Plasma Dickkopf-related Protein 1, an antagonist of the Wnt pathway, is associated with HIV-associated neurocognitive impairment

Abstract

Objective

DKK1 is a soluble antagonist of the Wnt pathway. It binds to and sequesters LRP5/6 away from Wnts. Because the Wnt pathway regulates synaptic transmission and plasticity, we hypothesized that increased DKK1 would increase the risk for neurocognitive impairment (NCI) in HIV+ individuals. We evaluated here the relationship between plasma DKK1 and global NCI.

Methods

Plasma samples and data from 41 HIV+ and 42 HIV− adults were obtained from the University of California, San Diego. Concentrations of DKK1 and a comparator protein, MCP-1, were quantified in plasma by immunoassay. All subjects completed a standardized comprehensive neuropsychological test battery and their performance was summarized using the global deficit score (GDS) method.

Results

A higher DKK1 level was associated with NCI among HIV⁺ participants (d=0.63, p=0.05), particularly among the 26 participants whose plasma HIV RNA level was suppressed (d=0.74, p=0.08). DKK1 level was not associated with NCI among HIV− participants (p=0.98). MCP-1 was not associated with NCI in either group. In HIV+ adults with suppressed plasma HIV RNA, a receiver-operator characteristic curve identified that a DKK1 level of at least 735 pg/ml had a positive predictive value of 83.3% for a diagnosis of NCI. This association did not weaken after accounting for the effect of AIDS, nadir CD4⁺ T-cell count, addictive drug use, or demographic characteristics.

Conclusion

DKK1 is a specific biomarker for NCI in HIV+ adults, implicating the Wnt pathway in HIV neuropathogenesis.

Introduction

The introduction of combined antiretroviral therapy (ART) transformed HIV to a chronic disease with several potential complications, one of which is HIV-Associated Neurocognitive Disorder (HAND). HAND refers to a spectrum of neurocognitive impairments associated with HIV neuroinvasion and neuroinflammation that leads to impairment in attention, memory, language, problem solving, and decision-making, which significantly impacts day-to-day function.[1] The spectrum of impairments includes Asymptomatic Neurocognitive Impairment (ANI), Mild Neurocognitive Disorder (MND), and HIV-Associated Dementia (HAD). Even when ART suppresses HIV RNA, HAND can still affect up to 50% of HIV+ adults [2, 3] and its prevalence is expected to increase with aging. This underscores the need to define biomarkers that can accurately predict the chance for developing HAND, reveal mechanisms that drive HAND, and ultimately devise novel strategies to prevent and/or treat HAND.

We assessed whether a molecule related to Wnt signaling is a clinical biomarker of HAND. The Wnt pathway plays critical roles in the developing[4] and adult brain[5], including regulation of synaptogenesis and plasticity[6]. Wnts are a family of 19 highly conserved, small secreted glycoproteins that bind to the seven-transmembrane Frizzeled receptors and its co-receptor, LDL receptor-related proteins 5 and 6 (LRP 5/6), to engage a signaling cascade that culminates in β-catenin dependent or independent signaling. Downstream genes of the Wnt pathway play critical roles in cell communication, cell cycle, cell survival, cell renewal/stemness, differentiation, and organogenesis, to name a few[7, 8]. Dysregulation in Wnt signaling has been linked to neurodegenerative diseases, such as Alzheimer’s disease, amyotrophic lateral sclerosis, and Parkinson’s disease[6, 7]. We established that Wnt/β-catenin is a restriction factor for HIV in astrocytes, monocytes, and T cells, and elucidated the relevant molecular mechanisms [9–14]. We also established that inhibition of β-catenin has a dramatic effect on the ability of astrocytes to scavenge for glutamate by inhibiting the expression of Excitatory Amino Acid Transporter 2 (EAAT-2/GLT-1 in rodents) and glutamine synthetase [15]. Our working model is that signals that diminish Wnt/β-catenin will compromise brain homeostasis and contribute to HAND.

