Expression of Interferon Regulatory Factor (IRF) Genes and Response To

Expression of interferon regulatory factor (IRF) genes and response to interferon-␣ in chronic myeloid leukaemia A Hochhaus1,2, XH Yan1, A Willer2, R Hehlmann2, MY Gordon1, JM Goldman1 and JV Melo1

1LRF Centre for Adult Leukaemia, Department of Haematology, Royal Postgraduate Medical School, London, UK; and 2III Medizinische Klinik, Klinikum Mannheim der Universita¨t Heidelberg, Mannheim, Germany

Interferon regulatory factors (IRF) 1 and 2 are DNA-binding pro- regulatory genes is in itself virus- and IFN-inducible; however, teins which control interferon (IFN) gene expression. IRF1 func- an increase in IRF2 transcription occurs only after induction tions as an activator for IFN and IFN-inducible genes, whereas IRF2 represses the action of IRF1. Expression of the two regu- of IRF1. The IRF2 protein is more stable than the IRF1 protein latory genes is itself IFN-inducible. Because therapeutic and in growing cells IRF2 is more abundant than IRF1. After responses of chronic myeloid leukaemia (CML) patients to IFN- stimulation by IFN or viruses, the amount of IRF1 increases ␣ may be determined by intracellular levels of these two mutu- relative to IRF2,8 suggesting that growth restraint by IFN ally antagonistic transcription factors, we have devised a com- depends on the balance between these two mutually antagon- petitive reverse-transcription polymerase chain reaction (RT- istic factors. This possibility is strengthened by the finding that PCR) assay which provides an estimate of the ratio of IRF1 to IRF2 expression in a given cell population. Analysis of periph- overexpression of IRF2 in NIH-3T3 fibroblasts leads to a trans- eral blood leucocytes from 25 normal individuals showed that formed tumorigenic phenotype, which can be reversed by the IRF1:IRF2 ratio varied between 1.13 and 2.30 (mean ± s.d. concomitant expression of the IRF1 gene.7 Taken together, the 1.49 ± 0.33). Similar values were obtained for normal bone mar- data identify IRF1 as a candidate tumour suppressor gene and row specimens, with no significant difference between CD34+ − IRF2 as a candidate proto-oncogene. and CD34 cells. In contrast, the IRF1:IRF2 ratio in leucocytes Recombinant IFN-␣ is useful in the treatment of Philadel- from CML patients showed a much wider variation (0.53–5.11). Eleven out of 130 patients in chronic phase had ratios above phia (Ph)-positive chronic myeloid leukaemia (CML) but the the normal range, whereas none of the 33 blast crisis samples response of CML patients to IFN treatment varies considerably had a ratio Ͼ2.5. Analysis of diagnostic specimens in 59 CML with approximately 70–80% of patients showing long-lasting patients treated subsequently with IFN-␣ showed a high haematological control and 20% being resistant. Patients in IRF1:IRF2 ratio of 5.11 in one of two patients who became com- early chronic phase (Ͻ12 months after diagnosis) respond bet- plete responders; all the 53 patients with minimal or no cyto- ter than patients in late chronic phase or in advanced disease.9 genetic response had ratios below 2.5. In a separate series of ␣ The reasons for this heterogeneity in response to IFN treatment 97 CML patients studied after IFN- therapy a highly significant ␣ correlation was found between the IRF1:IRF2 ratio and both the are unknown. In a considerable minority of patients IFN- cytogenetic and the molecular response (ie low concentration permits the re-emergence of a significant degree of Ph-nega- of BCR-ABL transcripts) to treatment: 53 out of 115 prospective hematopoiesis,9 but the mechanism of its apparently tively analysed samples of good cytogenetic responders had selective action in CML is unclear. Several groups have ratios above 2.0, as opposed to only 13 out of 91 samples from reported that prognostic discrimination by the conventional poor responders (P Ͻ 0.0001; ␹2 test). We conclude that a high Sokal index10 is less efficient in IFN-treated patients compared ratio of IRF1/IRF2 expression may be associated with good 11–15 cytogenetic and molecular response to IFN-␣ in CML. with patient populations on chemotherapy. In the individ- Keywords: interferon regulatory factors; interferon-␣; chronic ual patient haematologic and cytogenetic response is not pre- myeloid leukaemia dictable using initial clinical parameters. IFN-␣ has a direct antiproliferative effect on both normal