Factors Influencing the Measurement of Lysosomal Enzymes Activity in Human Cerebrospinal Fluid

Measurements of the activities of lysosomal enzymes in cerebrospinal fluid have recently been proposed as putative biomarkers for Parkinson's disease and other synucleinopathies. To define the operating procedures useful for ensuring the reliability of these measurements, we analyzed several pre-analytical factors that may influence the activity of β-glucocerebrosidase, α-mannosidase, β-mannosidase, β-galactosidase, α-fucosidase, β-hexosaminidase, cathepsin D and cathepsin E in cerebrospinal fluid. Lysosomal enzyme activities were measured by well-established fluorimetric assays in a consecutive series of patients (n = 28) with different neurological conditions, including Parkinson's disease. The precision, pre-storage and storage conditions, and freeze/thaw cycles were evaluated. All of the assays showed within- and between-run variabilities below 10%. At −20°C, only cathepsin D was stable up to 40 weeks. At −80°C, the cathepsin D, cathepsin E, and β-mannosidase activities did not change significantly up to 40 weeks, while β-glucocerebrosidase activity was stable up to 32 weeks. The β-galactosidase and α-fucosidase activities significantly increased (+54.9±38.08% after 4 weeks and +88.94±36.19% after 16 weeks, respectively). Up to four freeze/thaw cycles did not significantly affect the activities of cathepsins D and E. The β-glucocerebrosidase activity showed a slight decrease (−14.6%) after two freeze/thaw cycles. The measurement of lysosomal enzyme activities in cerebrospinal fluid is reliable and reproducible if pre-analytical factors are accurately taken into consideration. Therefore, the analytical recommendations that ensue from this study may contribute to the establishment of actual values for the activities of cerebrospinal fluid lysosomal enzymes as putative biomarkers for Parkinson's disease and other neurodegenerative disorders.


Introduction
Lysosomes are involved in the degradation of macromolecules, and the impairment of their functionality due to a genetic lack of acidic hydrolases is the cause of lysosomal storage diseases (1). In recent years, lysosomal dysfunction has also been shown in patients with neurodegenerative disorders, i.e., Parkinson's disease (PD) (2,3) and Alzheimer's disease (AD) (4,5), where the accumulation of undegraded and misfolded proteins occurs. Mutations in the GBA1 gene, which codes for the lysosomal glycohydrolase b-glucocerebrosidase (EC = 3.2.1.45), cause Gaucher's disease (6). Several reports have demonstrated that GBA1 mutations are an important risk factor for PD (7,8,9,10). Either the inhibition or down-regulation of b-glucocerebrosidase causes a-synuclein accumulation in cultured cells along with mitochondrial dysfunction (11). A self-propagating loop in which mutated b-glucocerebrosidase induces a-synuclein accumulation within the lysosomes that subsequently compromises lysosomal protein degradation has been postulated (12). As a further demonstration of a link between b-glucocerebrosidase deficiency and PD, decreased b-glucocerebrosidase activity has been described in the cerebellum and substantia nigra of PD patients, even in those without GBA1 mutations (13). Alterations in lysosomal chaperonemediated autophagy have been demonstrated to cause the reduction of b-glucocerebrosidase protein levels and activity in the brain of early stage PD patients, especially in brain regions showing increased a-synuclein levels (14). b-glucocerebrosidase activity has been found to be decreased in the CSF of PD patients compared to individuals affected by other neurological disorders (OND) (15). In another study that assessed the CSF b-glucocerebrosidase activity in patients with different dementia disorders, including synucleinopathy (dementia with Lewy bodies) and tauopathies (AD and frontotemporal dementia), a selective decrease was observed in patients diagnosed with dementia with Lewy bodies, while a-mannosidase (EC = 3.2.1.24) showed a marked decrease in all of the groups studied (16). Recently, increased b-galactosidase (EC = 3.2.1.23) and cathepsin E (EC = 3.4.23.34) activities and a reduced a-fucosidase (EC = 3.2.1.51) activity have been found in the CSF of PD patients compared to healthy controls, and b-glucocerebrosidase showed a trend toward reduction in the PD group (17). Most recently, significantly reduced b-glucocerebrosidase activity was found in a cohort of 71 PD patients compared to 45 neurological control patients. Interestingly, the b-glucocerebrosidase activity reduction was more pronounced in patients with an early stage of the disease (18).
The reasons for these different outcomes may be found both in the heterogeneity of the diagnostic groups studied (early vs late, de novo vs treated PD patients and healthy volunteers vs neurological patients as control groups) and in pre-analytical confounding factors, i.e., different conditions for the collection and storage of the CSF samples. Differences in the CSF handling and analytical procedures should play a negligible role in the total variability of these assays because all of the previously cited CSF studies were performed in a single laboratory.

