COR388 (atuzaginstat): an investigational gingipain inhibitor for the treatment of Alzheimer disease

Abstract

Introduction:

Evidence from in vitro and in vivo studies demonstrates that amyloid beta (Aβ) oligomers have potent, broad-spectrum antimicrobial properties created by fibrils that entrap pathogens and disrupt their membranes. Data suggest that Aβ may play a protective role in the innate immune response to microbial infections and that Aβ in the brain plays a damaging role when the inflammatory response is not well controlled.

Areas covered:

This paper describes the relationship between periodontal disease and Alzheimer disease (AD), the role of Porphyromonas gingivalis and its secreted gingipains in AD, and the potential of the gingipain inhibitor atuzaginstat (COR388) to modulate AD neuropathologies.

Expert opinion:

P. gingivalis is opsonized by Aβ42, is capable of entering the brain, and is an accelerant of neuropathologies in rodent models of AD. Thus, in our opinion, this bacteria is highly likely to be a pathogen capable of initiating or precipitating the progression of AD, which agrees with the pathogen hypothesis of clinical AD development.

1.0. Introduction

Until recently, amyloid beta (Aβ) 42 was considered exclusively as a peptide that is one of the principal pathogenic drivers of clinical Alzheimer disease (AD). This hypothesis was strongly supported by the observation of accelerated accumulation of amyloid plaques in the brains of persons with specific genetic mutations. These mutations affect proteins implicated in amyloidogenesis, such as mutations of the presenilin 1 (PSEN1), PSEN2, amyloid precursor protein (APP), and Down syndrome (also known as trisomy 21) genes, and they often lead to the presenile onset of AD [1]. In these autosomal dominant mutations and trisomy 21 (APP is located on chromosome 21), Aβ is frequently overexpressed and detectable at a young age. Persons who carry these mutations express symptomatic cognitive decline by the fourth or fifth decade of life, with increased Aβ loads detectable as early as the third decade in the cerebrospinal fluid (CSF) and in positron emission tomography (PET) imaging of the brain [2]. The long-term consensus formulated in the amyloid cascade hypothesis [3] has been that Aβ overproduction is an early event in AD pathogenesis that then drives downstream pathologies, including tau neurofibrillary tangles, central inflammation, excitotoxicity, and neurochemical changes [4].

Another emerging perspective on the role of Aβ is that the peptide is conserved evolutionarily and is present throughout life in cognitively normal persons. This concept, combined with the fact that APP knockout mice are viable, strongly suggests that Aβ also has physiologic roles that may overlap with other biological systems [5]. One putative role is that Aβ might serve as an antimicrobial peptide (AMP) [6,7] possessing fungicidal and virucidal activities [8]. AMPs are a class of innate immune defense molecules that use fibrillation to protect the host from a wide range of infectious agents. Soscia et al. [6] used in vitro assays to compare the antimicrobial activities of Aβ40, Aβ42, and LL-37, with LL-37 being an archetypical endogenous human antimicrobial protein. Their findings revealed that Aβ exerts antimicrobial activity against 8 common clinically relevant microorganisms. Soscia et al. [6] noted that Aβ42 was more potent than Aβ40 applied in the same concentrations. Moreover, they also demonstrated that AD brain homogenates that are preincubated with the anti-Aβ antibody 6E10 lose their antimicrobial activity. Similarly, Aβ may interact with viral coat proteins [9]. For example, Aβ was produced by human H4 neuroglioma cells upon exposure to the herpes simplex virus 1 (HSV1) [10].

Follow-up studies in animal models have confirmed in vitro observations. For example, a single intracerebral injection of Salmonella typhimurium bacteria caused seeding and acceleration of Aβ deposition in transgenic 5XFAD mice that closely colocalized with the invading bacteria [11]. Interestingly, 5XFAD mice showed a higher survival rate than wild-type littermates and APP knockout mice that also received S. typhimurium bacteria in brain injections, suggesting that Aβ has a protective role against bacterial infections in vivo. Similar results were observed with viral infections. Encephalitis occurred less often after HSV1 infection in 5XFAD mice than in wild-type mice [12].

