Opioid Modulation of Neuronal Iron and Potential Contributions to NeuroHIV

Abstract

Opioid use has substantially increased over recent years and remains a major driver of new HIV infections worldwide. Clinical studies indicate that opioids may exacerbate the symptoms of HIV-associated neurocognitive disorders (HAND), but the mechanisms underlying opioid-induced cognitive decline remain obscure. We recently reported that the μ-opioid agonist morphine increased neuronal iron levels and the levels of ferritin proteins that store iron, suggesting that opioids modulate neuronal iron homeostasis. Additionally, increased iron and ferritin heavy chain protein were necessary for morphine’s ability to reduce the density of thin and mushroom dendritic spines in cortical neurons, which are considered critical mediators of learning and memory respectively. As altered iron homeostasis has been reported in HAND and related neurocognitive disorders like Alzheimer’s, Parkinson’s, and Huntington’s disease, understanding how opioids regulate neuronal iron metabolism may help identify novel drug targets in HAND with potential relevance to these other neurocognitive disorders. Here, we review the known mechanisms of opioid mediated regulation of neuronal iron and the corresponding cellular response and discuss the implications of these findings for patients with HAND. Furthermore, we will discuss a new molecular approach that can be used to understand if opioid modulation of iron affects the expression and processing of amyloid precursor protein and the contributions of this pathway to HAND.

1. Introduction

Opioid compounds have been the standard of care for pain management and other related conditions for many years. However, these compounds have strong abuse liability, and the increased use of prescription and illicit opioid drugs over recent years has created an epidemic of opioid addiction and opioid-related overdose deaths in the United States [1,2]. The overuse of prescription opioid drugs is thought to lead to increased use of injection opioid compounds like heroin [2,3], and a recent study suggests that the use of heroin as an initiating opioid has also increased in recent years [4]. Importantly, individuals that share opioid injection equipment are more likely to become infected with bloodborne pathogens including hepatitis C virus and human immunodeficiency virus (HIV) [5–7], and HIV-infected individuals are more likely to use opioid drugs than the general population [8,9]. Opioid use and abuse also reportedly worsens the neurological complications of HIV-infection, also known as HIV-associated neurocognitive disorder (HAND) [10,11] or simply neuroHIV.

HAND is a spectrum of neurocognitive impairment that affects critical cognitive domains for everyday life, including working memory, executive function, and information processing speed [12,13]. HAND likely results from HIV infection of select non-neuronal cells in the central nervous system (CNS) in combination with the low-level expression of HIV-related neurotoxins and chronic inflammation [14]. In modern times, the severity of HAND has been dramatically decreased by antiretroviral therapies, which can reduce viral replication to undetectable levels in compliant patients and increase their lifespan to that of an uninfected individual [14]. However, treated patients with well-controlled infection can still develop HAND, although most cases present with mild or intermediate symptoms [14]. Additionally, since age is a risk factor for HAND, the increased life expectancy of these patients is also associated with an increased overall prevalence of the disorder [15]. There are currently no FDA approved therapies for HAND [16,17], demonstrating that our understanding of the pathology in the ART era is inadequate. Further, the mechanisms by which opioids exacerbate HAND symptoms remain obscure, and new research designed to uncover these mechanisms and their impact on neuronal structure and function in specific brain areas could lead to the discovery of novel and effective drug targets for opioid-using patients with HAND.

Several groups have reported that the μ-opioid agonist morphine reduces dendritic spine density of neurons from various brain regions [18], and since these spines are thought to facilitate learning and memory processes [19,20], their loss could underlie opioid-induced cognitive impairment in HAND patients. Building from these studies, our group showed that morphine specifically reduces mature types of dendritic spines on cortical neurons, which was dependent on intracellular iron stored in endolysosomes [21]. Shortly after morphine exposure, cortical neurons showed decreased endolysosomal iron levels and increased labile iron levels in the cytoplasm, which promoted cytosolic upregulation of the iron storage protein ferritin heavy chain (FHC) [21]. We previously characterized FHC as a critical mediator of morphine’s effects on cortical dendritic spine density [22,23], which is at least partially due to its ability to inhibit the homeostatic, neuroprotective, and dendritic spine promoting CXCL12/CXCR4 chemokine signaling axis [24,25]. Together, these data suggest that morphine and other μ-opioid agonists may contribute to cognitive impairment in HAND through their novel effects on neuronal iron metabolism.

This review will discuss opioids effects on neuronal iron metabolism in three major sections. First, we will present the current knowledge of opioid-mediated regulation of endolysosomal iron, briefly discuss the endolysosomal system in neurons, and explore potential mechanisms by which opioids could drive the efflux of iron from neuronal endolysosomes into the cytoplasm based on recent literature and our initial findings. Second, we will describe the currently known effects of morphine and other μ-opioid compounds on iron-related proteins in neurons and consider the potential pathways that underlie these effects. Additionally, we will discuss evidence suggesting that opioids and iron may impact the expression of amyloid precursor protein (APP), which could enhance amyloidogenesis in HAND patients. Third, we will discuss newly developed molecular tools to study APP processing changes by opioids and other stimuli. Understanding how μ-opioids modulate neuronal iron metabolism and APP processing may provide new insights about the mechanisms of cognitive impairment in HAND and a large list of other neurocognitive disorders that present with altered neuronal iron metabolism.

