical studies. In AD patients, increases in serum CTSB correlate with the extent of cognitive dysfunction [1]. CTSB is also elevated in brain [2] and cerebrospinal fluid (CSF) [3][4][5] of AD patients. Notably, CTSB accumulates with amyloid plaques in human AD brain [6].
These human AD findings of CTSB upregulation have been investigated in AD animal models by CTSB gene knockout, which demonstrate that CTSB participates in memory deficits [7,8]. Studies have also assessed amyloid-␤ (A␤) in several variant human amyloid-␤ protein precursor (hA␤PP) transgenic mouse models with conflicting results regarding A␤ production or degradation by cathepsin B. Therefore, the purpose of this review is to provide detailed analysis of the compiled CTSB data to understand its role in AD with respect to 1) elevation of CTSB in AD patients with the major sporadic population, 2) gene knockout of CTSB in AD animal models resulting in improved memory deficits, 3) participation of CTSB in A␤ production in hA␤PP models with wild-type (Wt) ␤-secretase site representing the major sporadic AD population, but not in hA␤PP models with the Swe mutant ␤-secretase site representing a minor portion of AD, and 4) the overall consilience of data showing participation of CTSB in memory deficits and A␤ production in hA␤PP models representing the major sporadic AD population.

ELEVATION OF CTSB IN AD PATIENTS
Numerous studies have demonstrated increased levels of CTSB in AD patients (Table 1). CTSB protein in the temporal cortex of human AD brains was increased by 80% compared to age-matched controls [2]. CTSB has been shown to accumulate in amyloid plaques in human AD brains [6]. CTSB protein levels in serum and plasma increased by 50% compared to controls (age-matched) [1,4]. Significantly, high CTSB levels in serum were strongly correlated with cognitive decline in AD patients [1]. Also, in CSF, CTSB protein levels were greater in AD patients compared to controls by proteomics and western blot assessments [3][4][5]. In chronic periodontitisassociated AD patients, CTSB in serum was higher than controls by 43% [9]. Notably, the higher levels of serum CTSB corelated with reduced Mini-Mental State Examination scores of cognitive function in these periodontitis AD patients [9]. These findings demonstrate upregulation of CTSB in brain and peripheral serum or plasma of AD patients.

CTSB GENE KNOCKOUT RESULTS IN IMPROVED MEMORY DEFICITS IN HUMAN A␤PP ANIMAL MODELS OF AD
CTSB deficiency by gene knockout (KO) resulted in improved memory deficits in a transgenic AD model expressing hA␤PP-695 [7] and in a non-transgenic chronic-periodontitis-associated AD model [8]. Improved memory deficits were associated with reduced A␤ levels by CTSB KO.

CTSB KO in the hAβPP-695/Wtβ-Lonγ model of AD improves memory deficits
The hA␤PP model of AD, expressing hA␤PP-695/Wt␤-Lon␥, displays memory deficits and amyloid plaque brain pathology [7]. CTSB KO in these AD mice resulted in significant improvement in memory deficits to nearly normal memory function, assessed by the Morris water maze test [7]. Improved memory by CTSB KO was also indicated by the increased time that mice spent in the quadrant with the submerged platform that the mice were trained to recall. Amyloid plaque pathology was significantly decreased by CTSB KO in these AD mice and was accompanied by decreased brain levels of A␤ 1-40 , A␤ 1-42 , pGlu-A␤  , and pGlu-A␤ 3-42 peptides [7,10]. These CTSB KO data indicate participation of CTSB in AD memory deficits and production of A␤ peptides [7].
In contrast, CTSB KO in a Swedish (Swe) mutant AD model expressing hA␤PP-695/Swe␤-Lon␥ had no effect on memory deficits [7]. The Swe mutant represents a minor portion of AD patients from one family [11]. CTSB KO also had no effect on A␤ peptide levels in the hA␤PP-695/Swe␤-Lon␥ mice.
These combined CTSB KO data in different hA␤PP models are consistent with the hypothesis that CTSB participates in regulating A␤ produced from hA␤PP with the Wt ␤-secretase site representing the major sporadic AD population, but not from hA␤PP with the Swe mutant ␤-secretase site that represents a minor percentage of AD. Further data in the field of hA␤PP models with the Wt or Swe mutant ␤-secretase sites in CTSB KO studies is discussed in this review.

