Address correspondence and reprint requests to Ruth G. Perez, PhD,Paul L. Foster School of Medicine, Center of Excellence in Neurosciences, 5001 El Paso Drive, MSB 1, Suite 4002, El Paso, TX 79905, USA. E-mail: firstname.lastname@example.org
Amyloid precursor protein (APP) and its secreted form, sAPP, contribute to the development of neurons in hippocampus, a brain region critical for learning and memory. Full-length APP binds the low-density lipoprotein receptor-related protein (LRP), which stimulates APP endocytosis. LRP also contributes to neurite growth. Furthermore, the receptor associated protein (RAP) binds LRP in a manner that blocks APP–LRP interactions. To elucidate APP contributions to neurite growth for full-length APP and sAPP, we cultured wild type (WT) and APP knockout (KO) neurons in sAPPα and/or RAP and measured neurite outgrowth at 1 day in vitro. Our data reveal that WT neurons had less axonal outgrowth including less axon branching. RAP treatment potentiated the inhibitory effects of APP. KO neurons had significantly more outgrowth and branching, especially in response to RAP, effects which were also associated with ERK2 activation. Our results affirm a major inhibitory role by full-length APP on all aspects of axonal and dendritic outgrowth, and show that RAP–LRP binding stimulated axon growth independently of APP. These findings support a major role for APP as an inhibitor of neurite growth and reveal novel signaling functions for LRP that may be disrupted by Alzheimer's pathology or therapies aimed at APP processing.
secreted amyloid precursor protein cleaved by α-secretase
secreted amyloid precursor protein cleaved by β-secretase
During development, axonal outgrowth connects various brain regions by forming synapses between neurons to create normal patterning and connectivity in the brain (Kimmel 1993). In neurodegeneration, synapse loss leads to widespread disconnection of the neuronal circuits, leading to a coordinate loss of normal brain function. This type of synaptic loss is widespread in Alzheimer's disease (AD) (DeKosky and Scheff 1990), a major brain disorder associated with severe cognitive loss, in which individuals lose memory, reasoning, language, and emotional control (Selkoe 2012).
The pathology in AD brain is characterized by neurofibrillary tangles and amyloid plaques. AD plaques contain high levels of the amyloid peptide, Aβ, a 4 kDa fragment of the amyloid precursor protein (APP) (Glenner and Wong 1984). Amyloid peptide is secreted following cleavage of APP by two secretases β-secretase, which releases the long N-terminal fragment sAPPβ, and γ-secretase, which releases Aβ. Non-amyloidogenic APP processing occurs by α-secretase, which releases sAPPα after which γ-secretase cleavage yields p3 (Selkoe et al. 1996). Abundant full-length APP resides at the cell surface, with α-secretase cleavage occurring mainly in this pool (De Strooper and Annaert 2000; Hui and Koo 2011), while β-secretase acts primarily on an endocytic pool of APP (Perez et al. 1999; Thinakaran and Koo 2008).
APP is a type 1 integral transmembrane protein (De Strooper and Annaert 2000; Hui and Koo 2011) that is highly expressed on neuronal precursors (Young-Pearse et al. 2007) and on cortical and hippocampal neurons (Löffler and Huber 1992; Gralle and Ferreira 2007). Although widely implicated in neurodegeneration, APP normally contributes to cell signaling, adhesion, axon branching, and synaptic plasticity (Mattson 1997; Perez et al. 1997; Wallace et al. 1997; Hui and Koo 2006; Ikin et al. 2007; Thinakaran and Koo 2008). In fact, APP knockout (KO) mice display brain gliosis (Zheng et al. 1995) and impairment of both hippocampal neuroplasticity and long-term potentiation, both of which are associated with memory (Dawson et al. 1999; Seabrook et al. 1999). Neuronal cultures from APP KO mice form abnormal axons and dendrites (Perez et al. 1997). Remarkably, the deficits found in APP KO mice can be rescued by knocking-in APP expression (Ring et al. 2007), supporting an important normal role for APP in brain.
