The proximate cause of neuronal dysfunction and death in Huntington Disease (HD) is the presence of soluble mutant huntingtin (_mtHtt) protein or its cleavage products [1] [2] [3] . Consequently, soluble _mtHtt is the most salient target for disease modifying therapies and measurement of soluble _mtHtt is essential for evaluating the success of therapies intending to affect _mtHtt levels. Using a cell based Homogeneous Time-Resolved Foerster Resonance Energy Transfer (HtrFRET) assay or HTRF, Weiss et al [1] have shown that _mtHtt was specifically detectable in human brain and blood cell sub-populations, as well as in tissues and blood from HD mouse models, in correlation with the disease. Furthermore, using properties of Lumi4-Terbium cryptate (Tb) Weiss et al [4] have developed a triplex assay, in which _mtHtt, as well as an epitope distant from the Htt N-terminus, hence an indication of total huntingtin (_tHtt), are measured concomitantly in the same sample.

This assay, applied to peripheral blood samples, hence easily accessible, from HD patients, could serve as a biomarker for assessing whether systemic therapies affect levels of _mtHtt. We chose buffy coat (BC) as a blood subpopulation showing measurable Htt levels (Weiss et al [5] and that is also relatively easy to quantify and work with. To provide the necessary consistency, reliability, and quality control essential for a clinical diagnostic bioassay, we have optimized and qualified the assay for BC samples and developed Standard Operating Procedures (SOPs), so that the assay can be conducted under Good Laboratory Practice (GLP) conditions and in compliance with Federal Drug Administration (FDA) requirements. We first optimized the assay to measure soluble _tHtt and _mtHtt in human brain lysates from control and HD subjects. Parameters providing the highest specificity and sensitivity for _mtHtt were selected and the assay was used to test sets of unblinded human BC samples from HD and control subjects, to qualify/ validate the assay in compliance with GLP. Following the assay qualification, a formal SOP consistent with FDA requirements for a potential diagnostic assay was issued. Using this SOP, we have started to test clinical samples to assess relative blood levels of _mt/tHtt using blood samples from active single and multi-center clinical research studies. Our results using this SOP confirm that the assay is sensitive, linear within the dynamic range and reliable and we are deploying it as a GLP assay to explore its potential as a peripheral biomarker for HD.

Antibodies: The monoclonal antibodies used in this cell-based HTRF assay are specific for selected epitopes on the Htt molecule. The antibodies include: 2B7 [5] ,which binds to the first 17 amino acids (aa) of normal and _mtHtt; MW1 [6] ,which is specific for expanded polyglutamine sequences (polyQ) and binds to _mtHtt, but not to normal Htt; and 2166 (Millipore Corp, Cat# MAB2166 [7] , which binds to a Htt epitope starting at aa 444, and recognizes both normal and _mtHtt. 2B7 was conjugated to Tb (Cisbio), to serve as the donor for HTRF, while MW1 was bound to AlexaFluor 488 (AF488, Alexa Fluor 488, Invitrogen, Cat#A20181) and 2166 was conjugated to d2 (Cisbio) to serve as acceptors and to enable multiplexing between the donor and the two acceptors. Additional antibody pairs were also evaluated (not shown), however the antibodies mentioned above showed the largest HTRF signals and discrimination between the controls and HD and were therefore selected for the optimization study.

Human Samples: Human blood samples were collected at the MGH Huntington’s Disease Center by standard phlebotomy from consenting HD and control subjects participating in the Partners IRB approved REVEAL-HD study. Genotyping was confirmed by the MacDonald laboratory (Molecular Neurogenetics Unit Genotyping Resource) at MGH. Twenty-four of these samples were from 12 individuals in whom venipuncture was repeated within six weeks. The peripheral blood samples, collected into EDTA tubes, were centrifuged at 4°C for 10 minutes (min) at 1000 x g. BC fractions were isolated and centrifuged again for 20 min at 4°C at 15800 x g, then stored at -80°C until their analysis. The BC pellets were pre-diluted 1:2 in “lysis buffer” and an additional two serial two-fold dilutions were performed.

