Development of a Sensitive and Specific in Situ Hybridization Technique for the Cellular Localization of Antisense Oligodeoxynucleotide Drugs in Tissue Sections

Abstract

A sensitive method has been developed for the identification and assessment of phosphorothioate oligonucleotide accumulation in dosed animal tissues using an in situ hybridization approach, which is both sequence specific yet adaptable to every antisense oligonucleotide (ASO), which has been tested to date. Hybridization is accomplished using a digoxigenin-tailed oligonucleotide probe complementary to the ASO target sequence on routinely processed paraffin sections which have been pretreated with a mild target retrieval solution. The DIG-labeled probe is amplified first with an anti-DIG:FITC antibody conjugate followed by an anti:FITC Alexa 488 antibody, then visualized using FITC epifluorescence microscopy. Fluorescent labeling of ASO drug in tissue sections by this method confirms that H&E basophilia previously observed in dosed tissues represents largely intact ASO. However, the fluorescent method enables a wider assessment of tissue distribution in a variety of tissue types due to increased sensitivity and lower signal to noise than can be obtained through an examination of H&E stained tissue sections alone.

Introduction

Antisense oligonucleotides are under investigation for a variety of therapeutic indications including cancer, inflammation, diabetes and cardiovascular diseases. Since ASOs represent a unique class of macromolecular drug, new analytical methods for analyzing the tissue distribution of these drugs need to be developed and evaluated. Hybridization ELISA methods have been successfully developed for the analysis of ASOs and other nucleic acid-based compounds in biological matrices (Yu et al., 2002; Brown-Augsburger et al., 2004). These assays rely on complementary base pairing of the target drug to labeled oligonucleotide probe sequences immobilized on a 96-well plate format and have been used primarily to quantify drug levels in plasma and serum. Tissue levels of ASO have been measured by quantitative autoradiography of labeled ASO (Rappaport et al., 1995; Sawai et al., 1995; Phillips et al., 1997), capillary gel electrophoresis (CGE) (Sewell et al., 2002; Geary et al., 2003; Yu et al., 2004a, 2004b), and more recently, by mass spectrometry (Murphy et al., 2005). In both the CGE and mass spectrometry methods, nucleic acid extraction must be performed from homogenized tissue extracts prior to analysis, which can be labor intensive and subject to nonspecific interferences. The main limitation of tissue quantitation techniques such as CGE, mass spectrometry, radiolabeling, and hybridization ELISA assays is that they do not identify specific cellular distribution of the ASO drug.

Tissue accumulation of phosphorothioate oligonucleotides has been reported in a variety of species with slow elimination from tissues (Geary et al., 2003; Yu et al., 2004b). The greatest amount of ASO accumulation has been demonstrated in kidney and liver, and to a lesser extent, in lymph nodes, bone marrow, and spleen. These experiments have largely been conducted using CGE techniques to measure drug levels associated with various tissues (Yu et al., 2004a, 2004b). Some studies evaluating the cellular localization and distribution of ASOs have been reported as part of the histopathological assessment of H&E tissue sections from animal toxicology studies of first and second generation ASO compounds (Henry et al., 1997, 2001; Monteith et al., 1999). In H&E sections, basophilic granules believed to represent ASO accumulation were identified in tissues, in particular the Kupffer cells of the liver and renal proximal tubule cells. By electron microscopy, these granules were present in intracytoplasmic vacuoles, as well as contained within possible endosomal or lysosomal vesicles or free in the cytoplasm (Monteith et al., 1999).

