Neurofilament light: a possible prognostic biomarker for treatment of vascular contributions to cognitive impairment and dementia

Background Decreased cerebral blood flow and systemic inflammation during heart failure (HF) increase the risk for vascular contributions to cognitive impairment and dementia (VCID) and Alzheimer disease-related dementias (ADRD). We previously demonstrated that PNA5, a novel glycosylated angiotensin 1–7 (Ang-(1–7)) Mas receptor (MasR) agonist peptide, is an effective therapy to rescue cognitive impairment in our preclinical model of VCID. Neurofilament light (NfL) protein concentration is correlated with cognitive impairment and elevated in neurodegenerative diseases, hypoxic brain injury, and cardiac disease. The goal of the present study was to determine (1) if treatment with Ang-(1–7)/MasR agonists can rescue cognitive impairment and decrease VCID-induced increases in NfL levels as compared to HF-saline treated mice and, (2) if NfL levels correlate with measures of cognitive function and brain cytokines in our VCID model. Methods VCID was induced in C57BL/6 male mice via myocardial infarction (MI). At 5 weeks post-MI, mice were treated with daily subcutaneous injections for 24 days, 5 weeks after MI, with PNA5 or angiotensin 1–7 (500 microg/kg/day or 50 microg/kg/day) or saline (n = 15/group). Following the 24-day treatment protocol, cognitive function was assessed using the Novel Object Recognition (NOR) test. Cardiac function was measured by echocardiography and plasma concentrations of NfL were quantified using a Quanterix Simoa assay. Brain and circulating cytokine levels were determined with a MILLIPLEX MAP Mouse High Sensitivity Multiplex Immunoassay. Treatment groups were compared via ANOVA, significance was set at p < 0.05. Results Treatment with Ang-(1–7)/MasR agonists reversed VCID-induced cognitive impairment and significantly decreased NfL levels in our mouse model of VCID as compared to HF-saline treated mice. Further, NfL levels were significantly negatively correlated with cognitive scores and the concentrations of multiple pleiotropic cytokines in the brain. Conclusions These data show that treatment with Ang-(1–7)/MasR agonists rescues cognitive impairment and decreases plasma NfL relative to HF-saline-treated animals in our VCID mouse model. Further, levels of NfL are significantly negatively correlated with cognitive function and with several brain cytokine concentrations. Based on these preclinical findings, we propose that circulating NfL might be a candidate for a prognostic biomarker for VCID and may also serve as a pharmacodynamic/response biomarker for therapeutic target engagement.


Introduction
Increases in brain and systemic inflammation, and decreased brain blood flow, as seen during heart failure (HF) [1], are strongly correlated with the development of vascular contributions to cognitive impairment and dementia (VCID) and Alzheimer's disease-related dementias (ADRD) [2,3]. With an aging population and a strong link between cardiovascular and cerebrovascular disease and subsequent cognitive impairment and dementia, there is a growing need for safe and effective therapies to protect cognitive function in individuals at risk for VCID [4].
Angiotensin 1-7 (Ang-(1-7)) is a clinically safe peptide [5] that decreases inflammation and increases cerebral blood flow in our VCID preclinical model [6][7][8]. Through targeting Mas receptors (MasR), located on neurons, microglia, and vascular endothelial cells in the brain [9], Ang-(1-7) decreases brain inflammation and increases cerebral circulation [9]. MasRs that regulate vasodilation are highly expressed in the hippocampus and perirhinal cortex, which makes them an ideal target for treating brain disease related to hypoxia and inflammation that impair memory, such as VCID and ADRD [6,7]. Our collaborative team has developed and optimized a novel, synthetic Ang-(1-7)-derived glycopeptide, PNA5, that selectively activates MasR and has improved bioavailability and brain penetration compared to the native Ang-(1-7) peptide. We have previously shown that PNA5 targets MasR, reverses cognitive impairment in our preclinical VCID model, and inhibits circulating and brain inflammatory cytokine production [7].
Neurofilament light protein (NfL) is an intermediate filament protein that is a component of the cytoskeleton of neurons and is abundantly expressed in axons [10,11]. NfL levels are increased in the brain following axonal damage and neurodegeneration [10,11] and increases in serum NfL correlate with cerebrospinal fluid (CSF) levels and increased cognitive impairment [12]. NfL has been found to be elevated in subjects suffering from multiple sclerosis [12,13], as well as traumatic brain injury [14,15], hypoxic brain injury [16], and cardiac disease and related surgeries [17].
In the present study, we hypothesized that (1) treatment with Ang-(1-7)/MasR agonists will rescue cognitive impairment and decrease VCID-induced NfL levels as compared to HF-saline treated mice, and (2) NfL levels will correlate with measures of both cognitive function and VCID-induced increases in brain cytokines. Our results indicate that circulating NfL might serve as a possible prognostic biomarker of cognitive impairment in VCID. Further, NfL might be a candidate biomarker to identify target engagement following treatment with Ang-(1-7)/MasR agonists in VCID.