Subjects
Twenty-eight subjects were included in this study. Twelve of the subjects were diagnosed as having PD according to the United Kingdom Parkinson's Disease Society Brain Bank criteria (19), whereas the remaining 16 patients were affected by other minor neurological disorders (OND) requiring lumbar puncture (LP) for diagnostic purposes (epileptic seizures, transient global amnesia,

Ethics Statement
The study was approved by the ethical committee of the Umbria region. None of the subjects included in the study presented with neurological deterioration. LP was performed on subjects in the OND group for diagnostic purposes. Patients were included only if they were able to understand the study aim and procedures. Written informed consent was obtained from all of the subjects included in this study.

CSF collection
The LPs were performed between 8:00 and 10:00 AM. The CSF samples (10 ml) were collected in sterile polypropylene tubes, centrifuged for 10 minutes at 2,0006g, and stored in aliquots under the various conditions described below.

Lysosomal enzyme assays
The activities of the CSF lysosomal enzymes were determined with the corresponding fluorigenic substrate and reaction buffer ( Table 2). To measure the b-hexosaminidase, a-fucosidase and bmannosidase, 10 ml of CSF were incubated with 40 ml of the substrate solution for 10 min at 37uC. In the a-mannosidase and b-galactosidase assays, 20 ml of CSF were incubated in the presence of 40 ml of the corresponding substrate solution for 3 h. The cathepsin D and cathepsin E activities were determined using MOCAc-Gly-Lys-Pro-Ile-Leu-Phe-Phe-Arg-Leu-Lys(Dnp)-D-Arg-NH 2 as a substrate dissolved in 50 mM sodium acetate buffer with a pH = 4.0. The samples were incubated with the substrate at a concentration of 30 mM for 1 h at 40uC in the presence and absence of the cathepsin D inhibitor pepstatin A (15 mM). The reactions were stopped by adding 5% (w/v) trichloroacetic acid to a final volume of 3 ml and were measured with a Jasco FP-750 spectrofluorometer at an excitation wavelength of 328 nm and an emission wavelength of 393 nm. The value obtained in the presence of pepstatin A corresponded to the cathepsin E activity. The cathepsin D activity was calculated by subtracting the value obtained in the presence of pepstatin A from that measured in the absence of the inhibitor.
The standard curves have been created using serial dilutions of the fluorochromes 4-methylumbelliferone and Mca-Pro-Leu-OH peptide (for cathepsins). In order to minimize the standard curves variability, they have been created by the same operator using the same stock reagents solutions. Moreover, substrates were prepared Table 3. Within-and between-run variabilities of the activities of CSF lysosomal enzymes.

Enzyme
Within-assay CV (%) Between-assay CV (%) Total variability (%) The within-assay and the between-assay coefficients of variation (CVs) together with the total variability are reported for each enzyme. The CSF aliquots stored at -80uC were measured across a ten day period. doi:10.1371/journal.pone.0101453.t003 and frozen into aliquots, so that in every assay the substrate was used after one freeze/thaw cycle. One unit (U) of enzyme activity was defined as the amount of enzyme that hydrolyses 1 mmol of substrate/min at 37uC. All of the activities were measured in triplicate. All of the samples included in this study were assayed for the activities of the lysosomal enzymes within 30 min from the collection (fresh CSF). The remaining volume of each sample was used to assess the confounding factors.