Observations in human samples indicate that a variety of microbial species are present in Aβ plaques and AD brains, including HSV1 [13], Borrelia [14], and Chlamydia pneumoniae [15]. The levels of soluble APP or Aβ have been reported to decrease in the CSF during microbial infections and to return to normal values after the resolution of infection, suggesting that these proteins may be sequestered in the brain during infectious events [16]. These data strongly suggest that infections of the brain, both in humans and in animal models, frequently lead to increased amyloidogenic processing of APP and result in fibrillary aggregates of Aβ; these data form the basis for the “pathogen hypothesis” of AD [17].

How can Aβ inhibit the multiplication of microbes? Moir et al. [7] posited that mechanistic Aβ oligomerization, a behavior traditionally viewed as intrinsically pathologic, may be necessary for the antimicrobial activities of Aβ. They suggested that 3 mechanisms are active: (1) soluble Aβ oligomers bind to microbial cell wall carbohydrates via a heparin-binding domain; (2) developing protofibrils inhibit pathogen adhesion to host cells; and (3) propagating β-amyloid fibrils mediate agglutination and final entrapment of microbes. These mechanisms resemble the opsonization mechanism of the innate immune system upon encountering pathogens, which is to recruit adaptive immune cells to eliminate the pathogens [18].

In summary, evidence from both in vitro and in vivo studies demonstrates that Aβ oligomers have potent, broad-spectrum antimicrobial properties as they form fibrils that entrap pathogens and disrupt their membranes. These data suggest that Aβ may play a protective role in the innate immune response to microbial infections and suggest a dual protective–damaging role for Aβ in the brain when the inflammatory response is not well controlled [4,7]. Microbes that may be implicated in the etiology of AD but have not been fully studied include the pathogens that induce periodontal disease. This avenue of research is important to explore to develop disease-modifying treatments for AD.

2.0. Linking periodontal disease to AD

A recently discussed theory is that the exact etiology of AD is highly likely to involve multiple pathologies converging in the production of Aβ in the brain [4]. However, several epidemiologic studies have pointed out that a possible cause of AD or an enhancer of AD pathologies could be the pathogens that induce periodontal disease. In the longitudinal Nun Study, elderly participants 75 to 98 years old who had the fewest teeth had the highest prevalence and incidence of dementia [19]. Similarly, in a Veterans Administration study of almost 600 men aged 28 to 70 years, tooth loss was associated with a 10% increase in cognitive decline. The risk of cognitive decline in older men increased as more teeth were lost, suggesting periodontal disease as a risk factor for cognitive decline [20]. In a study assessing the link between periodontitis and AD progression in 60 community-dwelling patients, researchers found that the presence of periodontitis at baseline was not related to baseline cognitive state but was associated with a 6-fold increase in the rate of cognitive decline as assessed by the Alzheimer’s Disease Assessment Scale–Cognitive Subscale (ADAS-Cog) over a 6-month follow-up period [21]. These initial observations were recently complemented by similar results from a study conducted using a database of more than 4 million participants with both documented dental care and an AD diagnosis or treatment status (N=121,168 diagnosed with AD) [22]. The researchers observed that the fewer teeth patients had, the more likely they were to be diagnosed with AD. The Chinese Longitudinal Healthy Longevity Survey, which started in 1998 and conducts follow-up examinations at regular intervals of 11,862 participants, showed that tooth loss in the elderly may increase their risk of developing cognitive impairment [23]. Furthermore, a meta-analysis of 14 studies with more than 34,000 participants, including more than 4,400 with diminished cognitive functioning, determined that patients with advanced tooth loss were almost 50% more likely to have developed cognitive impairment and 30% more likely to be diagnosed with dementia than those without advanced tooth loss [24]. Periodontal disease is the most likely connection between tooth loss and cognitive decline. Periodontitis is a diffuse oral pathology in which Porphyromonas is involved, and affected populations are often aged 20 to 50 years [25]. Tooth loss and dementia are potentially related in individuals with poor personal hygiene, possibly earlier in life, and in individuals considered more vulnerable to neurodegenerative disorders. Thus, periodontitis might be a seminal trigger of the AD cascade.