2. μ-opioid regulation of neuronal iron

Our recent studies show that morphine-mediated activation of the μ-opioid receptor Gαi-protein pathway alters neuronal iron metabolism for an extended period after drug exposure, and this is critical for morphine’s ability to regulate dendritic spine density and morphology [21]. In primary cortical neurons, morphine decreased endolysosomal iron levels as early as 30 minutes after exposure and increased labile iron levels in the cytoplasm at the same time, suggesting that morphine promoted endolysosomal iron efflux to the cytoplasm [21]. Labile cytoplasmic iron remained elevated after 30 minutes. At the same time we observed a morphine-mediated upregulation of the iron storage proteins FHC and FLC [21]. Using iron chelators with different properties, we demonstrated that endolysosomal iron is required for morphine to upregulate FHC and reduce dendritic spine density, while extracellular iron does not seem to impact this pathway. Therefore, endolysosomal iron stores appear to be a critical and relatively unexplored mediator of morphine’s actions on neurons, and our evidence also suggests that this finding may also extend to other cell types. For example, morphine reduced endolysosomal iron and increased cytosolic labile iron in cultured rat hippocampal neurons and human U87-MG glioblastoma cells. Follow up experiments in U87-MG cells confirmed that also morphine’s effects in non-neuronal cells were mediated by opioid receptors and blocked by chelation of endolysosomal iron [21].

2.1. Opioid modulation of neuronal endolysosomes

Although we will discuss the essential elements of endolysosomal physiology and their relationship to opioid use here, recent reviews have explored the neuronal endolysosomal system in greater depth [26] as well as its roles in HAND and Alzheimer’s disease [27]. The term endolysosome encompasses all the distinct endosome and lysosome-like organelles expressed throughout the cell. Neurons can create early endosomes through endocytosis of the cell membrane, and these organelles are then sorted into several pathways that can result in recycling of their cargo back to the plasma membrane or transport into the cell via late endosomes. The sorting process is influenced by specific protein associations with endosomes, as early endosomes develop into late endosomes following exchange of the endosomal protein Rab5 for Rab7 [28,29]. Rab7 expressing late endosomes are transported towards the soma, where they may fuse with acidic lysosomes to degrade their cargo [30]. In line with this, one group showed that silencing of the late endosomal protein Rab7 in peripheral neurons improved opioid responsiveness in a rat model of diabetes by preventing the degradation of internalized μ-opioid receptors [31]. Endolysosomes in neurons are spatially organized with more acidic organelles prevalent in the soma and proximal processes and less acidic organelles prevalent in distal processes [32,33]. This organization is also observed in non-neuronal cells [34], suggesting that general properties of the system may be preserved in different cell types. However, neurons also express a small fraction acidic endolysosomes in the distal processes and near dendritic spines, which may be capable of small-scale protein degradation [35].

Our studies show that morphine reduces iron content of endolysosomes that express lysosomal-associated membrane protein 1 (LAMP1), which are mostly localized in the soma and proximal processes [21,33]. This led to increased iron levels in the neuronal cytoplasm and later upregulation of FHC. Interestingly, our immunofluorescence studies suggest that morphine increases FHC levels in the soma as well as more distal neuronal processes [21], which could be due to multiple factors. The iron released from endolysosomes could diffuse from the soma into the processes over time, as we did not observe FHC upregulation until at least 6–24 hours after morphine treatment. Additionally, morphine could release iron from a subset of LAMP1 expressing endolysosomes that localize in neuronal processes [33], or even promote a redistribution of soma-localized endolysosomes towards the processes due to its ability to increase endolysosomal pH. Importantly, morphine’s effects on endolysosomal pH were blocked by the opioid antagonist naloxone, demonstrating that increased endolysosome pH and iron efflux both occur via activation of opioid receptors [21]. This pathway may also be relevant in other cell types, as naloxone also blocked morphine effects on endolysosomal pH and iron release in U87-MG glioblastoma cells [21]. The pH of endolysosomes also plays a critical role in their normal functions [36], suggesting morphine’s effects on endolysosomal physiology may have further reaching effects than altered iron metabolism.