Because β-catenin is an intracellular protein, it is not an ideal biomarker candidate. For this reason, we turned to Dickkopf-related protein 1 (DKK1). DKK1 is a secreted soluble antagonist of the Wnt pathway. It binds to the co-receptor of Wnts, LRP 5/6 and sequesters it away from Wnts. We showed that an inflammatory signal (e.g., interferon-γ) inhibits Wnt/β-catenin signaling in astrocytes through induction of DKK1 in a STAT3-dependent manner[12]. Recent studies demonstrate that DKK1 induces the rapid disassembly of synapses in mature neurons, and it mediates synaptic loss induced by Amyloid-β [16]. These emerging studies point to a relationship between DKK1 and neurodegenerative disorders. We assessed here whether DKK1 levels in plasma are associated with worse global neurocognitive (NC) performance in a cohort of HIV+ and HIV− adults.

Methods

Human research ethical statement

The study was conducted in accordance with the Declaration of Helsinki. The Human Research Protections Programs of the University of California San Diego and Rush University Medical Center approved the human rights and protection aspects of the study and all participants provided written, informed consent.

Patient and Sample Selection

Participants were recruited as part of the Translational Methamphetamine AIDS Research Center (TMARC) project, an ongoing multidisciplinary investigation of the effects of HIV infection and methamphetamine abuse on NC performance and other outcomes. HIV status and seronegative status for Hepatitis C virus (HCV) were confirmed by immunoassay (MedMira Inc., Nova Scotia, Canada). Blood was collected by standard methods and current CD4+ T-cell count (cells/μL) was determined by flow cytometry at a Clinical Laboratory Improvement Amendments (CLIA)-certified lab. HIV RNA levels were measured in plasma by reverse transcriptase-polymerase chain reaction (Roche Amplicor, v. 1.5, lower limit of assay at 50 copies/ml). Nadir CD4+ T-cell count and duration of HIV disease were obtained by self-report.

Eligible participants were at least 18 years of age, without a prior head injury resulting in loss of consciousness for more than 30 minutes; previous cerebrovascular event; or a prior or current seizure disorder, or demyelinating disease. Chronic renal or pulmonary disease or other medical conditions that could confound NC performance also were exclusionary. Exclusion criteria also included lifetime diagnosis of schizophrenia or other psychotic disorder, as assessed with the Composite International Diagnostic Interview (CIDI version 2.1)[17] (Kessler & Ustun, 2004) using DSM-IV criteria (American Psychiatric Association, 2000).

NC Testing

All subjects were assessed using a standardized comprehensive NC test battery that assessed seven cognitive domains and adhered to recommended guidelines.[18] NC performance was summarized using the Global Deficit Score (GDS) method. GDS was calculated for each participant based on demographically corrected T–scores. The battery and its GDS summary score have previously demonstrated sensitivity to the effects of HIV on neuropsychological performance [19].

DKK1 and MCP-1 Immunoassays

The plasma samples were diluted four times with Calibrator Diluent and DKK1 amount measured in duplicates using the Human DKK1 Quantikine ELISA Kit (DKK100), according to the manufacturer’s instructions (R&D Systems, Minneapolis, MN). Bound DKK1 signal was developed with 100μl of tetramethylbenzemidine microwell peroxidase substrate system (KPL, #50-76-00). The reaction was stopped with addition of 100μl 1M H₃PO₄. The optical density was acquired in a microplate reader set to 450 nm and a correct set to 540 nm. DKK1 levels were calculated from a standard curve generated from DKK1 standard in the same plate. Human plasma MCP-1 levels were similarly measured using Human MCP-1 Quantikine ELISA Kit (DCP00, R&D Systems) per manufacturer’s instruction.

Statistical Analysis

Data distributions were inspected and transformed as necessary to improve symmetry. For example, log₁₀-transformed DKK1 concentrations were used in analyses. Comparisons were made using linear (e.g., continuous GDS) or logistic (e.g., binary NCI) regression. Planned stratified analyses were performed based on HIV serostatus (HIV+ vs. HIV-) and HIV RNA suppression (HIV RNA ≤ 50 copies/mL vs. HIV RNA > 50 c/mL). Since the small number of HIV+ participants limited use of multivariate regression in this subgroup, the influence of individual covariates was evaluated by analysis of covariance. Considering the limited power of this analysis, no adjustment for type I error was imposed.