and CML progenitor cells in vitro, but no selective anti-leu- Introduction kemic activity on progenitor cell proliferation has been demonstrated.16,17 It is possible that therapeutic responses to IFN- Type I interferons (IFN-␣ and IFN-␤) are cytokines that exert ␣ can be identified by the relative intracellular levels of IRF1 antiproliferative activity on normal and transformed cells, and and IRF2, since an IFN-induced increase in IRF1 or decrease are able to block growth factor-stimulated cell cycle tran- in IRF2 will result in rise in the IRF1:IRF2 ratio which may be sitions. IFNs are induced by growth factors and viruses, which a critical event in the regulation of cell growth. In the present suggests that they take part in a feedback mechanism that study we have addressed this question by devising a sensitive regulates cell growth.1 Interferon regulatory factors (IRF) 1 and method for measuring this ratio and analysing its pattern in 2 are DNA-binding proteins which control IFN gene CML patients treated with IFN-␣. expression.2 The two factors are structurally related, parti- cularly in the amino-terminal region, which confers DNA- binding specificity, and they independently bind to a pro- Materials and methods moter element shared by the IFN-␣ and IFN-␤ genes, as well as many IFN-inducible genes.3–5 Patients and normal individuals Gene transfection studies have demonstrated that IRF1 func- tions as an activator for IFN and IFN-inducible genes, whereas Peripheral blood (PB) and/or bone marrow (BM) specimens IRF2 represses the action of IRF1.6,7 Expression of the two were obtained from 260 CML patients, and from 25 healthy volunteers aged 24 to 56 years, with informed consent as required by the Declaration of Helsinki. Correspondence: JV Melo, Dept of Haematology, RPMS, Ham- The CML series consisted of 227 patients studied at chronic mersmith Hospital, Ducane Road, London W12 0NN, UK phase (CP) and 33 at blast crisis (BC). All CML patients Received 14 February 1997; accepted 4 April 1997 expressed the BCR-ABL gene as assessed by reverse-transcrip- IRF expression in CML A Hochhaus et al 934 tion polymerase chain reaction (RT-PCR),18 and all but six fugation (CsCl gradient). A negative control in the form of were Philadelphia (Ph)-chromosome positive. To evaluate the 500 ␮l guanidinium-thiocyanate solution containing only IRF test system, random CML samples from 163 patients (130 5 ␮g E. coli tRNA but no cells was included in each batch of CP, 33 BC) were investigated first. In a second step, diagnostic RNA extraction. The RNA was reverse-transcribed into cDNA specimens from 59 of the chronic phase patients who were with MoMuLV reverse transcriptase using random hexamers subsequently treated with IFN-␣ were analysed separately. as primers.18 Finally a total of 226 samples from 97 patients on IFN-␣ therapy were tested. Five patients were investigated before and on IFN treatment. One to eight (median 2) follow-up samples PCR amplification were tested. Within the group analysed for response to IFN-␣, 145 Ph- PCR co-amplification of IRF1 and IRF2 cDNA sequences was + Ј Ј positive patients could be assessed for cytogenetic response done with primers IFS (5 -CAACttcATCCCGGGGCTCA-3 ) − Ј Ј and six Ph-negative patients for BCR gene rearrangement by and IFE (5 -ccctGaGGtAGCATgCGGTA-3 ). As shown in quantitative Southern blot.19 These patients received recombi- Figure 1, these primers were designed to span regions of high nant IFN-␣2a, b or c, or lymphoblastoid IFN-␣n1, either alone homology between the two genes; in positions where the two or in combination with hydroxyurea, busulphan or cytosine sequences were mismatched, a base (indicated in lower case) arabinoside. IFN-␣ therapy was usually started in the first year was introduced that was unrelated, ie non-homologous, to from diagnosis, with an initial dose ranging from 3 × 106 IU either IRF1 or IRF2. Because each primer can anneal equally 3 days per week to 9 × 106 IU 7 days per week. The dose to both IRF1 and IRF2 sequences, this is essentially a competi- was adjusted to reach leucocyte counts of Ͻ5 × 109/l. tive PCR assay in which the relative amount of amplified pro- Cytogenetic response to IFN-␣ was assessed at 12–28 duct from each cDNA species provides an estimation of the months (median 30 months) after initiation of treatment in all proportional concentration of the RNA transcripts in the test ␮ but 