CTSB KO in chronic periodontitis-associated AD model improves memory deficits
Infection by Prophyromonas gingivalis, the major periodontal bacteria, has been shown to be positively linked to AD and cognitive dysfunction [12,13]. Prophyromonas gingivalis lipopolysacharide (PgLPS) has been found in human AD brain [14]. Notably, cognitive deficits of periodontitis patients correlate with increased levels of cathepsin B [9]. Therefore, cathepsin B knockout in a mouse model of periodontitis was assessed for predicted improvements in memory deficits [8]. Significant results showed that CTSB participates in PgLPS-induced periodontitis and memory deficits [8]. KO of the CTSB gene in the periodontitis model of AD (PgLPS mice) resulted in significant improvements in memory deficits in middle-aged mice of 12 months old, but not in young 2-month-old mice, treated with PgLPS for 5 weeks. The PgLPS induction of CTSB in brain hippocampus was consistent with its blockade by CTSB gene KO that alleviated memory deficits. CTSB KO blocked PgLPS-induced activation of the inflammatory factors IL-1␤ and toll-like receptor 2. CTSB KO also blocked PgLPS-induced increases in A␤ 42 in brain. These data illustrate participation of CTSB in memory deficits, inflammation, and A␤ production from Wt A␤PP in the PgLPS model of periodontitisassociated AD memory deficits [8]. KO IN HA␤PP  MOUSE MODELS: CONFLICTING DATA  EXPLAINED BY VARIANT HA␤PP  ISOFORM MODELS, TRANSGENES,  NEURONAL VERSUS GLIA EXPRESSION,  AND WT OR SWE MUTANT  ␤-SECRETASE SITE OF HA␤PP CTSB KO studies in AD mouse models have investigated regulation of A␤ [7,10,[15][16][17]. The Hook group showed that CTSB participates in A␤ production, shown by reduced A␤ levels in CTSB KO AD mice compared to controls [7,10,15]. However, the Gan group suggests that CTSB participates in A␤ degradation, shown by small increases in A␤ in CTSB KO mice in different A␤PP models [16,17].

A␤ REGULATION BY CTSB
These results by the Hook and the Gan groups appear to be conflicting, but the different findings can be explained by differences in hA␤PP isoforms expressed in the mouse models. These models differ in hA␤PP-695 or hA␤PP-751/770 isoforms expressed, transgenes, alternative RNA splicing that generates hA␤PP isoforms, neuronal compared to glia expression of hA␤PP, and Wt compared to Swe mutant ␤-secretase sites of hA␤PP isoforms.
To provide the field with an understanding of these different hA␤PP models, detailed analysis of these variant hA␤PP models is explained in the next section "AD models expressing variant human A␤PP isoforms." Then, evaluations are provided for CTSB KO in Wt ␤-secretase site hA␤PP models representing the major sporadic AD population [18], and CTSB KO in Swe mutant ␤-secretase site hA␤PP models which represent a minor portion of AD patients in one family [19,20]. Overall, data from hA␤PP models representing the major AD sporadic condition show that CTSB KO results in decreased levels of brain A␤ peptides.

AD models expressing variant hAβPP isoforms
The six different hA␤PP models of AD used in CTSB KO studies differed in hA␤PP isoforms, gene constructs, promoter driven expression for natural or abnormal expression of hA␤PP isoforms in neurons compared to glia, and hA␤PP containing Wt or Swe mutant ␤-secretase sites. These features of the hA␤PP mouse models are summarized in Table 2. These models consisted of four different hA␤PP-695 isoform mouse models used by the Hook group [7,10,15], and two variant hA␤PP-751/770 models used by the Gan group [16,17]. Details of the different features of these variant hA␤PP models are provided in the following sections.

Normal hAβPP gene transcription and alternative splicing generates hAβPP isoforms of 695, 751, and 770 residues
The hA␤PP gene is composed of 18 exons that undergo alternative splicing to generate isoforms of hA␤PP-695, hA␤PP-751, and hA␤PP-770 (Fig. 1). These three hA␤PP isoforms differ in their amino acid lengths of 695, 751, and 770 residues, which are expressed in brain at approximate relative ratios of 20:10:1 [21] (Fig. 1). All hA␤PP isoforms contain the A␤ domain. The hA␤PP-751 and hA␤PP-770 isoforms contain the kunitz protease inhibitor (KPI) domain. The hA␤PP-770 isoforms also include the Ox-2 domain.

