Tau expression in model adenocarcinomas correlates with docetaxel sensitivity in tumour-bearing mice.

Docetaxel is a new taxoid with clinical activity in breast and lung cancer. Using docetaxel-sensitive and -refractory mammary and pancreatic murine tumours, as well as human-derived neoplasms, we investigated if a determinant of docetaxel sensitivity could be found at the level of its mechanism of action. Because microtubules represent the cellular targets of the drug, we studied their heterogeneity in the tumour models to try to explain the differences in drug sensitivity. Reverse transcription-polymerase chain reaction (RT-PCR) analysis of the expression of microtubular components showed that levels of Mbeta4-tubulin and Tau mRNAs were higher in the murine sensitive neoplasms than in the refractory ones. It was also found that Tau protein levels differed markedly among the tumours. In the human-derived sensitive neoplasm, beta-tubulins and some Tau isoforms were found to be more abundant than in the resistant one. Western blot analysis of MAP2 revealed the presence of several immunoreactive species. Some of these polypeptides were also found in higher amounts in the docetaxel-sensitive tumours. The possible meaning of these correlations is discussed in connection with the regulation of microtubule dynamics. ImagesFigure 1Figure 2Figure 3Figure 4Figure 5Figure 6

Docetaxel is a new taxoid which is currently tested in phase III clinical trials. In preclinical exvaluation. docetaxel w-as found to hax-e a broad spectrum of efficacx (Van Oosterom et al. 1995). It exhibits hiah lexels of activitv in both firstand second-line breast cancers. including anthrac -cline-resistant neoplasms and nonsmall-cell lung cancers. Howexer. wxithin a aiven indication. not all patients respond to therapy (Bisserx et al. 1991(Bisserx et al. . 1995a(Bisserx et al. . 1995b).
Taxoids. represented by docetaxel (Taxotere) and paclitaxel (Taxol). haxe the propertx of increasinc microtubule stabilitx to lexels incompatible with normal cell metabolism. P-Tubulins are thought to be the major targets of these drugs (Combeau et al. 1994) and. consequently. several prexvious studies have reported alterations in the expression level of ,B tubulins in cell lines resistant to paclitaxel. Jaffrezou et al (1995) reported the oxerexpression of 5-tubulin (class IVa) in cells derixed from an er-throleukaemic cell line (K562) that w-ere selected for their resistance to paclitaxel. Haber et al (1995) found increased lexels of MP32-tubulin in a series of paclitaxeland docetaxel-resistant J774.2 cell lines. However. the formal implication of tubulin oxerexpression in resistance to taxoids has not been firmlx established.
In addition. it has been show-n that taxoids. as well as other microtubule-damaging drugs. are able to induce Bc12 phosphorylation and apoptosis in cancer cells. Bc 12 has thus been defined as the -uardian of microtubule intet'itx- (Haldar et al. 1997).
We have identified responsixe. refractory and resistant model tumours w ithin breast and pancreatic cancers. Refractorx tumours do not respond to therapy (innate resistance). whereas the resistant ones are neoplasms that initially responded to therapy and then acquire resistance to the agent (Bisserx et al. 1991(Bisserx et al. . 1995a(Bisserx et al. . 1995b). Using sex-eral freshly explanted docetaxel-sensitive. refractorx and resistant tumours. our goal w-as to in estigate wxhether an alteration of microtubule components could explain their different drus sensitix ities. We haxe therefore studied the expression of sexeral microtubule-associated proteins (Tau. MAP2 and MIAP4) and 5tubulins bv usinc a semiquantitatix e non-competitixe RT-PCR approach and by Western blot analx-sis.