16 patients in whom the response was assessed between sample. The PCR was set up in 25 l volume containing ␮ 6 and 12 months on IFN. Three of these 16 patients had a 10 mM Tris pH 8.3, 50 mM KCl, 1.5 mM MgCl2, 250 M each ␮ Ј Ј good response within this period. The remaining 13 patients dATP, dCTP, dGTP and dTTP, 0.25 M each 5 and 3 primer, ␮ from this group were also included in the analysis, in order 0.5 U Taq polymerase and 1 l cDNA. All PCR experiments to avoid selection of patients who had survived more than 12 included a negative-control reaction without DNA template. months on IFN-␣, which would introduce a bias towards the Rigorous precautions were taken to avoid the possibility of 22 analysis of patients with better prognosis. In all but four of the false-positive results, as described elsewhere. Thermocy- ° 91 Ph-positive patients at least 20 metaphases were analysed, cling conditions were: three cycles of denaturing at 96 C for ° ° and the cytogenetic response was defined as: good, compris- 30 s, annealing at 46 C for 50 s, extension at 72 C for 1 min, + + ° ing complete (0 Ph metaphases) and partial (1–34% Ph followed by 30 cycles of denaturing at 96 C for 25 s, ° ° metaphases) responses, or poor, including patients with a annealing at 60 C for 50 s, extension at 72 C for 1 min, fol- + ° minor (35–94% Ph metaphases) and no response (95–100% lowed by a final extension at 72 C for 10 min. Four microliters Ph+ metaphases). Among the four patients with less than 20 of reaction products were electrophoresed through ethidium- analysable metaphases, two had a good and two a poor cyto- bromide stained 1.5% agarose mini-gel, visualised and photo- genetic response. The response status in these four patients graphed under UV light (Figure 1a). was confirmed by molecular analysis (see below), which showed results within the same categories. In Ph-negative Identification of IRF1 and IRF2 transcripts patients response was determined by Southern blot analysis. Absence of the rearranged BCR band was treated as equival- A unique EcoRI site present in the IRF2, but absent from the ent to complete cytogenetic response. Molecular response IRF1 cDNA (Figure 1), was used as a restriction fragment was assessed by estimation of the concentration of BCR-ABL length polymorphism (RFLP) to distinguish the amplified tran- transcripts and/or of the BCR-ABL:ABL transcript ratio by a 20 scripts from the two genes. Nine microliters of PCR product competitive PCR titration assay as previously described. In were digested with EcoRI in a 15 ␮l reaction (Figure 1b). Ten the majority of samples the quantitative PCR assay was perfor- microliter products of the EcoRI digestion were electrophor- med contemporaneously with the cytogenetic analysis. The esed through a 2% agarose gel, transferred by Southern blot- molecular response was defined as good (BCR-ABL:ABL ratio ting to a nylon membrane, prehybridized and hybridized at Ͻ у 14%) or poor (BCR-ABL:ABL 14%) as described else- 42°C overnight to the IFCO oligonucleotide probe. This 20- 20 where. mer synthetic oligonucleotide (5Ј-GATCCCAAa/gACa/ gTGGAAGGC-3Ј) is degenerate at the two positions of non- homology (indicated in lower cases) between IRF1 and IRF2. Processing of samples The membrane was washed in 2 × SSC at room temperature and exposed to radiographic film at −70° for 2–8 h. Densito- PB and BM leucocytes from normal individuals and from CML metric analysis of the 290 bp and 201 bp hybridisation bands patients were isolated by red-cell lysis of centrifuged buffy- (Figure 1c) was done by the NIH Image program. The ratio coat preparations. For some experiments, the leucocytes were IRF1:IRF2 represents the relative intensity of the upper band separated by gradient centrifugation over Ficoll–Hypaque to (IRF1 product) to the lower band (IRF2 product). yield granulocytes (GR) and mononuclear (MN) cells. Specific subpopulations such as CD34+ and CD34− cells were obtained by immunomagnetic separation on Mini-MACS col- Statistical analysis umns (Miltenyi Biotec Inc, Sunnyvale, CA, USA) according to the manufacturer’s recommendations. Total RNA was Differences of the IRF1:IRF2 ratios according to cytogenetic extracted from the cell suspensions by the acid- and molecular response groups were calculated by Mann– guanidinium/phenol/chloroform method21 or by ultracentri- Whitney test and ␹2 test. IRF expression in CML A Hochhaus et al 935