Different hAβPP isoforms and transgene constructs of AD models used in CTSB KO studies by the Hook group compared to the Gan group
AD models used by the Hook and Gan groups differ in hA␤PP isoforms expressed (Fig. 2). The Hook group utilized hA␤PP-695 models expressing the cDNA of hA␤PP-695 for direct mRNA expression and protein translation, without RNA splicing ( Fig. 2a) [7,10,15]. This model represents hA␤PP-695 as the major hA␤PP isoform in brain [21,24].
In the hA␤PP transgenes used by the Hook group (Table 2) [7,10,15] the PDGF promoter drives neuronal expression of hA␤PP-695 ( Fig. 3b.i) which represents its normal endogenous neuronal cell localization ( Fig. 3a.i). Thus, the four PDGF hA␤PP-695 models used by the Hook group (listed in Table 2) study normal hA␤PP-695 expression and function in neurons.
However, the PDGF promoter in the minigene used by the Gan group (Table 2) resulted in abnormal neuronal expression of hA␤PP-751/770 (Fig. 3b.ii) as these isoforms are normally expressed in glia cells (Fig. 3a.ii). Thus, the PDGF hA␤PP-751/770 models (listed in Table 2) represent abnormal expression of hA␤PP-751/770 in neurons. The processing of hA␤PP-751/770 in the abnormal neuronal location will likely differ from glia cells, since each cell type has its distinct trafficking and proteolytic systems.
Human AβPP models with the Wt β-secretase site represent the major sporadic AD population, while hAβPP isoforms with the Swe mutant β-secretase site represent a small number of AD patients from one family Models expressing hA␤PP result in its proteolytic processing to generate A␤ in brain. A␤ is generated from A␤PP by proteolytic cleavage at the ␤-secretase and the ␥-secretase sites that flank A␤ (Fig. 1).
The Wt ␤-secretase of hA␤PP is expressed by the major sporadic AD population [18]. Thus, models expressing hA␤PP with the Wt ␤-secretase site are important to gain understanding of the major sporadic AD population possessing no known mutations.
The Swe mutant hA␤PP is expressed in one AD family and represents a very small number of AD patients [11]. Studies of the Swe mutation of hA␤PP in mouse models have been of interest to understand mechanisms of elevated A␤ and impaired memory that occurs in AD. The Swe hA␤PP mutation is  [22][23][24], and the hA␤PP-751/770 isoforms are normally expressed in glia cells [27,28]. (b) Transgenic expression of hA␤PP-695 in its normal neuronal cell type, but abnormal expression of hA␤PP-751/770 in neurons. Expression of the hA␤PP-695 under the control of the PDGF promoter results in expression in neurons, the normal cell type for this isoform as conducted by the Hook group [7,10,15]. But PDGF driven expression of hA␤PP-751/770 results in abnormal expression in neurons [16,17,23], rather than in the normal location of glia cells [27,28]. relevant to the AD family possessing this inherited genetic mutation associated with AD [11].
Models utilized in CTSB KO studies consisted of those expressing hA␤PP-695 and hA␤PP-751/770 with the Wt and the Swe mutant ␤-secretase sites ( Table 2). Human A␤PP models with the Wt ␤secretase site provide analysis of A␤ production by Wt ␤-secretase activity. The Wt ␤-secretase activity represents the majority of the AD population.

Models with γ-secretase mutations of hAβPP display amyloid plaques and memory deficits
Human A␤PP contains the ␥-secretase site sequence which is cleaved after ␤-secretase cleavage by the ␥-secretase complex to produce A␤. While most AD patients express hA␤PP with the Wt ␥sequence, familial mutations near this site occurs and hA␤PP models with such familial ␥-sequence site mutations overproduce A␤ and develop amyloid plaque with memory deficits but have Wt ␤-secretase activity. These models provide assessment of A␤, amyloid plaques, and memory deficits.