MATERIALS AND METHODS Tumour models and mice
Docetaxel wxas exaluated for its anti-tumour actixitx in xixo against sexven tumour models including mammarx adenocarcinomas MA 13/C. NMA 16/C. MA44 of murine origin. the human Calc l 8 and analy ses A-ere performed according to protocols described prexiously (Bissery et al. 1991(Bissery et al. . 1995a(Bissery et al. . 1995b(. The actixitend point used to assess subcutaneously implanted solid tumours A-as tumour growth inhibition (TIC, Where T and C are the median tumour weight of the treated and the control groups respectixelx ) In cases of hinh anti-tumour actixity. the txwo following end points were also used: the tumour growth delay [T-C. A-here T and C are the median times. in day s. required for the treatment group and the control group tumours to reach a predeterrnined size (750-1000 mgh] and the log, cell kill. which is the logarithm of the total number of cells killed bv treatment. A compound is considered highly actixe if the locg cell kill total is > 2.8 and is considered inactixe if the log cell kill total is < 0.7. For advanced-stage tumours. regressions were either partial (more than 50%7e reduction in tumour mass) or complete (rearessions below-the palpation limit). Complete regressions were included in the partial ones. Toxicity Nx-as based on drug deaths (> 20%7c) or a wei2ht loss in excess of 20%/s.
Total RNA extraction RNA w-as extracted from about 0.5 , of sexeral tumour fragments using, a guanidinium thiocyvanate-acid phenol extraction method (Chomczy-nski and Sacchi. 1987). The RNA pellets were resuspended in 0.5 ml of sterile water. The RNA was treated with DNAase I (IOU m-RNA) for 1 h at 37 C to eliminate any possible DNA contamination (RNAase-free DNAase I. Boehringer. Mannheim. German ).
cDNA synthesis cDNA sxnthesis >-as carried out from 2.5 tg of DNAase I-treated RNA (from at least tw-o independent tumour preparations) using the Superscript II enzyme (Life Technologies. Gaithersburg. MD. USA) follo ing the manufacturer's instructions. After an ethanol precipitation step. the cDNA A-as resuspended in 300 .l of sterile w-ater.

RT-PCR
All primers for PCR were designed from the sequence data available in the literature and the Genbank database (accession numbers in brackets). F and R stand for fonAard and reverse primers respectively. Sequences are gixen in the 5' -s 3' direction and the expected lengths of the different amplicons are indicated in brackets. Absence of genominc DNA contamination w as verified bv RT-PCR with pnrmers (ClInhF: 5'-TGGCCTCCAGGCTGACCC-CACTGA-3' and ClnhR: 5'-TGTFTATTGTGATG(GCTACACT-GGT-3' able to amplihf both cDNA and genomic DNA coding for the munine Cl inhibitor. The expected lengths of the products are 221 bp for amplified cDNA and 1200 bp for genomic DNA. The primers used to amplify the different M>-tubulin isoforms were those described by Haber et al ( 1995). w-ith the exception of the rexerse primer for MfVt-tubulin. Another specific forward primer for this tubulin (TuB4F*) w xas used to confirm the results discussed belowx. The expected amplicon lengths for M1-tubulins were 146. 171. 145. 572 and 194 from M1l to MB5-tubulin respectixely.
Non-competitixve RT-PCR w-as performed on a Perkin Elmer thermal cycler using the following conditions: for Tau and Cl Inh. 5 min at 94 C then 1 min at 94 C. 1 min at 58 C and 1 min at 72-C for a number of cycles that wxill be indicated w-hen necessarfollow-ed by a final extension step for 10 min at 72:C: for MAP2.
MAP4 and all tubulins. 5 min at 94 C then 30 s at 95 C. 30 s at 55-C  and 30 s at 72'C for a number of cvcles that A-il be indicated %x-hen necessarv. followed by a final extension step for 10 min at 72'C.