Animal
Three-month-old (~ 25-30 g) C57BL/6 male mice (Jackson Laboratories, Bar Harbor, Maine) were ordered and housed 3 per cage in a temperature and humidity-controlled facility on a 12 h light/dark cycle. Mice are acclimated upon receival for a week before undergoing MI. Water and chow were available ab libitum for the duration of the experiments. Experiments were performed with adherence to guidelines approved by the Institutional Animal Care and Use Committee at the University of Arizona and in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. The number of animals needed per experimental group was determined using a G*Power analysis.

VCID animal model
Mice were weighed prior to surgery and anesthetized with 2.5% isoflurane in a mixture of air and O 2 . Left coronary artery (LCA) ligation was used to induce myocardial infarction (MI) using the procedure described in Gao et al. [18]. In short, a left-sided thoracotomy was performed at the fourth intercostal space and the LCA was sutured to induce a permanent ligation and MI. Occlusion of the LCA was confirmed by observing myocardial blanching of the left ventricular anterior wall. Control mice underwent sham surgeries, undergoing the same procedure except for the permanent LCA ligation.

Echocardiography
Transthoracic, high-resolution ultrasound was performed the day of, or up to 2 days prior to, tissue harvesting using a Vevo 2100 High-Resolution Imaging System (Visual Sonics, Toronto, ON, Canada) and a 25-MHz transducer. Mouse chest hair was removed using a chemical depilatory. Echocardiographs were performed on anesthetized mice under 2% isoflurane and an O 2 /air mixture. Echocardiographic images were taken in B mode from the parasternal angle both on long and short-axis view at three levels starting at the apex, mid-ventricle, and top of the ventricle. The thickness of the left ventricular wall, cardiac chamber dimensions, interventricular septum, left ventricular posterior wall thickness, and left ventricular internal dimension were measured using Vevo 2100 ® analytic software (Visual Sonics, Toronto, ON, Canada). Only MI animals showing evidence of a significant MI, defined by significant wall motion abnormalities, and reduced contractile function, were included in the study.

Subcutaneous injections
Daily treatment was delivered subcutaneously either at the scruff or inguinal area for 24 days using a 15-gauge needle. All prepared drugs were kept in the 4 °C refrigerator before use and transported on ice before injection. All procedures were performed within the housing facility to minimize stress.