CTSB KO reduced Aβ and amyloid plaques in the hAβPP-695/Wtβ-Lonγ AD mice
In the hA␤PP-695/Wt␤-Lon␥ AD model, CTSB KO substantially reduced A␤ peptide levels and amyloid plaque pathology [7,10]. Absence of CTSB resulted in decreased brain A␤ 40 and A␤ 42 by ∼85% and reduced amyloid plaques by ∼85%. CTSB KO reduced the ␤-secretase product CTF␤ by 60% and increased the non-amyloidogenic product sA␤PP␣ by 60% [7]. CTF␤ and sA␤PP␣ are biomarkers of ␤-secretase activity; their changes that occurred in the CTSB KO condition indicate reduced ␤-secretase activity. Furthermore, CTSB expression increased brain levels of A␤ 40 and A␤ 42 by 150% and 200%, respectively. compared to controls of 100% [10]. These data demonstrate participation of CTSB in the upregulation of ␤-secretase activity for A␤ production.
Additional evidence for CTSB KO reduction of A␤ was demonstrated in mice expressing Wt hA␤PP695 having no mutations (hA␤PP695Wt) [15]. The absence of CTSB resulted in reduced human A␤ 40 and A␤ 42 by 70%, reduced CTF␤ by 40%, and increased sA␤PP␣ by 160% compared to controls of 100% [15]. These results provided further support for CTSB participation in A␤ production in brain.

CTSB KO in the hAβPP-751/770/Wt model resulted in no change in hippocampal Aβ and a small elevation in cortical Aβ in mouse brain
While CTSB KO in human A␤PP-695/Wt␤-Lon␥ AD mice resulted in lowered A␤ species [7,10,15], CTSB KO in hA␤PP-751/770/Wt mice had no effect on hippocampal A␤ 42 and increased A␤ 1-x by 18% above controls, and resulted in small increases in A␤ 42 and A␤ 1-x in brain cortex of 20% and 24% above controls [17]. Further studies examined consequences of elevating CTSB by overexpression or by deleting cystatin C, an endogenous inhibitor of CTSB. These conditions of increased CTSB in hA␤PP-751/770Wt mice resulted in no change in A␤ 1-x and decreased A␤ 42 by 12% in hippocampus compared to controls. Increased CTSB had no effect on levels of CTF␤, CTF␣, and sA␤PP␣. These data in hA␤PP-751/770 mice suggest in these conditions, CTSB may be involved in degradation of brain A␤ [17].
Significantly, memory function was not assessed [17] and, thus, findings of the relationship of small increases in A␤ resulting from CTSB KO with memory function are unknown.
It is important to utilize the model that best represents the normal production of A␤ in neurons from hA␤PP-695. Therefore, the hA␤PP-695/Wt␤-Lon␥ model logically represents normal endogenous production of A␤ in neurons involving CTSB.

CTSB KO in hAβPP-695 mice with Swe mutant β-secretase site had no effect on Aβ, but CTSB KO in hAβPP-751/770 mice with Swe mutation resulted in a small elevation of Aβ
As discussed above, CTSB KO in hA␤PP-695 models with the Wt ␤-secretase site resulted in reduced A␤ and reduced Wt ␤-secretase activity [7,10,15]. But in hA␤PP-695 models with the Swe mutant ␤-secretase site, CTSB KO had no effect on A␤ levels [8,15]. These different results may be due to CTSB having or effecting cleavage of the Wt ␤secretase site, but not the Swe ␤-secretase site.
In the hA␤PP-751/770 model with the Swe mutant ␤-secretase site, CTSB KO resulted in a small elevation of A␤ [16]. The reason for the no effect versus slight increase in A␤ between the Swe mutant hA␤PP-695 models versus Swe mutant hA␤PP-751/770 is unclear but may result from the abnormal neuronal hA␤PP-751/770 expression.

CTSB KO in hAβPP-695/Sweβ-Lonγ AD mice had no effect on Aβ or Swe mutant β-secretase activity
The Swe mutation consists of Asn-Leu instead of the normal Lys-Met amino acid sequence at the ␤-secretase site of hA␤PP [11]. Studies of CTSB KO in Swe mutant hA␤PP mice expressing hA␤PP-695/Swe␤-Lon␥, which mimicked normal neuronal expression, showed no effects on A␤ 40 , A␤ 42 , CTF␤, sA␤PP␣, and memory deficits [7]. These data showed that CTSB was not involved in production of A␤ from Swe mutant ␤-secretase activity. However, CTSB participates in A␤ production from hA␤PP-695 having the Wt ␤-secretase site and utilizing Wt ␤-secretase activity [7].