Figure 1 Competitive RT-PCR assay strategy for assessing the ratio of IRF1:IRF2 expression. The IRF1 and IRF2 partial cDNA sequences were aligned based on data deposited in GenBank, accession no. X14454 (IRF1) and X15949 (IRF2). Indicated on the sequences are the positions of the sense IFS+ and the antisense IFE− primers used for PCR ampliﬁcation; the IRF2 discriminating EcoRI site; and the IFCO oligonucleotide probe used for Southern hybridization. Illustrative results of the assay are shown in (a) agarose-gel electrophoresis of the PCR products, showing a 290-bp predominant band (arrow), as well as minor spurious PCR fragments; (b) agarose-gel electrophoresis of EcoRI digestion products, where two major bands can be resolved (arrows): a 290-bp fragment corresponding to the uncut IRF1 product, and a 201-bp fragment representing the largest of the two IRF2 digests; (c) autoradiograph of Southern hybridization with the IFCO oligonucleotide probe which identiﬁes the two discriminating bands (arrows) for densitometry analysis. The right-most lane on both (b) and (c) corresponds to a control undigested PCR product. IRF expression in CML A Hochhaus et al 936

Figure 2 (a) Ratio IRF1:IRF2 in PCR-ampliﬁed leucocyte cDNAs from 25 normal individuals as compared to 163 CML patients. (b) Distribution of samples from CML patients studied in chronic phase and in blast crisis according to the ratio IRF1:IRF2. The horizontal dotted lines represent the upper and lower limits of the range of IRF1:IRF2 ratios observed in leucocytes from 25 normal individuals.

