Abstract
Plucked human hair follicles have been proposed as a potential surrogate for tumour tissue for measuring the effect of drugs on pharmacodynamic biomarkers in drug intervention studies. We describe a new technique of embedding plucked hair follicles in the acrylic resin, methyl methacrylate, and the immunohistochemical demonstration of six potential biomarkers (Ki67, EGFR, phospho-p27, phospho-histone H3, phospho-MAPK and phospho-Rb) in de-plasticised sections. The advantages of this technique over those that have been used in support of clinical drug trials, such as skin and tumour biopsies, whole blood and whole hair samples is discussed.
Keywords
Introduction
Early phase clinical trails have traditionally been used to investigate the safety, and dosage levels, of new drugs in volunteers. During these studies there is a regulatory and scientific desire to establish that the drug in question is acting in the expected way in inducing its therapeutic effect. Hence modern drug development frequently involves incorporation of a pharmacodynamic (PD) marker as evidence that the mode of action of the drug on human tissues is via the proposed pathway. Whilst PD effects seen in diseased tissue are likely to relate most closely to clinically relevant outcomes (Baselga, 2003) there is a need, within oncology drug development, to use normal tissue as a surrogate for tumour tissues for two reasons; firstly, healthy volunteers do not have tumours and, secondly, tumour heterogeneity and accessibility in patient volunteers may prove problematic.
For a tissue to be a suitable cancer surrogate it has to have similar characteristics in expressing a relatively high rate of proliferation, express the target pathway of the drug, be easily accessible and the sampling be well tolerated. Normal tissue-based oncology biomarker studies have been performed with success using peripheral blood mononuclear cells (Cohen et al., 2003; Peralba et al., 2003), exfoliated buccal squamous cells (van de Vaart et al., 2000; Adjei et al., 2003), punch biopsies of the skin (Albanell et al., 2002; Malik et al., 2003; Vanhoefer et al., 2004) and punch biopsies of buccal mucosa (Camidge et al., 2005a).
The sheath around the base of a plucked scalp hair contains proliferating cells (Moll, 1996; Gho et al., 2004), is easily accessible and the plucking is relatively well tolerated. We have previously described a method for performing immunohistochemistry (IHC) on whole plucked human hairs (Camidge et al., 2005b) but this methodology has several drawbacks.
In this method the IHC is performed on the intact hair, which is subsequently sectioned to permit morphometric analysis. With this approach only one biomarker can be demonstrated per hair and appropriate controls cannot be performed on the same hair. We determined that multiple hairs needed to be stained for each biomarker to reduce statistical variability. If a panel of biomarkers were to be employed the number of plucked hairs required would soon become impractical and the exercise prohibitively expensive. The IHC staining of intact hairs also had to be done within a limited time after sampling and a significant number were lost or damaged (18%), failed to stain (15%) or showed high and unacceptable levels of background staining (5%).
Attempts have been successfully described for sectioning hair shafts (Kempson et al., 2002) but the preparations were not subsequently useful for IHC. Previous attempts in our laboratory to reliably section formalin fixed, paraffin embedded (FFPE), hairs in a longitudinal direction, as is necessary for locating the proliferative areas in the follicle (Ito et al., 2002), have proved extremely difficult mainly due to the difference in hardness between the hair shaft and the wax embedding medium. Resin embedding would appear to provide a solution and the epoxy resin, Araldite, has been used successfully to section hair (Pepper and Lantis, 1977; Rosen and Kerley, 1971) although the inability to subsequently remove the resin from the hair without severe alkaline pretreatment renders the technique unusable for most IHC methodologies (Sato et al., 1995).
Many commercially available resins are based on glycol methacrylate (GMA) but reports of attempted IHC staining of tissues embedded in this material are mixed (Beckstead el al., 1986; Archimbaud et al., 1987; Islam et al., 1988; Burgio et al., 1991; Vincic et al., 1990). Chemically polymerised methyl methacrylate (MMA), on the other hand, has been used successfully for tinctorial stains, IHC (Blythe et al., 1997; Hand et al., 1996) and non-isotopic in situ hybridisation (Church et al., 1997). MMA has the advantage over other acrylic and epoxy resins in that sections may be deplasticised by immersion in xylene, similar to the procedure used with wax sections, resulting in the removal of significant stearic hindrance by the embedding medium.