Quantification of RT-PCR products
In order to achieve an accurate quantification of mRNA. the number of PCR cxcles w as chosen in a w av that w-ould preserx e a linear relationship betx-een input cDNA and final RT-PCR

RESULTS
In vivo anti-tumour efficacy of docetaxel In this study. docetaxel w-as exaluated i.%-. acainst murine mammarv (MA 13/C. MA 16/C and MA-H) and pancreatic (P02 and P03) tumours. Their characteristics and response to chemotherapy are summarized in Table 1. The human-derived breast tumours Calc 18 and Calc 1 8/TIXT. also used in our studies. and which are currentlI beine characterized. came from cell lines sensitive and resistant to docetaxel respectively (Riou et al. 1994). Docetaxel is a schedule-independent drug (Bisseret al. 1991(Bisseret al. . 1995b) that was found to be clearly active on three out of four mammaix tumours (the murine MA 16/C. MA13/C and the human Calc 18) and one out of tx o murine pancreatic tumours (P03 . w-ith a high rate of complete tumour regressions of advanced-stage disease (100%7c for MA16/C. 60%7 NMA13/C and 83%e for P03) ( Table 2). Two of the murine tumour models w-ere found to be refractory to docetaxel. the mammarv MA44 and the pancreatic P02. when treated at an early stage (Table 1). It is interesting to note that P02 is also refractors to docetaxel-unrelated drugs (Table 1). Howexver. the P-glycoprotein (involved in producing a multidrug resistance phenotype: Arceci et al. 1993) has not been shown to be relex-ant for this model (Kessel andCorbett. 1985: Priebe et al. 1992).

Analysis of tubulin expression in the murine tumours
The RT-PCR analysis of the expression of tubulin isotypes show-ed no differences in the mRNA levels correspondinc to tubulins M132.

MP3 and M05 among the different tumours. when Mcland
Mat2-tubulins w-ere taken as a reference for the quantification. In contrast. the sensitix e tumours w-ere characterized by higher lex els aT/C (0o). for solid tumours = 100 x median tumour weight of the treated/median tumour weight of the controls. nT-C (days) = median time in days required for the treatment group T and the control group C tumours to reach a predetermined size. :log cell kill = T-C in days/3.32 x tumour doubling time of control mice.