As a further approach to characterize the distribution of ASO in tissue sections, Butler et al. (1997a, 1997b) describe the use of monoclonal antibody 2E1-B5 for the localization of first-generation phosphorothioate antisense molecule ISIS 2105. This monoclonal antibody was generated against a KLH conjugate of ISIS 2105 (a 20-mer full phosphorothioate oligonucleotide targeting human papillomavirus) and recognizes the 3′-end of this ASO. The antibody also demonstrates limited specificity towards other first and second generation ASOs (personal communication, Richard Geary, ISIS Corporation), but extensive cross-reactivity with all compounds of this class has not been demonstrated. Using the 2E1-B5 antibody, ASO that had been dosed by the inravenous route was localized in a variety of rodent tissues by immunohistochemistry. Labeled cell types included the proximal tubule cells of the kidney and the endothelial and Kupffer cells of the liver.

Label appeared to be granular and concentrated within vesicles. Qualitative assessment of immunolabeled ASO in mouse liver appeared to be dependent on both dose administered and time following administration: at 30 minutes, a large quantity of drug could be observed in the liver sinusoids, but by 24 hours the majority of the reaction product was present in the Kupffer cells (Butler et al., 2000). In monkey kidney, labeling in the proximal tubule cells ranged from punctate to diffuse and was concentration dependent (Monteith et al., 1999). Overall, data obtained using this monoclonal antibody is consistent with the histopathological observations of basophilic granules in H&E-stained tissue sections. However, there remains the limitation that the intact or functional nature of the material labeled by the antibody is unclear.

While the work by Butler et al. has demonstrated that an immunohistochemical technique is applicable to evaluation of ASO cellular localization in tissue sections, there may be limitations to the widespread use of this method. It appears that the specificity/cross-reactivity of antibodies generated against a single ASO sequence must be determined on a case-by-case basis. Additionally, the generation of monoclonal antibodies to antisense targets is labor intensive, expensive, and time-consuming. As an alternative to antibody labeling, the methods described in this paper take advantage of the specificity of oligonucleotide hybridization interactions, and the learning obtained from the development of hybridization ELISA techniques.

In situ hybridization methods in tissue sections are widely established for the localization of endogenous DNA and mRNA targets, as well as for infectious organisms present in cells and tissues. To date, the use of an in situ hybridization method for the analysis of distribution of an antisense therapeutic has not been reported. Here we describe a modified in situ hybridization procedure for the localization of ASO drug in tissue sections that is both sensitive and specific for individual ASO sequences.

Materials and methods

Antisense oligonucleotides used in this study were of the “gap-mer” configuration, consisting of a central length of phosphorothioate modified bases flanked on both the 3′ and 5′ ends by five 2′-O-(2-methoxyethyl) modified (2′-MOE) nucleosides on each of the terminal nucleotide sugars. These modifications increase target mRNA binding affinity and reduce endogenous exonuclease degradation while maintaining a high degree of target specificity through Watson-Crick base pair interactions (Sewell et al., 2002). ASOs used in this report include molecules directed against the murine survivin sequence ISIS 114926 (5′-GAACCAAGACCTTGCACAGG-3′) and ISIS 114953 (5′-CATCGGGTTCCCAGCCTTCC-3′), the mouse glucagon receptor (GCGR) sequence 180475 (5′-GAGCTTTGCCTTCTTGCCAT-3′) and the monkey GCGR sequence 315163 (5′-ACCTGGAAGCTGCTGTACAT-3′). All studies were conducted in accordance with local Animal Care and Use Guidelines, and were approved by local committees.

Mouse GCGR ASO Time Course Study

CD1 mice were dosed with either saline vehicle or 25 mg/kg GCGR ASO 180475 formulated in physiological saline solution once every 3–4 days by the iv route. Animals were sacrificed 24 hours following 1, 2, 4, or 8 doses. Tissues were collected in neutral- buffered formalin and embedded in paraffin using standard methods. For comparison purposes, a few animals were fixed by cardiac perfusion with freshly prepared 4% paraformaldehyde or 4% paraformaldehyde and 0.2% glutaraldehyde in PBS. Individual tissues from these animals were postfixed for 2 hours in the same fixative, then transferred into ethanol for paraffin processing.