Results genetic responders the IRF1:IRF2 ratio prior to treatment was 0.90 and 5.11, respectively, with the four partial responders Validation of the assay exhibiting ratios between 1.05 and 1.56. All the 53 patients with minor or no cytogenetic response had initial ratios In order to measure the ratio of IRF1:IRF2 gene expression, below 2.5. we designed an assay system (Figure 1) based on RT/PCR co- In a separate series of 226 samples from 97 CML patients amplification of transcripts from both genes, followed by studied at various time-points after starting IFN-␣ therapy a restriction digestion of the PCR products and Southern highly significant correlation was found between the hybridization with an oligonucleotide probe capable of quan- IRF1:IRF2 ratio and the cytogenetic response to IFN-␣ treat- titatively identifying the two cDNA sequences, ie the uncut ment: whereas 46% (53/115) of samples from patients who IRF1 and the largest of the two digested IRF2 fragments. Densi- achieved a good response (Ͻ35% Ph-positive metaphases) tometric analysis of the scanned autoradiographic images pro- had IRF1:IRF2 ratios above 2.0, only 14% (13/91) of those vided an estimate of the level of expression of each gene, as with a poor cytogenetic response (у35% Ph-positive measured in arbitrary units, from which the ratio IRF1:IRF2 metaphases) showed IRF1:IRF2 ratios above 2.0 (␹2 test: was derived. P Ͻ 0.0001) (Tables 1 and 2). Similarly, there was a highly Several experiments were set up initially to test the repro- significant correlation between the IRF1:IRF2 ratio and the ducibility and sensitivity of the PCR/RFLP/hybridization assay. molecular response (ie low ratio of BCR-ABL:ABL transcripts) These experiments showed that there was little difference in to IFN-␣ therapy in all 199 samples investigated for BCR- the IRF1:IRF2 ratio obtained from replicate reactions of the ABL:ABL and IRF1:IRF2 ratios simultaneously in the latter ser- same cDNA sample, as indicated by the low coefficient of ies of patients: 51 out of 106 samples showing good response variation (CV) for a seven-replicate assay on eight samples (CV had ratios above 2.0, as opposed to only 18 out of 93 samples 3, 3, 3, 5, 5, 6, 8 and 11%, respectively). Similarly, six half- showing poor response (P Ͻ 0.0001; ␹2 test) (Table 3). In six log serial dilutions of the cDNA templates up to 1:1000 did Ph negative patients, two complete responders had maximum not introduce significant variation in the IRF1:IRF2 ratio (CV IRF1:IRF2 ratios of 5.0 and 8.2, while four poor responders 4, 5, 5 and 9% for four samples). had ratios between 1.5 and 2.0 (Table 1). Overall analysis of the 97 CML patients in relation to the highest IRF1:IRF2 ratio achieved after IFN-␣ therapy showed that this was above 2.0 IRF1:IRF2 ratio in normal and CML specimens in 30 of the 43 good responders and in only 12 of the 54 poor responders (Figure 3). Analysis of peripheral blood leucocytes from 25 normal indi- Five patients were analysed before and during IFN-␣ treat- viduals showed that the IRF1:IRF2 ratio varied between 1.13 ment. Only one of these patients achieved a complete cyto- and 2.30, with a median of 1.35 and a mean ± s.d. of 1.49 ± genetic response: he had a high ratio (5.11) at diagnosis and 0.33 (Figure 2a). Similar values were obtained for six normal on two other occasions 2 and 5 months after initiation of IFN- bone marrow specimens, with no significant difference ␣, but specimens tested at 18 and 24 months showed ratios between granulocytes and mononuclear cells or, among the in the normal range (1.76 and 2.2, respectively). One other latter, between CD34+ and CD34− cells (range of all measure- patient, who achieved a partial response, had a low (1.11) ments, 0.86–2.57). IRF1:IRF2 ratio at diagnosis, high ratios 6 and 12 months after In contrast, the IRF1:IRF2 ratio in leucocytes from CML IFN-␣ was started (4.82 and 3.25, respectively), and lower patients (n = 166) showed a much wider variation (0.53–5.11) ratios in the range 1.13–2.19 on four samples tested sub- (Figure 2a). Eleven out of 130 patients in chronic phase had sequently at 15, 16, 17 and 26 months. In each of these two ratios above the normal range, whereas none of the 33 blast patients the respective cytogenetic response was detected crisis samples had a ratio Ͼ2.5. (Figure 2b). approximately 12 months after initiation of IFN-␣ therapy and was sustained throughout the follow-up period. The remaining three patients in this group showed no response to IFN-␣ and IRF1:IRF2 ratio and response to IFN-␣ exhibited low ratios before and throughout or after treatment. It should be noted, however, that for two of the latter patients Analysis of diagnostic specimens from 59 CML patients only one post-IFN-␣ specimen (approximately 24 months after treated with IFN-␣ showed that in the two complete cyto- initiation of therapy) was available for tests. IRF expression in CML A Hochhaus et al 937 Table 1 Ratios IRF1:IRF2 in healthy donors, CML patients at diagnosis and CML patients on therapy with IFN-␣

Patients Samples Total IRF1:IRF2 ratio

Range Median Range Median (per patient) (per patient)

Healthy donors 25 NA NA 25 1.13–2.30 1.35 CML patients at diagnosis 59 NA NA 59 0.66–5.30 1.14 Patients on IFN-␣ therapy 97 1–8 2 226 0.64–8.16 1.60 Complete respondersa 22 1–6 3 67 0.73–6.59 1.96 Partial respondersa 19 1–7 2 48 0.76–6.19 1.81 Minor respondersa 11 1–5 2 25 0.73–5.41 1.37 Nonrespondersa 39 1–4 1 66 0.64–3.39 1.34 Ph negative patients, complete responseb 2 3–8 NA 11 0.69–8.16 1.82 Ph negative patients, poor responseb 4 1–3 3 9 0.90–1.97 1.18 aCytogenetic response, as assessed by the proportion of Ph-positive metaphases. bSouthern blot analysis for the presence of BCR gene rearrangement. NA, not applicable.

Table 2 Comparison of the ratio IRF1:IRF2 and cytogenetic response in 206 samples from 91 Ph positive patients on IFN-␣ therapy

Ratio IRF1:IRF2

Ͻ2.0 Ͼ2.0 Total

‘Good’ response (Ͻ35% Ph+ metaphases) 62 53 115 ‘Poor’ response (у35% Ph+ metaphases) 78 13 91 Total 140 66 206

␹2 test P Ͻ 0.0001.