Compared with full-length APP, unique roles have been proposed for the long N-terminal secreted fragments sAPPα and sAPPβ. Data from cells expressing wild-type APP or Swedish mutant APP showed that sAPPα has neurotrophic effects while sAPPβ does not, although the high levels of Aβ secreted by mutant cells may have contributed to the sAPPβ effects (Li et al. 1997). Our earlier work revealed that both sAPP and cell-surface APP contribute to neurite growth (Perez et al. 1997). Newer data demonstrate that sAPPα and sAPPβ both contribute to axon growth (Chasseigneaux et al. 2011). Others have shown that sAPPα modulates axon growth, but only in the presence of full-length APP, occurring by cell-surface APP and sAPPα competing for Integrin ß1 binding (Young-Pearse et al. 2008).
In addition to APP, another type 1 transmembrane protein that is highly expressed on hippocampal and cortical neurons is the low-density lipoprotein receptor-related protein (LRP). LRP, sometimes called LRP1, a member of the low-density lipoprotein (LDL) receptor family (Jordan et al. 1998; Van Uden et al. 2000; Lillis et al. 2008). LRP binds lipoproteins such as apoE and α2m to regulate their uptake by endocytosis. Curiously, adding α2m and/or apoE to neurons in culture raises LRP protein levels and redistributes LRP to neurites (Qiu et al. 2004). In addition to lipoproteins, LRP also binds to cell-surface APP751 through the Kunitz protease inhibitor (KPI) domain, which stimulates APP endocytosis (Kounnas et al. 1995). The receptor-associated protein (RAP) normally assists in the intracellular transport of endocytic receptors such as LRP (Willnow 1998), but when added extracellularly, RAP binds LRP in a manner to block the uptake of all known LRP ligands (Bu and Rennke 1996). Cultured cortical neurons treated with RAP redistribute LRP out of the neurites and back into the soma (Qiu et al. 2004); however, RAP effects on hippocampal neurons have never been studied. It is noteworthy that hippocampal neurons express similar levels of both APP695, which lacks the KPI domain, and APP751 (Johnson et al. 1990).
Studies of LRP and APP often focus on Aβ production in search of novel targets for AD therapy. However, the normal physiology of APP and LRP might be inadvertently dysregulated by therapies that target amyloid production. Thus, in this study, we explored the contributions of full-length APP, sAPPα, and LRP on axonal and dendritic growth from hippocampal neurons from APP WT and APP KO mice. The neurons were grown in varying concentrations of sAPPα and in the presence or absence of RAP, to block sAPP binding and uptake via LRP (Kounnas et al. 1995). The extracellular regulated kinase (ERK) contributes to axon elongation in both chick retina and mouse hippocampus (Perron and Bixby 1999; Wang et al. 2011) and ERK2 can be activated by LRP ligands that regulate neurite growth (Qiu et al. 2004), and therefore, we also measured ERK2 activation in our hippocampal cultures.
The use of hippocampal neurons for our studies is important because hippocampus (i) regulates short-term memory (Suzuki 2007), (ii) undergoes continued stem cell renewal (Eriksson et al. 1998; von Bohlen und Halbach 2011), and (iii) is a brain area that undergoes dynamic synaptic plasticity (Ho et al. 2011), processes that are all severely impaired in AD. Our findings confirm an inhibitory role on axonal growth for full-length APP; a novel role for full-length LRP in signal transduction that may be co-regulated by APP, and novel effects on minor process (dendritic) growth in response to APP, LRP, and sAPPα interactions. These data provide evidence that neurons with too little full-length APP could form axons that are overly long and may bypass normal post-synaptic targets. The data further suggest that hippocampal neurons with too much full-length APP could produce short axons with underdeveloped terminal arbors, again producing synaptic disconnection. Changes resulting in abnormal APP expression may ultimately lead to impaired plasticity and further affect memory formation. Cumulatively, these findings suggest that ideal AD therapies should act to preserve normal APP levels to sustain the optimal synaptic contacts that subserve normal memory function.