We report here a method for the fixation, processing, sectioning and immunohistochemical staining of a number of oncology-relevant antibodies on plucked human hair, using a modification of a method previously described by Blythe et al. (1997), and discuss the possible application of the methodology within clinical trials.
Materials and Methods
Sampling and Fixation
Hairs were plucked from the parietal region of the scalp using blunt-nosed forceps. The hairs were gripped close to the scalp and plucked with a single, sharp action, checked for the presence of sheath, trimmed to approximately 1 cm in length containing the bulb region and dropped into a 1.5 ml microfuge tube (Fisher Scientific, Loughborough, UK) containing 1 ml 10% neutral-buffered formalin. The optimal period of fixation was determined by a time course experiment fixing the hairs for 1, 2, 4, 6, 12 and 24 hours, processing and staining with our most fixation sensitive antibody (anti-phospho-p27, Upstate, Charlottesville, VA) and was found to be 4 hours at room temperature (20°C).
Processing
To enable the automation of the processing using a Leica EM TP electron microscopy tissue processor (Leica Microsystems GmbH, Wetzlar, Germany) the fixed hairs were pre-embedded in 3% low melting point agarose (Sigma-Aldrich, Gillingham, Dorset, UK) in phosphate buffered saline (PBS), pH7.4, melted and cooled to 37°C. The agarose blocks were formed in silicone rubber flat embedding moulds (Agar Scientific Ltd, Stansted, UK) and placed on ice to cool and solidify. Between 1 and 5 hairs were embedded in one agarose block.
The agarose blocks were trimmed to approximately 6 × 4 × 3 mm and placed in the processing baskets of the Leica EM TP and processed according to the following schedule: PBS, 30 min; PBS, 30 min; 50% Ethanol, 1 hr; 70% Ethanol, 1 hr; 90% Ethanol, 1 hr; 100% Ethanol, 1 hr; 100% Ethanol, 1 hr; 100% Ethanol, 1 hr; Catalysed MMA, 1 hr; Catalysed MMA, 16 hrs.
The catalysed MMA infiltration solution consisted of 75 ml methyl methacrylate, 25 ml dibutyl phthalate and 5 g dried benzoyl peroxide (all Sigma-Aldrich). Benzoyl peroxide is potentially explosive in the dry state and is supplied dampened; aliquots are dried overnight in an oven (Memmert GmbH, Schwabach, Germany) at 37°C.
Embedding
Processed blocks were embedded in the polymerisation solution which consisted of 10 ml catalysed MMA plus 125 μl N,N-dimethylaniline (Sigma-Aldrich, Gillingham, Dorset, UK).
Blocks were embedded in 8 mm diameter, flat-ended, embedding capsules (Agar Scientific Ltd), which were filled with the polymerisation solution and then closed. At this point it was possible to orientate the hairs into their desired position for sectioning (longitudinally or transversely). The capsules in a rack were placed on a tray within a desiccator (Fisher Scientific, Loughborough, UK). To dissipate the heat generated be the exothermic polymerisation of the resin, water was added to the tray at a level half way up the capsules. The desiccator was closed and the air flushed out with oxygen-free nitrogen. The polymerisation of the resin takes approximately 2½ hours after which the resin blocks were removed from their capsules and stored at room temperature before sectioning.
Sectioning
The blocks were sectioned on a Leica Microsystems RM2165 (Leica Microsystems GmbH, Wetzlar, Germany) motorised microtome using a glass knife. 4 μm thick sections were picked up using fine-nosed forceps and floated out onto a drop of water on a SuperFrost Plus®(Thermo Electron, Runcorn UK) electrostatically charged glass slide. The slide was placed on a hotplate set at 100°C and the section observed until it flattened out. The slide was then removed from the heat and the excess water carefully tipped away. The sections were placed in an incubator and allowed to dry at 37°C overnight.