Results

Demographic and clinical characteristic of study participants

Eighty-three participants were included: 41 HIV+ and 42 HIV− adults between the ages of 18 and 68 (Table 1). The HIV+ and HIV− groups did not differ in age, ethnicity, education, illicit drugs such as methamphetamine dependence, or Hepatitis C virus (HCV) serostatus (Table 1). Among HIV⁺ adults, NCI was associated with AIDS, ART use, and a trend toward lower nadir CD4+ T-cell count (Table 1).

Table 1.

Demographic and clinical characteristics of all study participants.

All Participants (n=83)	HIV−	HIV+	p value	Unimpaired	Impaired	p value
	42	41		58	25
Age (Years)	38.8 (12.7)	42.6 (8.9)	0.12	41.3 (10.9)	39.3 (11.6)	0.45
Sex (Women)	9 (21.4%)	6 (14.6%)	0.42	8 (13.8%)	7 (28.0%)	0.13
Ethnicity (non-White)	20 (47.6%)	13 (31.7%)	0.40	22 (37.9%)	11 (44.0%)	0.90
Education (Years)	12.9 (1.7)	13.2 (2.4)	0.56	13.0 (1.9)	13.3 (2.4)	0.61
Methamphetamine Dependence	20 (47.6%)	22 (53.7%)	0.58	25 (43.1%)	17 (68.0%)	0.06
Urine Drug Screen Positive	1 (2.4%)	0 (0%)	0.25	1 (1.7%)	0 (0%)	1.00
HCV Seropositive	0 (0%)	0 (0%)	1.00	0 (0%)	0 (0%)	1.00
HIV+ Participants (n=41)		41		24	17
AIDS Diagnosis	-	26 (63.4%)	-	10 (41.7%)	16 (94.1%)	0.0008
Duration of HIV Disease (Months)	-	129.2 (97.4)	-	115.9 (95.1)	148.0 (100.5)	0.31
Current CD4+ T-cells (/μL)	-	422.6 (192.5)	-	445.9 (200.9)	387.6 (179.8)	0.34
Nadir CD4+ T-cells (/μL)	-	198.4 (148.0)	-	234.2 (155.0)	147.8 (124.8)	0.06
Current ART Use	-	29 (70.7%)	-	13 (54.2%)	16 (94.1%)	0.006
Plasma HIV RNA ≤ 50 c/mL among ART Users	-	25 (86.2%)	-	12 (92.3%)	13 (81.2%)	0.60

Values are either mean (standard deviation) or number (proportion).

Plasma DKK1 levels are higher among HIV+ participants with NCI

Plasma DKK1 levels did not differ by HIV serostatus (656.7 pg/ml vs. 598.0 pg/ml; p=0.54; Fig. 1A). Plasma DKK1 levels also did not differ by NC performance among the combined HIV+ and HIV− groups, either when expressed as continuous GDS (p=0.46) or binary NCI (Fig. 1B). Planned stratified analyses identified that DKK1 levels did differ by NCI among HIV⁺ participants (mean 813 pg/ml vs. 549 pg/ml, p<0.05; Fig. 1C), particularly those taking ART with HIV RNA levels ≤ 50 c/mL (d=0.74, p=0.079; Fig. 1D). In this subgroup, a receiver-operator characteristic (ROC) curve identified that DKK1 levels ≥ 735 pg/mL were highly specific for NCI (specificity 92.3%, odds ratio=7.5, p=0.054, Fig. 2) with high positive predictive value (83.3%), although DKK1 levels were not sensitive for detecting NCI (38.5%). ANCOVA identified that the relationship between the binary DKK1 variable and NCI was not significantly weakened by accounting for the covariance of AIDS status, nadir or current CD4+ T-cell count, addictive drug use variables, age, sex, or ethnicity (data not shown).

Figure 1. Plasma DKK1 levels are higher among HIV+ patients who are neurocognitively impaired.

Plasma DKK-1 levels were measured by ELISA from 43 HIV− and 41 HIV+ patients obtained from the University of California, San Diego. A) Plasma DKK1 level among HIV− and HIV+ subjects. B) Plasma DKK1 levels among all subjects (HIV+ & HIV-) classified as unimpaired or impaired. C) Plasma DKK1 levels among HIV+ subjects classified as unimpaired or impaired. D) Plasma DKK1 levels among HIV+ subjects with undetectable viral load (≤ 50 copies /mL) classified as unimpaired or impaired. Median value of DKK1 (pg/ml) is shown. p<0.05 denote statistical significance, as measured by analysis of covariance (ANCOVA).