Table 3 Comparison of the ratio IRF1:IRF2 and contemporaneously determined molecular response in 199 samples from 91 Ph positive and six Ph negative patients on IFN-␣ therapy

Ratio IRF1/IRF2

Ͻ2.0 Ͼ2.0 Total Figure 3 Highest ratio IRF1:IRF2 in 97 CML patients on IFN-␣ therapy divided according to cytogenetic response: good, comprising ‘Good’ response 55 51 106 complete or partial (Ͻ35% Ph-positive metaphases); poor, comprising (Ͻ14% BCR-ABL:ABL ratio) minimal or no response (у35% Ph-positive metaphases). ‘Poor’ response 75 18 93 (у14% BCR-ABL:ABL ratio) Total 130 69 199 of the criterion used to assess response, it is clear that CML patients are heterogeneous and that, in some cases, IFN-␣ ␹2 test P Ͻ 0.0001. administration does not control the disease. Early identification of these patients would therefore be useful, as it would avoid the need for a long-lasting, expensive and potentially Overall, no obvious correlation was observed between the toxic therapy, and accelerate the administration of alternative IRF1:IRF2 ratio and time on IFN-␣ or changes in IFN-␣ dose forms of treatment. Conversely, detection of patients who are in patients for whom multiple follow-up samples could be likely to benefit significantly from IFN-␣ could justify the analysed. Similarly, we did not detect significant numbers of decision to include this drug as a first-choice treatment for a patients with persistently high ratios. Thus, whereas most of longer period of time. the good responders had high IRF1:IRF2 ratios on at least one The gradual unravelling of the IFN-␣ signalling pathways occasion (Figure 3), earlier or later samples from the same provides a biological framework for investigating the varia- patients could be found with ratios in the normal range. bility of clinical responses to IFN-␣ treatment in CML. In the present study we focused on the possibility that deviations in either the steady-state or in the induced mRNA levels of the Discussion antagonistic transcription factors IRF1 and IRF2 could dictate the ultimate effect of IFN-␣ on the CML clone. Our overall The response to IFN-␣ is an important prognostic indicator for results suggest that there is a correlation between the ratio of survival in CML. Currently, response is determined by haema- IRF1:IRF2 expression in CML leucocytes and the response to tological, cytogenetic or molecular parameters.23 Regardless IFN-␣ therapy. Patients who exhibit a high IRF1:IRF2 ratio IRF expression in CML A Hochhaus et al 938 (Ͼ2.0) during IFN-␣ treatment are more likely to experience References a partial or complete cytogenetic and/or molecular remission 1 Tanaka N, Taniguchi T. Cytokine gene regulation: regulatory cis- than those whose IRF1:IRF2 ratio is persistently low (Figure 3). elements and DNA binding factors involved in the interferon sys- It is unlikely that the constitutive IRF1:IRF2 ratio, ie before tem. Adv Immunol 1992; 52: 263–281. IFN-␣ therapy, shows a similar correlation, since the two com- 2 Harada H, Fujita T, Miyamoto M, Kimura Y, Maruyama M, Furia plete responders had IRF1:IRF2 ratios corresponding to the A, Miyata T, Taniguchi T. Structurally similar but functionally dis- upper and the lower extremes of the CML range (0.90 and tinct factors, IRF-1 and IRF-2, bind to the same regulatory elements 5.11). However, more patients for whom tests on diagnostic of IFN and IFN-inducible genes. Cell 1989; 58: 729–739. 