2.2. Potential mechanisms driving opioid-mediated iron efflux from endolysosomes

The molecular mechanism leading to morphine-mediated endolysosomal iron release is still unclear, but some initial experimental observations have shed light on the pathway. At the beginning of the pathway, morphine’s effects on endolysosomal iron efflux were completely blocked by the opioid receptor antagonist naloxone [21], demonstrating that opioid signaling is required. Downstream of iron efflux, the upregulation of FHC and subsequent dendritic spine deficits were blocked by the specific μ-opioid receptor antagonist CTAP and by the Gαi-protein inhibitor pertussis toxin [21], suggesting that activation of the μ-opioid receptor G-protein pathway is the specific driver of initial endolysosomal iron efflux. Beyond this point, the signaling pathway is not completely worked out. The μ-opioid receptor signals through several kinases and modulates several membrane channels [37], either of which could play a role in the pathway. Additionally, other factors could contribute to the initial part of this pathway, including μ-opioid receptor recruitment of β-arrestin after G-protein activation [38], expression of μ-opioid receptor splice variants [39], and dimerization of the μ-opioid receptor with an increasing list of interacting receptors [40].

An additional complication is that the μ-opioid receptor can be activated in several distinct subcellular regions. Cell membrane bound μ-opioid receptors transmit the signals of opioid compounds in the extracellular space, but the endocytosed receptor may interact with different proteins and activate additional signaling pathways [41]. Interestingly, one group showed that different types of μ-opioid agonists activate μ-opioid receptors expressed in distinct cellular organelles. Using a cellular system that expressed a nanobody that binds activated μ-opioid receptors, their study showed that the peptide μ-opioid agonist DAMGO activated μ-opioid receptors at the cell membrane and again in early endosomes, while the small-molecule μ-opioid agonist morphine additionally activated a separate collection of μ-opioid receptors in the Golgi apparatus [42]. Another group used a proteomic approach to show that morphine and DAMGO alter the distribution of μ-opioid receptors expressed in HEK293 cells, which was associated with spatially distinct downstream signaling profiles [43]. Therefore, it is possible that plasma membrane-localized μ-opioid receptor signals as well as endolysosomal or trans-Golgi localized μ-opioid receptor signals could contribute to endolysosomal iron efflux. However, morphine and DAMGO both upregulate FHC in cortical neurons [24], suggesting that opioids effects on neuronal iron metabolism are due to a conserved μ-opioid receptor signaling pathway at or near the cell membrane as opposed to an alternative pathway driven by receptors with a specific subcellular localization.

μ-opioid receptor effectors could facilitate endolysosomal iron efflux by modulating one or more endolysosomal resident channels and transporters (figure 1). These protein conduits regulate the efflux of iron, calcium, and other cations, control endolysosomal luminal pH, and play critical roles in homeostatic endolysosomal functions [44,45]. Morphine-mediated efflux of endolysosomal iron through a transporter that uses the endolysosomal proton gradient could lead to simultaneous de-acidification of endolysosomes - or - morphine-mediated de-acidification of endolysosomes could be the driver of iron efflux through several conduits. Therefore, morphine’s effects could be mediated by one or several endolysosomal channels and transporters. The first potential mechanism involves two-pore channels (TPCs), which have been previously reported to play important roles in endolysosomal calcium efflux, trafficking, and morphology [46]. These channels were recently reported to transfer iron out of the endolysosome when treated with the TPC agonist nicotinic acid adenine dinucleotide phosphate (NAADP), and this iron efflux was blocked by the TPC antagonist Ned-19 or by TPC knockdown [47]. Further, cytotoxicity from iron efflux through TPCs was dependent on activity of the small GTPase Rab7a, as a constitutively active Rab7a increased apoptosis in iron loaded cells, while cytotoxicity was reduced in these cells by the Rab7 inhibitor CID1067700, a dominant negative Rab7a, or a TPC mutant that does not bind Rab7a [47]. Rab7a induced cytotoxicity was also reduced by the TPC antagonist Ned-19, suggesting that Rab7a activation occurs downstream of TPCs [47]. Importantly, activation of TPCs can also lead to proton release from endolysosomes, thereby increasing their luminal pH [48,49]. The findings in these reports are mostly consistent with those of the morphine pathway, suggesting that morphine could activate endolysosomal TPCs through a yet to be determined mechanism.

Figure 1.

Signaling pathways that may underlie morphine-mediated efflux of endolysosomal iron. A. μ-opioid receptor (μOR) signaling may lead to iron efflux through endolysosomal two-pore channels (TPC) by enhancing levels of their endogenous agonist NAADP or by promoting their modulation by the small GTPase Rab7a. B. μ-opioid activation of neuronal nitric oxide synthase (nNOS) could enhance iron flux through endolysosomal divalent metal transporter-1 (DMT-1) by directly S-nitrosylating the transporter as well as the small GTPase dexras1, which then interacts with DMT-1 and increases its efficiency. C. Morphine’s ability to increase endolysosomal pH could be the driver of endolysosomal iron efflux via TPCs, DMT-1, or other conduits. D. Signaling from μ-opioid receptors that were internalized or localized to endolysosomes could induce iron efflux through one or several nearby conduits.