Statistical analysis
The statistical analysis was carried out with R v.2.15 (20). The descriptive statistics are presented as the mean 6 SD for the continuous variables and as percentages for the categorical variables. To analyze the variability of the enzymes' activities, mixed-effects regression models were used for taking into account the repeated measurements of samples between storage techniques, freeze-thaw cycles and time points (21). The statistically significant level is considered as p,0.05.

Results
Analysis of the effect of the diagnosis on pre-analytical factors has been performed, showing no significant differences between OND and PD groups for the measured enzyme activities (data not shown). Therefore, the results are presented considering the whole group of samples independently from the diagnosis. Precision of the assays for evaluating the activities of lysosomal enzymes in CSF Eight CSF samples were flash-frozen in liquid nitrogen in ten separate aliquots. One aliquot was thawed and used for enzymatic assays each day for ten consecutive days. Each enzyme was tested in duplicate while each sample was measured in triplicate. The within-and between-run assay coefficients of variation (CVs) were calculated according to guideline EP5-A2 from the National Committee for Clinical Laboratory Standards (NCCLS) (22). As an internal control, a pooled CSF sample was run in each experiment to ensure that the measured samples were within an acceptable range, which was defined as 62 times the standard deviation (SD) from five consecutive measurements of the same pool.
The values of the within-and between-run CV for each enzymatic assay are reported in table 3. The within-run CVs were in the range of 3.31-7.43%, while the range for the between-run variability was 5.21-9.21%. Globally, the enzyme that showed the highest total variability was cathepsin E (16.64%), while the one with the lowest total variability was b-hexosaminidase (8.63%). Table 4 reports the median and the ranges of the enzymatic activities studied in fresh centrifuged CSF for all of the subjects enrolled in the study. The highest median activity was reported for b-hexosaminidase (3811.63, range = 1745.40-6919.98 mU/L), while b-glucocerebrosidase had the lowest median activity (6.52, range = 1.46-15.20 mU/L).

Variability related to the freezing conditions
A previous study found an increased stability in the activities of lysosomal enzymes when the CSF was frozen in liquid nitrogen (23). To test which freezing conditions should be recommended for the measurement of the activities of lysosomal enzymes, three conditions were compared: flash freezing in liquid nitrogen, direct freezing at 280uC and direct freezing at 220uC. Twelve CSF samples were aliquoted, placed on ice and frozen within 30 min after the lumbar puncture by placing them at 220uC, 280uC or in liquid nitrogen and then stored at 280uC. Within 3 days, the aliquots were thawed on ice and assayed.
Analysis of the variance among the three conditions tested showed that only the a-mannosidase activity was significantly reduced when the sample was frozen at 220uC compared to the fresh sample (p,0.001), flash freezing the sample in liquid nitrogen and directly freezing the sample at 280uC (p,0.01). The activities of all of the other enzymes showed no significant differences among the three freezing conditions (Fig. 1A).
Pre-storage conditions: the time delay before freezing and the temperature Several pre-storage conditions, resembling the most common procedures for CSF storage before freezing, were tested and compared to the respective fresh CSF sample. Twelve CSF samples were divided into aliquots, which were kept at 4uC or RT (22uC). One aliquot of each group was flash-frozen after 1, 4, 24, and 48 hours and stored at 280uC. After 1 week from the last freezing, all the aliquots were thawed on ice and used to determine the activities of the lysosomal enzymes. The activity of each enzyme in each sample was compared to that of the respective fresh sample. Fig. 1B shows the effect of storage at 4uC before freezing for time delays up to 48 hours. The most stable enzymes were bhexosaminidase, b-mannosidase and cathepsin D, which showed no significant changes when stored at 4uC for up to 48 hours. Alpha-fucosidase and cathepsin E showed no significant changes    Table 5). The effect of the storage at room temperature before freezing was more evident for all of the enzymatic activities tested ( Fig. 1C and Table 5). The enzymes that were most affected by the storage at room temperature were b-glucocerebrosidase and a-mannosidase, and after 4 hours, they showed decreases in their activities of 260.92615.21% and 2 13.94610.53%, respectively (Fig. 1C).