3 Sims SH, Cha Y, Romine MF, Gao PQ, Gottlieb K, Deisseroth AB. specimens and follow-up information are available will be A novel inteferon-inducible domain: structural and functional needed to confirm this impression. analysis of the human interferon regulatory factor 1 gene pro- A single previous report24 investigated whether unbalanced moter. Mol Cell Biol 1993; 13: 690–702. expression of the IRFs as well as other IFN-␣ inducible genes 4 Harada H, Takahashi EI, Itoh S, Harada K, Hori TA, Taniguchi T. (2Ј–5Ј oligoadenylate synthetase, p68 kinase and RNaseL) had Structure and regulation of the human interferon regulatory factor any bearing on the clinical response of CML patients to IFN- 1 (IRF-1) and IRF-2 genes: implications for a gene network in the ␣ ␤ interferon system. Mol Cell Biol 1994; 14: 1500–1509. or IFN- treatment. Constitutive and IFN-induced transcript 5 Cha Y, Deisseroth AB. Human interferon regulatory factor 2 gene: levels of these genes were assessed by Northern blot tech- intron-exon organization and functional analysis of 5Ј-flanking niques in 16 CML patients, and the authors concluded that region. J Biol Chem 1994; 269: 5279–5287. neither the magnitude of induction of each gene, nor the 6 Abdollahi A, Lord KA, Hoffman-Liebermann B, Liebermann DA. IRF1:IRF2 mRNA balance differed between the chronic phase Interferon regulatory factor 1 is a myeloid differentiation primary patients who responded and those who failed to respond to response gene induced by interleukin-6 and leukemia inhibitory ␣ factor: role in growth inhibition. Cell Growth Diff 1991; 2: 401– IFN- treatment. The discordance between this report and our 407. results on patients tested during IFN-␣ therapy may be due to 7 Harada H, Kitagawa M, Tanaka N, Yamamoto H, Harada K, Ishi- either the significant difference in size of the CML population hara M, Taniguchi T. Anti-oncogenic and oncogenic potentials of analysed, and/or to the different sensitivities of the methods interferon regulatory factors-1 and -2. Science 1993; 259: 971– utilised for analysis of gene expression, namely, Northern 974. blotting24 vs competitive RT-PCR. 8 Sato T, Selleri C, Young NS, Maciejewski JP. Hematopoietic inhibition by interferon-␥ is partially mediated through interferon regu- The mechanisms linking the ratio of IRF1:IRF2 expression latory factor-1. Blood 1995; 86: 3373–3380. in CML leucocytes to the response to IFN-␣ therapy as 9 Kantarjian HM, O’Brien S, Anderlini P, Talpaz M. Treatment of detected in our study are unknown. It is possible that in chronic myelogenous leukemia. Current status and investigational responsive patients, IFN-␣-induced overexpression of IRF1 by options. Blood 1996; 87: 3069–3081. the leukemic clone ensures the availability of IFN-␣ binding 10 Sokal JE, Cox EB, Baccarani M, Tura S, Gomez GA, Robertson JE, sites on other IFN-inducible genes that are directly involved Tso CY, Braun TJ, Clarkson BD, Cervantes F, Rozman C. Prognos- tic discrimination in ‘good-risk’ chronic granulocytic leukemia. in cell growth inhibition. In this scenario, by competing out Blood 1984; 63: 789–799. IRF2, the relative excess of IRF1 protein would act as a cofac- 11 Kloke O, Niederle N, Qiu JY, Wandl U, Mortiz T, Nagel Hiemke tor with IFN-␣ in activating a set of genes whose products M, Hawig I, Opalka B, Seeber S, Becher R. Impact of interferon are required for the negative regulation of cell proliferation. alpha-induced cytogenetic improvement on survival in chronic Alternatively, overexpression of IRF1 may by itself exert an myelogenous leukaemia. Br J Haematol 1993; 83: 399–403. antiproliferative effect on the leukemic clone through a path- 12 Ozer H, George SL, Schiffer CA, Rao K, Rao PN, Wurster Hill DH, ␣ Arthur DD, Powell B, Gottlieb A, Peterson BA, Rai K, Testa JR, way unrelated to that of IFN- . In this case, the presence of LeBeau M, Tantravahi R, Bloomfield CD. Prolonged subcutaneous a high IRF1:IRF2 ratio in some CML patients could represent administration of recombinant alpha 2b interferon in patients with an independent prognostic marker of controllable disease. previously untreated Philadelphia chromosome-positive chronic- Regardless of the mechanism, the correlation between phase chronic myelogenous leukemia: effect on remission dur- IRF1:IRF2 ratio and response to IFN-␣ treatment observed in ation and survival: Cancer and Leukemia Group B study 8583. the present study suggests that measurement of this ratio may Blood 1993; 82: 2975–2984. 