Novel Object Recognition (NOR) test
Mouse cognition was evaluated using the Novel Object Recognition test. For this test, the arena was an evenly illuminated chamber (12 × 12×12 cm) placed in a light and sound-controlled room. The walls of the arena were white, and the floor had a grid to ensure that the placement of the objects remained the same between phases of the test. Exploratory behavior was measured using a digital camera and evaluated by calculating the time spent exploring each object. Orientation of the nose toward an object at a distance of ∼2 cm or less was considered exploratory behavior [19] Fig. 1 Experiment timeline. Three-month-old male mice were acclimated for one week before MI or Control surgeries. 5 weeks following recovery, mice were treated with Saline, Ang-(1-7) or PNA5 at 2 concentrations (50 and 500 ug/kg) for 24 days. During the last 3 days of treatment mice underwent Novel Object Recognition testing. Following Novel Object Recognition testing, mice underwent echocardiographs before killing. At killing, the plasma, brain, and heart were collected. MI myocardial infarct, S.C. subcutaneous injection rearing on the object and resting against the object were not considered exploratory. Importantly, we have demonstrated that exploration is similar in both Control-Saline and HF-Saline treated mice, indicating that MI does not have a significant impact on mouse movement after 8 weeks of HF [6,7]. Individuals scoring mouse exploration were blinded to treatment group. Three sets of distinct objects were used, variable in size, color, and shape, and made of either plastic, glass, or wood. For the habituation phase of the test, mice were habituated to an empty testing arena for 10 min each day for 2 days. For the learning phase of the test, two identical objects were placed in the testing arena and the mouse was allowed to explore the objects for 6 min. After the 6 min of exploration, mice were returned to their home cage for 2 h. For the memory testing phase, one of the objects was replaced with a novel object and the mice were returned to the testing arena and allowed to explore the objects for 2 min. In between each phase, the testing boxes were cleaned with 70% ETOH to prevent the influence of olfactory cues. Analysis: Recognition memory was scored using a discriminatory ratio (discrimination ratio) calculated by: discrimination ratio = the time spent exploring the novel object (t novel ) minus the time spent exploring the familiar object (t familiar ) divided by the total exploration time: A positive discrimination ratio score indicates that the mouse spent more time at the novel object than the familiar object, while a negative score indicates that the mouse spent more time with the familiar object than the novel object. A zero discrimination ratio score indicates a null preference. Typically, in healthy states, mice will spend a greater amount of time with the novel object than the familiar object. Discrimination ratios between groups were analyzed using ANOVA, followed by Dunnett's post with significance set at p < 0.05.

Measurement of neurofilament light (NfL)
Upon killing, blood was collected via heparinized syringe from the neck following decapitation. Collected blood was centrifuged and the plasma stored at -80 ͦ C before shipment to PBL Assay Science for measurement of NfL using a Quanterix-Simoa assay (NF-LIGHT ® 103186, PBL, Neurofilament-Light Advantage Assay Kit).

Brain lysis
Mice were anesthetized with isoflurane and rapidly decapitated. Brains were separated from the skull and quickly cut into two hemispheres. The halves of the brain (1) Discrimination ratio = (t novel − t familiar )/ (total exploration time).
were then rapidly frozen in corning tubes in liquid nitrogen and stored at -80° C. One half of each brain was lysed using 1 mL RIPA buffer (1% Triton X-100, 0.1% SDS, 1 × PBS) with protease (protease inhibitor cocktail 100x [Sigma P8340]) and phosphatase inhibitors (phosphatase inhibitor cocktail 100x [Sigma P5726]). Tissue was then sonicated using an ultrasonic cell disruptor at 30% 40% amplitude for 1 min. Samples were returned to ice in between sonication to prevent the tissue from warming. Samples were then centrifuged at 4 °C or 15 min at 13,000 × g. The supernatant was collected and stored at − 80 °C. Total protein was measured in the range of 5-10 mg/mL. Brain lysates were used for multiplex immunoassays.

Heart tissue fixation
Mice were anesthetized with isoflurane and rapidly decapitated. The heart was exposed by opening the chest cavity and excised. The heart, while still beating, was then (2) z score = (individual value − mean value)/SD. washed in ice cold PBS, dried, and weighed. Hearts were then cut longitudinally, exposing the interventricular septum and both ventricle chambers, and immersed in 10% formalin overnight at 4 °C. Hearts were then dehydrated in methanol with at least two methanol changes before being cleared with xylene for 45 min. Hearts were then wax permeabilized and paraffin embedded.

Histochemistry
Paraffin-embedded hearts were sectioned on a microtome to a thickness of 5-microns and mounted onto polarized slides. They were then stained with hematoxylin and eosin to visualize heart morphology and cellular infiltrates.

Statistical analysis
Data were analyzed using GraphPad Prism 9.0. Values are expressed as mean ± SE. The Shapiro-Wilk W-test was used for determining normal distribution of the data. Differences between multiple groups were analyzed with ANOVA followed by appropriate post hoc group comparisons where indicated. Comparison of NfL levels between the Control-Saline and HF-Saline groups were analyzed with an unpaired t-test. Statistical significance was set at p value < 0.05. Association between NfL levels and discrimination ratio scores and the brain cytokines were analyzed using Pearson correlation. Statistical significance was set at p value < 0.05. The discrimination ratio vs NfL linear regression and brain cytokine vs NfL linear regression was used to generate a best-fit line.

