• Open Access

Oxidative damage in cultured human olfactory neurons from Alzheimer's disease patients


George Perry, PhD, Institute of Pathology, Case Western Reserve University, 2085 Adelbert Road, Cleveland, OH 44106, USA. Tel.: +1 216 368 2488; fax: +1 216 368 8964; e-mail: george.perry@case.edu


Oxidative abnormalities precede clinical and pathological manifestations of Alzheimer's disease and are the earliest pathological changes reported in the disease. The olfactory pathways and mucosa also display the pathological features associated with Alzheimer's disease in the brain. Olfactory neurons are unique because they can undergo neurogenesis and are able to be readily maintained in cell culture. In this study, we examined neuronal cell cultures derived from olfactory mucosa of Alzheimer's disease and control patients for oxidative stress responses. Levels of lipid peroxidation (hydroxynonenal), Nɛ-(carboxymethyl)lysine (glycoxidative and lipid peroxidation), and oxidative stress response (heme oxygenase-1) were measured immunocytochemically. We found increased levels for all the oxidative stress markers examined in Alzheimer's disease neurons as compared to controls. Interestingly, in one case of Alzheimer's disease, we found hydroxynonenal adducts accumulated in cytoplasmic lysosome-like structures in about 20% of neurons cultured, but not in neurons from control patients. These lysosome-like structures are found in about 100% of the vulnerable neurons in brains of cases of Alzheimer's disease. This study suggests that manifestations of oxidative imbalance in Alzheimer's disease extend to cultured olfactory neurons. Primary culture of human olfactory neurons will be useful in understanding the mechanism of oxidative damage in Alzheimer's disease and can even be utilized in developing therapeutic strategies.


Oxidative damage and antioxidant responses are the earliest features of vulnerable neurons in the brain of patients with Alzheimer's disease (AD), with damage occurring to every category of biological macromolecule prior to lesion formation (Perry et al., 1998). Moreover, a number of lines of evidence now indicate that oxidative damage and antioxidant responses precede clinical and pathological manifestations of AD (Nunomura et al., 2000, 2001).

The human olfactory system is also targeted in the early stages of AD, with the olfactory epithelium displaying oxidative damage (Perry et al., 2003). Neuronal populations that receive olfactory connections are among the earliest brain regions to develop pathological lesions (Pearson et al., 1985; Reyes et al., 1987). The olfactory bulb, for example, contains significant numbers of neurofibrillary tangles early in the course of AD (Talamo et al., 1989; Tabaton et al., 1991; Trojanowski et al., 1991). Together with recent findings of amyloid-β deposition in the lens (Goldstein et al., 2003), extension of the central nervous system (CNS) to the exterior offers a window to understanding brain changes in disease.

Because olfactory neurons regenerate and can be sampled from living patients, we hypothesized that olfactory neurons, which can be maintained in culture, could provide a valuable means of studying early biochemical changes in AD. Extending our previous findings on oxidative damage in AD to cultured olfactory neurons provides a cellular model system in which to study the molecular basis for oxidative imbalance in AD. When we examined cultured olfactory neurons from AD and control patients immunocytochemically for biochemical indicators of oxidative imbalance, we found many of the changes characteristic of AD are maintained in cultured neurons. These studies open the way to examine the mechanism responsible for oxidative imbalance in AD directly, as opposed to genetically modified systems, as well as to use these cells to screen potential therapeutic agents.

Materials and methods

Cell culture

Olfactory neuroblasts were generated as described previously (Wolozin et al., 1992; Johnson et al., 1994). Biopsies of olfactory epithelium were obtained from donors with probable AD (n = 4) based on NINCDS–ADRDA criteria for ‘probable AD’ (McKhann et al., 1984) and age-matched control donors (n = 3).