Brain Tissue Samples: Postmortem frontal cortex samples from HD and control patients obtained from the Brain Bank of the Alzheimer Disease Research Center (ADRC) at MGH were homogenized in 10x volume “lysis buffer” as above. Stocks of lysates were prepared, their protein concentrations determined and the stocks aliquoted and stored at -80 ^° C for use as quality control (QC) samples and for assay development and troubleshooting purposes.

HTRF Assay: Our optimization process examined plate formats; the performance of various plate readers and filters, reaction volumes, buffer compositions, cell lysate and antibody concentrations, quality control and normalization approaches, and reaction incubation times. Optimal conditions utilized white, 384 well/plate (Perkin Elmer, Cat# 6008280), and a total volume of 15 µL/well. The cell lysates at 12.5 µL/well were derived from a 1:1 dilution of the BC samples and the two subsequent two fold dilutions, or from 9 µg/well of each control and HD brain lysate diluted in the lysis buffer. The fluorophore-conjugated antibodies were in a 2.5 µL/well reaction buffer, that included a final dilution of 1.25ng/well 2B7-Tb cryptate-labeled antibody and 2.35ng/well MW1-AF (different lot than previously mentioned) and 13ng/well 2166-d2-labeled antibodies in 50 mM NaH₂ PO₄ (NaH₂ PO₄ -H₂O, Sigma Aldrich, S9638), 200 mM KF (Sigma-Aldrich, Cat# 402931-100G), 0.1% BSA (Sigma, Cat# A-7906) and 0.05% Tween 20 (Fisher, Cat# BP337-500). The concentrations of conjugated-antibodies are lot dependent, therefore their concentration is adjusted with each new lot to yield similar ratio of _mt/tHtt of HD/control human brain lysates. The plates were spun for 3 min at 500 x g then incubated in the dark on an orbital shaker at 4°C for 2 hours (hr). The HTRF signals were read on a VICTORX5 plate reader (Perkin Elmer). The emission from Tb was captured at 615 nm, following optimization and we term the values, for simplicity, maximum (Max) signal. The ratio of signals at the 665/615 nm and 510/615 nm, respectively, represents a specific, artifact corrected determination of the two antibody pairs: 2166-d2/2B7-Tb and MW1-AF/2B7-Tb, simultaneously bound to Htt and reveal information about _tHtt/Max and _mtHtt/Max signal, respectively.

dsDNA Content was measured in all the wells (including the dilutions of each sample) of the assay following the reading of the HTRF signals using PicoGreen (Quant-iT PicoGreen dsDNA Quantitation Kit, Invitrogen, Cat# P11496). A standard curve of dsDNA was generated on every plate, to make sure that the samples measured are within the dynamic range of the standard curve. The plate was read at an excitation of 485 nm and an emission of 535 nm.

The protein concentrations of the human brain lysates and BC samples were determined using DC Bio-Rad, Protein Assay Reagents, per the kit instructions.

Purified 548Q25, 548Q46 and 548Q72 Htt fragments: Purification of recombinant huntingtin protein fragments was performed as previously described (Weiss et al [ [5] [4] ). We thawed the cryo-stored purified Htt fragments at room temperature, prior to spiking them into the control buffy coat sample at 1:1 ratio, yielding Htt fragments concentration of 300-0 ng/mL and a 1:4 final dilution of the control BC used. The HTRF assay of the spiked samples and controls proceeded as described above.

Tb has high stability and long-lived fluorescence emission over a wide spectrum (480-720 nm) following its excitation at 340 nm. Its emission signal is captured at 615 nm-maximum (Max) signal. The energy transferred from the Tb excites the two acceptors: d2, with the peak emission at 665 nm (red) and AF488 (AF), with its peak emission at 510 nm (green). These filters were fine tuned as a result of optimization of the initial assay (Weiss et al [5] ). Time resolved detection based on Tb characteristics eliminates the background fluorescence, resulting in a more specific signal. The interaction between the fluorophores reflects the proximate binding of antibodies that are conjugated to their specific epitopes on the Htt molecule (Weiss et al [4] ) in this triplex assay. The antibodies we used in the HTRF assay, their conjugated fluorophores and their peak emissions and the antibodies’ epitopes on the Htt molecule are represented schematically in Figure (Fig.) 1:

HTRF Assay Schematic Summary

To qualify the HTRF assay and validate soluble _mtHtt as a potential clinical biomarker, we developed and optimized a protocol based on the Weiss et al assay [5] , following GLP, to insure: standardization, reproducibility, clinical utility, the potential for regulatory acceptance, and to provide a basis for future modifications or variations. The immediate technical goal of the optimization process was to develop a protocol that provides the strongest detection of _mtHtt (MW1/2B7) and _mt/tHtt (MW1/2166) HTRF signal ratios, using human brain lysates as QCs. Several of the parameters evaluated experimentally for the clinical assay optimization are listed below.

Plate Format and Plate Color: We determined that 384 well plates gave a stronger specific signal compared to 96 well plates (results not shown), which is related to the optical interaction of the light, the fluorophores and the height of the reaction fluid (Cisbio, technical information). We compared black and white 384 well plates and evaluated the impact on the relative HTRF signals. We had hypothesized that black plates would reduce the fluorescent background signal better than white plates, however white 384 well plates showed consistently higher HTRF signals compared to black plates. This suggests that the time-resolution fluorescence conferred by the Tb, reduces the background sufficiently, irrespective of plate color and therefore white plates were used from that point forward.

Reaction Volume and Intra-assay Reproducibility : We compared (Fig. 2) the effects of different reaction volumes in the triplex HTRF using: 12 µL/well versus 15 µL/well in HD compared to control samples. Our results, represented as average (AVE) of HTRF raw signals ± standard deviation (STDEV) of the triplicates, indicate that at 15 µL/well there is better intra-assay reproducibility than at 12 µL/well for all the antibodies/fluorophores combinations: 2166-d2, 2B7-Tb and MW1-AF488, measured in the same well. The smaller coefficients of variation (CV) between the replicate wells increase confidence in the accuracy of the measurement and the signal seems more specific showing a higher signal with MW1-AF488 for HD compared to controls. Therefore, going forward we used 15 µL/well as the reaction volume.

Reaction Media (KF concentration): We compared (Table 1) concentrations of 83 mM KF to 200 mM KF in the reaction media (concentrations in use by different labs) and we found that 200 mM KF gave higher signal ratios, due to reduced non-specific binding in the control brain lysates, therefore we adopted 200 mM KF for our reaction media in the HTRF assay.

HTRF Reaction Time Course : Evaluation of the time course of the HTRF assay (Fig. 3) revealed an increase in the _mt/tHtt (MW1-AF/2166-d2) of HD/control HTRF signal peaking at 2-3 hrs, followed by a decrease in this ratio from 9.2 at 3 hrs to 5.0 at 23 hrs, due mostly to an increase with time of the non-specific binding of MW1-AF488 in the CTRL brain. We therefore chose a 2 hr reaction time for the SOP. Stability of the signal within the 2- to 3 hrs incubation interval, indicates that the assay is robust.

Quality Control:

A. Consistency of Signal across the Plate: Plate based assays can provide inconsistent signals across the plate: top to bottom and/or left to right. Since our goal is to detect small changes in subjects with disease progression or changes due to a _mt Htt altering therapy it is important to minimize and account for such potential variability. To this end we evaluated 384 well white plates from different manufacturers and we also assessed this parameter when we evaluated the different plate readers by testing brain lysates from control and HD subjects in triplicates at the four corners of the plates. The signal consistency is evaluated as the Z’ defined as:

Z’=1-[(3xSTDEV-HD)+(3xSTDEV-CTRL)]/(AVE-HD-AVE-CTRL)]

For HTRF assays a Z’ value ≥0.5 is considered satisfactory.

Although minimal, we observed that the left to right variability was lower than top to bottom variability in the plates we determined experimentally to be best for our application. We therefore use only the top 60% of the plate, and the entire plate from left to right. To minimize and control for potential plate related variability we measured Htt in QC samples, control and HD brain lysates, in triplicates on both left and right extremes of each plate when testing clinical samples and calculated the Z’ based on a total of 6 replicates each for control and HD. A Z≥0.5 was determined as a QC’ acceptance criteria for the assay.