Ang-(1-7)/MasR activation rescues cognitive impairment in VCID mice
As illustrated in Fig. 3A, treatment with Ang-(1-7)/MasR agonists rescues cognitive impairment in our mouse model of VCID. Cognitive function was measured using the NOR test following 24 days of treatment with either 50 or 500 µg/kg/day of either Ang-(1-7) or PNA5. Cognitive function was quantified using the Discrimination ratio score, with a negative, or zero Discrimination ratio indicating impaired cognitive function. HF mice treated with saline had an average negative Discrimination ratio of − 0.026 ± SE 0.07 that was significantly lower than Control-Saline treated mice (mean 0.590± SE 0.13, p < 0.001, ANOVA, Dunnett's post). This indicates, as previously described, that in our model of VCID, HF results in a decrease in cognitive function [6,7].

NfL plasma levels in VCID mice and effects of Ang-(1-7)/ Mas receptor activation
Levels of NfL were higher in the HF-Saline-treated mice relative to levels in the Control-Saline mice. The mean NfL level in the HF-VCID mice was 381.6 ± SE 30.0, (n = 10) as compared to Control-Saline-treated mice at 301.0 ± SE 32.0, (n = 12). Student's t-test resulted in p = 0.08 and did not reach significance (Fig. 4A). To assess the relationship between cognitive impairment observed in our VCID model and circulating NfL, NfL levels were correlated to the discrimination ratio score of all mouse groups (Fig. 4C). Plasma NfL was significantly negatively correlated with the discrimination ratio score (Pearson's r = − 0.43, p = 0.03), demonstrating that cognitive abilities decrease in this model as neuronal injury markers increase.
The relationship between levels brain cytokines and NfL was analyzed using Pearson correlation of NfL concentrations and brain cytokines ( Table 2, Fig. 7). IL-2, IL-13, IL-17, and the combined cytokine composite score were significantly negatively correlated with NfL plasma concentrations. Figure 8 is a Pearson's correlation matrix illustrating the correlation coefficients between NfL and those brain cytokines described in Table 2. These data show that (1) NfL is most strongly negatively correlated with the cytokine-composite score and, (2) the levels of IL-2, IL-17 and IL-13 are all strongly positively correlated with each other.

Discussion
Vascular contributions to cognitive impairment and dementia (VCID) and Alzheimer's disease-related dementias (ADRD) significantly contribute to the 47 million people world-wide who suffer with dementia. This number is estimated to increase to over 130 million people by 2050. A number of studies have shown that VCID and conversion to ADRD are strongly correlated with vascular disease, inflammation and decreased cerebral brain blood flow [21][22][23][24][25][26][27][28]. The relationship between vascular disease, cognitive function and progression to dementia and possible AD have been recently reviewed [4,29,30]. These authors successfully make the case for a close relationship between cardiovascular risk factors and risk for VCID and ADRD. Furthermore, conversion rates of VCID to dementia has been reported to be within 40-46% within 5 years of diagnosis of VCID [31,32]. There is an urgent unmet medical need for therapeutics to prevent cognitive decline in individuals at risk for VCID.
In this study, we demonstrate that treatment of our VCID mice with Ang-(1-7)/MasR agonists reverse cognitive dysfunction and decrease circulating NfL as compared to HF-saline-treated animals. This change in NfL level is significantly negatively correlated with cognitive function. These early preclinical data suggest that NfL levels might serve as a prognostic to identify changes in cognitive function in VCID. In addition, NfL significantly correlated with treatment induced changes in brain cytokines further suggesting that NfL may potentially also serve as a pharmacodynamic/ response biomarker for therapeutic target engagement and disease modification. Extensive studies on levels of NfL in humans with VCID are needed to confirm the utility and reliability of NfL as a biomarker for cognitive impairment in individuals at risk for VCID.