In a direct correlational study, periodontal disease and brain amyloid loads measured by 11C-PiB PET imaging were evaluated in 38 cognitively normal elderly patients to test the hypothesis that periodontal disease, as assessed by clinical tooth attachment loss, was associated with increased brain amyloid loads. After adjustment for confounders, clinical tooth attachment loss (≥3 mm), representing a history of periodontal inflammatory or infectious burden, was associated with a significant increase in 11C-PiB uptake in amyloid-vulnerable brain regions (p=0.002), which strongly suggests a direct correlation between the 2 conditions [26].

3.0. The role of Porphyromonas gingivalis and its secreted gingipains in AD

Although these observational studies are useful in establishing a connection between periodontal disease and AD, they do not demonstrate causation. What is the mechanistic link, if any, between these 2 diseases? Periodontitis, which is caused by periodontopathic bacteria, is a major cause of tooth removal and loss. Several pathogens have been implicated in periodontitis [25], but the most prevalent is Porphyromonas gingivalis, which is found in 80% of patients with periodontal disease [27]. P. gingivalis is a nonmotile gram-negative anaerobic rod bacterium. P. gingivalis resists most broad-spectrum antibiotics, such as amoxicillin [28] and moxifloxacin [29], and it expresses several virulence factors, including a lipopolysaccharide on its outer membrane [25]. Critically for AD, P. gingivalis produces and partially secretes gingipains, which are cysteine proteases often referred to as trypsin-like enzymes, in the extracellular milieu. These proteases are called gingipains R (Rgp) and gingipains K (Kgp) because they cleave their substrates after arginine (R) and lysine (K) residues. Gingipains R are further subdivided into RgpA and B, which possess slightly different enzymatic properties. Gingipains are able to catabolize extracellular matrix components, fibrinogen, cytokines, immunoglobulins, and complement factors of the immune system [30]. More precisely, gingipains can catabolize and inhibit AMPs, including complement factors C3 and C4, cathelicidins, LL-37, and neutrophil-derived α-defensins. They can also inhibit T-cell receptors CD4 and CD8, as well as the innate immune receptor CD14, which blocks macrophages from responding to bacterial infections [31,32]. Chewing causes P. gingivalis to migrate into the bloodstream, as does routine dental care such as brushing, flossing, and dental procedures. Its virulence factors not only can invade CSF and multiple tissues, such as arteries, liver, and brain [33,34], but also can escape the host’s immune response. For example, one small study found that P. gingivalis was present in the arteries of all patients with cardiovascular disease [35], thereby confirming an association between oral bacteria and atherosclerosis.

The strongest evidence of a link between P. gingivalis and AD pathology comes from human postmortem studies and mechanistic animal studies. The lipopolysaccharide of P. gingivalis has been identified in the brains of patients with AD but not in controls [36]. Immunohistochemistry and quantitative polymerase chain reaction (PCR) were used to study gingipains and the P. gingivalis–specific gene hmuY, which were identified in the hippocampal and cortical neurons and in some astrocytes of more than 90% of the AD brains that were examined, and which were present at significantly higher loads than in control brains [29].

These observations are supported by several animal studies. First, using 8-week-old C57BL/6 wild-type mice that received chronic oral inoculations of P. gingivalis (strain W83) 3 times per week over 22 weeks, Ilievski et al. [37] demonstrated via several methods that the bacteria (1) migrate to the brain; (2) induce neurodegeneration in the hippocampus, which is highly affected in AD; (3) induce the overexpression of Aβ42 in the hippocampus; (4) increase the levels of phospho-tau (p-tau Ser396) in the hippocampus; and (5) stimulate microgliosis and astrogliosis in the hippocampus. Second, Dominy et al. [29] studied 44-week-old female BALB/c mice orally infected every other day for 6 weeks with P. gingivalis (W83 strain) and found detectable bacterial DNA and increased levels of Aβ42 in the brain. Third, the chronic oral inoculation of ApoE−/− mice with P. gingivalis for 12 and 24 weeks resulted in the detection of bacterial genomic DNA in the brain [38]. Fourth, Sprague-Dawley male rats that received an acute oral infection of different strains of P. gingivalis for 45 days displayed (1) brain infection detected by PCR, (2) a decreased performance in spatial learning and memory on the oasis maze, (3) increased cytokine levels in the CNS and in serum, (4) glial activation in the hippocampus, and (5) increased immunoreactivity for p-tau S404 in the hippocampus [39].