Immunohistochemistry
Sections were deplasticised by immersion in xylene for a minimum of 6 hours at room temperature, preferably overnight. The sections were taken through 2 changes of xylene and 3 of industrial methylated spirits (IMS) before re-hydration in distilled water. Heat-mediated antigen retrieval (HMAR), if required, was performed in a Milestone RHS-2 microwave (Milestone, Sorisole, Italy) at 110°C for 2 minutes in 10 mM citrate buffer, pH 6.0, and immunohistochemical staining was performed, at room temperature, on a Lab Vision Autostainer 720 (Lab Vision, Newmarket, UK). The antibodies used are detailed in Table 1. The reagent and wash buffer was 0.05 M Tris buffered saline plus 0.05% Tween 20 (TBST).
Endogenous peroxidase activity was quenched by incubation in 3% hydrogen peroxide in TBST for 10 minutes. Slides were then washed and incubated for 20 minutes with 5% normal goat serum in TBST. Excess blocking serum was blown off and the slides were incubated with the primary antibodies for 60 minutes.
Following washing the slides were incubated for 30 minutes in either rabbit- or mouse-specific EnVision® + System–HRP (Dako UK Ltd, Ely, UK), depending on the source of the primary antibody, and visualised by incubation in diaminobenzidine (DAB), from the EnVision® + kits, for 10 minutes. The slides were counterstained for 1 minute using Carazzi’s haematoxylin (Clin-Tech, Guildford, UK). Staining was negatively controlled by substituting mouse isotype immunoglobulin (Ig) or rabbit Ig fraction, diluted to the same concentration, for the primary antibodies. Sections of SW620 human colorectal xenograft tumours, grown in Nude mice, which had previously been successfully stained with these antibodies, were used as positive controls. Following staining between 4 and 10 hair sections for each antibody were assessed microscopically by eye for the intensity and the approximate percentage of cells stained was estimated.
Result
Using this method up to 50 section of 4 μm thickness could be obtained and successful immunohistochemical staining was achieved with all six antibodies (Figure 1).
Ki67 (Fig 1a)
The proliferation marker showed strong staining of approximately 10% of the nuclei around the periphery of the hair sheath, an area known as the outer root sheath (ORS), which is considered to derive from epithelial stem cells and be actively proliferative (Krause et al., 2006).
EGFR (Fig 1b)
The cell membranes of the majority of the cells within the sheath stained positive for epidermal growth factor receptor, the strongest staining being apparent in the ORS.
Phospho-p27 (Fig 1c)
Staining for the phosphorylated form of p27 was mainly confined to the inner root sheath (IRS). Over 50% of the nuclei within the inner route sheath stained strongly positive for this cell cycle related protein. Occasional nuclei within the ORS also showed weak positivity.
Phospho-Histone H3 (Fig 1d)
The Histone, H3, is phos-phorylated during mitosis. Mitotic figures were rarely seen in the hair sheath outside of the bulb area and this was reflected in the observed staining pattern. From 10 hair sections only three instances of positive phospho-histone H3 staining were observed, all of which were in the ORS.
Phospho-p44/42MAPK (Fig 1e)
Variable intensities of nuclear staining were seen within the IRS and expression of phospho-p44/42 MAPK was seen in approximately 50% of the nuclei. An area of cells within the IRS also exhibited positive staining for the phospho-p44/42 MAPK protein in the cytoplasm and cell membrane. The cytoplasmic and membrane staining coincided in cells with the most intense nuclear staining and appeared as a zone of positive immunoreactivity in all samples.
Phospho (S249/T252)-Rb (Fig 1f)
Variable levels of signal were seen in over 75% of the nuclei within the IRS. Occasional nuclei within the ORS showed pale staining but all nuclei throughout the hair expressed this protein to some extent.