Briefly, the neurons and subsequent passages were maintained in culture as previously described. Prior to thawing the olfactory neuroblasts, Matrigel basement membrane (Biowhittaker, Rockland, ME, USA) was reconstituted at a 1 : 10 dilution in phosphate-buffered saline (PBS). The olfactory neuroblasts were plated on this basement membrane and incubated in 4506 medium containing Ham's F12-based medium (neuroblast formulation; Gibco, Rockville, MD, USA), supplemented with 6% fetal bovine serum (Gibco), 150 µg mL−1 bovine hypothalamus extract, 50 µg mL−1 bovine pituitary extract (Gibco), 1 µg mL−1 insulin (Sigma, St Louis, MO, USA), 5 µg mL−1 human transferrin (Gibco), 10 nmol L−1 hydrocortisone (Sigma), 40 pg mL−1 thyroxin, 2.5 ng mL−1 sodium selenite (Gibco) and 1% antibiotic–antimycotic (Gibco).


Olfactory neuroblasts were plated on coverslips coated with poly-d-lysine (25 µg mL−1) (Sigma) and incubated overnight to allow the cells to adhere, followed by fixation with methacarn (methanol–chloroform–acetic acid; 6 : 3 : 1) for 1 h at room temperature. The coverslips were then incubated for 15 min in 10% normal goat serum followed by incubations with primary antibodies for 16 h at 4 °C. The cells were stained with the peroxidase antiperoxidase method (Sternberger, 1986) with 3,3-diaminobenzidine (DAB) and 0.015% H2O2 as co-substrates.

The markers used were antiserum against heme oxygenase-1 (HO-1) 1 : 100 (Smith et al., 1994); antiserum to the lipid peroxidation adduct hydroxynonenal (HNE) pyrrole (1 : 100) (Sayre et al., 1996); and Nɛ-(carboxymethyl)lysine (CML) 1 : 100 (Castellani et al., 2001).

Neuronal phenotype was confirmed morphologically and by antiserum against the neuron-specific marker PGP9.5 (1 : 1000), which recognized over 99% of the cells growing in culture.


Intensity of immunoreaction for the markers used was measured using an Axiocam digital camera and KS300 image analysis software (Carl Zeiss, Inc., Thornwood, NY, USA). The cells were outlined manually and the computer-generated optical density values determined. Background values were subtracted and the mean densities determined for each case. Values from the AD cases were compared with values obtained from neurons from control cases. A nested analysis of variance (with neurons nested within subjects) was used to compare AD patients with controls (Neter et al., 1996).


The cell lines used for biochemical analysis in the present study were prepared from olfactory mucosa of four probable AD patients and three controls taken at biopsy. The cells grew slowly in culture. The cultures yielded over 99% olfactory neurons, confirmed by morphology (neuritic outgrowth) and the neuronal marker PGP9.5. Neurons continued to proliferate beyond the 2–3 weeks used for this analysis with similar growth rates between the AD and control neurons, with the daughter cells displaying the same features as the initial cultures.

Neurons from patients with AD displayed a higher level of oxidative damage than neurons from controls. Quantitative analysis showed the difference reached significance for HNE (P ≤ 0.0001), HO-1 (P = 0.0002) and CML (P = 0.0002) (Fig. 1).

Figure 1.

Immunocytochemical quantification of HNE, HO-1 and CML. CML and HO-1 showed the most significant differences between AD and control (P = 0.0002 and 0.0002, respectively). HNE was also significant (P ≤ 0.0001). Results were further analysed for statistical significance using Student's t-test.

Lysosome-like structures containing HNE-adducts were prominent in neurons from one patient with AD, with 19% of the neurons containing them (Fig. 2).

Figure 2.

Immunocytochemical analysis of HNE and HO-1. Hydroxynonenal (a,b) and HO-1 (c,d) are both elevated in cells derived from AD patients. Although hydroxynonenal was increased in AD (a), the most striking feature is the prominent lysosome-like cytoplasmic material seen in some AD neurons (inset) that was absent from all the control cases (b). The intensity of HO-1 was greater in the AD case (c) as compared with the control case (d).