In addition to TPC-mediated iron efflux, another potential mechanism by which morphine may enhance endolysosomal iron efflux involves the well-characterized iron and heavy metal transporter divalent metal transporter-1 (DMT-1) [50,51]. There are several different isoforms of DMT-1 that are expressed in distinct subcellular locations and are differentially expressed in response to labile iron [52,53]. Interestingly, DMT-1 isoforms that are expressed in response to labile iron localize to endolysosomes, while DMT-1 isoforms that are insensitive to iron are mostly expressed in the nucleus and on the cell membrane [54]. Additionally, DMT-1 is an iron and proton symporter in acidic environments [55], so iron efflux through DMT-1 on acidic endolysosomes may also increase their luminal pH. Several reports suggest that the optimum pH for DMT-1 iron transport is 5.5 [52,53], which is very similar to what we observed in endolysosomes of morphine treated cortical neurons [21]. Further, some recent studies have highlighted that various stimuli can activate or increase the efficiency of endolysosomal DMT-1 iron flux in neurons. In one study, enhancing neuronal activity by adding KCl to culture media caused a DMT-1 mediated efflux of iron from lysosomes to the cytoplasm, which correlated with a reduction of neuronal activity [56]. Mechanistically, KCl increased the activity of neuronal nitric oxide synthase (nNOS), which led to S-nitrosylation and activation the small guanosine triphosphatase (GTPase) dexamethasone-induced Ras-related protein 1 (dexras1). Dexras1 then associated with DMT-1 via the linker protein acyl-CoA binding domain containing 3 (ACBD3), which increased DMT-1 throughput, labile iron levels in the neuronal cytoplasm, and reactive oxygen species [56]. Another study showed that DMT-1 can be directly S-nitrosylated, which also increases the throughput of the transporter [57]. This study identified two specific S-nitrosylation sites, and mutagenesis of these sites prevented the NO-mediated increased throughput in substantia nigra neurons administered with bacterial lipopolysaccharide [57]. Therefore, several stimuli may activate the nNOS/DMT-1 mediated pathway of endolysosomal iron efflux, and this includes μ-opioid receptor signaling. As rat cortical interneurons highly express the μ-opioid receptor [58,59], opioid agonist-mediated dampening of gamma aminobutyric acid (GABA)ergic interneuron firing could result in increased neuronal activity and nNOS/DMT-1 activation. Additionally, since select GABAergic interneurons express nNOS [60,61], μ-opioid agonists could directly induce nitric oxide (NO) signaling in these cells if they also express μ-opioid receptors. Furthermore, other reports show that μ-opioid agonists can activate nNOS or facilitate NO signaling in various cells [62–64], and NO signaling is closely related to iron homeostasis [65], suggesting that NO signaling plays an important role in morphine-mediated efflux of endolysosomal iron.

3. Neuronal responses to increased labile iron levels

Although all cells require iron for proper function of various proteins, enzymes, organelles, and cellular processes, excess labile iron can lead to oxidative stress and even cell death through production of reactive oxygen species (ROS) [66,67]. In order to minimize ROS production and maintain homeostasis, cells have developed various mechanisms to tightly regulate the amount of labile iron available to proteins and other organelles [68,69]. Some of these mechanisms are activated by morphine-induced efflux of endolysosomal iron, and this has several important implications that will be covered in this section.

3.1. FHC upregulation and downstream consequences

In addition to altering iron metabolism, morphine upregulates ferritin heavy chain (FHC) and ferritin light chain (FLC) protein in cortical neurons [21]. FHC and FLC are well-known for their ability to self-assemble into 24-mer ferritin cages that sequester and store excess iron [70], and our results suggest that neurons produce these proteins in response to the efflux of endolysosomal iron caused by morphine [21]. The individual ferritin subunits have distinct functions, as FLC is thought to facilitate nucleation of iron inside of ferritin [71] and electron transfer reactions [72], while FHC functions as a ferroxidase that converts reactive ferrous iron into ferric iron for storage inside the ferritin complex [71]. Additionally, our group and others have shown that FHC has important functions outside of iron metabolism [73,74], including the ability to interact with and block the activation of the homeostatic and neuroprotective chemokine receptor CXCR4 by its natural chemokine ligand CXCL12 [22–24]. This is thought to at least partially underlie opioids ability to worsen HAND symptoms, as neuronal CXCR4 signaling has well-known functions in the brain during development and regulates a host of critical neuronal functions during adulthood. These include regulation of neuronal progenitors [75,76] and oligodendrocyte progenitors [77], modulation of glutamatergic and GABAergic transmission [78,79] as well as other types of neurons [80], and activation of survival-associated pathways that are protective against excitotoxicity and various neurotoxins [79,81–83]. Additionally, CXCR4 activation by CXCL12 in vitro and in vivo increases the density of dendritic spines on cortical neurons [23], which are small dendritic protrusions on excitatory neurons that act as the post-synaptic components of excitatory synapses [84]. Importantly, dendritic spine density in specific brain areas like the prefrontal cortex (PFC) is thought to underlie learning and memory performance [20], as several studies have shown associations with spine density and cognitive performance in small animal models [22,85], non-human primates [86,87], and human patients [88–91]. In our study of postmortem brain tissue from humans and macaques with HIV/SIV ± use of μ-opioid agonists, opioid use alone or in combination with HIV/SIV infection increased FHC levels and decreased CXCR4 activation in PFC neurons, and FHC expression was positively correlated with the extent of cognitive impairment in humans [23]. These lines of evidence converge to suggest that μ-opioid agonist-mediated increased neuronal iron levels and upregulation of FHC leads to decreased PFC dendritic spine deficits and cognitive impairment in human patients with HAND.

