Plant Molecular Biology 17: 49-60, 1991. © 1991 Kluwer Academic Publishers. Printed in Belgium. 49 Transgene expression variability (position effect) of CAT and GUS reporter genes driven by linked divergent T-DNA promoters Cindy Peach 1"3 and Jeff Velten 1'2'3. Graduate Program in Molecular Biology (* author for correspondence); 2plant Genetic Engineering Laboratory, and 3 Chemistry Department, New Mexico State University, Las Cruces, NM 88003-0001, USA Received 14 January 1991; accepted in revised form 5 March 1991 Key words: CAT, GUS, mas promoters, position effect, reporter genes, transgene expression Abstract Forty-five individually transformed clonal tobacco callus lines were simultaneously assayed for both chloramphenicol acetyltransferase (CAT) and/3-glucuronidase (GUS) activity resulting from expression of introduced reporter genes driven by the adjacent and divergent mannopine (mas) promoters. Excluding lines in which one or both of the enzyme activities was essentially zero, the activities of the reporter genes varied by as much as a factor of 136 (CAT) and 175 (GUS) between individual transformants. Superim- posed upon the high degree of inter-clonal expression variability was an intra-clonal variability of 3-4-fold. The observed degree of intra-clonal reporter gene activity may be more extreme because of the regulatory characteristics of the mannopine promoters, but must still be addressed when considering the limitations of reporter gene-based analysis of transgene function and structure. There was no consistent correlation between the expression levels of the introduced CAT and GUS genes since the ratio of GUS to CAT activities (nmol min- 1 mg- 1) within individual lines varied from 0.05 to 49. Even divergent transcription from two directly adjacent promoter regions (both contained within a 479 bp TR-DNA fragment) is insufficient to guarantee concurrent expression of two linked transgenes. Our quantitative data were compared to published data oftransgene expression variability to examine the overall distribution of expression levels in individual transformants. The resulting frequency distribution indicates that most transformants express introduced transgenes at relatively low levels, suggesting that a potentially large number of Agrobacterium-mediated transformation events may result in silent transgenes. Introduction The development of efficient procedures for intro- ducing in vitro manipulated DNA into higher eukaryotic cells has contributed greatly to recent progress in understanding many important devel- opmental, cellular and molecular processes of eukaryotes. The same techniques have also allowed researchers to introduce, into both plants and animals, engineered genes designed to augment the normal genetic content of the target organisms by providing desirable traits difficult or impossible to obtain using more traditional proce- dures. However, very little is currently known about the actual molecular processes acting upon the foreign DNA during uptake, intracellular
`
`Inari Ex. 1018
`Inari Agric. v. Corteva Agriscience
`PGR2023-00022
`Page 00001
`
`