A mechanistic link between periodontal disease and AD is further supported by a 2021 study of an AD transgenic mouse model (APP/PS1) with periodontitis triggered by P. gingivalis infection that exacerbated learning and memory impairment and augmented Aβ and neuroinflammatory responses [40]. Furthermore, Ishida et al. [41] conducted a study of 62-week-old hAPP-J20 female mice (a transgenic model of AD amyloid pathology) that received an acute oral inoculation of P. gingivalis mixed with carboxymethyl cellulose and were sacrificed 5 weeks later. They observed reduced learning ability in bacteria-treated mice on the novel object recognition test and found increased brain plaque loads and Aβ40 and Aβ42 levels.

Additional mechanistic observations include the antibacterial effects of Aβ42, but not Aβ40 or scrambled Aβ42, on P. gingivalis [29]. Both P. gingivalis and its associated gingipains were able to cleave recombinant human tau expressed in SH-SY5Y cells in multiple locations. This result is complemented by the observation in Sprague-Dawley rats that P. gingivalis induces tau hyperphosphorylation and attenuates the activity of PP2A by triggering systemic inflammation and neuroinflammation [42]. A rigorous review of the alterations of tau by P. gingivalis was published in 2020 [43]. P. gingivalis and gingipains, but not dominant negative gingipains, have been found to fragment apolipoprotein E (ApoE) [44]. Gingipains preferentially cleave ApoE4>3>2 because the known 3-dimensional structure of ApoE4 seems to make it more prone to proteolysis [45]. Prior reports showed that ApoE fragments tend to be neurotoxic, whereas full-length ApoE is protective [46]. Both microglia and inflammasomes are activated in the AD brain and by gingipains [47]. In addition, complement is dysregulated in the AD brain, which may be mediated by gingipains [47].

These data strongly argue for the participation of P. gingivalis and gingipains in AD pathogenesis. Given the epidemiological link between periodontal disease and AD, one has to speculate whether P. gingivalis and gingipains are an early trigger of AD pathology and neuroinflammation in the nascent amyloid-driven phase of AD because periodontitis would be able to affect individuals at approximately 50 years of age or might be considered a risk factor or, better, a contributor to further degeneration. However, this bacteria is resistant to broad-spectrum antibiotics; thus, a therapeutic approach would not be viable to inhibit the deleterious effects of P. gingivalis on the brain and other tissues. This paradigm has led to clinical trials of small-molecule drugs to potentially inhibit gingipains (Figure 1).

Figure 1. Proposed mechanistic targets of atuzaginstat.

Used with permission from Cortexyme, San Francisco, California.

4.0. The gingipain inhibitor, atuzaginstat (COR388), could modulate AD neuropathologies

Atuzaginstat is a novel small molecule that inhibits Kgp [29]. It binds covalently to the active site of the lysine gingipains, thereby permanently inactivating them. It is orally bioavailable, brain penetrant, selective, and potent, with an IC₅₀ of <50 pM.

Atuzaginstat blocks gingipain toxicity, reduces bacterial load, and normalizes immune function. In mice exposed to oral P. gingivalis for 5 weeks before randomization to COR388 or vehicle twice daily, COR388-treated mice had lower levels of amyloid and reduced tumor necrosis factor alpha; COR388 prevented the death of hippocampal GABAergic interneurons [29]. Furthermore, treating cell cultures with COR388 not only protects neurons from synaptic loss but also protects against deficits in decreased synaptic transmission, apoptosis, and other disease-associated pathways in P. gingivalis–infected co-cultures [48]. Induced pluripotent stem cell studies show that ApoE proteolysis in astrocytes induced by P. gingivalis is blocked by COR388 [45].

Exposures based on animal models and mechanism of action suggest that the optimal dosing of COR388 in humans is twice daily. Investigational new drug–enabled phase 1 and phase 2/3 clinical trials have been completed. In the phase 1 trial (ClinicalTrials.gov: NCT03418688), COR388 was well tolerated without severe adverse events in the single ascending dose, in the 10-day multiple ascending dose in healthy elderly patients, and in the 28-day study in AD patients. Drug-related treatment-emergent adverse event rates were similar between placebo recipients (20%) and COR388-treated patients (14%); the only drug-related adverse event that occurred was dizziness (twice, once at 50 mg and once at 100 mg). No dose-limiting toxicity was identified. No clinically significant trends were observed on laboratory tests or electrocardiograms, blood pressure, heart rate, or temperature [49].