APP knockout mice
Mice lacking APP (APP KO) (Zheng et al. 1995) were re-derived on a C57BL6 background by Taconic Farms (Germantown, NY, USA) and wild-type C57BL6 mice (APP WT) served as controls. Timed breeding was done at Taconic and animals were handled consistent with USPH Policy in Humane Care and Use of Laboratory Animals and NIH guidelines on protocols approved by the Institutional Animal Care and Use Committee at both Allegheny General Hospital and the University of Pittsburgh. To minimize pain or discomfort mice were killed by CO2 inhalation followed by decapitation. Experiments were performed according to the guidelines of Animal Research: Reporting of In Vivo Experiments (ARRIVE).
Triplicate neuron cultures were plated 160 000 cells/well in 24-well plates pre-treated with poly-l-lysine. Media were gently changed 4 h before subsequent treatments. Cells were given vehicle alone, 150 nM RAP alone, or U0126 (U120; Sigma/Aldrich, St Louis, MO, USA) for 15 min. Controls were pre-treated with U0126 1 h prior to RAP. Media were aspirated and lysates prepared by application of boiling 2X Laemmli buffer. Proteins were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and ERK activation was determined from western blots reacted with antibodies specific for phosphorylated ERK1/2 (Cell Signaling, cat # 9101, Danvers, MA, USA) and total ERK1/2 (BD-Pharmingen, cat # 554095, San Jose, CA, USA). Densitometry was performed as before (Perez et al. 1997).
Primary hippocampal cultures and sAPP and RAP treatments
Hippocampi dissected from embryonic day 16 mice were prepared as described (Goslin and Banker 1991). Briefly, hippocampi were suspended in trypsin-EDTA 15 min, 37°C, washed with calcium-and magnesium-free Hanks' buffered saline solution, and triturated with a fire polished pipet to dissociate. Cells were plated onto 12-mm round glass coverslips (Carolina Biological Supply, Burlington, NC, USA) pre-treated with poly-l-lysine hydrobromide (1 mg/mL, Sigma, St Louis, MO, USA) in 0.1 M borate buffer (pH 8.5). Cells attached for 3–5 h in minimal essential medium (MEM)/10% horse serum/0.6% glucose (Plating-MEM), at 37°C 5% CO2 and then transferred to APP KO glial-conditioned N2.1 medium. Freshly re-suspended sAPPα was added to media at 0–20 nM. In pilot studies, 75–600 nM RAP (recombinant RAP, gift of Dr. Rebeck, Georgetown University) was applied to APP WT and APP KO cultures with optimal viability noted in 150 nM RAP, the concentration selected for our experiments. Cells were fixed in 4% paraformaldehyde/4% sucrose phosphate buffered saline 24 h after treatments.
Primary astrocyte cultures
Glial cultures enriched in type 1 astrocytes (> 95%) were prepared using cortical tissue from post-natal day 1 mice, as described (Goslin and Banker 1991). Briefly, APP KO cortex was dissected free of choroid and meninges and dissociated using trypsin and trituration. Astrocytes were grown in 75-mm tissue culture flasks with MEM, 10% horse serum, 0.6% glucose (Glial-MEM), 37°C, 5% CO2 until confluent. For astrocyte-conditioned medium, cultures were rinsed 1x with unsupplemented MEM and changed to N2.1 supplemented MEM and allowed to condition the medium with secreted factors for 6–7 days. Media were collected, cells eliminated by centrifugation, and aliquots stored at −80°C.
For morphometric analyses, neurons were labeled with anti-α-tubulin (DM1A antibody; Sigma/Aldrich) and anti-mouse IgG-HRP (Amersham, Arlington Heights, IL, USA) reacted with diaminobenzidine (DAB) (Sigma) for bright field microscopy as before (Perez et al. 1997).