`

`50 transport and stable integration into the genomes of eukaryotic cells. It is clear that genetic material newly introduced into eukaryotic cells must be subject to considerable modification and pro- cessing between its entry into the target cell (usually as DNA of prokaryotic or synthetic origin) and its subsequent expression as part of the structurally organized chromatin of the resulting transgenic cell or organism. Of specific importance to the genetic engineer is the question of how the process of genetic transformation affects both the character and stability of expres- sion of chromosomaUy integrated foreign genes. One such widely reported effect is a seemingly random clonal variability in the level of expression of newly introduced transgenes, each containing initially identical regulatory and structural DNA sequences. Expression level variability between different transgenic cell lines or organisms has been ob- served after introduction of many unrelated genes, both natural and chimeric, into numerous plants species [4, 12, 13, 14, 15, 18, 19, 20, 22, 23, 29, 41, 45, 52, 57, 62, 65]. The observed variability has often been referred to as 'position effect', based on the as yet unproved assumption that expression levels of the introduced genes are di- rectly influenced by host DNA sequence or chro- mosomal structure/composition at or near to the site of integration. Despite the nearly ubiquitous occurrence of 'position effect', the nature of the molecular fac- tors contributing to transgene expression varia- bility remains elusive. In general, transgene varia- bility has failed to correlate with the copy number of stably integrated transgenes [30, 41, 57, and this paper], although a significant correlation between gene copy number and transgene expres- sion has been described [22]. Co-transformation of up to 23 kb of plant DNA flanking a petunia ribulose bisphosphate carboxylase (rbcS) gene does not appear to influence the level oftransgene variability upon reintroduction into tobacco plants [ 15]. Some indication of the molecular resolution of the processes producing transgene variability is given by investigation of expression variability of two linked genes co-transferred on the same T-DNA. Expression levels of linked nopaline syn- thase (nos) and octopine synthase (ocs) genes [30], as well as closely adjacent neomycin phos- photransferase II (NPTII) and CAT reporter genes [4] were found to vary independently between individual transformants. However, sig- nificant co-variation was reported between inde- pendent transgenotes containing linked CAT and GUS genes driven by the the 35S promoter of the cauliflower mosaic virus [20]. Interestingly, co- variance of linked genes driven by two rbcS pro- moters and divergently expressed chlorophyll a/b- binding protein (Cab) genes was found to be greatly influenced by either the particular combi- nation of promoters used [14] or the location of the transgenes within the T-DNA of the plant transformation vector [ 18, 23 ]. In this paper we report the quantitative analysis of simultaneous independent transgene expres- sion level variability using two reporter genes (CAT and GUS) fused to an extremely closely linked (479 bp, ATG-ATG) divergent promoter pair, the mannopine promoters (mas) from Agro- bacterium tumefaciens. To date, function of the mannopine promoters in plants has only been examined either separately [16, 53, 60] or under conditions in which simultaneous activity of both promoters is required for reporter gene activity (the luxA and luxB genes [33]). Materials and methods DNA manipulation and cloning Figure 1 shows pGC4-OO and pGC4-NP. The pGC40-OO plasmid, a binary vector, contains a pair of divergently oriented reporter genes, chloramphenicol acetyltransferase (CAT) from Tn9 [65] and/%glucuronidase (GUS) encoded by the uidA locus of Escherichia coli [27] driven by the two divergent mannopine promoters, 1' (Pmasl') and 2' (Pmas2'), isolated from the TR-DNA of A. tumefaciens [66]. This construc- tion includes the 1.0 kb Cla I-Eco RI fragment from pCAP212 which contains the CAT coding
`
`PGR2023-00022 Page 00002
`
`

`

`51 BamHI /~,,x P GC4"00 BamHI ,.~//Pnos-NPTII N~c~:~ " I I pGC.4.'NP ~. ~/B~, ,, ~ ... (1 b Y,b) t~ L ,~Eco RI I~ oriV -- t-j ~Sal I Fig. 1. The circular map shown, pGC4-NP, is based upon the binary vector, pGG102 (pGA470 [3] in which a Bgl II linker has been inserted into the unique Hind III site (W.M. Ainsley, personal communication)). Inclusion of the mas dual promoter fragment [66] at the indicated Bam HI-Cla I sites creates pGC4-00. Construction of the CAT .-- Pmasl'- Pmas2' ~ GUS cassette is described in [47]. Symbols: Tc, tetracycline resistance gene (from pTJS75); CAT, chloram- phenicol acetyltransferase (Tn9) coding region fused to the g7 polyadenylation signal [65]; GUS, fl-glucuronidase coding region fused to the nos 3' polyadenylation signal [27]; Pnos- NPTII, nopaline synthase promoter fused to the NPTII (kanamycin resistance) gene of Tn5 and the nopaline synthase 3' polyadenylation/termination signal; BR, the right border of T-DNA (from pTiT37); BL, the left border of T-DNA (from pTiT37); oriV, origin of vegetative growth (pRK4); oriT, origin of transfer (pRK4). the CAT and GUS reporter genes. The negative control plasmid, pGC4-NP (no promoter), is identical to pGC4-OO (Fig. 1) except that it lacks the Pmas l'-Pmas2' promoter fragment. Plant transformation and ma&tenance The plasmids, pGC4-OO and pGC4-NP, were moved into A. tumefaciens strain C58C1(r/f) containing the pGV3850 Ti plasmid [69] by the freeze/thaw method of An et al. [2] and the struc- ture of the T-DNA confirmed by Southern hybridization of restriction-digested total A. tu- mefaciens DNA. MesophyU protoplasts were iso- lated as described previously [55] from Nicotiana tabacum cv. Petit Havana SR1 [34] plants sterile- ly maintained on 1/2 MS hormone-free agar media [40]. Regenerating protoplasts were trans- formed by co-cultivation with Agrobacterium har- boring pGC4-OO or pGC4-NP [37, 66, 68]. Micro-calli embedded in agarose were cultured on liquid K3 media [41] plus sucrose (0.4 M ~ 0.05 M) supplemented with 1 mg/1 naphthaleneacetic acid (NAA), 0.2 mg/1 kinetin, 100 #g/ml kanamycin and 500 #g/ml cefotaxim (Claforan, Hoechst Chemicals). Individual trans- formed micro-calli appeared in 6-8 weeks. Only well separated micro-calli were further pro- pagated for analysis. region and polyadenylation signal from TL-DNA gene 7 [65]. The GUS gene was obtained from pRAJ275 [27]. The 2.1 kb Bam HI-Eco RI GUS cassette includes Kozak's transcriptional initiator [32] 5' to the GUS coding sequence and 3' nopaline synthase termination signal. These frag- ments were directionally ligated into pGG102 (pGA470 [ 3 ] modified to contain a Bgl II linker at the unique Hind III site (W.M. Ainley, personal communication)), a binary vector containing right and left borders of T-DNA, suitable for Agrobac- terium-mediated transformations, pGC4-OO contains the Pmasl' and Pmas2' dual promoter fragment from pOP4434 [65] inserted between CAT, GUS and protein assays Chloramphenicol acetyltransferase activity was assayed by a modification of that described by Neumann et aL [44]. A more detailed description of the CAT kinetic assay employed is given by Peach and Velten [46]. A previously reported spectrophotometric assay [28] was used to measure GUS activity using the the substrate p-nitrophenyl fl-D-glucuronide. Both CAT and GUS activity values for individual extractions were normalized to total protein content as deter- mined by the method of Bradford [7].
`
`PGR2023-00022 Page 00003
`
`