B. Signal Normalization: Since the binding to Htt of all the antibodies involved occurs simultaneously in the same cell lysate/well, we can normalize the signal for _mtHtt to the signal for _tHtt resulting in a ratio-_mt/tHtt that identifies each sample. Furthermore, these ratios allow comparisons between different samples to assess the relative levels of _mt/tHtt at the time of measurement. Since there is variability in the concentration of cells/mL of blood based on processing variations, sex, age, and other factors we make every attempt to compare the _mt/tHtt ratios between samples corresponding to similar dsDNA content and/or protein concentrations, which are valuable QC factors. This is facilitated by measuring dsDNA, using PicoGreen, in all the wells, following the reading of the HTRF signals and by measuring the protein concentration of the lysates and comparing the resulting HTRF ratios at the sample dilutions corresponding to similar dsDNA and protein concentrations.

Plate Reader : We evaluated the performance of five different plate readers from three manufacturers and found differences (not shown) of up to 46% in regard to the highest HTRF signal ratios for the HD/control brain lysates and signal consistency across the plate (Z’ values). In addition, cost and technical support played a role in the decision process. We selected the VICTORX5 (Perkin Elmer) as our plate reader, and the assay qualification/validation and the analyses of the BC and QC samples that followed were performed using it. We started to test the clinical samples following the installation of Security Software and the implementation of Installation Qualification (IQ) and Operational Qualification (OQ) by the VICTOR5X manufacturer, as required by the FDA for a potential diagnostic assay.

Assay Qualification/Validation Using Clinical Buffy Coat and QC Samples: Using the optimized protocol and the VICTOR5X we proceeded to confirm the specificity, linearity and sensitivity of the assay. We then proceeded with assay qualification/validation to assess the technical and biological reproducibility of the assay using blood samples such as those intended for clinical use of the assay.

Blood samples from eight individuals including HD subjects and negative controls were divided into three equal sample sets. Each blood sample from each set was processed to obtain the BC and the samples were tested in triplicates along with two sets of QC samples on both the left and right edges of each plate. The three assays were performed over several days, by two operators, using replicate aliquots of the reagents.

Assay Specificity: We confirmed the assay’s specificity, using our GLP SOP protocol, as shown in Fig. 4 for human BC samples and the brain QC samples, and demonstrated a significant difference between HD and control subjects in regard to the _mtHtt and _mt/tHtt HTRF signal ratios.

Assay Linearity: Every BC sample was analyzed at 3 dilutions, corresponding to approximately 1:2, 1:4 and 1:8 final dilutions (Fig. 5). The control and HD brain lysates were used at a constant protein concentration (9 µg/mL) on every plate, a concentration that was determined experimentally (not shown) to be within the linear range for these samples and within the range of values obtained for BC samples in HTRF, dsDNA and protein measurements. The decrease in the signal for the control BC samples correlates most likely to the dilution of the matrix effect (the influence of other cell lysate components). We noticed a linear decrease of signal in HD samples with lower _mt/tHtt. In samples with moderate to low Htt levels, the signal decreases in a linear fashion with the sample’s dilution, however, in samples with high Htt levels, the signal is usually inhibited at the lowest (1:2) dilution of the lysates, it peaks at the 1:4 dilution and then decreases, resulting in a linear dose response within 1:4 to 1:8 dilutions. The existence of a linear dose response curve in the HD BC samples provides the dynamic range necessary for the relative quantitation, in conjunction with QC criteria (dsDNA and protein concentration mentioned above). The inhibitory effect of concentrated lysates on the _mt/tHtt HTRF signal is most likely due to excess of antigen (Htt) and/or matrix effect, as is often seen in such situations in ELISA and other immunoassays.