CTSB KO in hAβPP-751/770/Sweβ-Indγ AD mice had no effect on Swe mutant β-secretase activity and produced a small increase in Aβ and amyloid plaque
Studies of mice expressing hA␤PP-751/770Swe␤-Ind␥, representing abnormal neuronal rather than glia expression, showed that CTSB KO had no effect on CTF␤ levels in brain which indicates no effect on Swe ␤-secretase activity [16]. These findings are consistent with CTSB KO having no Swe ␤-secretase activity in mice expressing hA␤PP-695/Swe␤-Lon␥ [7]. However, KO of CTSB in the hA␤PP-751/770Swe␤-Ind␥ mice resulted in a small increase in the ratio of A␤ 42 /A␤ 1-x and increased amyloid plaque, suggesting that CTSB may be involved in degradation of A␤ [16]. Lentiviral CTSB expression reduced preexisting amyloid deposits, also suggesting CTSB degradation of A␤.
CTSB cleaves the Wt β-secretase site but not the Swe mutant β-secretase site CTSB KO data supports participation of CTSB in regulating Wt ␤-secretase activity to generate A␤ peptides but CTSB does not participate in Swe mutant ␤-secretase activity in A␤ production ( Table 2). To test the hypothesis that CTSB may function as an alternative ␤-secretase, CTSB cleavage of the Wt ␤secretase site of the model Z-Val-Lys-Met-↓AMC substrate was assessed. CTSB has high activity for cleaving the Wt ␤-secretase site (Table 3) [32]. However, CTSB showed almost no cleavage of the Swe mutant ␤-secretase site of the Z-Val-Asn-Leu-↓AMC (Asn-Leu is the Swe mutation) substrate (Table 3) [32]. CTSB displayed a 2,735-fold higher rate of cleaving the Wt over the Swe mutant substrates (Table 3). CTSB clearly prefers to cleave the Wt Z-Val-Lys-Met-↓AMC substrate compared to the Swe mutant substrate Z-Val-Asn-Leu-↓AMC. These results demonstrate CTSB as an alternative Wt ␤secretase in addition to the well-established BACE1 ␤-secretase [33][34][35].

CHRONIC PERIODONTITIS-ASSOCIATED AD AND NEURODEGENERATION MODELS HAVE SHOWN CTSB PARTICIPATION IN A␤ PRODUCTION VIA WT ␤-SECRETASE
Models of chronic periodontitis-associated AD, advanced glycation end (AGE) products, and Mucopolysaccharidosis type I (MPSI) described in this section provide evidence for participation of CTSB in A␤ production in AD-related neurodegenerative conditions.

Chronic periodontitis-associated AD
Clinical evidence indicates a positive link between periodontitis and AD with respect to cognitive dysfunction and inflammation [12,13]. CTSB KO in the neuroinflammatory periodontitis model of AD showed that CTSB participates in neuronal A␤ production and drives memory deficits [8]. CTSB KO blocked PgLPS-induced elevation of A␤ 42 in mouse brain, indicating that A␤ 42 production is dependent on CTSB [8]. These data show that CTSB regulates Wt ␤-secretase activity for conversion of mouse A␤PP to A␤. These mouse studies are relevant to clinical periodontitis, since periodontitis patients display elevated serum CTSB that correlates with cognitive deficits [8,9]. In cellular neuroblastoma studies, inhibition of CTSB with the selective inhibitor CA-074Me reduced PgLPS-induced increases in A␤ 40 and A␤ 42 [9]. These results demonstrate participation of CTSB in A␤ production generated from A␤PP by Wt ␤-secretase activity.