`

`52 DNA isolation and analysis Total DNA was prepared from individual callus tissues by the method of Doyle and Doyle [ 17] and further purified by cesium chloride-ethidium bromide density gradient centrifugation [35]. This procedure yielded 1-2 #g DNA per g fresh tissue weight. Total callus DNA was digested with restriction enzymes, electrophoretically se- parated on 1 ~o agarose gels and transferred to Zeta-Probe blotting membrane (Bio-Rad) for Southern hybridization analysis [ 58]. Results Activities of the Pmasl '-CA T and Pmas2'-GUS reporter genes vary extensively among clonal callus lines Transcriptional activities of the mannopine pro- moters in regenerated transformed plants are known to display tissue specificity, hormone sen- sitivity and wound inducibility [33, 47, 53, 60]. Additionally, the expression levels of transgenes within regenerated plants have been reported to show considerable, and difficult to control, envi- ronmental and developmental dependence [ 15]. Based upon the presumption that reporter gene activity within relatively homogeneous, undif- ferentiated callus tissue (grown under controlled tissue culture conditions) is less subject to envi- ronmentally and developmentally related gene regulation, we chose to use independently trans- formed, clonal tobacco callus lines for our analy- sis. Differences in reporter gene activity among individual callus clones was expected to pre- dominantly reflect 'position effect' or general inter-clonal variability in transgene expression levels. Clonal callus lines were produced by co-culti- vation of protoplasts with Agrobacterium har- boring a binary T-DNA vector containing the dual reporter gene construct and a nos promoter- NPTII kanamycin (Km) resistance marker gene (see Figure 1). Transformed protoplasts were embedded in agarose and incubated in liquid media under continuous Km selection until small, well separated micro-calli developed. Due to con- tinuous uniform exposure of the co-cultivated protoplasts to kanamycin, each separated micro- calli has a high probability of being clonally derived. The resulting micro-calli were indi- vidually propagated and assumed to be the result of an independent transformation event. To reduce the number of variables and to minimize assay inaccuracy, the activities of both reporter genes were measured from the same extract and were determined by enzyme kinetic analysis instead of single-point assays. Enzyme activities were normalized to total soluble protein in each common extract. Both the CAT and GUS assays are linear with respect to added extract and are highly reproducible, displaying standard deviations of 1.5~o of mean and 2~o of mean, respectively (a more detailed description of accu- racy of the CAT kinetic assay is published else- where [46]). The measured activities of both reporter genes within 45 clonal lines are presented in Table 1 (the values given are the mean of 2 or more indepen- dent assays). The largest observed inter-clonal differences in activities were 136-fold for CAT and 175-fold for GUS (Table 1). When different portions of the same clonal callus line were inde- pendently assayed, it was noted that essentially all the lines show a 3-4-fold variability in CAT and GUS activities (e.g. Table 2). Intra-clonal varia- bility of phenotypes within the same cell line has been reported for different traits (e.g. [5, 50]) and may result from micro-heterogeneity in general cell physiology or ploidy levels within each callus. In our case the intra-clonal transgene expression variability is much smaller in magnitude than the inter-clonal variability and, considering the hor- mone dependence of the mannopine promoters [33 ], may result from differential exposure of cal- lus to hormones within the the media. Consistent with the findings of others [30, 57], we found no correlation between observed trans- gene activity and DNA content within the clonal callus lines. Transgene DNA dosage within twelve individual callus lines was estimated by densiometric scannings of autoradiograms from
`
`PGR2023-00022 Page 00004
`
`