Morphometry and statistical analyses
An imaging system equipped with AIS software (MCID, Brock, ON, CAN) was used to measure neurite growth. Briefly, microscopic images of isolated neurons were displayed on a video monitor and captured to a computer. Outgrowth was traced using a mouse and AIS software. Neurons were evaluated for axon growth, minor process growth, axon and minor process branching, minor process numbers, minor process length, average axon branch length, total neurite growth, and main axonal process growth. The main axonal process refers to the length of the single longest neurite extending from the neuronal cell body. Axon outgrowth refers to the length of the single longest neurite from the neuronal cell body plus the sum of all processes emanating from the longest process. All neurons analyzed had a single axon. At 1 day in vitro, dendritic-like growth is referred to as a minor process. Minor process growth consists of the sum of the lengths of all non-axonal processes emanating from cell body including the sum of the lengths of side processes arising from minor processes. Total outgrowth is the sum of all axonal and minor process growth. Branches refer to the small processes extending laterally from an axon or minor process. Minor process number refers to the total number of processes extending directly from the neuronal cell body, not including the main axonal process. Data from 10 independent experiments measured outgrowth from a minimum of 37 neurons per treatment condition, and most experiments analyzed 50 neurons per condition. Data collected as Excel files were analyzed by one way anova (Instat3; GraphPad, San Diego, CA, USA), with significance set to p < 0.05 or better, and data given as mean ± the standard error of the mean (SEM).
Purification of sAPPα
We previously showed that CHO cells transfected with APP751 lacking the last 43 amino acids secrete sAPPα (Perez et al. 1999). Cells were grown to 90% confluence in Dulbecco's modified Eagle's medium, 10% fetal bovine serum, G418 then changed to serum-free defined media for 48 h. Conditioned media were collected, supplemented with protease inhibitors, cooled and spun at 2000 g at 4°C, 30 min to eliminate cells. sAPPα was purified by Affigel-Heparin followed by Mono-Q FPLC as previously described (Kametani et al. 1993). Purity of sAPP was confirmed by silver staining and immunoblotting with antibody 8E5, which recognizes residues 444–592 of human APP. Purified sAPPα fractions were pooled, protein concentrations determined, and aliquots lyophilized and stored at −80°C.
Full-length APP decreases total axon outgrowth
Hippocampi were dissected from APP WT and APP KO E16 embryos, and dissociated neurons were cultured in APP KO glial-conditioned medium for 1 day in 0 nM, 10 nM, or 20 nM sAPPα bearing a KPI domain, in the absence (0 nM) or presence of RAP (150 nM). We then evaluated hippocampal neurons for differences in total axon outgrowth and noted that sAPPα alone had limited effects (Fig. 1a). Bigger effects were noted for full-length APP. In all conditions, APP KO neurons had significantly more total axonal outgrowth compared with APP WT neurons (Fig. 1a, 1b). Remarkably, addition of 150 nM RAP to APP KO neurons grown in 0 nM and 10 nM sAPPα induced a further increase in already robust axon outgrowth, while outgrowth was similar for APP KO neurons in both 20 nM sAPPα and 20 nM sAPPα + RAP (APP KO; Fig. 1a). With regard to APP WT neurons, there was a slight tendency for sAPPα at 10 nM, with or without RAP, to decrease total axon outgrowth although not significantly (Fig. 1a). These data suggest that full-length APP plays an inhibitory role on axon outgrowth. They further suggest that RAP–LRP interactions, in the absence of full-length APP, have outgrowth inducing effects that are independent of full-length APP.