Freeze-thaw cycles
Five CSF samples were flash frozen in liquid nitrogen and stored at 280uC. The samples were thawed in ice altogether, assayed for the enzyme activities and frozen again at 280uC. Five cycles were performed using this procedure.
The mean percentage activity changes of the lysosomal enzymes for up to 5 freeze and thaw (F/T) cycles with respect to fresh CSF are reported in Fig. 2 and Table 6. The cathepsins and bgalactosidase were the most stable enzymes, and remarkably, the cathepsin E activity was not influenced by the F/T cycles. The bgalactosidase and cathepsin D activities showed significant changes only after 5 cycles. The b-glucocerebrosidase activity showed a moderate decrease (214.68612.82%) after 2 cycles, while the activity significantly dropped after 5 cycles (245.4265.23%, p, 0.01). The b-mannosidase, b-hexosaminidase and a-fucosidase activities significantly increased after 2 F/T cycles (by up to 120-140%, Table 6). The a-mannosidase activity decreased by 2 18.1862.90% after the first cycle, and had a maximal decrease of 245.6366.73% after 5 cycles (p,0.001).

Effect of blood contamination
To test if a traumatic LP had a direct influence on the activities of lysosomal enzymes in CSF, three CSF samples were spiked with the blood of the same patient to obtain the following scalar erythrocyte concentrations: 50,000, 5,000, 1,000, 500 and 250 erythrocytes/ml. The spiked CSF samples were assayed for the different enzyme activities, and the results were compared with the respective neat sample.
Blood contamination of the CSF samples with up to 50,000 erythrocytes/ml did not significantly modify the activity levels of any of the lysosomal enzymes in a dosage-dependent way (Fig. 3).

Longitudinal evaluation of the activities of lysosomal enzymes
Aliquots of 8 CSF samples were flash frozen and then stored at 220uC or 280uC. Four samples were used to assess the longitudinal stability of the enzymes at each temperature. One aliquot of each sample was thawed in ice and used for enzymatic assays after 1, 2, 3, 4, 8, 16, 32 and 40 weeks of storage. The differences between the enzyme activities of fresh CSF and the sample stored for 1 week were also calculated. To estimate the longitudinal stability of the activities of the enzymes, the samples stored for 1 week either at 220uC or 280uC were chosen as reference samples, and the percentage difference between each time point and the respective reference sample was calculated.
Initially, the change in the activity percentage between fresh CSF and the same sample stored for 1 week at -80uC was calculated for each enzyme. The results are reported in Fig. 4. The a-mannosidase enzyme showed the most significant changes, with a reduction of 253.48635.17% at 220uC and 217.9464.03% at 280uC. The other enzymes showed changes between 2 and 10%  when the samples that were stored for 1 week were compared to fresh CSF.
The tested activities had different degrees of variability when assayed longitudinally. Fig. 5 shows the percentage changes of the activities of the tested enzymes for samples stored at 220uC or 2 80uC for a period of 40 weeks compared to the same sample after 1 week of storage (100%). Storage at 220uC was the most unstable freezing condition, with almost all of the enzymes showing a large reduction in their activities starting from the second week of storage. Only b-hexosaminidase and cathepsin D showed a high degree of stability at 220uC for up to 32 and 40 weeks, respectively (Fig. 5). Storage at 280uC allowed for more reliable measurements of the enzymatic activities. b-galactosidase and afucosidase were the most unstable enzymes when stored at 280uC, and they showed increases in their mean percentage activities of 54.9638.08% after 4 weeks and 88.94%636.19% after 16 weeks, respectively. The b-glucocerebrosidase activity change with respect to the first week of storage was not significant up to 32 week, when the mean percentage activity decreased by 2 22.36618.18% (p,0.05). After 40 weeks of storage, b-glucocerebrosidase showed a mean percentage decrease of 2 19.60619.11%. The CSF b-hexosaminidase activity had a significant increase after 16 weeks of storage with respect to the reference sample (13.5568.56%, p,0.05). b-mannosidase, cathepsin D and cathepsin E did not show any significant changes up to 40 weeks of storage when stored at 280uC (Fig. 5 and Table 7).