`

`53 Table I. Pmasl' ~ CAT and Pmas2' --, GUS activities within clonal transgenic tobacco callus lines. Callus line identifier CAT activity ~ GUS activity ~ Ratio of activities 2 (nmol min- 1 mg- l ) (nmol min- 1 mg- 1 ) GUS/CAT GC4-NP 78.99 0 NC GC4-00.1 29.52 15 0.19 GC4-00.2 41.80 285 9.65 GC4-00.3 34.86 1434 34.30 GC4-00.4 25.84 980 28.11 GC4-00.5 2.43 609 23.55 GC4-00.6 60.13 16 6.53 GC4-00.7 1.86 1298 21.58 GC4-00.8 71.51 15 8.02 GC4-00.9 98.59 272 3.81 GC4-00.10 0 485 4.92 GC4-00.11 40.91 0 NC GC4-00.12 56.94 1109 27.11 GC4-00.13 60.87 597 10.49 GC4-00.14 53.40 192 3.16 GC4-00.15 22.30 18 0.34 GC4-00.16 14.45 0 NC GC4-00.17 0 419 29.03 GC4-00.18 73.29 0 NC GC4-00.19 30.89 578 7.88 GC4-00.20 247.39 1513 48.99 GC4-00.21 97.60 14 0.66 GC4-00.22 0 625 6.40 GC4-00.23 33.89 0 NC GC4-00.24 65.39 731 21.58 GC4-00.25 0 631 9.65 GC4-00.26 65.39 0 NC GC4-00.27 23.00 471 20.49 GC4-00.28 26.55 328 12.36 GC4-00.29 62.29 1337 21.46 GC4-00.30 56.13 1423 25.35 GC4-00.31 38.16 1214 31.80 GC4-00.32 3.49 0 NC GC4-00.33 135.11 33 0.25 GC4-00.34 67.39 805 11.95 GC4-00.35 36.21 1290 35.63 GC4-00.36 49.48 849 17.16 GC4-00.37 43.26 847 19.58 GC4-00.38 54.48 2447 44.92 GC4-00.39 253.33 14 0.05 GC4-00.40 62.16 1492 24.01 GC4-00.41 0 0 NC GC4-00.42 4.67 0 NC GC4-00.43 17.36 810 46.66 GC4-00.44 36.18 881 24.34 GC4-00.45 57.80 0 NC i Values indicated by '0' were less then two times the background values nmol min - 1 mg- 1 and GUS < 4.5 nmol min ~ mg- a ). 2 GUS/CAT ratios were not calculated (NC) when one or both activities for the no promoter (GC4-NP) callus (CAT < 0.12 were zero.
`
`PGR2023-00022 Page 00005
`
`