NfL as a biomarker in neurodegenerative diseases
NfL is released into the CSF and blood upon axonal damage [10]. Increases in CSF and blood concentrations of NfL have been found in multiple neurodegenerative diseases [33] as well as traumatic brain injury [14], hypoxic brain injury [16], and cardiac disease [17]. Blood NfL levels have been proposed as a biomarker of cognitive decline in Alzheimer's disease (AD) and Parkinson's disease (PD) patients [33, 34. It has been demonstrated that patients experiencing mild    Relationship between brain cytokine levels and NfL levels. A-E The relationship between brain cytokines IL-2, IL-5, IL-13, IL-17, IP-10, and plasma NfL levels were analyzed using Pearson correlation. F Brain cytokine composite score of IL-2, IL-13, IL-17 were significantly negatively correlated with NfL plasma concentrations cognitive impairment and dementia stages of AD have increased blood and CSF NfL levels [35], from these results it has been previously proposed that NfL may be able to help predict the progression of AD dementia in patients [33,34].
Additionally, it has been demonstrated that CSF NfL levels are increased in patients with vascular dementia [36,37] and that increased NfL levels are correlated to lower MMSE scores [37]. Further, it has been shown that patients with vascular dementia demonstrate a correlation between increased NfL serum levels and decreased cognitive impairment [38]. Studies including CADASIL (a genetic disorder affecting the small blood vessels in the brain), and small vessel disease (SVD) VCID patients also demonstrated increases of NfL correlated to worsened processing speed [39]. In the present study, we provide preclinical support for the hypothesis that NfL may be a prognostic biomarker, as described by the FDA BEST Biomarker Working Group [40] for the identification of cognitive impairment in individuals at risk for VCID.
In the present study we have also shown that treatment with Ang-(1-7)/MasR agonists can decrease NfL concentrations in our VCID model as compared to HF-saline treated animals and this decrease in NfL is inversely correlated to several pleiotropic cytokines found in the VCID mouse brain. The correlation of blood NfL levels with neurodegenerative treatment protocols has been demonstrated in previous human studies. In MS patients, studies have shown that blood levels of NfL correlate with MS related disability and can predict the longitudinal course of the disease as well as response to treatment [41][42][43].