Discussion
The potential diagnostic value of the activities of CSF lysosomal enzymes for different neurological diseases has been investigated  since the end of the seventies (24,25,26,27). Nevertheless, due to the thermal instability of these enzymes in biological fluids, this measurement has never been considered for clinical utilization (26,28). A renewed interest in these enzymes has emerged as a consequence of recent evidence highlighting the link between the impairment of the lysosomal system and PD pathology (11,29,30,31). Interestingly, some recent studies have shown that the activities of some lysosomal enzymes in the CSF of patients with PD and other synucleinopathies is altered compared to control patients or patients with other neurodegenerative diseases (15,16,17,18). Accordingly, the opportunity to use CSF lysosomal enzymes as biomarkers for PD deserves attention (32,33).
Here, we systematically evaluated a series of pre-analytical factors that may affect the measurement of the activities of some lysosomal enzymes in CSF. Our results show that the total variability of these assays is minimally influenced by the withinand between-run variability, which ensures day-to-day repeatability. Other analytical conditions affected each enzyme in a different fashion.
The activities of some enzymes seem to be weakly influenced by pre-analytical factors. In particular, the activity levels of cathepsins D and E showed elevated resilience to F/T cycles, high longitudinal stability up to 40 weeks and no significant change when kept at 4uC for at least 24 hours before freezing. Accordingly, the measurement of their activities in CSF may represent a promising biomarker when also considering their possible involvement in the pathogenesis of PD and AD (17,34,35). To our knowledge, this is the first study assessing the longitudinal stability of the cathepsin D and E activities in CSF. The b-mannosidase activity was found to be stable when the CSF samples were kept at 4uC for up to 24 hours and then stored at 2 80uC for up to 40 weeks similarly to the cathepsins, but its activity has a marked fluctuation after F/T cycles. Among the other lysosomal enzymes, the a-mannosidase activity was altered by almost all of the pre-analytical factors tested. For instance, after only 1 week of storage the a-mannosidase activity showed a large decrease (,20% at 280uC and ,60% at 220uC), which confirmed the findings of a previous study (23). The bhexosaminidase activity also showed marked fluctuations after F/T cycles. The b-galactosidase and a-fucosidase activities had similar inconsistencies when evaluated longitudinally. These changes could be explained by the presence of unstable enzymatic modulators in the CSF (36). Moreover, some enzymes are part of multi-enzyme complexes that could be disjointed during freezing steps, which could increase the affinity of each enzyme to the synthetic substrate (37).
Our results also showed that the CSF b-glucocerebrosidase activity is influenced by pre-analytical factors. A 40% decrease of the CSF b-glucocerebrosidase activity was reported when the samples were directly frozen at 280uC with respect to samples snap-frozen in liquid nitrogen (23). In this study, we were not able to find significant difference in the CSF b-glucocerebrosidase activity between the two conditions after 1 week of storage (same time frame as in the previous study). Furthermore, at 280uC, bglucocerebrosidase was stable for up to 32 weeks. These differences may be due to the time-delay and the storage temperature before freezing, which were not specified in the previous study and significantly affect the CSF b-glucocerebrosidase activity. Moreover, the two studies are not directly comparable because in Goi et al., the b-glucocerebrosidase enzymatic assay was performed in absence of TDC. Taurodeoxicolic acid enhances the specificity of the assay inactivating specific isoenzymes, as it has been shown in leucocytes (38).
The final recommendations for the proper assessment of the activities of CSF lysosomal enzymes emerging from this work are: i) the CSF samples should be centrifuged and kept at 4uC immediately after the LP and stored at 280uC within 1 h of the collection; ii) the b-mannosidase, cathepsin D and cathepsin E activities should be tested within 40 weeks of the LP, and the GCase activity should be tested within 32 weeks; and iii) the number of F/T cycles do not significantly affect cathepsins D and E for up to four cycles, but they should be avoided for bglucocerebrosidase and the other enzymes.
In conclusion, we have shown that after applying the correct pre-analytical operating procedures, the measurement of bmannosidase, b-glucocerebrosidase, cathepsin D and cathepsin E in CSF can be reliable and reproducible. These conditions are mandatory to correctly compare the results from different cohorts aimed at confirming the role of the activities of CSF lysosomal enzymes as potential biomarkers for neurodegenerative disorders, with a special interest on Parkinson's disease.