Monkey Study

Cynomolgus monkeys received 15 subcutaneous doses of GCGR ASO over a 7-week period. Three 20 mg/kg doses were administered in week one followed by twice-weekly doses of 10 mg/kg in weeks 2–7. Tissue samples of pancreas were collected in 10% neutral-buffered formalin 2 days following the last dose and were processed for paraffin histology using standard methods.

Mouse Survivin ASO Study

CD1 mice were administered 13 doses of 50 mg/kg survivin ASO every 3–4 days over a 4-week period. Animals were sacrificed following a 3-week recovery period and tissues were fixed in 10% neutral-buffered formalin then processed for paraffin histology.

Fluorescence in situ Hybridization (FISH) Procedure

Four-micron paraffin sections of selected tissues were cut and placed on Superfrost Plus slides (Fisher). For detection, unmodified DNA oligonucleotides complementary to the antisense phosphorothioate ASO sequences were synthesized commercially (Qiagen Operon). These oligonucleotide probe sequences were tailed with digoxigenin at the 3′-end using the DIG Oligonucleotide Tailing Kit, 2nd Generation (Roche) following the standard protocol described in the package insert. Briefly, the kit utilizes terminase transferase to incorporate DIG-dUTP bases at the 3′-end of the probe which enables immunologic detection using an anti-digoxigenin antibody. Based on information provided in the Roche insert, tailing using the standard protocol results in a probe with a hypothetical average 3′ tail length of 50 nucleotides and incorporates five copies of digoxigenin. Immediately prior to use, DIG-tailed probes were diluted in 5X SSPE buffer containing 0.4% Tween-20. The starting probe concentration was 400 pmol of tailed oligonucletide per ml of SSPE/Tween-20 buffer. Further dilution of probe to 50–200 pmol/ml was usually possible without affecting the sensitivity of the hybridization reaction.

For hybridization, sections were deparaffinized in xylene, rehydrated to distilled water, and placed into 1X PBS. Sections were placed in a heated (80–90°C) target retrieval solution consisting of 0.1% sodium citrate and 0.1% Triton X-100 for 8 minutes. Following 3 PBS rinses, 50–100 μl hybridization probe was applied to each section and slides were covered with a hybridization well or coverslip. Sections were incubated for 3 hours at 37°C in a humidified chamber.

Following hybridization, sections were rinsed 3 times in PBS and incubated with anti-digoxigenin-fluorescein (Roche) diluted 1:200 in PBS with 1% BSA for 30 minutes. Sections were rinsed and additionally amplified using the Alexa Fluor 488 anti-fluorescein/Oregon Green antibody (from the Alexa Fluor 488 Signal-Amplification Kit for mouse antibodies from Molecular Probes) diluted 1:200 in PBS with 1% BSA for 30 minutes. Sections were rinsed and coverslipped with aqueous mounting media and images were acquired using a Nikon Eclipse TE 2000-U inverted fluorescence microscope equipped with a FITC HyQ filter cube (Ex: 480/40; Em: 535/50). Controls included tissues from animals dosed with saline vehicle as well as animals dosed with a particular ASO but probed with a noncomplementary ASO sequence. On additional control slides, the hybridization probe was eliminated and only detection reagents were applied.

Results

ASO was successfully visualized following FISH analysis in specific cells of the kidney, liver, spleen, pancreas, and other mouse and monkey tissues when animals were exposed to multiple doses of targeted ASO. Tissue fixation in paraformaldehyde by either the perfusion or immersion methods enabled good localization of ASO in paraffin sections with strong signal to noise. In contrast, perfusion of tissue with paraformaldehyde-glutaraldehyde fixation was less optimal due to increased background levels and was not pursued after the initial attempt (data not shown).