In this study, we validate that neurons from olfactory mucosa can be successfully cultured and maintained as a primary cell line for studies of oxidative stress. The neurons maintain neuronal phenotype morphologically, with the presence of neuritic cytoplasmic processes, and by expression of the neuron-specific marker PGP9.5. Olfactory neurons are particularly appealing for the study of the in vivo mechanisms responsible for AD as they are normally capable of entering the cell cycle and dividing, unlike the vast majority of neurons within the CNS. Consistent with this notion was our finding that primary olfactory neurons are sustainable for longer periods of time than primary neuronal cultures derived from other CNS structures. In addition, because we are examining differences between AD and controls without genetic modification, the approach has the promise to reveal the pathophysiology of AD. In addition, because olfactory neurons represent an interface between the CNS and the external environment, they could theoretically provide meaningful information regarding the impact of acquired stressors on AD pathogenesis. Disease-specific biochemical changes found in the cortex and other affected brain areas can also be found in olfactory neurons. Unfortunately, olfactory neurons are difficult to obtain and have proven hard to maintain in culture, thereby limiting the number of cases available as we found in our own study.

We find here that biochemical markers of oxidative stress, specifically HNE-pyrrole, CML and HO-1, well characterized as representing early pathological changes in the AD brain (Smith et al., 1994; Sayre et al., 1997; Castellani et al., 2001), are significantly increased in cultured AD neurons compared with controls. Although the changes are significant, they are less striking than the changes seen in brain of a greater range of oxidative abnormalities (Perry et al., 1998; Nunomura et al., 2001). Because the olfactory system and its projections are among the earliest structures affected by neurofibrillary pathology in AD, these findings of increased oxidative damage substantiate the accumulating data that oxidative imbalance is an early pathogenic factor (Nunomura et al., 2001). This study also suggests, as we have previously asserted, that amelioration of oxidative stress represents an important treatment and preventative strategy (Perry et al., 1998). Additionally, and also of potential therapeutic note, it should be pointed out that recent studies show that neurogenesis is increased in AD (Chen et al., 2003), and, given the role of HO-1 in neurogenesis, the increase in HO-1 reported here may be a response to enhanced cell cycling (Raina et al., 2000; Bowser & Smith, 2002) in addition to oxidative stresses.

In addition to developing biochemical changes consistent with the changes found in the AD brain, the cultured AD neurons develop similar morphological structures. The accumulated granular lysosome-like structures within the cytoplasm of cultured olfactory neurons from one of the AD cases are similar to those found in the pyramidal neurons in the AD hippocampus when HNE-pyrrole adducts are also seen (Sayre et al., 1997). Although only 19% of the olfactory neurons in this case display the lysosome-like structures, previous studies have shown increased levels of lipid peroxidation in vulnerable pyramidal neurons and neurofibrillary pathology in AD. Because these structures contain lipid peroxidation adducts, they may represent degradation products of membrane lipids that are increased in AD. Whether these structures contain evidence of mitochondrial turnover secondary to oxidative damage remains to be determined, although it is noteworthy that we have previously demonstrated residual mitochondrial protein within lipofuscin and other secondary lysosomes, in the perikaryal cytoplasm of neurons in AD (Hirai et al., 2001).

In summary, we demonstrate increased levels of HNE-pyrrole, CML and HO-1, well-known biochemical markers of oxidative damage, in neuronal cultures derived from olfactory mucosa in AD, substantiating our previous assertions that oxidative imbalance is an important early feature of AD pathophysiology. The use of cells showing abnormalities defining actual patients with AD offers opportunity to dissect the nature of oxidative imbalance relevant to AD patients directly. Further studies will utilize fibroblast cultures, which are easier to obtain and still allow for individual patient analysis. This approach offers direct analysis of therapeutic strategies, including therapies aimed at diminishing the effects of oxidative imbalance.


This research was supported by Panacea Pharmaceuticals, Inc. G.P. and M.A.S. are compensated consultants and own equity in Panacea.