`

`54 Table 2. Repeated GUS and CAT assays from separate extracts of callus line GC4-00.2. Callus line identifier CAT activity GUS activity Ratio of activities (nmol min- 1 mg 1) (nmol min- 1 mg- l) GUS/CAT GC4-00.2A 1 35.93 313 8.70 GC4-00.2A2 40.16 192 4.77 GC4-00.2A3 22.74 227 9.97 GC4-00.2A4 24.15 264 10.93 GC4-00.2A5 26.74 280 10.48 GC4-00.2B 1 13.52 283 20.96 GC4-00.2B2 27.59 240 8.70 GC4-00.2B3 22.54 352 15.63 GC4-00.2B4 29.40 309 10.51 GC4-00.2B5 19.74 360 18.24 Southern blots in which total callus DNA was digested with Eco RI and hybridized to the 479 bp dual promoter fragment (which detects a 2.9 kb DNA band containing all of the GUS gene, both promoters and part of the CAT gene; see Figure 1). Based upon the lack of any secondary bands on the Southern blot, none of the lines analyzed showed any evidence of gross rearrange- ment within the Eco RI fragment probed (data not shown). When the transgene dosage of the twelve clonal lines (transgene DNA content varied 20-30-fold between lines) was compared with both GUS and CAT activities, correlation coefficient of 0.172 (CAT activity vs. transgene content) and 0.116 (GUS activity vs. transgene DNA content) were obtained. For example, callus lines GC4-00.4 and GC4-00.11 both contained approximately 10-12 integrated copies of the dual promoter, yet dis- played vastly different GUS (GC4-00.4 = 980 versus GC4-00.11 = 0 nmol min-1 mg-~) and CAT (GC4-00.4 = 34.86 versus GC4-00.11 = 0 nmol min- 1 mg- 1) activities (see Table 1). Activities of the Pmasl'-CAT and Pmas2'-GUS reporter genes vary independently between clonal callus lines Somewhat surprisingly, examination of the data presented in Table 1 clearly indicates that, even discounting lines in which one or both of the activities are essentially zero, the activities of the two reporter genes show a large degree of inde- pendent variability, with the ratio of GU S to CAT in separate clonal lines ranging from 0.05 to 49 (Table 1). Not all of the inter-clonal variation between the linked reporter gene activities (corre- lation coefficient of 0.165) can be directly attri- buted to 'position effect' since repeated assays of different samples collected from the same callus tissue also show a lower level of independent variation of GUS and CAT activities (Table 2). Ratios of GUS to CAT activities ranging from 4.8 to 21 (Table 2) were observed within different samples from a single callus line (GC4-00.2), giving a correlation coefficient between GUS and CAT activities of only 0.35. Smaller numbers of repetitive assays from other clonal lines often, but not always, displayed a similar range of GUS- CAT ratios (data not shown). It would, thus, appear that enzymatic activities from the two re- porter genes can respond differentially to what- ever factors contribute to the observed micro- heterogeneity of transgene activities within the same clonal transgenic callus line. The frequency distribution of transgene expression level variability Published quantitative examinations of inter- clonal transgene expression level variability have measured reporter gene activities at the level of
`
`PGR2023-00022 Page 00006
`
`

`

`55 20 Q) C o~5 "T E [] [This paper] {Pmas2'->GUS} [] [This paper] {Pmasl'->CAT} [] Ref. [4] {Pnos->CAT} [] Ref. [14] {rbcS} [ [] Ref. [57] {legA} ~ ,~ Ref. [29] {Pcab->ocs} 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Transgene activity [relative to the mean value for each experimental group} 4.0 4.5 5.0 5.5 60 I 30 $ El.. E o "G o Fig. 2. Frequency distribution oftransgene expression variability. The number of clones displaying relative activities within the specified range have been plotted against the corresponding activity ranges. Activity values are relative to the average activity for each experimental group and are clustered into ranges of 0 to 0.249, 0.250 to 0.499, 0.500 to 0.999, etc. The line plot represents a summation of all experimental data groups. mRNA [14, 18, 19, 23, 29, 30], enzyme activities [4, 13, 22, 52, 53] and immunologically detectable protein [57]. We felt it would be of interest to compare the overall frequency distribution of these independent measurements of transgene ex- pression variability to our own data. In order to compare the different data sets, it was necessary to present transgene activity values as the ratio of each measured value to the mean for that data group. The results of the comparison are pre- sented in Fig. 2. The distributions for all the data groups (measured activities for the directly selected marker gene, NPTII [4], were not included due to probable selection bias) are quali- tatively the same, with the number of clones dis- playing each activity range decreasing steadily as relative transgene activity levels increase. Sum- mation of the results from all data groups follows essentially the same distribution as the individual data groups (Fig. 2). Discussion Transgene expression variability of reporter genes driven by the mannopine synthase (mas)promoters: inter-clonal ('position effect') verses intra-clonal variability Our results clearly indicate that, similar to results reported for many other genes introduced into plants, expression of both the co-transferred mas promoter-driven reporter genes are subject to a high degree (136-175-fold) of variability between independently transformed transgenic callus lines. Our original intent was to confirm the appli- cability of the CAT and GUS reporter genes to an analysis of the mas dual-promoter system. Clonal callus lines were chosen for this work in order to better focus on the inter-clonal 'position effect' and to avoid difficult to control tissue, develop- mental and environmental regulatory effects on gene expression levels known to occur in regener- ated plants. It is clear, however, that at least with respect to CAT and GUS activities driven by the mas promoters, even clonally derived tobacco callus is not a simple homogeneous collection of uniform cell types. Nearly all of the 45 indepen-
`
`PGR2023-00022 Page 00007
`
`