13 Hehlmann R, Heimpel H, Hasford J, Kolb HJ, Pralle H, Hossfeld result in the early detection of CML patients who will derive DK, Queisser W, Loffler H, Hochhaus A, Heinze B, Georgii A, long-term benefit from treatment with IFN-␣. Prospective sur- Bartram CR, Grieβhammer M, Bergmann L, Essers U, Falge C, veys based on the RT-PCR assay described here should aim Queiβer U, Meyer P, Schmitz N, Eimermacher H, Walther F, Fett at elucidating several issues which could not be addressed in W, Kleeberg UR, Ka¨bisch A, Nerl C, Zimmermann R, Meuret G, this analysis, such as possible correlations between the Tichelli A, Kanz L, Tigges F-L, Schmid L, Brockhaus W, Tobler A, IRF1:IRF2 ratio and the time-interval from injection of IFN-␣ Reiter A, Perker M, Emmerich B, Verpoort K, Zankovich R, Wus- ␣ sow PV, Pru¨mmer O, Thiele J, Buhr T, Carbonell F, Ansari H and and collection of the blood sample, the effect of IFN- dosage, the German CML Study Group. Randomized comparison of inter- and the time-point along treatment when the results are most feron-alpha with busulfan and hydroxyurea in chronic myelogen- representative and could be used as a check-point. ous leukema. Blood 1994; 84: 4064–4077. 14 Kantarjian HM, Smith L, O’Brien S, Beran M, Pierce S, Talpaz M and the Leukemia Service. Prolonged survival in chronic myelogenous leukemia after cytogenetic response to interferon-␣ ther- Acknowledgements apy. Ann Intern Med 1995; 122: 254–261. 15 Hasford J, Ansari H, Pfirrmann M, Hehlmann R and the German CML Study Group. Analysis and validation of prognostic factors This study was supported by the Leukaemia Research Fund, for CML. Bone Marrow Transplant 1996; 17: (Suppl. 3): S49–S54. the Dr Mildred Scheel Stiftung and the Forschungsfonds der 16 Galvany D, Cawley C. Mechanism of action of ␣-interferon in Fakulta¨tfu¨r Klinische Medizin Mannheim der Universita¨t Hei- chronic granulocytic leukaemia: evidence for preferential inhi- delberg. We thank the clinicians and cytogeneticists partici- bition of late progenitors. Br J Haematol 1989; 73: 475–479. 17 Dowding C, Guo AP, Osterholz J, Siczkowski M, Goldman J, Gor- pating in the UK Medical Research Council CML trials and in don M. Interferon-alpha overrides the deficient adhesion of the German CML trials for sending samples and providing chronic myeloid leukemia primitive progenitor cells to bone mar- clinical cytogenetic data. row stromal cells. Blood 1991; 78: 499–505. IRF expression in CML A Hochhaus et al 939 18 Cross NCP, Melo JV, Feng L, Goldman JM. An optimised multiplex 22 Cross NCP. Detection of BCR-ABL by RT-PCR in haematological polymerase chain reaction (PCR) for detection of BCR-ABL fusion malignancies. In: Cotter FE (ed). Molecular Diagnosis of Cancer. mRNAs in haematological disorders. Leukemia 1994; 8: 186–189. Humana Press: New Jersey, 1996, pp 25–36. 19 Reiter A, Skladny H, Hochhaus A, Seifarth W, Heimpel H, Bartram 23 Hochhaus A, Lin F, Reiter A, Skladny H, Hehlmann R, Goldman CR, Cross NCP, Hehlmann R. Molecular response of CML patients JM, Cross NCP. Quantitative molecular methods to monitor the treated with interferon-␣ monitored by quantitative Southern blot response of CML patients to interferon-␣. Bone Marrow Transplant analysis. Br J Haematol 1997; 97: 86–93. 1996; 17 (Suppl. 3): S41–S44. 20 Hochhaus A, Lin F, Reiter A, Skladny H, Mason PJ, van Rhee F, 24 Fischer T, Aman J, van der Kuip H, Rudolf G, Peschel C, Autlitzky Shepherd PCA, Allan NC, Hehlmann R, Goldman JM, Cross NCP. WE, Huber C. Induction of interferon regulatory factors, 2Ј–5Ј Quantification of residual disease in chronic myelogenous leuke- ␣ oligoadenylate synthetase, P68 kinase and RNase L in chronic mia patients on interferon- therapy by competitive polymerase myelogenous leukaemia cells and its relationship to clinical chain reaction. Blood 1996; 87: 1549–1555. responsiveness. Br J Haematol 1996; 92: 595–603. 21 Chomczynski P, Sacchi N. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 1987; 162: 156–159.