Analysis of the expression of microtubule-associated proteins in the munne tumours
The expression of microtubule-associated proteins (MAPs) was also considered at both the mRNA and the protein level. The transcript analysis by RT-PCR of mammary and pancreatic tumours demonstrated that the mRNA coding for MAP2 (Figure 2). considered to be neuron specific (Lewis et al. 1986). was also expressed in these non-neural tissues. As shown in Figure 2. the amount of MAP2 mRNA did not correlate with the docetaxel sensitivity of the neoplasms. This was also true for MAP4 (results not shown). To study Tau mRNA expression. we designed primers that were able to amplify most of the isoforms: two variants (Genbank Al: U12915 and U12914) expressed in the liver (Kenner et al, 1994) and two other isoforms (M18775 and M18776) known to be expressed in mouse brain ). Interestingly. docetaxel-sensitive tumours showed higher levels of Tau mRNA ( Figure 2). These results were ftuther confirmed with the forward primer TF: 5-GCTCGTGTGGCCAGCAAA-3 which is able to amplify all known Tau isoforms in combination with TR. to yield a Figure 3 Tau protei from munne mamnary and pancreabc tumours detected by Westem blot analysis. Volumes of RIPA extacts containin similar amount of a-bibuin (also detected by ECL) were baded in each Lane. Tau proteins were detected with polycona antbodis directed against neuronal Tau. (A) Antbody produced as descrbed by Vantard et al (1991); (B) antbody TRS 1-2. The docetaxel-refractory neoplasms are distinguished by vertcal arrows. SAd: Adult rat brain crude exta product of 120 bp. The relative amount of Tau proteins was measured by analysing tumour extracts (classical RIPA) containing similar amounts of a-tubulin. Immunoblotting with a polyclonal antibody directed against brain Tau (Vantard et al, 1991) revealed a higher level of Tau-related polypeptides in the sensitive neoplasms (MA13/C. MA161C and P03) than in the refractory ones (MA44 and P02) ( Figure 3A and C). This result was conffimed with the polyclonal antibody TRS 1-2 (Cambiazo et al. 1995) ( Figure 3B and C). To investigate the possibility that some Tau proteins might remain insoluble in the extracts obtained with the classical RIPA buffer. tumours were also treated with a detergent-enriched RIPA buffer. Some Tau-related polypeptides were found to be significandy more abundant in the tumours sensitive to docetaxel. As shown in Figure 4A and B, the higher levels of Tau-like polypeptides in the sensitive mammary adenocarcinomas consisted mainly of a species of high molecular weight (125 kDa). However, in the refractory tumour (MA44). some polypeptides located towards the higher molecular weights within the Tau region were more abundant than in the sensitive neoplasms. By using a monoclonal antibody directed against brain MAP2. the presence of related polypeptides was established in the different murine tumours. As shown in Figure 5A-C, the classical 300-kDa MAP2 was not found. probably because of its proteolytic fragmentation. and the relative amounts of the different polypeptides British Joumal of Cancer (1998) 78 (7) Analysis of microtubular proteins in human-derived tumours sensitive to, or with acquired resistance to, docetaxel Mammary tumours of human origin either sensitive to (Calc 18) or resistant to (Calcl8/17XT) docetaxel were also analysed after extraction of proteins with a detergent-rich RIPA buffer (Figure 6). Tau proteins displayed a complex pattern involving molecular species which were expressed in different amounts im the two types of tumour. The monoclonal antibody Taul. which does not recognize the highly phosphorylated Tau of the Alzheimer disease tangles (Kosik et al. 1988). revealed a major band which was more abundant in the sensitive neoplasm. The MAP2-related fragments were also found to be different in the two types of tumour (not shown). In addition. the global amount of I-tubulin was found to be more abundant in the sensitive tumour (Calc 18) after normalization for equal amounts of a-tubulin. DISCUSSION Docetaxel (Taxotere) is an anti-microtubular agent with clinical activity against various cancers. Using docetaxel-sensitive and -refractory mammary and pancreatic murine tumours and humanderived neoplasms. our goal was to investigate if a determinant of docetaxel sensitivity could be found at the level of its mechanism of action. In the case of the docetaxel non-sensitiv-e models. we investigated the drug-refractorv tumours P02 and MA44 and a tumour model with acquired resistance to docetaxel.