In the multiple ascending dose study in healthy older participants, COR388 was absorbed rapidly with a median T_max of 0.5 to 1.5 hours. The target therapeutic level was achieved with doses as low as 25 mg twice daily at steady state with a mean C_max of 25 ng/mL, a mean AUC_0-24 of 178 h●ng/mL, and clearance T_½ of 4.5 to 4.9 hours at steady state. COR388 was detected in CSF, and pathologic ApoE fragments in CSF were reduced by COR388 in the 28-day study of AD patients. Clinically detectable signals were identified, with trends favoring the benefit of the investigational product over placebo on exploratory cognitive measures, including the Mini-Mental State Examination (MMSE), the Cognition Cambridge Neuropsychological Test Automated Battery, and the Winterlight speech assessment [50].

The phase 2 GingipAIN Inhibitor for Treatment of Alzheimer’s Disease (GAIN) study (ClinicalTrials.gov: NCT03823404), which was completed on January 1, 2022, randomized 643 patients with mild to moderate AD dementia to placebo, 40-mg COR388 twice daily, or 80-mg COR388 twice daily. The co-primary endpoints were changes in scores on the ADAS-Cog11 and ADCS-ADL (Alzheimer’s Disease Cooperative Study activities of daily living), and the secondary endpoints were changes in scores on the Clinical Dementia Rating Scale Sum of Boxes, the MMSE, and the Neuropsychiatric Inventory. To identify the best method of diagnosis and treatment response, the study prespecified biologically relevant subgroups based on markers of P. gingivalis infection, including saliva P. gingivalis DNA, anti-P. gingivalis IgG in serum, and anti-P. gingivalis IgG in CSF. In addition to P. gingivalis markers, biomarker endpoints included volumetric changes on magnetic resonance imaging and changes in CSF measures of Aβ, tau, p-tau. The data for the phase 2/3 GAIN trial are being analyzed, and the results should be released in 2022. A dose-dependent hepatotoxic signal was identified that prompted a full clinical hold by the US Food and Drug Administration [51]. Preliminary results from a subpopulation of the GAIN study noted that patients with P. gingivalis DNA detected by oral swab were more likely to benefit from atuzaginstat treatment [51]. On January 25, 2022, the GAIN study was halted by the U.S. Food and Drug Administration because of hepatotoxicity, which caused the sponsor to suspend drug development [52].

5.0. Conclusions

The gingipain inhibitor atuzaginstat is a novel protease inhibitor that blocks a neurotoxic virulence factor known to trigger AD pathogenesis. Because the link between periodontal disease and AD is mechanistic and not incidental, it seems logical to develop drugs that target the virulence factors of P. gingivalis infection, such as gingipains. Atuzaginstat is appealing because it is orally bioavailable and blood–brain barrier penetrant. Clinical trials of atuzaginstat have been completed, with no evidence of amyloid-related imaging abnormalities. However, a hepatotoxic signal has emerged that requires monitoring and could potentially influence further development of this potential treatment. Since the patients most likely to benefit from the results of the clinical trials underway are those with mild-to-moderate AD, more studies are being planned. These patients would have almost no overlap with patients eligible for antiamyloid monoclonal antibodies. Atuzaginstat can address the ongoing need for treatment for most patients for whom antiamyloids are not suitable.

6.0. Expert opinion

Because P. gingivalis is opsonized by Aβ42, is capable of entering the brain, and is an accelerant of neuropathologies in rodent models of AD, this bacteria is highly likely to be one of the pathogens capable of initiating or precipitating the progression of AD, which is a finding in agreement with the pathogen hypothesis of clinical AD development. Therefore, using non-Aβ therapeutics to curb the onset and acceleration of AD pathologies is highly warranted in both prodromal and early AD patients. Because P. gingivalis is resistant to broad-spectrum antibiotics, the next-best therapeutic approach is to inhibit its virulence factors known to suppress both the innate and adaptive immune responses, namely, gingipains.