Full-length APP decreases the number of axon branches
Next, we sought to determine the effects of full-length APP, sAPPα and RAP on axonal branching. Branching was similar for APP WT and APP KO neurons grown without sAPPα or RAP (Fig. 2). However, APP WT and APP KO neurons grown without RAP but in the presence of sAPPα exhibited significant differences in axonal branching (Fig. 2, 10 nM and 20 nM sAPPα, compare white to gray bars). APP KO neurons cultured in 150 nM RAP, with or without sAPPα, also had significantly more axon branches compared with APP WT neurons in the same conditions (Fig. 2, compare cross-hatched to black bars). Indeed, for APP WT neurons sAPPα had no significant effect on axonal branching in the absence of RAP, while in 150 nM RAP with 0 nM or 10 nM sAPPα, APP WT neurons had significantly less axon branching (Fig. 2, compare white bars to cross-hatched bars). Much like total axonal outgrowth, these findings again support a major inhibitory role for full-length APP on axonal branching.
RAP and full-length APP modulate axon growth in multiple ways
Side branching off the main axonal process as measured by total axon outgrowth less the longest axonal process
As we found the most robust changes in axon growth associated with the loss of full-length APP and in response to RAP, we sought to further scrutinize the contributions of APP and LRP by measuring subtler aspects of axon outgrowth. We first looked at total axon outgrowth less the longest axonal process, to determine if APP and RAP may have affected axon branch outgrowth independently of effects on the main axonal process. This type of outgrowth was significantly shorter in APP WT neurons compared with APP KO neurons grown without RAP (Fig. 3a, 0 nM RAP). RAP significantly increased branching off the main axonal process for APP KO neurons, but not for APP WT neurons (Fig. 3a, 150 nM RAP). These RAP results revealed that axon branching (not including the longest axonal process) significantly increased for cells lacking APP, suggesting that RAP binding to LRP alone can transduce a signal into neurons to stimulate axonal branching.
Length of the main axonal process
A measure of the longest axon process showed that APP WT and APP KO neurons were similar when grown in medium without sAPPα or RAP (Fig. 3b, left side). In 150 nM RAP, APP WT neurons main axon processes were significantly shorter compared with parallel cultures grown without RAP (Fig. 3b, compare white bars), while APP KO neurons had similar main axonal length when grown in the presence or absence of RAP (Fig. 3b, gray bars). As RAP treatment significantly reduced the length of the main axonal process of APP WT neurons, that reduction produced a significant difference in the length of the main axonal process between APP WT and APP KO neurons grown in the presence of RAP (Fig. 3b, compare white to gray bars in 150 nM RAP).
Length of the axonal side branches
To further scrutinize the source of differences in outgrowth, we measured the length of each axonal side branch. We found that RAP significantly increased axon branch lengths of APP WT neurons compared with parallel WT cultures grown without RAP (Fig. 3c, white bars). RAP had no effect on APP KO axon branch lengths (Fig. 3c, gray bars). From these results, we conclude that the more robust total axon outgrowth noted for APP KO neurons (Fig. 1) was likely because of increased numbers of axon branches (Fig. 2), not increased axon branch lengths.
Ratio of total axon outgrowth to main axon process length
As a measure of full-length APP effects on axon growth, we also calculated the ratio of total axon outgrowth compared with the length of the main axon process for APP WT and APP KO neurons grown without RAP. This ratio was significantly greater for APP KO compared with APP WT neurons (Fig. 3d, 0 RAP). We also calculated the ratio of total axonal outgrowth compared with growth of the main axonal process in response to 150 nM RAP. RAP tended to increase this ratio for all neurons, but only APP WT neurons had a significant increase (Fig. 3d, white bars), a result that might have been expected as RAP significantly decreased the length of the main axon process (Fig. 3b), whereas RAP significantly increased axon branch lengths for APP WT neurons (Fig. 3c). From these results, we infer that full-length APP primarily affects axon growth by modulating branch numbers, not the length of each axon branch. This suggests that RAP–LRP interactions that block the binding of LRP to full-length APP, act to inhibit the growth of the main axon process (Fig. 3b), while RAP in the absence of full-length APP, increases the total number of axonal branches (Figs 1, 2).