`

`56 dently transformed callus lines were found to dis- play a 3-4-fold intra-clonal variation in reporter gene expression levels superimposed upon the much larger inter-clonal variability. Our data are consistent with the overall premise of reporter gene usage, i.e. that the large variability of inter-clonal reporter gene enzyme activity is representative of differences in trans- gene transcription levels. The observed intra- clonal variability is more likely to reflect physio- logical or biochemical micro-heterogeneity within the callus lines, especially considering the hor- mone-responsive nature of at least one of the mas promoters [ 33]. There are reported precedents for micro-heterogeneity in callus gene expression characteristics (e.g. [5, 49]). In our case, because the co-transformed genes in each clonal callus line were assayed at the level of CAT and GUS en- zyme activities, intra-clonal variability could re- sult from localized (within the callus) differences in transcription, mRNA stability, translation, protein stability or overall cellular protein concen- tration (since the enzyme activities were normal- ized to total soluble protein within a common extract). The fact that the CAT and GUS activities from different samples of the same callus line failed to co-vary, essentially eliminates normalization to total soluble protein as the basis for intra-clonal variability, and indicates that lo- calized differences in the other potential mecha- nisms are not general effects, but instead differen- tially influence the two reporter genes, or their gene products. Whatever the basis of the observed intra-clonal variability, it must be taken into con- sideration when using reporter gene enzyme activities to compare transgene expression levels. Resolution of the molecular factors producing trans- gene expression level variability: transgene expres- sion from divergent promoters within 479 bp of DNA vary independently In this and previous reports, independent expres- sion level variability has been documented using several different co-transformed transgenes (Table 3). It is apparent that the molecular factors contributing to transgene expression variability can differentially affect even 5' adjacent genes such as the divergent petunia Cab genes [14, 23] and mas promoter driven CAT and GUS chime- ric genes (Table 1). The ability of co-transformed genes to exhibit coordinated transgene expression levels in plants seems to be sensitive to as yet poorly defined variables related in some way to the nature of the two promoters being compared and the location of the transgenes within an artifi- cial T-DNA [ 18, 23]. To the best of our know- Table 3. Co-transferred transgenes reported to display independent transgene expression level variability. Transgene Promoters Reporter Inter-gene distance Inter-promoter Reference arrangement genes (inclusive) 1 distance (inclusive) 2 Tandem nos CAT ca. 6 kb ca. 4.5 kb [4] ( ~ --, ) nos NPTII Tandem rbcS 301 ocs ca. 10 kb ca. 9 kb [14] ( ~ --,) rbcS 301 CAT Divergent cab 21 Cab 21 ca. 4 kb ca. 1.2 kb [18] (~ ~) cab 22 cab 22 Divergent mas 1' CAT ca. 3.5 kb ca. 0.5 kb ( ~ ~ ) mas 2' NPTII [This paper] The inter-gene distance is the length of DNA between and including both transgenes (Tandem = 5'-gene 1 to 3'-gene 2, Divergent = 3'-gene 1 to 3'-gene 2). 2 The inter-promoter distance is the minimum DNA length between and including both promoters.
`
`PGR2023-00022 Page 00008
`
`