Our group and others have provided evidence of the protein-level interactions and regulatory framework involving FHC and CXCR4. The first study to report an interaction between these proteins used various cell lines to show that FHC co-localized and immunoprecipitated with CXCR4 after CXCL12 treatment [92]. They hypothesized that the interaction between CXCR4 and FHC may be direct, as GST-labeled fusion proteins of the CXCR4 N and C-termini both pulled down FHC when incubated in protein lysates from HEK293 cells [92]. Based on these initial findings we postulated that FHC could be involved in morphine-mediated inhibition of CXCR4. Indeed, we found that morphine and DAMGO upregulated FHC in cortical neurons and increased the amount of FHC that immunoprecipitated with CXCR4. This outcome was associated with CXCR4 inhibition, as CXCL12 treatment did not activate the CXCR4 downstream mediators ERK1/2 and Akt when FHC protein levels were upregulated [24], and FHC knockdown completely blocked morphine-mediated dendritic spine deficits in these neurons [23]. Together, these data suggest that FHC may be involved in a feedback mechanism that controls CXCR4 signaling in several cell types, and that this mechanism can be commandeered by μ-opioid agonists. Moreover, evidence suggests that FHC can also inhibit the chemokine receptor CXCR2, as FHC overexpression prevented CXCR2-mediated phosphorylation of ERK1/2 in HEK293 cells stably expressing the receptor, which was also reversed by FHC knockdown [92]. This suggests that the regulatory capacity of FHC is not restricted to CXCR4 and could have broader relevance. One possibility is that FHC may specifically regulate signaling from CXC family chemokine receptors by binding to a common structural motif on these receptors. However, if FHC binds to a scaffold protein or second messenger that interacts with other GPCRs, FHC upregulation could produce a broader regulation of GPCR signaling. Therefore, morphine regulation of neuronal iron and FHC protein levels may have wide ranging effects on neuronal signaling pathways, neurotransmission, and overall cellular homeostasis.

3.2. Potential mechanisms leading to FHC upregulation

We recently demonstrated that morphine post-transcriptionally increased FHC expression in cortical neurons [21], suggesting that several previously characterized mechanisms could be involved in this pathway. The first possibility involves the iron regulatory protein (IRP) – iron responsive element (IRE) regulatory system, which is a post-transcriptional system that allows neurons and other cells to tune iron availability by controlling the production of proteins involved in iron metabolism [93,94]. Transcripts for ferritin subunits and other iron-related proteins contain an IRE sequence in either their 5’ or 3’ untranslated region, which is a stem loop structure that can be bound by an IRP in low iron conditions [95,96]. There are two isoforms of IRPs (IRP1 and IRP2), which are regulated by different mechanisms [97,98] and are differentially expressed in various tissues with IRP2 notably enriched in the brain [99].

Importantly, both isoforms ability to bind IRE sequences is controlled by labile iron. IRP binding leads to stabilization and translation of transcripts with 3’ IREs and blocks ribosomal association for transcripts with 5’ IREs. When labile iron levels are increased, IRPs dissociate from IREs, which destabilizes and prevents translation of transcripts with 3’ IREs while allowing transcripts with 5’ IREs to associate with the ribosome for translation. As such, proteins that are translated in response to iron possess 5’ IREs, while proteins that are translated in low iron conditions possess 3’ IREs.

As the FHC transcript contains a 5’ IRE [100], morphine-mediated efflux of endolysosomal iron may lead to IRE detachment and translation of FHC protein. If this is the case, then morphine may have additional effects on other transcripts that are regulated by IRPs. For example, FLC and the iron exporter ferroportin also possess 5’ IRE sequences, while proteins involved in iron intake, such as transferrin receptor and select DMT-1 isoforms possess 3’ IRE sequences [101]. In line with this regulatory framework, morphine does increase FLC levels [21]. To our knowledge, morphine’s effects on other IRP-regulated proteins involved in iron metabolism has not been examined and is an area we are currently investigating. If morphine only regulates the level of ferritin proteins, this could be due to specific regulation of these transcripts by IRP2, which binds more strongly to a unique structural motif in ferritin IREs [102], and may be modulated in response to morphine-mediated NO signaling [103].