In mouse tissues which received 1, 2, 4, or 8 iv doses of a GCGR ASO, and which were processed and imaged under identical conditions, the relative differences in ASO levels between doses is clearly observed (Figure 1). In the kidney, ASO is discernable by epifluorescence microscopy in the epithelial cytoplasm of the proximal convoluted tubules 24 hours following a single 25 mg/kg dose of ASO (Figure 1A). Although labeling is faint, a granular accumulation of drug is evident. By 2 doses, this granular accumulation of ASO in the proximal convoluted tubules is pronounced (Figure 1B). By 4 doses, label is most abundant in the cytoplasm of the proximal convoluted tubule epithelial cells (Figure 1C), but also evident in distal tubule epithelial cells, collecting duct epithelial cells, and scantily within glomeruli (data not shown). At eight doses, fluorescence levels saturate the cytoplasm of these cells when viewed and imaged under identical conditions, while cell nuclei remain unlabeled (Figure 1D).

Similarly, in the liver, evidence of ASO drug can be visualized following a single dose of compound (Figure 1E). By 2 doses, strong fluorescence is observed in the Kupffer cells and sinusoidal lining cells (Figure 1F). At 8 doses, greater accumulation of ASO in the Kupffer cells is noted (Figure 1G), resulting in an enlarged appearance of these cells over the time course. When the FISH method is performed on undosed tissues, background levels are generally low (Figure 1H). Background levels are comparable between tissue sections from undosed animals and those where the hybridization probe is eliminated.

Slight autofluorescence is observed in most tissue types and is most pronounced in red blood cells, although microscopically, this autofluorescence is clearly distinguished from the specific granular labeling observed in dosed tissues. While there is some suggestion of hepatocyte labeling in multidosed animals such as Figure 1G, the labeling intensity is not sufficiently above background to indicate its unequivocal presence. If ASO is present in the hepatocytes, its appearance is diffuse rather than granular, and much less intense than that observed in the Kupffer cells.

Individual hybridization probes are specific for their complementary ASO sequences and ASO drug in tissue sections did not hybridize to alternate probe sequences (Figures 2A–2D). Specific labeling in the mouse spleen was prominent in the endothelial cells lining the red pulp sinusoids with a specific survivin probe (Figure 2B). However, when the same tissue was probed with an alternate survivin sequence, some non-specific red cell autofluoresence is observed (similar to undosed controls) in the red pulp, but the intense labeling of sinusoids is absent (Figure 2A). Similarly, in the pancreas when ASO drug and probe sequence were matched (Figure 2C), bright fluorescence is observed in the interstitial mononuclear cells, however, when an alternate probe sequence is used, labeling is absent (Figure 2D).

Labeling of interstitial cells presumed to be macrophages or phagocytic dendritic cells was observed in every tissue examined using the FISH method. Such labeling is observed in the heart (Figure 2E) and the lung (Figure 2F). In the lung, coarse granular labeling was observed in the alveolar macrophages, with finer labeling present in the alveolar septae. Tissues depicted in Figures 2A–2F were assayed from animals which were administered the last dose of ASO drug 3 weeks prior to sacrifice. Although labeling was observed in some endothelial (spleen sinusoids) or epithelial cells (kidney proximal convoluted tubules) at this time interval, in other tissues such as the pancreas, heart, or lung, the dominant fluorescent labeling was observed in the macrophages.

In addition to rodent studies, the FISH method was applied successfully to monkey tissue. Figure 2G shows labeling of the GCGR ASO in macrophages of the monkey interstitial pancreas, comparable in appearance to those observed in mouse (Figure 2C). In lymph nodes adjacent to the monkey pancreas, medullary and cortical sinuses contained enlarged macrophages filled with granular intracytoplasmic accumulation of ASO (Figure 2H).

Discussion

Antisense oligonucleotides are a newer class of drug molecule designed to exert a therapeutic effect on their target as a result of nucleic acid base pairing. Nucleotide modifications in second-generation phosphorothioate ASOs provide enhanced stability of these compounds in vivo while assuring strong affinity to the target nucleotide sequence. It is recognized that one reproducible “class effect” resulting from the administration of antisense compounds in all species examined to date includes the granular accumulation of material in the cytoplasm of various tissues (Farman and Kornbrust, 2003).