Discussion
The importance of being able to monitor the effects of therapeutic drugs on cancer cannot be over stated (Severino et al., 2006; Colburn, 2003; Lee et al., 2007). In the field of oncology clinical trials, this has most commonly involved the sampling of biopsy material from either the cancer itself, or by taking blood, and analysing for changes in the function of the putative target (Nishio et al., 2005; Cummings et al., 2004; Elyin et al., 2004).
Whilst a shift in the expression of biomarkers in the tumour itself remains the ideal goal, and would be more indicative of drug activity in the disease target (Baselga, 2003), obtaining tumour material is not always possible, practicable or tolerable. Healthy volunteers in phase 1 clinical trials will, by definition, have no tumour to biopsy making the use of a surrogate unavoidable. In patient volunteers it may be difficult to obtain permission for multiple biopsies pre- and post-dose. Additionally, tumour heterogeneity may pose problems when trying to compare paired biopsies.
Also, these techniques are generally invasive to varying degrees, involve a certain degree of risk to the subjects, and are variably tolerable by those undergoing the procedure (Duran et al., 2006; Albanell et al., 2002; Soo et al., 2006; Ma and Chan, 2006; Kindler et al., 2004; Stathopoulos et al., 2005). Although peripheral blood cells are readily accessible, their low intrinsic proliferation rates make them a poor marker for the effects of anti-proliferative, anti-cancer, agents unless ex vivo induction of cell proliferation, using mitogenic agents such as phytohaemagglutinin, tetanus toxoid, or interleukin-2 (Rosato et al., 2001; Ingham et al., 1991) are employed.
Plucked human hair offers a minimally-invasive, easily sampled and well-tolerated source of cells for the investigation of the pharmacodynamic (PD) effects of specific anti-proliferative drugs on human tissue. We have shown that a number of important biomarkers of cell cycling, in its broadest sense, may be demonstrated immunohistochemically on plucked and sectioned hair follicles using this technique and the current methodology has a number of advantages over methods previously described (Camidge et al., 2005b).
Immunohistochemistry, as with any assay, needs to have the appropriate controls, both positive and negative (Leong and Leong, 2006; Chan et al., 2000; Maxwell and McCluggage, 2000). This was difficult to achieve with the whole hair methodology as each hair follicle could only be stained once. The current method allows multiple serial sections from the same hair to be used for both the novel immunohistochemical marker and the appropriate controls.
The ability to derive up to 50 sections from each plucked hair follicle also enables multiple assays to be performed on a single hair, reducing the overall number of hairs required and leading to a reduction in both the experimental costs and patient discomfort. An additional advantage is the ability, once the samples have been processed and embedded, to store the blocks for additional investigations at a later date. This is not possible with the whole hair method (Camidge et al., 2005b) where the whole sample has to be used for each stain employed. The new methodology subsequently permits considerably greater flexibility in workload scheduling by avoiding the necessity for multiple sampling from each volunteer at each time point in the study.
The great advantage of methyl methacrylate in immunohistochemistry over other commonly used resins is the ability to remove the resin by immersion in xylene (Britten et al., 1993; Howat et al., 2005), the consequence being that IHC methods developed on FFPE tissues may be transferred to MMA embedded tissues with little or no modification. This is advantageous to the investigator, saving time in method development and strengthening confidence in the assay. Furthermore having sections equivalent in most respects to de-waxed FFPE sections enables the application of haematoxylin as a nuclear counterstain, which helps place the IHC into context and as a means of normalising any subsequent image analysis.
In conclusion, the methodology described has the potential to make hair sampling a valuable and versatile tool for monitoring the effects of anti-cancer drugs on cell proliferation, and on the upstream signalling molecules that control cell proliferation. Obtaining hair samples is minimally invasive and sectioning of the hair allows a large number of samples to be obtained for subsequent immunohistochemistry. It should be a valuable additional tool to the range of techniques currently being applied in early phase clinical trials of potential drugs.