Relative Assay Sensitivity : Spiking of control BC sample with purified full-length _mtHtt is not useful for absolute quantitation, since Htt will be affected by the aggregation of the mutant and normal protein in our testing conditions. Therefore, this is a semi-quantitative assay, in which we can only evaluate relative signals in HD samples compared to controls. To gain information on the recovery of full-length _mtHtt and relative sensitivity of the measured _mtHtt in our assay we spiked control brain lysates with increasing amounts of HD brain lysates to yield ratios as shown in Fig. 6 (ratios of CTRL: HD from 90:10 to 10:90). It is difficult to calculate the exact % recovery and implicit Limit of Quantitation (LoQ) sensitivity, since the control sample contributes as well to the signal, however a signal recovery of approximately 80-100% is apparent. The dynamic range resulting in a linear response occurs within 20-60% HD spiked into the control sample, again emphasizing the linear nature of the assay within this range.

To obtain an estimated numeric value for our HTRF measurements and to examine the effects of differing CAG lengths, we spiked purified N-terminus 548 aa Htt fragments: 548Q25, 548Q46, 548Q72, respectively into control BC. The final concentrations reached ranged between 300-0 ng/mL and the control BC final dilution was 1:4. We tested in parallel our QC set of control and HD brain lysates, as well as a known HD BC sample. Although these are just Htt fragments, and not the full-length molecule, it indicates the relative effect of the PolyQ length on the resulting HTRF signal ratios, as shown in Fig. 7.

As a result of spiking of the 548Qn fragments into a matrix similar to the clinical matrix, we observed that the dynamic range for _mtHtt (lack of nonspecific signal in 548Q25) is within<100-16.1 ng/mL for 548Q46 and<100-16.1 ng/mL for 548Q72. The dynamic range for _t Htt is 100-3.7 ng/mL for 548Q25, and 300-16.1 ng/mL for 548Q46 and 548Q72, a PolyQ dependent shift to higher concentrations. However, the dynamic range for _mt/tHtt, as shown in Fig. 7is 33.4-16.1 ng/mL (the lower limit is imposed by the lower limits of quantitation for both _mtHtt and _tHtt), which is a narrow range.

Intra-Assay Variability : Fig. 8 is a representative example of the three assays performed for the assay qualification, demonstrating the reproducibility of the triplicate wells measured in the assay. The raw HTRF signals have a maximal Coefficient of Variation (CV) of <7%, which is low for a cell-based assay.

The Inter-Assay Variability : Fig. 9 illustrates the average of MW1/2B7, 2166/2B7 and MW1/2166 ratios from three experiments. The STDEV for the HD samples is higher than that for controls, and its CV < 28%, which is satisfactory for a cell-based assay. The Z’ for _mt/tHtt (MW1-AF488/2166-d2) calculated from the corresponding values of two sets of control and HD brain lysates/plate and including each of the three experiments was Z’=0.6. This Z’, is therefore based on a total of six determinations/QC sample/plate/experiment x 3 experiments = a total of 18 replicates/QC sample, which is very satisfactory. The results from the intra and inter-assay variability analyses indicate high technical reproducibility.

Biological Reproducibility : We examined the biological reproducibility of the assay by measuring Htt in blood samples collected twice within a few weeks from the same individuals:

The results shown in Fig. 10 indicate excellent biologic consistency among the samples of the pairs tested, as confirmed by Pearson’s coefficient=0.88 and Lin’s concordance coefficient =0.87.

Following the assay qualification, we proceeded to write and issue a GLP compliant SOP. Using this SOP, we analyzed BC samples from 8 negative controls and 30 HD subjects (at different disease stages) in parallel to QC brain samples to demonstrate that this assay can differentiate between HD and controls based on the _mt/tHtt levels.

Measuring Clinical Buffy Coat Samples using SOP:

Using the GLP-SOP, we were able to confirm our original findings (Weiss et al [4] ) (Fig. 11) that the median _mt/tHtt signals in BC samples from 30 HD subjects are higher than those measured in 8 control subjects. It is noteworthy that the samples tested were drawn from subjects at a single clinical site, ensuring minimal technical processing variability. The high standard deviation in the HD group reflects heterogeneity within this group which we hope to understand further by examining larger sample sets and clinical and laboratory covariates.}