An additional IRP - IRE independent pathway could also be involved in morphine-mediated upregulation of FHC protein. The FHC transcript contains an internal ribosome entry site (IRES) downstream of the IRE sequence, which could allow the transcript to bypass the traditional cap-dependent translation process [104]. Select proteins may be translated via an IRES even when global protein synthesis is shut down, for example during viral infection [105,106]. One group showed that when protein synthesis was shut down by several stimuli, cells exposed to iron can still produce FHC and FLC [104]. Using a bicistronic positive feedback vector, they demonstrated that the FHC transcript contains an IRES in the 5’ untranslated region, and they suggest that this mode of translation may be prevalent in stress conditions or during IRP1 overexpression [104]. Therefore, it is also possible that morphine may increase FHC expression through an IRES-mediated mechanism, and due to the similarity of the FHC and FLC 5’ UTR [104], this translation may also extend to FLC. In line with this, several studies have shown that morphine increased the expression of RNA-binding proteins reported to facilitate IRES translation of select transcripts in neurons and non-neuronal cells, including heterologous nuclear ribonucleoprotein k (hnRNP k) and poly(C)-binding protein 1 (PCBP1) [107,108]. Both proteins may play important roles in the morphine signaling pathway or neuronal response to iron prior to FHC upregulation, although their specific contributions may differ. One group showed that morphine activation of μ-opioid receptors increased hnRNP k levels in primary rat cortical neurons through an interesting post-transcriptional mechanism where hnRNP k bound to the 5’ UTR of its own transcript, inducing IRES-mediated translation [107]. Upon upregulation, hnRNP k protein bound to the 5’ UTR of the μ-opioid receptor transcript, which led to altered μ-opioid receptor translation and altered analgesic responses in animals [107]. Although hnRNP k may induce an IRES-mediated post-transcriptional regulation of FHC, the reported induction of hnRNP k protein expression by morphine was modest and transient, and it occurred much earlier than FHC upregulation [107]. Therefore, if hnRNP k is involved in FHC upregulation, then it may play a role in an earlier phase of the pathway.

In another study by the same group, morphine upregulated PCBP1 at the same time as hnRNP k, although this was reported in NMB human neuroblastoma cells with and without μ-opioid receptor overexpression [108]. Evidence suggests that PCBP1 interacts with FHC transcripts and regulates their translation [109] and plays important roles in iron homeostasis in various cell types [110]. An initial report showed that cultured yeast expressing human FHC and FLC increased ferritin iron storage upon overexpression of PCBP1, and PCBP1 directly interacted with iron and immunoprecipitated with ferritin [111]. Additionally, they showed that knockdown of PCBP1 resulted in reduced iron storage in ferritin and increased cytosolic labile iron levels in human Huh7 cells [111]. Together, these experiments suggest that PCBP1 acts as an iron chaperone that facilitates the transfer of free labile iron into ferritin complexes. The same group followed up these initial findings in several subsequent studies characterizing the iron chaperone activities of other PCBP1 and other isoforms [112,113]. Notably, the group demonstrating that morphine upregulated PCBP1 levels [108] showed that this occurred at the same time as morphine-induced endolysosomal iron efflux [21], suggesting that morphine’s effects on cellular iron metabolism may also drive a modest upregulation of PCBP1.

3.3. Iron-mediated regulation of amyloid precursor protein expression and processing

Through IRP or IRES-mediated translation, morphine and other μ-opioid agonists may affect the expression of FHC and other proteins involved in cellular iron metabolism, including the amyloid precursor protein (APP) (figure 2). Many studies indicate that the iron accumulation can significantly contribute to Alzheimer’s disease pathogenesis [114,115], which is characterized by increased production and deposition of amyloid β in the brain via amyloidogenic cleavage of APP [116]. In line with this, iron accumulation can increase APP levels, accelerate Aβ production, and promote neurodegeneration in Alzheimer’s disease through altered IRP-IRE signaling [117,118]. The APP transcript contains an IRE sequence in its 5’ UTR [119], as well as a 5’ IRES [120], suggesting that iron may control APP expression through several mechanisms. One group demonstrated that the APP IRE sequence is able to specifically bind an IRP [121], and this binding was disrupted when an APP cRNA probe was mutated in the core IRE domain [119]. A follow-up study characterized the potential of four metal chelators to limit APP expression in SH-SY5Y cells, including deferoxamine, clioquinol, VK-28, and piperazine-1 [122]. The APP 5’ UTR element was more selective for iron over copper and was unresponsive to zinc, suggesting that iron chelator based therapeutic strategies may be useful for slowing the progression of amyloid pathology in Alzheimer’s disease.

Figure 2.

