FHIT Loss of Function in Human Primary Breast Cancer Correlates with Advanced Stage of the Disease¹

Breast cancer is still the most frequent malignancy among women of western countries (1). The genetic abnormalities that occur in breast cancer involve multiple genetic alterations, which are responsible for the progression from normal breast epithelia to invasive cancer cells.

Chromosomal region 3p14.2 is a frequent target for genetic and cytogenetic alterations in a wide range of solid tumors (for review see Ref. 2), leading to the search for a tumor suppressor gene in this region. The tumor suppressor gene FHIT, located at chromosome 3p14.2, is more than 1 Mb in size and encodes a 1.1-kb cDNA with 10 small exons; exon 5 is the first protein coding exon and is flanked in intron 4 and intron 5 by the most common fragile site in the human genome, FRA3B (3). The FHIT gene belongs to the histidine triad (4) superfamily and encodes a cytoplasmic M_r 16,800 protein with diadenosine triphosphate (Ap3A) hydrolase activity. The conserved histidines are required for full enzymatic activity (4). The gene is inactivated by deletions in cancer-derived cell lines and primary tumors of the lung (5), head and neck (6), breast (7), stomach and colon (3), and other organs.

Analysis of FHIT gene in a series of human primary breast cancer revealed allelic loss in 25% (7) and abnormal transcripts in approximately 30% of the cases (7, 8). FHIT homozygous deletions in samples with 3p14.2 aberrations were also found in the benign breast lesions of two women with familial predisposition to breast cancer (9). Another study of normal breast epithelium, breast preneoplastic lesions, and invasive tumors reported the loss of heterozygosity of the FHIT locus in two patients with intraductal hyperplasia (10). Studies correlating FHIT gene alterations with its protein level are required to determine the grade of inactivation of the gene in breast cancer disease. To address this question, we analyzed 29 cases of primary breast tumors for normal and abnormal FHIT transcripts and for the level of expression of Fhit protein in the normal breast epithelia and breast epithelia tumor of the same patient. Down-modulation or absence of Fhit protein was also evaluated in a series of 156 consecutive patients with primary breast carcinomas. In both groups, Fhit protein levels were reduced or absent in almost 70% of the breast cancer samples, whereas aberrant FHIT transcripts were detected in only 31% of the samples. Moreover, down-regulation of Fhit protein expression is associated with highly proliferative and large tumors.

The study included two series of primary breast carcinoma obtained from patients surgically treated at Istituto Nazionale Tumori (Milan, Italy). The first (Series N. 1) included 23 DCs³ and 6 LCs selected for the availability of fresh tumor specimens. The second (series N. 2) included 156 consecutive cases of ductal, lobular, or mixed infiltrating breast cancer, retrospectively examined. Primary tumor diameter and axillary nodal status were obtained from histopathological reports. Histological grading was performed according to Pereira et al. (11), based on tubule formation, nuclear morphology, and number of mitoses. Hormone receptors expression was determined by the DCC adsorption technique (12).

Tumor specimens were frozen immediately after surgical resection. Total RNA was extracted from the frozen tumors using RNA-zol (Tel-Test, Friendswood, TX) according to the manufacturer’s instructions.

cDNA was synthesized from 1 μg of total RNA. RT was performed in a 20-μl volume of 1× first-strand buffer (Life Technologies, Inc.), 10 mm DTT, 500 μm each dNTP, 0.3 mg/ml random primer (Life Technologies, Inc.), and 300 units of SuperScript II reverse transcriptase (Life Technologies, Inc.). Samples were incubated at 70°C for 5 min and then at 42°C for 1 h before 1 μl of RNase A was added for 30 min.

One μl of cDNA was used for the first PCR amplification with primers EX2F and 06 (CTTTGAAGCTCAGGAAAG and CTGTGTCACTGAAAGTAGACC, respectively) from FHIT exons 2–9 in a volume of 25 μl containing 20 pmol of each primer, 200 μm each dNTP, 1× reaction buffer (Boehringer-Mannheim), and 1.25 units Taq polymerase (Boehringer-Mannheim). PCR cycling conditions were: (a) initial denaturation at 95°C for 3 min followed by 28 cycles of 30 s at 94°C; (b) 30 s at 57°C and 2 min at 72°C; and (c) final extension at 72°C for 5 min, using a Perkin-Elmer-Cetus PCR Thermocycler. Amplified products were diluted 20-fold and 1 μl was used for a second PCR amplification with primers UR5 and 752R (CTGTAAAGGTCCGTAGTG and CTGCCATTTCCTCCTCTGAT, respectively) from FHIT exons 3–9. The nested PCR amplification was carried out for 22 cycles under the same conditions as the first PCR, and products were resolved on a 1.3% agarose gel.