APP inhibits minor process outgrowth
In addition to measuring all aspects of axonal outgrowth, we also measured the effects of full-length APP, sAPPα and RAP on minor processes (presumptive dendrites). As shown in Table 1 (upper three rows), neurons which lack APP (APP KO) had significantly more minor processes, increased minor process branching, and increased minor process outgrowth compared with APP WT neurons. We observed similar effects on minor processes in the presence or absence of sAPPα. As might be expected, the enhanced minor process growth noted for APP KO neurons, when summed with axonal outgrowth, accounted for significantly greater total neurite outgrowth for APP KO neurons compared with APP WT neurons (Table 1, fourth row). When we compared within groups effects of RAP on total neurite outgrowth, we noted significant differences for both WT and KO neurons. RAP treatment decreased total neurite outgrowth for APP WT neurons, but increased total neurite outgrowth for APP KO neurons (Table 1, bottom row). These results imply that full-length APP inhibits minor process growth and that RAP blockade of an APP–LRP binding interaction further accentuates the inhibitory effects of APP, while RAP treatment of neurons lacking APP enhances their already robust minor process growth; raising the possibility that a growth-associated-signal transduced by RAP was different in neurons that express APP compared those from APP KO mice, which lack APP. ERK2 has been implicated in neurite outgrowth (Perron and Bixby 1999; Wang et al. 2011), and thus, we did experiments to evaluate RAP effects on ERK2 as follows.
Table 1. Minor process outgrowth and total neurite outgrowth for APP WT and APP KO hippocampal neurons grown without (w/o RAP, 0 nM) or with RAP (w/ RAP, 150 nM). At the developmental stage evaluated, presumptive dendrites are called minor processes. APP KO neurite growth was significantly greater than APP WT growth in all conditions
Between groups measures
WT w/o RAP
KO w/o RAP
WT w/ RAP
KO w/ RAP
In the presence of RAP, total neurite outgrowth was even more pronounced for WT and KO neurons, however, in opposite the direction. APP, amyloid precursor protein; RAP, receptor associated protein; KO, knockout; WT, wild type.
Minor process number
5.200 ± 0.3
7.516 ± 0.5
3.925 ± 0.2
7.211 ± 0.4
Minor process branches
6.373 ± 0.6
18.122 ± 1.8
4.050 ± 2.9
18.688 ± 1.4
Minor process outgrowth (μm)
218.16 ± 8.8
312.11 ± 13.8
184.22 ± 11.4
329.63 ± 16.0
Total neurite outgrowth (μm)
411.79 ± 14.0
553.82 ± 17.4
282.11 ± 18.7
640.18 ± 20.2
Within groups measures
WT w/o RAP
WT w/ RAP
KO w/o RAP
KO w/ RAP
Total neurite outgrowth (μm)
411.79 ± 14.0
282.11 ± 18.7
553.82 ± 17.4
640.18 ± 20.2
ERK2 response to RAP is different in APP WT and APP KO neurons
We evaluated signaling effects of RAP associated with blockade of APP–LRP binding (APP WT) or associated with RAP–LRP binding in the absence of APP (APP KO). Neurons were grown 1 day without or with RAP after which ERK2 activation was measured (Fig. 4). Controls included vehicle alone for baseline ERK2 activity; and in response to U0126, an inhibitor of MAP kinase kinase 2 (MEK2), to block ERK2 activity. Immunoblots were probed for total ERK2 and phosphorylated ERK2 and quantified using established methods (Ugarte et al. 2003; Lin et al. 2008). In four independent experiments, we noted a trend toward lower baseline ERK2 activation in neurons without APP (APP KO) compared with APP WT neurons, though the difference was not significant (Fig. 4, white bars). As expected, U0126 treatment reduced ERK2 activity (Fig. 4, dark gray bars) as also noted when U0126 was given prior to RAP treatment (not shown). Opposite effects of RAP on ERK2 were noted in APP WT and APP KO neurons; whereby RAP diminished ERK2 activity in APP WT neurons, presumably by blocking the interaction between APP and LRP; but enhanced ERK2 activity in APP KO neurons that lack full-length APP (Fig. 4, shaded bars). These findings provide the first evidence that binding of RAP directly to LRP in the absence of APP activates ERK2 in an APP-independent manner.