While the identification of basophilic granules in H&E sections may be indicative of ASO accumulation, this observation may be subtle or equivocal in situations where ASO dose is low or tissue accumulation is slight. Generation of specific anti-ASO antibodies is time consuming and expensive and sensitivity to different ASO drugs may vary. By taking an in situ hybridization approach, we hoped to be able to develop a screening method using readily available reagents that could be applied to analyze the distribution of multiple ASO drugs in a variety of tissue types and species.

Since ASO drug does not represent an endogenous DNA target, it was unknown whether these introduced modified oligonucleotide sequences would be able to be recognized from routinely processed paraffin tissue sections. We had our first confirmation that these sequences could be identified in paraffin sections when we tried a TUNEL apoptosis method (which typically detects 3′ DNA strand breaks in apoptotic cells) to look for ASO on the basis of 3′-end recognition (data not shown). While we were convinced that the specific fluorescence we observed by the TUNEL method in cell types such as the Kupffer cells of liver represented the accumulation of ASO rather than the identification of apoptotic cells, we wished to clearly discriminate ASO distribution from apoptosis and prepared sequence-specific hybridization probes. By modifying buffer composition and stringency washes from more traditional in situ hybridization protocols and including a mild target retrieval step, we were able to generate a basic protocol that worked successfully with a variety of different ASOs and tissue types.

The distribution of ASO drug evaluated by FISH in this study for a variety of tissue types using several different ASOs is consistent with previous descriptions of basophilic granule accumulation observed in H&E-stained sections from animal studies of first and second generation ASOs (Henry et al., 1997, 2001; Monteith et al., 1999). By FISH, intracytoplasmic granule accumulation is strongly evident in the epithelial cells of the kidney proximal convoluted tubules, and in macrophages and phagocytic dendritic cells, including the Kupffer cells of the liver, the splenic sinusoids, lymph node macrophages, and histiocytes throughout organ stromal and connective tissues.

The relative abundance of ASO identified in these various tissue types correlates with both ASO dose and time following administration, which is also consistent with previous studies. Furthermore, ASO distribution is also consistent with immunohistochemical localization of ASO in tissues dosed with ASO recognized by the 2E1-B5 antibody (Butler et al. 1997a, 1997b, 2000). Given the ability of the ASO in tissue sections to specifically bind the appropriate probe oligonucleotide, it is probable that ASO identified in tissue sections using the FISH method represents largely intact ASO rather than smaller metabolites. In hybridization ELISA studies using oligonucleotide probes similar to those employed in the FISH method, hybridization to truncated ASO metabolites is relatively inefficient (Brown-Augsburger, 2004). This result is also consistent with data from tissue distribution/metabolism studies with second generation ASOs, demonstrating that the majority of tissue associated drug is intact, full length ASO (Yu et al., 2004b).

While some assessment of ASO distribution can be made through the examination of basophilic granules in H&E-stained tissue sections, the FISH method provides several advantages. The FISH method is very sensitive with a high signal-to-noise ratio. This enables clear observation of ASO distribution in tissues where ASO is less abundant such as the collecting ducts of the kidney or the septal cells of the lung, or in tissues assessed following a longer postdose interval. The FISH method has been applied successfully to more than 12 ASO compounds tested to date from mouse, monkey, and human tumor xenograft studies (data not shown). Each FISH probe recognizes only a single oligonucleotide sequence and no cross-reactivity has been observed between individual ASOs and alternate probe sequences. In addition to the sensitivity and specificity of the FISH method, the technique can be performed with readily available commercial reagents without the time needed to generate more expensive ASO-specific antibodies.