DNA bands corresponding to the normal and abnormal size FHIT transcripts were excised from the gel, purified using the Quick Gel extraction kit (QIAGEN), and sequenced on the Applied Biosystems model 373A and 377 DNA sequencers.

Immunoperoxidase assay was carried out on paraffin-embedded sections of primary breast carcinomas. Briefly, 1–2 μm consecutive sections of formalin-fixed, paraffin-embedded tissue were cut and mounted in poly-l-lysine (Sigma, St. Louis, MO) coated-slides, deparaffined in xylene, and rehydrated in grade alcohol. Endogenous peroxidase activity was blocked by treatment with 0.3% hydrogen peroxide in methanol for 30 min. Fhit antigen enhancement was performed by staining the sections at 120°C in 10 mmol/liter sodium citrate buffer (pH 6.0) for 2 min. After treatment with normal goat at 1:50 for 30 min at room temperature, slides were incubated overnight with different antibodies. Antibodies reactivity was detected using biotinylated goat antirabbit IgG (Dako, Glostrup, Denmark) followed by incubation with streptavidin-conjugated horseradish peroxidase (Dako), and peroxidase activity was detected by amino-ethyl carbazole (AEC). Staining without antibody was performed as a negative control.

The following panel of monoclonal antibodies was applied: (a) rabbit polyclonal antibody anti-GST Fhit serum (1:4000); (b) anti-c-erbB-2 MAbcB11 (1:60; Ylem, Avezzano, AQ, Italy); (c) anti-p53 MAb DO7 (1:500; Novocastra, Newcastle-upon Tyne, United Kingdom); (d) anti-bcl2 Mab 100 (1:20; a kind gift from Dr. David Mason, John Radcliffe Hospital, Headington, Oxford, United Kingdom); and (e) anti-progesterone receptor MAb1A6 (1:20; DBA, Segrate, MI, Italy) evaluated as described previously(13). Fhit immunostaining was assessed based on cytoplasmic labeling.

χ² analysis was used to evaluate differences in frequencies of the various parameters.

Nested RT-PCR analysis of 29 primary breast tumors (23 DCs and 6 LCs) and normal breast epithelium (as a control) using primers within FHIT exons 3 and 9 revealed aberrant transcripts in 7 cases and absence of FHIT products in 2 cases (Table 1; Fig. 1). Sequence analysis of the aberrant transcripts showed exon deletions involving exon 5 (containing the initiating methionine) in 6 cases—specifically, deletions of exons 4–7 (sample 15), exons 5–8 (samples 8 and 14), exons 4–8 (sample 26), exons 4–6 (sample 27), and exons 5–7 (sample 24). All of the 6 cases also showed the normal-sized fragment of 577 bp, with the correct FHIT sequence. Sample 6 showed a larger abnormal FHIT transcript in the absence of the normal FHIT product; sequence analysis revealed a 65-bp insertion between exon 4 and 5 that leaves the coding region intact. Besides the normal and deleted FHIT transcripts, a larger band was detected in samples 2, 22, and 24 and indicated an ∼100-bp insertion between noncoding exons 3 and 4 of sample 24, whereas the products were not resolved in sample 2 or 22. In samples 9 and 11, FHIT transcripts were not detected.

Immunohistochemical analysis of formalin-fixed and paraffin-embedded breast tumor and normal epithelium sections of the first series served to define three groups of tumors based on the reactivity of anti-Fhit polyclonal antibody (Table 2). The first group of the first series included nine tumors in which immunostaining was as intense as in the matched normal breast epithelium (e.g., Fig. 2,a). This cytoplasmic staining was uniform and “strong” (Fig. 2,b and 2,c). All 9 cases displayed only the normal-sized Fhit transcript (Table 1).

The second group included nine cases characterized by homogeneous reactivity on the tumor, but immunostaining intensity was much lower than in normal epithelial cells in the same sample (Fig. 2, d and e); five of these cases showed only normal FHIT transcripts by RT-PCR, whereas the other four samples showed both normal and abnormal transcripts (deletions of exons 5–8, 4–7, and 4–8; Table 1).