Figure 2. Plasma DKK1 level is associated with Global Deficient Score among HIV⁺ subjects on ART.

HIV⁺ subjects who are on suppressive ART (HIV plasma viral load ≤ 50 copies /mL) were grouped as those diagnosed with HAND, based on GDS score, or those without HAND and portioning analysis (DKK1 ≥ 735 pg/ml) applied to plasma DKK1 level. OR denotes odds ratio. p=0.05 as measured by analysis of covariance (ANCOVA).

Among the seven cognitive domains that were assessed, higher DKK1 levels were specifically associated with worse working memory (d=0.77, p=0.05). P values for comparisons with impaired performance in other domains were > 0.20 except for motor performance (d=0.69, p=0.149).

Plasma MCP-1 is higher among HIV⁺ than HIV⁻ subjects, but is not associated with NC status

MCP1 is a soluble chemoattractant protein that is produced by activated astrocytes to recruit monocytes/macrophages into CNS [20, 21]. We measured plasma MCP-1 as a comparator biomarker for this DKK1 study. Plasma MCP-1 levels were higher among HIV+ than HIV− subjects (mean 208 pg/ml vs. 173 pg/ml; p= 0.02; Fig. 3A). In contrast to DKK1, MCP-1 levels were not associated with GDS or NCI either in the entire group (Fig. 3B), in the HIV+ subgroup (Fig. 3C), or in those with HIV RNA ≤ 50 c/mL (Fig. 3D). An ROC curve failed to identify a diagnostically useful threshold value for MCP-1 (238 pg/mL: sensitivity 69.2%, specificity 30.8%, positive predictive value 50.0%).

Figure 3. Plasma MCP-1 level is higher among HIV+ than HIV− subjects, but is not associated with HAND.

Plasma MCP-1 was measured by ELISA and compared between HIV⁺ (n=41) and HIV⁻ subjects (n=43) in (A) and between those who are unimpaired vs. impaired in (B).

Discussion

HIV invades the CNS within two weeks of infection and sets the stage for inflammation and dysregulation in the brain that ultimately leads to HAND in a substantial proportion of HIV+ adults. While HAND may take years to present, advances in neuroimaging techniques demonstrate considerable loss in brain volume as early as a year after HIV infection [22]. Neuroimaging (e.g. fMRI and PET) is expensive, underscoring the need for biomarkers of HAND which can reveal mechanisms driving HAND, inform therapeutic strategies, and point-of-care monitoring, especially in the settings of limited resources.

Recent data showed that plasma HIV DNA levels are associated with mild NCI [23] as is differential methylation of DNA loci in monocytes [24]. Putative markers such as MCP-1, CSF albumin and neopterin correlate with HIV encephalitis (HIV-E)[3]. HIV-E, however, has a clear clinical manifestation that does not require a biomarker, and HIV-E is also less prevalent in the era of ART. Further, other biomarkers that have been linked to ANI/MND such as those reflecting cell stress (e.g., ceramide, sphingomyelin), oxidative stress (e.g., protein carbonyls and hemoxygenase), energy metabolism (e.g., Krebs cycle substrate), and glutamate regulation (e.g., glutamine) require lumbar puncture [3], which is less practical in the clinical setting than venipuncture. Further, most HIV+ patients experience mild to moderate NCI that nonetheless significantly impacts their day-to-day function [3, 25] and ANI has a 6-fold odds ratio of predicting symptomatic neurocognitive decline [26]. These ongoing considerations demand biomarkers that are inexpensive and less invasive, which can diagnose or predict the milder forms of HAND.

Although plasma sCD14 and sCD163 are elevated in ANI/MND, these markers are indicative of monocyte/macrophage inflammation which is well documented in HAND pointing perhaps to global inflammation rather than specific mechanisms driving neuroinflammaton in HAND. Likewise, blood neurofilament-light protein was reported to be associated with lower blood CD4+ T cell count and HAD and with cognitive functions[27]. Again, while this link points to neuronal injury, it does not inform the mechanisms driving HAND.

