FFPE Stored Samples

Tissues coming from surgery have been long collected and stored under formalin-fixed paraffin-embedded (FFPE) form. Nowadays, the existence of this huge and easily accessible library of clinically relevant samples, leads scientist to ask a big question: is it possible to turn out this practically unused material in a convenient way to fit the needs of modern analytical techniques like MALDI-TOF-IMS?

The chance to get access to such a large samples collection would open new frontiers to the knowledge of the molecular features of a lot of diseases. This is especially relevant for cancer disease, as it would allow performing longitudinal studies on existing samples.

MALDI-TOF Imaging MS is a powerful technique that allows the localization of proteins and small molecules on tissue surface [1], but it implies samples to be freshly collected to get optimal results. Moreover, new laser generation has dramatically changed the analytical performances of MALDI-TOF-IMS, resulting in higher protein resolution and a three fold increase in sensitivity [2].

We have pursued a technological improvement of FFPE tissue preparation specifically for MALDI-IMS. In particularly we aimed to develop protein unlocking procedures which might enable a suitable recovery of polypeptides for MS analysis of FFPE specimens.

Experimental Strategy

Albumin embedded in tissue surrogates and fixed with formalin was digested and analyzed by MALDI-TOF-MS to evaluate PMF (Peptide Mass Fingerprint) quality from cross-linked proteins. A shotgun proteomics approach was then applied on FFPE human breast cancer tissue samples to choose the best unlocking protocol among different unlocking procedures, combined or not with the Heat Induced Antigen Retrieval (HIAR) protocol. Collected evidences have been shifted to MALDI-TOF-IMS to implement tissue sample preparation. FFPE unlocking treatment at high temperature [3] was followed by on target trypsin digestion and classical spray matrix deposition (sinapinic acid 10 mg/mL in MeCN/H₂O 80:20 0.1% TFA).

MS spectra collected from FFPE tissue MALDI-TOF-IMS session have been clustered following their histological area and subjected to supervised Principal Component Analysis (PCA) in order to generate a model.

Its external validation was executed on spectra belonging to similar histological regions.

Tissue Localisation of Signals

The developed strategy to unlock proteins and peptides for MALDI-IMS analysis has been taken advantage by the combination of heat induced antigen retrieval techniques (HIAR) and on tissue trypsin digestion. HIAR EDTA treatment is the most effective as it allows tissue morphology to be retained; nevertheless a further digestion step is required to obtain satisfactory MALDI-IMS performances on FFPE stored samples.

Figure 1 shows the optical image of the investigated breast cancer section (coloured according to the classical H&E method) compared to the ion density map of signals at m/z=6024 and m/z=3447. The former is most intense in the tumoural tissue (the dark blue area in H&E staining) while, the latter has a specific localisation in the oedematous area.

Supervised data analysis confirm the retention of tissue morphology: to perform a principal component analysis (PCA), MS spectra collected from FFPE tissue MALDI-IMS session have been clustered into seven groups, depending on the histological region they belong to (adipose tissue 1, 2 and 3, tumoural region 1 and 2, oedema region and connective tissue) and exported to Clinprotools software from Bruker-Daltonics for the statistical analysis. PCA analysis shows a strong clustering of spectra acquired in the selected areas (fig. 2A). An external validation of the model was executed taking spectra belonging to the regions that were left out from the generation of model and let the model classify them. In panel B of figure 2 the results of the classification model are shown. Red, blue and green coloured plume represent adipose tissue, connective tissue and tumoural region respectively. Black plume in each panel represent the predictions on external datasets of tumoural region 1, connective tissue and adipose tissue 1 and 2 respectively.

Conclusions

Our results outline the possibility to obtain peptide mass spectra qualities comparable between the Fresh frozen and FFPE tissue preparation enabling direct peptide sequencing experiments. The potential application of clinical Proteomics investigation to FFPE tissue samples have been particularly interesting in the latest years, due to the possibility opened in the biomarker discovery by the direct employment of collections of biopsies. In this regard, the remarkable works of Hood et al. [4], Lemaire et al. [5] and Crockett et al. [6] have clearly demonstrated the possibility to apply a shotgun proteomics approach in FFPE protein profiling from clinical samples and to identify proteins extracted from FFPE tissues by nanoLC-nanoESI-qQTOF-MS/MS.