AGE products in aging and neurodegeneration
AGE involves reaction of glucose or other sugars with proteins that induce neuronal toxicity through the AGE receptor [2]. In cortical neurons, AGE increased CTSB and A␤ 42 ; furthermore, the cathepsin B inhibitor reduced A␤ 42 . These results indicate that CTSB participates in A␤ 42 production [2]. nd [17] All human A␤PP models utilized C57BL/6 mouse strain (adult ages 3-12 months) using equal numbers of male and female mice. Swe, K670N/M671L/Lon V717I/Ind V717F; nd, not determined; n/a. not applicable.  [32]. These Z-peptide-AMC model substrates mimic the ␤-secretase cleavage site within the amyloid-␤ protein precursor (A␤PP).

Mucopolysaccharidosis type I (MPS I)
MPS I is a rare neurologic disease resulting from a genetic deficiency of ␣-L-iduronidase (IDUA) involving impaired lysosomal catabolism and neurodegeneration [36]. The MPS I mouse model, generated by KO of the IDUA gene, displays increased levels of CTSB and elevated A␤ in brain. The study indicated that CTSB provides an alternative amyloidogenic pathway for A␤ production [36].

Evidence for CTSB as an alternative Wt β-secretase for Aβ production
Overall, the studies described here have demonstrated a role for CTSB in A␤ production from A␤PP having the Wt ␤-secretase site, indicating CTSB involvement in Wt ␤-secretase activity. CTSB participates in A␤ production in neurodegenerative disease models of periodontitis, AGE, and MPS I expressing Wt A␤PP [8,2,36]. Evidence supports CTSB as an alternative Wt ␤-secretase to generate A␤ from Wt A␤PP [7,10,15,32,37] which is expressed in the major sporadic population of AD patients. Consideration of CTSB as an alternative Wt ␤-secretase contributes to the established role of the BACE1 ␤secretase [38][39][40], combined with recently studied proteases with ␤-secretase activity of meprin [41,42], delta-secretase [43,44], and matrix metalloproteinases [45].

BACE1 ASPARTYL PROTEASE PREFERENTIALLY CLEAVES THE SWE MUTANT ␤-SECRETASE A␤PP SITE, COMPARED TO THE WT ␤-SECRETASE SITE, FOR A␤ PRODUCTION
A␤ 40 and A␤ 42 are known to be produced by the ␤-secretase BACE1 [33,[46][47][48]. BACE1 has different proteolytic cleavage capability for Wt versus Swe ␤-secretase site sequences. The BACE1 protease inefficiently cleaves the WT ␤-secretase site and effi-ciently cleaves the Swe ␤-secretase site [32,[49][50][51]. Cathepsin B has been postulated as a ␤-secretase and it differs from BACE1 in cleavage properties since cathepsin B efficiently cleaves the Wt ␤-secretase site but inefficiently cleaves the Swe ␤-secretase site [32]. A possible mechanism by which cathepsin B may augment production of A␤ 40 and A␤ 42 in models expressing Wt ␤-secretase activity may involve regulation of BACE1 activity, direct cleavage of the WT ␤-secretase site, or mechanisms yet to be defined.
A distinction between BACE1 and cathepsin B is their role for A␤ production in the constitutive versus regulated secretory pathways of neurons. Neurons possess the regulated secretory pathway that is utilized for activity-dependent secretion of the majority of neurotransmitters [60]. Basal secretion of a small portion of neurotransmitters occurs through the constitutive secretory pathway [60]. BACE1 was identified as a ␤-secretase for A␤ production through cleavage of the Swe mutant ␤-secretase site of A␤PP that functions in the constitutive secretory pathway of human embryonic kidney cells [49][50][51]. Cathepsin B was discovered by purification of Wt ␤-secretase site cleaving activity in regulated secretory vesicles for production of A␤ [61]. These regulated secretory vesicles produce multiple A␤ species of A␤ 40 and A␤ 42 as well as the truncated pGlu-A␤ 3-40 and pGlu-A␤ 3-42 [62]. The pGlu-A␤ 3-40/42 peptides accumulate in human AD brains and promote neurotoxicity through oligomerization of A␤ peptides [30,31,63].
Significantly, BACE1 does not appear to produce pGlu-A␤ 3-40/42 peptides [10] that are likely the neurotoxic species that promotes oligomeriza- Fig. 4. Cathepsin B participates in memory deficits of the major sporadic Alzheimer's population. (a) CTSB elevation in Alzheimer's disease (AD) patients correlates with cognitive deficits. Increased levels of CTSB were observed in sporadic AD [1,2,4], the major population of AD. Significantly, elevated CTSB was found to be significantly correlated with cognitive decline in AD patients [1]. (b) CTSB gene knockout in animal models of AD results in improved memory deficits. In the AD mouse model expressing hA␤PP-695, CTSB gene knockout resulted in substantial improvement in memory deficits [7]. Furthermore, knockout of CTBS in the periodontitis model of AD resulted in improved memory deficits in middle-aged mice [8].
tion of A␤ peptides involved in causing AD [30,31]. However, cathepsin B participates in the production of pGlu-A␤ species in models expressing the WT ␤-secretase site sequence [10] that is found in most AD patients. While BACE1 inhibitors have been effective in the clinic to reduce A␤ 40 and A␤ 42 [64][65][66], unfortunately, such inhibitors have not significantly improved cognitive deficit of AD [64,66]. That may be due to BACE1 inhibitors not affecting pGlu-A␤ 3-40/42 production. Immunotherapy by aducanumab targeting A␤ was not efficacious for improving cognition in AD patients [67]. However, donanemab immunotherapy targeting pGlu-A␤ resulted in improved cognition in AD patients [68,69], showing the importance of pGlu-A␤ in AD. An exciting possibility is that cathepsin B inhibitors may prove useful in the clinic by reducing these pernicious pGlu-A␤ 3-40/42 species of A␤. More research and development on inhibitors of cathepsin B is warranted.