We evaluated here a putative soluble biomarker for HAND, DKK1. DKK1 is an antagonist of the canonical Wnt/β-catenin pathway. It binds to Wnt co-receptor LRP5/6 to prevent LRP5/6 interaction with Wnt-Frizzled complexes. Dkk1 also reduces cell surface levels of LRP5/6 by binding to Kremen-1 or -2 and triggering the internalization of LRP5/6 from the cell surface. DKK1 is a potential biomarker for ankylosing spondylitis[28] and osteoarthritis[29], as well as cancers[30]. In our study, the association between DKK-1 and NCI points to dysregulation of Wnt signaling as a critical mechanism driving HAND. This revelation can set the stage for novel therapeutic to target normalization of Wnt signaling as a novel therapeutic strategy in HAND.

Since Wnt signaling is critical for synapse development and maintenance[8], antagonizing Wnt signaling through DKK1 leads to synapse disassembly[16]. Further, soluble DKK1 infusion impaired cognitive functions in mouse models of hippocampal memory consolidation [31]. We showed here that plasma DKK1 levels could serve as a biomarker for HAND. Plasma DKK1 levels were not correlated with HIV status. However, among HIV+ subjects, cognitively impaired subjects have significantly higher levels of plasma DKK1 than unimpaired subjects. Even in this small sample, the clinically relevant subgroup of HIV+ participants who had suppressed HIV RNA and DKK1 levels more than 735 pg/ml had a high positive predictive value for NCI. Importantly, in the current study the association between elevated DKK1 and HAND did not significantly weaken after accounting for several influential characteristics. In contrast, MCP-1, a biomarker comparator in this analysis, was not associated with NCI. Moreover, the high positive predictive value and high specificity suggest that DKK1 may be useful to screen for elevated HAND risk, providing that the current study results can be replicated in a larger sample.

The premise then is that higher DKK1 levels during ART will lead to reduction in Wnt/β-catenin signaling. Dysregulated Wnt/β-catenin signaling including elevated levels of DKK1 has been reported in a number of neurological diseases,[7] such as autism [32], schizophrenia[33, 34] and Alzheimer’s disease[35, 36]. SNP analysis showed nominally significant association with late onset Alzheimer’s disease for rs1881747 near DKK1 (P = 0.011, odds ratio = 1.24).[37] DKK1 was detected in degenerating neurons in the brain from Alzheimer’s patients, where it co-localized with neurofibrillary tangles and dystrophic neuritis [36]. DKK1 promotes neuron degeneration. Expression of DKK1 is required in the ischemic and excitotoxic neuronal death [38]. DKK1 Treatment caused degeneration of the excitatory synapses in in vitro hippocampus slice culture [16]. In an inducible transgenic mouse model, DKK1 expression in striatum caused degeneration of corticostriatal glutamatergic synapses and decrease in D1 and D2 dopamine receptor clusters, and impaired striatal-regulated behaviors, e.g., a failure to respond to amphetamine stimulation[39]. In vivo infusion of DKK1 into mouse hippocampus caused cognitive impairment/object cognitive memory consolidation [31], conversely activation of Wnt pathway in hippocampus enhanced cognitive functions [40]. Further, DKK1 inhibits neurogenesis in brain. Forced expression of DKK1 severely reduces neurogenesis in the developing mouse hippocampus [41]. DKK1 accumulates in the aging dentate gyrus, in contrast, loss of DKK1 restores neurogenesis in hippocampus and counteracts cognitive decline in old mice [42]

Increased levels of DKK1 may disrupt neuronal function directly (e.g. degeneration of synapses) and/or through impairing astrocyte functions. Astrocytes are in close contact with synapses, astrocyte processes form tripartite synapse together with pre-synaptic region and post-synaptic density. Astrocytes are implicated in Wnt signaling-mediated neuroinflammation and consequently brain damage[43]. Lastly, plasma DKK1 may indicate a heightened level of peripheral inflammation that impairs neuronal/glial health.

Our data suggests that DKK1 is a promising biomarker for HAND that may be distinct from those that reflect monocyte activation (e.g., MCP-1). Larger, longitudinal studies are warranted to assess whether DKK1 is a prognostic biomarker for HAND progression. Nonetheless, our findings provide intriguing human data that support substantial in vitro and animal experimental data on DKK1 and Wnt signaling in neurodegenerative disorders.