Post-transcriptional regulation of FHC and APP by morphine and iron. The 5’ untranslated regions (UTR) of the FHC and APP transcripts play a major role in their regulation by iron. Both transcripts possess an iron responsive element (IRE) in this region, where iron regulatory proteins (IRPs) bind in low iron conditions to block translation and are released in high iron conditions, allowing translation. Reports suggest that both transcripts also possess an internal ribosomal entry site (IRES) in their 5’ UTRs, which may be utilized when labile iron levels increase during conditions of cellular stress. The minimal sequence required for IRES translation of APP is the first 50 nucleotides from the 5’ cap (gray box), although IRES translation efficiency increases when the entire APP 5’ UTR is present (gray line). The FHC IRES has not been precisely mapped and may also encompass the entire 5’ UTR (grey line). Both UTRs also possess a GC-rich acute box (AB) domain upstream of the IRE. The AB and nearby areas on the FHC 5’ UTR are reported to bind poly(C) binding proteins, notably PCBP1, which is increased by morphine in neuroblastoma cells and plays a role in FHC translation.

The role of APP in cellular iron metabolism has also been investigated previously. Iron efflux from mammalian cells is facilitated by the synergistic actions of the iron exporter ferroportin and ferroxidase activity of extracellular hephaestin and ceruloplasmin that stabilize ferroportin in the plasma membrane and catalyze extracellular iron release. The membrane stabilization of ferroportin is also affected by its interaction with a 22-amino acid synthetic peptide based on a short sequence in the extracellular E2 domain of APP. One group investigated the interaction between ferroportin and APP using cyan fluorescent protein (CFP)-tagged ferroportin in conjunction with yellow fluorescent protein (YFP) fusions of hephaestin and APP family members APP, APLP1, and APLP2 in HEK293T cells. They used fluorescence and surface biotinylation to quantify ferroportin membrane occupancy and measured ⁵⁹Fe efflux. They showed that a ferroportin-targeting sequence, (K/R)EWEE, present in APP and APLP2, but not APLP1, helped to modulate ferroportin-dependent iron efflux in the presence of an active multicopper ferroxidase [123]. Moreover, another group provided evidence that APP plays a major role in the maintenance of normal iron levels in the brain and liver during aging and that its deletion leads to increased iron levels in these organs. The authors showed that APP deficient mice display a persistent age-dependent increase in brain iron levels that was associated with decreased brain levels of ferroportin. This is also consistent with a mechanistic role for APP to facilitate neuronal iron efflux through ferroportin stabilization at the cell surface. The group hypothesized that brain ferritin is less responsive to iron load compared to liver and that iron overload in APP-KO results in an age-dependent increase in the non-ferritin iron pool [124].

4. Novel molecular approaches to study the contributions of APP processing in HAND with opioid abuse

In the post-ART era, aging of patients infected with HIV likely contributes to the development or progression of HAND [125] and could also promote the development of age-related neurocognitive disorders like Alzheimer’s disease. Indeed, postmortem studies of patients with HAND have shown increased expression of amyloid β and tau tangles in the brain [126], although the expression and localization of these proteins did not exactly match that observed in Alzheimer’s disease. For example, HAND patients in the pre-ART era generally showed intra-neuronal accumulation of amyloid β as opposed to extracellular amyloid β plaques [127,128]. However, these immunohistochemistry-based studies may not necessarily be in line with neuroimaging studies from the ART era, which have shown mixed results on amyloid β expression in HAND patients [129]. Therefore, more work must be done to address the role of amyloid β in HAND pathology using a combination of neurocognitive testing, advanced imaging modalities, and ex vivo analysis of brain tissue from ART-treated patients. On the other hand, HAND patients that use opioids may be especially susceptible to developing amyloid pathology, which could be an avenue that contributes to the onset or progression of disease in these patients. Although there has not been a clinical study examining amyloid β expression in brain tissue of HIV+ opioid users, one clinical study suggested that opioid abusers without HIV may show a predisposition to develop changes in the brain that are related to Alzheimer’s disease, including increased expression of amyloid β protein [130]. Given that APP translation is regulated by labile iron levels, it is possible that opioid-mediated efflux of endolysosomal iron may increase APP translation, which would produce additional substrates for amyloidogenic processing. Since HIV infection may shift APP processing towards the amyloidogenic pathway [131], opioid-mediated production of APP could synergistically lead to an accumulation of amyloid β that would not necessarily be observed in a non-opioid using HAND patient. Therefore, future studies aimed at understanding opioid modulation of APP expression and processing will provide critical insights into HAND pathology in the post-ART era.

In order to address the points above, our lab has recently developed a novel molecular tool to study APP processing based on the herpes simplex virus-1 protein US9. US9 is a type II single transmembrane protein that normally facilitates the transport of viral cargo in neuronal processes and plays an important role in the spread of the virus to other neurons [132]. Our studies suggest that the US9 protein can be repurposed to deliver customized molecular cargo attached to its N or C terminus in various types of cells. US9 is expressed in several subcellular domains, including transport vesicles, the trans-Golgi network, endolysosomes, and the plasma membrane [133,134]. Further, we have shown that the intracellular localization and trafficking of US9 is an intrinsic quality of the protein and is independent of post-translational modifications, molecular cargo, and the cell type in which it is expressed [133]. Importantly, US9 expression in the absence of viral cargo does not display catalytic activity or cellular toxicity in vitro [133,135].