Many studies have evaluated APP effects on neuron development in both the central nervous system (CNS) and peripheral nervous system (PNS). Some of the most intriguing findings have come from neuronal cultures using APP KO mice. In our current study, we evaluated low density hippocampal cultures from APP WT and APP KO mice to further explore the impact of full-length APP and sAPPα on neurite growth as a model of hippocampal plasticity, controlling for sAPP effects using purified sAPPα and RAP. It is known that hippocampal neurons express similar levels of APP751, which contains the KPI domain, and APP695 (Johnson et al. 1990). Extracellular RAP, which blocks sAPPα uptake into neurons via LRP, was used as a control to help elucidate our earlier findings (Perez et al. 1997). We noted a significant increase in axon branching for APP KO neurons exposed to sAPPα (Fig. 2); however, there were very limited sAPPα effects overall, even in APP WT cultures, in which small amounts of endogenous sAPPα, sAPPβ, or Aβ may have been released by the neurons during the time after plating. Thus, we focused the remainder of our studies on the effects of full-length APP and on novel findings pertaining to RAP in both APP WT and APP KO neurons.
Abundantly apparent in our results is the finding that the absence of full-length APP on KO hippocampal neurons led to increased axon outgrowth, axon branching, minor process outgrowth, and total neurite growth, supporting a major inhibitory role by full-length cell-surface APP on neurite outgrowth. Remarkably, RAP potentiated already robust total axon outgrowth for APP KO neurons, suggesting that in cells lacking APP, the binding of RAP to LRP transduces a signal that enhances neurite growth. Moreover, for APP WT neurons, RAP reduced axon branching even further as described below, consistent with the findings of Qiu et al. (2004) using cortical neurons.
Our most robust findings revealed that full-length APP strongly inhibited axonal growth, much like results from other laboratories (Ikin et al. 2007; Young-Pearse et al. 2008). In the current studies, we measured all aspects of outgrowth including the main processes and all side branches of both axons and minor processes. We found that neurons expressing full-length APP (APP WT) had significantly less axon growth including fewer branches off the main axonal process (Fig. 2), which resulted in less total axon outgrowth (Fig. 1a). When APP WT neurons were exposed to RAP, we noted a further decrease in axon branch number per cell (Fig. 2) suggesting that RAP blockade of APP–LRP interactions further contributed to inhibition of already limited axon branching of APP WT neurons. A review of the literature allows us to propose the following explanation for these finding. Ikin et al. (2007) found that full-length APP binds to the cytoskeletal regulatory protein Mena via interactions with the adaptor molecule, Fe65, to form a complex with APP that inhibits branching of axons. Young-Pearse et al. (2008) found that full-length APP inhibits the growth of the main axonal process by interacting with Integrin ß1. In addition, Integrin ß1 can modulate ERK2 activation (Reyes-Reyes et al. 2002). Our results suggest that APP–LRP binding, when blocked by RAP, acts to limit growth-associated ERK2 signaling.