Previous reports noted ASO loss in paraffin processed tissues (Plenat et al., 1995), and perfusion fixation with paraformaldehyde/glutaraldehyde has been utilized in other distribution studies to prevent possible diffusion of ASO (Butler et al., 1997a, 1997b). We found routine immersion fixation with 10% neutral-buffered formalin was adequate for good retention of ASO integrity in the tissues examined. Perfusion fixation with paraformaldehyde or paraformaldehyde/glutaraldehyde did not alter the distribution or increase the signal intensity of the label, and glutaraldehyde increased background autofluorescence with this method.

In developing the FISH assay, it was found that it was not necessary to include steps frequently needed for successful localization of endogenous cDNA or mRNA targets, such as complex hybridization buffers or stringency washes. A critical step for the enhancement of sensitivity in our FISH assay was the inclusion of a mild target retrieval step involving a short incubation in a heated sodium citrate solution. Heat-induced retrieval has been applied successfully to modify nucleic acid-formalin interactions and improve nucleic acid extraction or hybridization from paraffin embedded material (Shi et al., 2001), though more rigorous conditions have often been required for these applications.

It is also possible that short, “free” ASO moieties are more readily available to interact with fixative than internalized sequences of DNA might be, and hence the conditions required to unmask and hybridize a probe to the target sequence are less stringent. While visualization of fluorescent signal was possible with application of a single anti-Dig:FITC antibody, the application of a second fluorescent antibody (anti-FITC:Alexa 488) increased the signal intensity and eliminated photobleaching. Section fluorescence remained intense with no decrease in sensitivity for a period of several weeks without the use of anti-fade reagents.

ASO accumulation in cells is most frequently granular in appearance. In the variety of tissues examined, ASO was often readily observed in phagocytic cells such as Kupffer cells and macrophages and has probably accumulated in vesicles or lysosomes within these scavenger cells where it may not be available for binding to intended target cell types. This is an important consideration in the evaluation of drug levels using methods such as CGE or mass spectrometry where tissues are homogenized prior to analysis. While ASO drug may be present at low levels in various target tissues, their concentrations relative to the nontarget phagocytic cell populations are low.

Similarly, the measurement of drug in tissues after a long postdose interval does not necessarily indicate the presence of drug available for binding to target cells. If drug is indeed present in other cell types in a more diffuse and less concentrated form than is detectable by the FISH assay, its concentrations may be overestimated. A future goal of drug distribution studies for compounds such as ASOs will be to quantify drug concentrations in individual cell types rather than in entire tissues. It may eventually be possible to use a method such as FISH to identify drugs in specific cells, then follow with an analytic method such as CGE or mass spectrometry to obtain quantitative data from these cells. Identification of additional methods involving laser capture microdissection of ASO positive cells identified by the FISH method might also be developed.

In conclusion, we have developed a sensitive method for the cellular assessment of the accumulation and distribution of ASO drugs in tissue sections. This method relies on the in situ hybridization of labeled oligonucleotide probe sequences to externally administered drug, unlike traditional in situ hybridization assays which target endogenous DNA or mRNA sequences. Since the assay relies on probe hybridization to the drug sequence, it uniquely recognizes individual ASOs. The method is broadly applicable to a variety of tissue types prepared using the routine paraffin histology methods employed in toxicology studies and has demonstrated good specificity in the rodent, monkey, and human tumor xenograft models which have been tested to date. The technique takes advantage of easily synthesized oligonucleotides and commercially available reagents. Patterns of ASO localization observed with this method are consistent with intracellular granular basophilia observed in routine H&E-stained tissue sections. However, the contrast and sensitivity afforded by epifluorescence microscopy enables a better assessment of the wide distribution of ASOs throughout each tissue, particularly in cell types where less drug is present, such as the renal collecting ducts or septal cells of the lung. While antibodies may be developed that are useful for analyzing the distribution of pan-ASO drugs in animal studies and clinical trials, the FISH method provides a useful tool for the analysis of tissue distribution of individual ASO drug sequences in studies where high specificity and sensitivity is warranted.