Systemic and brain inflammation
A number of different mechanisms have been suggested to contribute to VCID [44,45]. Systemic inflammation has been linked to impaired cognitive function [46,47]. Specifically circulating TNFα is increased in HF patients [48,49] and has been demonstrated to contribute to pathologic cognitive changes [47]. In this study we observed that TNFα was significantly increased in the HF-Saline treated mice compared to the Control-Saline treated mice, thereby possibly contributing to systemic inflammation in our VCID mouse model. Importantly, treatment with Ang-(1-7)/MasR agonists in VCID mice reversed not only cognitive impairment, but also decreased NfL levels, and systemic TNFα concentrations. These data provide further evidence that NfL might serve as a biomarker to indicate target engagement of the disease-modifying treatment with our Ang-(1-7)/MasR agonists.
We have also begun to identify longitudinal changes in circulating inflammatory cytokine profiles in our VCID mouse model. We compared changes in systemic cytokines in animals with 3 weeks of HF to animals with 8 weeks of HF. Previously, we have identified inflammatory biomarkers associated with the early progression of VCID following the onset of HF (3 weeks following MI) [6]. IL-1α, MIP-1α, MIP-2, and GCSF in HF-Saline mice were significantly increased compared to the Control-Saline treated mice 3 weeks following MI [6]. These data reveal a longitudinal decrease in IL-1α, MIP-1α, and MIP-2 which provides insight into the inflammatory mechanism of VCID development.
IL-1α has been shown to have brain protective qualities in mouse stroke models and promotes neurorepair when mice are treated with delayed administration of IL-1α [50]. Other studies have demonstrated that IL-1α induces angiogenesis in brain endothelial cells in vitro [51]. Further it was observed that IL-1α levels increase within the brain in post-stroke animal models during angiogenic periods [51]. In the current study IL-1α was significantly Pearson's correlation matrix of the relationship between plasma NfL and brain cytokine levels. The relationship between the brain cytokines IL-2, IL-5, IL-13, IL-17, IP-10, the composite cytokine z-score and plasma NfL levels were analyzed using Pearson correlation. Blue demonstrates positive correlation and red represents negative correlation. Values closer to 1 or -1 indicate more closely correlated variables lower in HF-Saline treated mice compared to Control-Saline treated mice, and that treatment of P50, or A500 increases IL-1α significantly in comparison to HF-Saline mice. Suggesting that treatment with Ang-(1-7)/MasR agonists may provide brain protection and neurorepair by increasing brain IL-1α levels.
IL-2 has also been shown to play a protective role in the brain. IL-2 has been shown to decrease amyloid plaque load, improve synaptic plasticity, and has been associated with memory recovery in AD mice [52]. We observed that IL-2 brain levels significantly decreased in our VCID model in comparison to controls. We further demonstrated that treatment with P50, or A500 significantly increased IL-2 levels in comparison to HF-Saline mice, and that IL-2 levels were significantly negatively correlated with NfL levels. These data suggest that one of the mechanisms by which Ang-(1-7)/MasR treatment protects cognitive function may include increasing IL-2.
IL-5 is a Th2 cytokine that can act as an immunosuppressor [53] and contributes to maintaining homogeneity of astrocyte activation states [54]. Elevated serum levels of IL-5 have been have been reported in both VCID patients and vascular encephalopathy (VE) [53]. It has been suggested that elevated levels of IL-5 in VCID and VE may contribute to disease reduction. In PC12 cells, treatments of IL-5 decreased neuronal cell apoptosis, and Aβ 25-35 induced tau phosphorylation [54]. In the present study we show that treatments with P50, or A50 significantly increase brain IL-5 levels compared to HF-Saline mice. These data suggest that elevated IL-5 levels may contribute to the mechanism by which Ang-(1-7)/MasR treatment protects cognitive function. IL-13 has been shown to play both a protective and injurious role in the brain. While IL-13 has been demonstrated to be neuroprotective by modulating microglia/ macrophage responses in traumatic brain injury [55], other studies have shown that IL-13 can be detrimental to neuronal survival [56]. Here we found that treatment with P50, A50, and A500 significantly increase IL-13 levels in comparison to HF-Saline treated mice. The IL-13 levels were significantly negatively correlated to NfL plasma concentrations. These data indicate that IL-13 may play a neuroprotective role following treatment with Ang-(1-7)/MasR agonists in our mouse model of VCID. IL-17 has also been shown to play both a protective and injurious role in the brain. Although IL-17 is typically linked to proinflammatory disease progression [57], it has also been shown to promote tissue repair via meningeal resident gamma delta T-cells and support long-term potentiation (LTP) and the plasticity of glutamatergic synapses during short-term learning [58]. We observed that HF-Saline treated mice had significantly lower levels of IL-17 than Control-Saline treated mice, and that treatments with P50, A50, or A500 normalized IL-17 brain concentrations. Further, levels of IL-17 were significantly negatively correlated with Nfl levels. These findings suggest that Ang-(1-7)/MasR agonists increase IL-17 levels in VCID mice, which may contribute to these peptides' cognitive protective effects.
IP-10 (CXCL-10) has been demonstrated to increase in the pathogenesis of AD [59]. In contrast, studies with AD patients demonstrate an increased in CSF and plasma IP-10 levels were independently associated with NfL levels [60]. In our model of VCID, we observed IP-10 brain levels were significantly higher in both Control-Salinetreated mice and Ang-(1-7)/MasR agonist-treated HF mice as compared to HF-Saline mice.
In total, these data suggest that in our VCID model there is a decrease in the levels of multiple pleiotropic cytokines in HF-saline-treated animals. Treatment with Ang-(1-7)/MasR agonists mitigated cognitive impairment in this model while simultaneously increasing the levels of these pleotropic, and some putative neuroprotective cytokines, some of which are correlated to decreases blood NfL levels.
Finally, while extensive human studies are needed to support this premise, we propose that NfL may be a potential predictive biomarker to identify individuals who are at risk for VCID and pharmacodynamic biomarker to indicate target engagement following Ang-(1-7)/MasR therapy to protect cognitive function.

Study limitations
One limitation of this study is the lack of data in female animals. We appreciate that effects of both HF and drug treatment may be different in female animals, and we have plans to fully investigate the effects of HF and Ang-(1-7)/MasR treatments in female mice in future studies. A second limitation of this study, is the lack of complete longitudinal data showing the change in circulating and brain cytokines over the course of development of the HF post-MI. While our results comparing 3-week to 8-week post-MI levels of cytokines in both plasma and brain suggest that the levels of these cytokines do change during disease progression a more extensive analysis of the time-course of these changes and the time-course of the effectiveness of treatment with Ang-(1-7)/MasR therapies will be needed to better understand the inflammatory mechanisms underlying the development of VCID.