The third group included 11 cases negative for Fhit protein expression in the tumor, whereas the normal epithelium in the same sample was highly positive (Fig. 2,f). Six of these tumors displayed normal FHIT transcripts. It is possible that the normal RT-PCR band derives from the normal cells contaminating the tumor tissue. Two cases showed a mixture of normal and aberrant transcripts (deletions exon 5–7 and 4–6). In one other sample (sample 6) only an abnormal transcript with an insertion that leaves the coding region intact was revealed; suggesting that this insertion was sufficient to abolish protein expression. In the remaining 2 samples (9 and 11), the absence of Fhit transcripts correspond to the lack of Fhit protein (Table 1). Thus, Fhit protein level were reduced or absent 69% (20 of 29) of the breast carcinoma samples tested.

Reduction and loss of Fhit protein expression was observed with similar frequency also on a retrospective series of 156 consecutive primary breast carcinomas (Table 3).

Analysis of the clinical, pathological, and biological characteristics of tumors in comparison with FHIT alterations at the mRNA or protein levels revealed no major differences between tumors with abnormal and normal FHIT transcripts. On the other hand, comparison of tumors with strong Fhit versus Fhit-negative tumors showed that the latter were more frequently large and poorly differentiated tumors (P < 0.05 by χ² analysis) based on both high tumor grade and lack of hormone receptor expression. Tumors scored as strongly positive for Fhit protein expression were mainly of small size and more differentiated. Tumors with weak immunostaining were between these two groups. These data were also confirmed on the larger series. It is noteworthy that the series N.2 consisted in smaller tumors, with a lower frequency of poorly differentiated tumors. Nevertheless, the association between loss of Fhit expression and large tumor size was confirmed (P < 0.05). In addition, association with a high number of mitoses was also found (P < 0.05). The association with tumor grade, found in the first series, was observed only as a trend. No association with HER2/neu overexpression, p53 alteration, and absence of hormone receptors expression was also observed.

In the present study, detection of abnormal FHIT transcripts was reported in one-third of the human breast carcinomas examined in keeping with our previous data (7). These alterations led to a decrease or absence of Fhit protein in the tumor cells. In addition, down-regulation or absence of Fhit protein was detected in another one-third of the samples displaying normal Fhit transcript. This leads to the conclusion of altered Fhit expression in a larger fraction of tumors (70%) than the one expected on the basis of gene alterations, which suggests that FHIT is one of the most frequent targets of alteration in breast cancer. Indeed, well-studied markers of breast carcinomas aggressiveness, such as HER2 and p53 were affected in less than 30% of our cases, as reported previously (13, 14, 15). The higher frequency of FHIT alteration in human breast cancer suggests the potential usefulness of this marker for prognostic purpose and as a therapeutic target in breast tumors.

In previous studies, loss of function of FHIT was attributed to intragenic deletions leading to the production of transcripts incapable of synthesizing the protein (16). In fact, the absence of the initiating methionine located in exon 5 (deleted in some of the samples examined in our study) as well as transfection experiments with constructs sequentially deleted in each FHIT coding exon,⁴ showed that some deletions did not allow Fhit protein expression. Similarly, sample 6 in our study, which displayed only the abnormal transcript with a 65-bp insertion between exon 4 and 5, did not express detectable amounts of protein, suggesting that even an insertion outside the coding region can abolish Fhit protein synthesis. However, our finding of absent Fhit protein in 11 cases and reduced level of the protein in 4 cases suggests that Fhit protein expression can be abrogated by additional mechanisms. Similarly, Hadaczek et al. (17) have reported no FHIT gene abnormalities and a very low number of cases with altered RT-PCR products in clear cell renal carcinoma where the great majority of clear cells showed reduced or absent Fhit protein. One possible reason for the lack of correlation between the data by RT-PCR and by immunohistochemistry observed in these 11 cases may rest in the exquisite sensitivity of RT-PCR in detecting very low-abundance transcripts whose products may not be detectable at the protein level; we cannot exclude the possibility that the normal transcript derives from normal ductal cells present in the tumor specimens (less than 1%) that are strongly positive by immunohistochemistry in all of the 11 specimens.

Finally, the decrease or absence of Fhit protein observed in these 29 cases and in a series of 156 consecutive breast carcinomas is clearly associated with a more aggressive disease, inasmuch as it is significantly associated with highly proliferative and large tumors, showing a tendency to be poorly differentiated.

In conclusion, our studies showed for the first time that Fhit inactivation is a very frequent event in breast carcinomas and identifies a more aggressive phenotype.

We thank Piera Aiello, Ghirelli Cristina, and Luisa Moiraghi for expert technical assistance.