CONCLUSION: CTSB PARTICIPATES IN MEMORY DEFICITS AND WT ␤-SECRETASE ACTIVITY FOR A␤ PRODUCTION IN HUMAN A␤PP MODELS REPRESENTING THE MAJOR SPORADIC AD CONDITION
Evaluation by this review of findings in the literature indicate that elevated CTSB correlates with cognitive deficits in AD patients. In fact, both AD patients and chronic periodontitis-associated AD patients display elevated serum CTSB that correlates with the extent of cognitive deficits [8,9].
CTSB participates in memory deficits and production of A␤ in AD animal models [7,10,15]. Among the six animal models utilized in CTSB KO studies (Table 2), the hA␤PP-695/Wt␤-Lon␥ AD model best represents the majority of the AD population expressing hA␤PP-695 as the primary brain A␤PP isoform present in neurons. Significantly, CTSB KO in the hA␤PP-695/Wt␤-Lon␥ AD mice results in substantial improvement in memory deficits to nearly normal values, reduced brain levels of A␤ peptides of A␤ 40 , A␤ 42 , pGlu-A␤ 3-40 , pGlu-A␤  , and reduced amyloid plaque load [7,10]. CTSB KO reduces the ␤-secretase cleavage product of CTF␤ generated from hA␤PP, suggesting that CTSB participates in Wt ␤-secretase activity.
CTSB KO in mice with the Swe mutant ␤-secretase of A␤PP (hA␤PP-695/Swe␤-Lon␥ mice) had no effect on memory deficits or A␤ peptides [7]. The Swe mutant hA␤PP represents only one AD family [11] and does not represent the major sporadic AD population.
The numerous studies of CTSB KO demonstrate that CTSB participates in memory deficits and A␤ production in hA␤PP-695 models, combined with clinical data showing correlation of elevated CTSB with cognitive deficits, support the conclusion that CTSB participates in AD memory deficits and pathology. CTSB participates in modulating Wt ␤-secretase activity for A␤ production in hA␤PP models representing the major sporadic AD population (Fig. 4).
These findings demonstrate CTSB as a logical drug target for development of therapeutic agents for AD.

ACKNOWLEDGMENTS
The authors have no acknowledgments to report.

FUNDING
This work was supported by NIH grants R01NS109075 (awarded to VH) and R41NS110147 (awarded to GH). Additionally, this work was supported by a Senior Research Career Scientist Award from the Veterans Administration, JK6 BX005239-01 (awarded to MSK).

CONFLICT OF INTEREST
V. Hook and G. Hook have equity positions at American Life Science Pharmaceuticals (ALSP) and are founders of ALSP. V. Hook is an advisor to ALSP. G. Hook at ALSP is vice president of research, corporate counsel, and member of the board of directors. V. Hook's conflict has been disclosed and is managed by her employer, the University of California, San Diego.