Another possibility to consider is that P. gingivalis triggers neuroinflammation, possibly mediated by the gingipains. P. gingivalis is known to trigger inflammatory processes throughout the body and is linked to heart disease mediated through inflammation. Neuroinflammation is complex but has been considered as a potential treatment target for AD for more than three decades. Early attempts to target inflammation focused on using non-steroidal anti-inflammatory drugs, including indomethacin, naproxen, and ibuprofen. Trials to treat AD symptoms also included prednisone. Prevention studies targeting inflammation before the onset of symptoms in at-risk populations were attempted including Naproxen, celecoxib and ibuprofen. Outcomes from treatment and prevention trials largely indicate that non-steroidal anti-inflammatories are ineffective in AD. Nevertheless, targeting neuroinflammation remains appealing. More recent attempts at targeting inflammation have focused on non-cyclooxygenase pathways and have pivoted to focus in TNFα. These include etanercept, thalidomide, and more recently lenalidomide. Targeting this pathway has theoretical appeal, but results are unavailable or inconclusive.

Development of a protease inhibitor that could potentiate antimicrobial activity and target neuroinflammation is appealing because it would act upstream from the inflammatory cascade, thus reducing the development of neuroinflammation. Creation of this new class of protease inhibitors, directed against periodontitis, might provide a pathway to be treat AD degeneration or inflammation.

Given the potential of P. gingivalis as a neuroinflammation trigger and the potential that the protease inhibitor atuzaginstat has therapeutic effects by modulating or inhibiting neuroinflammation, the question that needs to be considered is not whether AD should be targeted but when in the disease course should treatments be targeted. The current studies have focused on mild-to-moderate AD dementia, which, pathologically speaking, is several years into the disease course, and damage from neuroinflammation might already be significant and widespread. In the future, targeting preclinical or early symptomatic disease might make sense. Also, future considerations should focus on whether atuzaginstat would be appropriate and efficacious for all patients or the subset of patients who express P. gingivalis or have periodontal disease. Targeting a subset of patients with AD is termed segmentation. Segmentation is likely to become more common by target symptoms or genotype or other risk factor. In other words, single drugs may not be able to treat all subsets of AD patients.

The results of the GAIN trial are expected to determine whether atuzaginstat should continue being tested in the context of AD or whether more efficient analogs should be generated, given some concerns about hepatotoxicity. A sister product, COR588, is currently being tested in a phase 1 single ascending dose/multiple ascending dose study. As a final note, replacing diseased teeth with nontoxic prosthodontics made of nontoxic material might be a preventive approach that could lower the risk that cognitive impairment will develop at late ages [24].

Article highlights:

• Amyloid beta (Aβ) might have natural antimicrobial properties.

• The link between periodontal disease and Alzheimer disease (AD) might not be incidental and is likely to be mechanistic.

• Porphyromonas gingivalis and its proteases, called gingipains, are neurotoxic and trigger neuroinflammatory pathways.

• Atuzaginstat is the first-in-class gingipain protease inhibitor and has been shown to reduce AD pathology and inflammatory markers.

• Clinical trials of atuzaginstat have demonstrated efficacy in a subset of patients who express P. gingivalis DNA, but drug development has been hampered by emerging hepatotoxicity.

Drug summary box.

Drug name: Atuzaginstat

Phase: 2/3

Indication: Alzheimer’s disease dementia, mild to moderate

Pharmacology description: Atuzaginstat is a novel small molecule that inhibits gingipains K. It binds covalently to the active site of the lysine gingipains, thereby inactivating them permanently. It is orally bioavailable, brain penetrant, selective, and potent, with an IC₅₀ of <50pM.

Route of administration: Oral

Chemical structure:

Pivotal trial: GAIN study (ClinicalTrials.gov: NCT03823404).

Abbreviations

Aβ

amyloid beta

AD

Alzheimer disease

ADAS-Cog

Alzheimer’s Disease Assessment Scale–Cognitive Subscale

ADCS-ADL

Alzheimer’s Disease Cooperative study activities of daily living

AMP

antimicrobial peptide

ApoE

apolipoprotein E

APP

amyloid precursor protein

CSF

cerebrospinal fluid

GAIN

GingipAIN Inhibitor for Treatment of Alzheimer’s Disease

HSV1

herpes simplex virus 1

Kgp

gingipains K

MMSE

Mini-Mental State Examination

PCR

polymerase chain reaction

PET

positron emission tomography

PSEN

presenilin

Rgp

gingipains R