On the basis of the above-mentioned studies and our results, we propose a model whereby full-length APP and LRP both modulate axon growth (Fig. 5). With regard to full-length APP, interactions between APP and Integrin β1 at the cell surface combined with intracellular interactions between Fe65, Mena, and APP serve to limit axon growth and branching (Fig. 5a). The addition of RAP to APP WT neurons decreased the length of the longest axonal process (Fig. 3b) as well as axonal branches (Fig. 2). This may be because RAP bound to LRP and prevented LRP binding to full-length APP, diminishing APP endocytosis and increasing cell-surface APP levels (Ulery et al. 2000; Conboy et al. 2005), thus amplifying APP inhibitory effects on neurite growth (Fig. 5b). Lack of full-length APP on APP KO neurons changed outgrowth in a manner that also supports a role for APP as an outgrowth inhibitory molecule. APP KO neurons had more robust outgrowth that was primarily associated with increased axon branching, likely attributable to the absence of full-length APP (Fig. 5c). No significant differences were seen for APP WT and APP KO neurons with regard to axonal branch lengths or the length of the main axonal process.
We initially thought that administration of RAP to APP KO neurons might have no effect as there is no APP–LRP interaction to be blocked. Instead, we found that the administration of RAP to APP KO neurons resulted in increased axon outgrowth (Fig. 1), a tendency to increase axon branch numbers (Fig. 2) and an increase in branching off the main axonal process (Fig. 3a). Considering that APP is not present on APP KO neurons to interact with LRP, in that situation we might assume that an interaction between RAP and LRP initiated a signaling cascade that accentuated the growth of branches emanating from the main axonal process (Fig. 5d). In support of this, ERK2 is implicated in neurite outgrowth (Perron and Bixby 1999; Wang et al. 2011) and we anticipated that higher levels of activated ERK2 may be present in APP KO neurons, which also had the most total outgrowth of axons and dendritic minor processes (Table 1). In fact, APP KO neurons exposed to RAP had significantly elevated ERK2 activity (Fig. 4), suggesting that a RAP–LRP interaction itself transduces ERK2 signaling in hippocampal neurons when full-length APP is missing (Fig. 5d). LRP is associated with signaling (Herz 2001), and a study by Qiu et al. (2004) noted that RAP alone added to mouse cortical neurons stimulates ERK2, though their cells expressed APP and no increase in outgrowth occurred. Interestingly, we find that hippocampal and cortical neurons from APP KO mice display unique neurite outgrowth patterns (unpublished).
In conclusion, our data reveal that because APP plays such a major inhibitory role on neurite outgrowth, changes associated with AD pathology or therapies directed at APP, may tip the balance of APP in a manner to impair synaptic plasticity in hippocampus. It has long been appreciated that there is aberrant neurite growth in AD brain (Cotman et al. 1990; Masliah et al. 1991). Furthermore, levels of LRP decrease with age and a polymorphism in the LRP gene is linked to late-onset AD (Kang et al. 1997). Additionally, as LRP and its ligands modulate Aß synthesis and degradation, LRP is becoming an increasingly popular target for AD therapy (Vana et al. 2009; Zlokovic et al. 2010). Thus, obtaining a better understanding of normal APP and LRP functions in hippocampal neurite growth offers an opportunity for translating findings from the bench into new AD therapies.
We thank Bill Rebeck (Georgetown University) for RAP, Dominic Walsh (Harvard Medical School) for help purifying sAPPα, Merck Laboratories (Hui Zheng) for APP KO mice; Lupe Vidal, Ernesto Villanueva, and Sesha Krishnamachari for critical reading of the manuscript. Funding was provided by the Alzheimer's Association (to RGP) and the Alzheimer's Disease Research Center at the University of Pittsburgh [NIH P50AG005133, Seed Award to RGP]. The authors have no conflict of interest and declare that the funding agencies played no role in design, analysis, or interpretation of data. Author credits: RGP conceived of experiments and experimental design, prepared cultures, purified sAPPα, performed statistical analyses, supervised manuscript preparation, revised drafts for intellectual content, and had final approval of the submission. AJB performed literature searches, analyzed and interpreted data, prepared the figures, wrote the manuscript, and designed the hypothetical model. CY maintained cultures, performed immunostaining, did western blots, and worked on data acquisition; IB assisted with data acquisition.