Calc 1 8/EXT. The murine sensitive tumours studied (MA 16/C. MA 13/C and P03) were characterized by a higher level of expression of class IVa tubulin (MP4) mRNA. As the expression of class IV tubulin is thought to be restricted to neuronal tissue and tumour cell lines (Lee et al. 1984: Cowan et al. 1986). the increased level that we detected in the docetaxel-sensitive tumours is likely to originate from the malignant cells. We could not find relevant differences among tumours in the MP4tubulin protein level (data not shon).  Figure 6 Tubulin and Tau content in two human-derived mammary tumours. Volumes of extracts (detergent-enriched RIPA buffer) containing similar amounts of a-tubulin were loaded in each lane. Tau were detected with a polyclonal antibody (Vantard et ai, 1991) and the monocknW antibody Taul possible cross-reactions with other tubulin isotypes may have introduced a source of variability in our results. We examined the expression level of different MAPs by taking a-tubulin to normalize for the amount of microtubule transcripts and proteins. Although we cannot provide formal proof. the increased levels of Tau found in the docetaxel-sensitive neoplasms may represent one of the factors responsible for the drug sensitivity of the tumours. We have recently described the role of these proteins in modulating microtubule alterations induced by docetaxel (Fromes et al. 1996). In addition, it is known that the ectopic expression of Tau in non-neural cells increases microtubule stabilization (Lee et al. 1992). Furthermore, different levels of Tau proteins may alter the ratio of polymerized versus non-polymerized tubulin, and it has been shown that lowering the net amount of polymerized tubulin is sufficient to confer taxol acquired resistance (Cabral et al, 1989). Our data also showed a significant expression of a Tau-related polypeptide. of about 125 kDa. in the sensitive tumours. which may correspond to the high molecular weight isoform found in the mature peripheral nervous system and therefore may be involved in modulating microtubule stability (Couchie et al. 1992). The shift of several Tau isoforms to the higher molecular weights within the Tau region in the murine refractory tumour MA44 suggests that the state of phosphorylation of these proteins is more important in this tumour (Figure 4). Interestingly. highly phosphorylated Tau has a poor capacity to be inorporated in microtubules and may generate unstable structures that are less sensitive to the stabilizing effect of docetaxel (Lindwall and Cole. 1984;Correas et al. 1992). When comparing Figures 3 and 4. it can be seen that several Tau-related polypeptides present in the murine refractory tumours were extracted in detectable amounts only upon using a detergent-enriched REPA buffer. It is possible that cytoskeletal proteins of tumoral epithelial cells are not easily extractable because of the abundance of a fibrous and dense stroma surrounding tumoral cells. It is also likely that several Tau isofonns may aggregate by a self-association reaction similar to that described by de Ancos et al ( 1993).
A higher level of several MAP2-related polypeptides was detected in the docetaxel-sensitive tumours, particularly a fragment of approximately 36-38 kDa. Interestingly. the 36-kDa tubulin-binding domain of MAP2 has some Tau-like properties and has been shown to modulate the effect of some antimicrotubular drugs such as vinblastine (Fellous et al. 1994).
We have also investigated the behaviour of microtubule components in two types of tumours derived from a human breast carcinoma: Calc 18. sensitive to docetaxel. and Calc 1 8TfXT. resistant to the drug. The latter came from a cell line overexpressing the MDRJ gene and veraparnil was able to reverse its docetaxel resistance (Riou et al, 1994). Consequently. the tumour Calcl8/1XT moderately expressed the P-glycoprotein (data not shown). In spite of this. we showed that frtubulin was less abundant in the resistant neoplasm when using a-tubulin to normalize for the amount of microtubule proteins. This may reflect the decreased level of >tubulin mRNAs observed in the Calc 18/XT cell line with respect to the parental Calc18 (Riou et al. 1994). Interestingly, decreased levels of tubulins have been shown to confer resistance to taxol in certain cellular models (see Cabral et al. 1989). Using our polyclonal antibody directed against Tau (Vantard et al. 1991). we also observed differences in the expression of Tau-related polypeptides in both types of tumours. The monoclonal antibody Tau 1. sensitive to the phosphorylation state of the protein, revealed striking differences between the neoplasms. but a certain variability that was noticed in the results was probably due to uncontrolled changes of the Tau phosphorylation state. Immunohistochemical studies demonstrated that the polypeptides recognized by Taul were expressed mainly in the sensitive tumours and the epithelial malignant cells (data not shown).
Although the nature of the decreased docetaxel sensitivity of the tumours refractory (innate resistance) and resistant (acquired resistance) to the drug are conceptually different, both types of tumour displayed differences in the expression and modification of some Tau isoforms. However, other alterations of microtubular components specific to each type of tumour may also contribute to produce the resistant phenotype. i.e. lower levels of J-tubulin in Calc I 8frXT.
In conclusion, the levels of several Tau isoforms in both murine and human-derived freshly explanted docetaxel-sensitive tumours and. to a lesser extent. the mRNA levels for MV-tubulin in the murine tumours can be regarded as markers of sensitivity to the drug. From a practical point of view. the finding that some Tau isoforns are not only present in different amounts but also probably in different phosphorylation states. may allow the use of specific antibodies on tumour biopsies and may help decide upon the worth of a taxoid treatment. These findings are now being evaluated on a broader panel of human tumour samples.