`

`ledge, no 'locus control regions' able to eliminate 'position effect' with animal fl-globin genes [21, 24, 51, 61] have been yet identified in plants. Based upon reported data [30, 57] and limited Southern analysis of the transferred genes in our callus lines, differences in copy number or rearrangement of the transgenes cannot account for the complete lack of coordination in the ex- pression levels of the linked CAT and GUS re- porter genes. Both of mas promoters are con- tained within a 479 bp (start codon to start codon) DNA fragment (the actual regulatory sequences are more likely to be confined to the central 360 bp of DNA, Pmasl'-transcription start to Pmas2'-transcription start (unpublished data from this lab and Winter et al. [67]). It would, thus, appear that the molecular factors contri- buting to transgene variability are either able to efficiently discriminate between two adjacent pro- moters, or, alternatively, produce the observed expression level variability through changes to, or interactions with, other portions of the affected transgenes. If one assumes that modification of transgene promoter function is the primary basis of expression level variability, then the observed independent variability of the mas promoter driven reporter genes provides useful insight into potential molecular mechanisms of 'position effect'. Inherent in the term 'position effect' is the con- cept that some characteristic (or characteristics) of the genetic material in the vicinity of the trans- gene insertion site produce the observed expres- sion level variability. The high resolution of trans- gene variability suggests the phenomenon does not result from some generalized effect such as chance insertion of the foreign DNA in the neigh- borhood of general transcriptional enhancers. This conclusion is supported by the results of Dean etal. [15] in which inclusion of 23 kb of flanking plant DNA with a petunia rbcS gene failed to influence the degree of transgene expres- sion level variability observed in transformed tobacco plants. In the context of transgene variability, it is im- portant to consider that virtually all the sources of DNA used for plant genetic engineering share a 57 common lack of any predetermined chromatin structure and/or pattern of DNA modification normally provided by the parental gametes (e.g. genomic imprinting [39, 50, 54]). During the pro- cess of genetic transformation, foreign DNA enters the cell normally as DNA of prokaryotic origin and therefore lacks any of its eventual chro- matin structure of eukaryotic-specific modifica- tion (e.g. CG or CXG cytosine methylation [25]). Foreign DNA stably integrated in the nuclei of plants cells has been found to display normal chromatin structure (e.g. [11, 59]), and to re- spond to changes in DNA methylation [ 1, 26, 48, and unpublished data from our lab]. Thus, the eventual state, both biochemical and functional, of the inserted transgene must result from inter- actions between the initially 'naked' foreign DNA and host (and donor?) cell proteins and enzymes present during the process of transformation, in- tegration and eventual gene expression. It is certainly conceivable that flanking host DNA, and its current chromatin content or pattern of modification, could influence the final state of the introduced transgenic DNA. However, essen- tially random patterns of modification and/or as- sociation of incoming DNA with chromatin pro- teins during transformation, but prior to inte- gration, are also possible. Inter-clonal expression level variation could result from such random fac- tors and be essentially unrelated to the eventual site of transgene integration. Differences in either local chromatin fine-struc- ture or DNA methylation patterns have sufficient molecular resolution to functionally discriminate between the two divergent mas promoters. DNA methylation has been clearly demonstrated to be capable of affecting local patterns of gene expres- sion in native plant genes (e.g. transposon activity [6, 8, 9, 10, 36, 56] and integrated foreign DNA [38]). Apparent regulatory interaction between independently introduced foreign genes (co-sup- pression [43, 63 ]) has been correlated with DNA methylation changes within promoters [38], and must be considered in the context of transgene expression level variability. Based on currently available data, it is prema- ture to attempt to assign any specific molecular
`
`PGR2023-00022 Page 00009
`
`

`

`58 mechanism or mechanisms as the primary basis of transgene variability. It seems likely that the phenomenon of transgene expression level varia- bility or 'position effect' will be found to have multiple underlying molecular origins, at least some of which reflect higher-order gene regulatory mechanisms normally active in plants. The frequency distribution of inter-clonal transgene expression levels indicates high activity levels are considerably less common than low or no transgene activity The distribution presented in Fig. 1 clearly indi- cates that a majority of transformants express introduced transgenes at levels well below the potential maximum expression levels. This obser- vation is clearly pertinent in considering how many different transformants need to be examin- ed in order to obtain one or more clones ex- pressing the introduced gene(s) at desired (usually high) levels. The observed distribution is consis- tent with the possibility that many plant cells re- ceiving new DNA may fail to express introduced genes or express them at very low levels. In functional terms, it is possible that introduced foreign DNA either only rarely escapes inacti- vating modification or, conversely, is only rarely activated to maximal levels of expression. A more trivial explanation in which a majority of the population of low to zero expressing clones have simply failed to receive, or have reorganized, the transgene or transgenes displaying reduced activi- ty, cannot be completely ruled out without exten- sive analysis of the T-DNA structure within each clonal line. However, when such an analysis was performed on T-DNA containing a chimeric ocs gene, a group of nineteen low expressing trans- genic plants (expression levels from undetectable (13 of 19) to 0.25 times the group mean activity), contained only two lacking the predicted ocs gene structure [ 30]. It will be interesting in terms of a basic under- standing of overall plant gene regulation, and im- portant to the long-term successful application of plant genetic engineering, to better define the molecular nature of transgene expression level variability. Acknowledgements We would like to thank Angelica Duarte for enormous patience and diligence in her care of our precious callus and plant lines. C.P. was s