The expression profile of US9 is consistent with that of APP, suggesting that US9 cargoes may be able to modulate APP processing. Since amyloidogenic processing of APP is thought to occur in lipid rafts [136,137], US9’s natural localization to these membrane microdomains [135,138] provides a way to target molecular cargoes directly to the sites of amyloid β production. We performed proof of concept studies examining the functional activity of US9 cargo by designing a functional assay based on a series of US9 fusion proteins (figure 3). These proteins either expressed an exogenous protease or the specific sequence cleaved by the protease, and each element was placed either on the N or C terminus of US9. In this system, US9 containing the exogenous protease only cleaved US9 containing the exogenous cleavage sequence if both elements were expressed on the same side of the membrane, suggesting that US9 can transport functional cargo in a cell. Next, we examined the potential of US9 cargoes to interact with APP and BACE1 by creating new US9 fusion proteins containing the exogenous cleavage sequence and the transmembrane domains of either APP or beta-secretase 1 (BACE1). US9 containing the protease was able to cleave both US9 substrates with altered transmembrane domains, suggesting that US9 colocalizes with BACE1 and APP, and US9 cargoes can interact with these proteins [135]. Therefore, US9-derived fusion proteins may provide a precisely targeted approach to study APP processing alterations by opioids and if altered APP processing contributes to opioid-mediated reduction of dendritic spines and cognitive impairment in HAND.

Figure 3.

US9-based molecular tools. A. US9 containing a N-terminal protease will cleave a US9 driven reporter construct if it expresses a protease cleavage site (blue bar) on the cytosolic side of the membrane. B. Likewise, US9 containing a C-terminal protease will cleave a US9 driven reporter construct if it expresses a protease cleavage site in the endolysosomal (EL) lumen. Configurations in A and B suggest that US9 cargoes can functionally target other proteins expressed on a specific membrane leaflet. C. US9 containing a C-terminal protease can cleave a reporter construct driven by the beta-secretase 1 transmembrane domain (BACE1 TM - purple) if a cleavage site is inserted on the luminal facing domain. D. Further, the same US9 driven protease can also cleave a reporter construct driven by the amyloid precursor protein transmembrane domain (APP TM - yellow), again if a cleavage site is inserted on the luminal facing domain. C and D suggest that US9 can target functional cargoes to subcellular areas where BACE1 and APP are expressed and amyloidogenic processing occurs.

5. Conclusions

Although the symptoms and pathology of modern HAND and Alzheimer’s disease may start to converge as ART-treated patients age, the current clinical presentations of these diseases show distinct differences. However, HAND, Alzheimer’s disease, and several other neurocognitive disorders share a common thread of altered brain iron metabolism, which is thought to at least partially contribute to the presentation of symptoms. As such, the modulation of neuronal iron metabolism by morphine, and presumably other licit and illicit μ-opioid agonists, may play an important role in worsening cognitive symptoms in patients infected with HIV. Future studies will determine the extent to which morphine-mediated dysregulation of iron contributes to neurocognitive decline in HAND, the precise molecular mechanisms that underlie morphine-mediated efflux of endolysosomal iron, and the downstream consequences of opioid use on iron metabolism, APP production and processing, and other critical cellular functions.

Beyond activation of the μ-opioid receptor G-protein pathway, the molecular mediators that facilitate morphine-induced endolysosomal iron efflux are unknown. Although several groups have reported interesting signaling pathways leading to endolysosomal iron efflux, it is unclear if morphine activates these same pathways or promotes iron efflux through a pathway that is currently undefined. Since endolysosomes express a host of channels and transporters [44], opioids effects could be mediated by a specific protein conduit for iron or by a complex interaction of several different endolysosomal transporters and channels. Future work designed to tease out the complexities of this signaling pathway will provide important insights about the inner workings of the endolysosomal system and its role in opioid abuse and neurocognitive disease pathogenesis.

Opioids effects on endolysosomal physiology could also directly contribute to production of amyloid β, given the importance of the endolysosomal system for APP processing. Further, opioids could induce translation of APP after endolysosomal iron is released to the cytoplasm, and both mechanisms could converge to lead to increased amyloidogenic processing in HAND patients. As our studies also implicate the iron storage protein FHC as a correlate of cognitive decline in brain tissue from HAND patients, it is possible that opioid-induced changes in neuronal iron metabolism can worsen HAND symptoms through multiple downstream pathways. Additionally, these iron-dependent outcomes may not be limited to neurons, as multiple CNS cell types express μ-opioid receptors and our studies showed that U87 glioblastoma cells released endolysosomal iron in response to morphine. Thus far, the downstream effects of opioid-mediated efflux of endolysosomal iron in CNS cells are relatively unexplored, so new studies would greatly benefit our understanding of opioids physiological effects in the brain and their specific contributions to HAND pathology. These studies could also guide future therapeutic strategies for opioid-using patients with HAND.