Modulation of the Murine CD8 Gene Complex Following the Targeted Integration of Human CD2-Locus Control Region Sequences

The human CD2 (hCD2) locus control region (LCR) inserted in the mouse CD8 gene complex activates expression of the CD8 genes in T cell subsets in which the CD8 locus is normally silenced (e.g., CD4+ single-positive T cells). In this article, we show that, in conditional mCD8/hCD2-LCR (CD8/LCR) knock-in mice, the continuous presence of the hCD2-LCR is required for this effect. Deletion of the inserted hCD2-LCR in a developmental stage and cell lineage-specific manner revealed that the temporary presence of the LCR during early development does not permanently alter the expression pattern of the CD8 genes. As a result, cells that have been affected by the insertion of the LCR can convert to their destined phenotype once the LCR is removed. DNaseI hypersensitive sites 1 and 2 of the hCD2-LCR influence the expression of the CD8 genes in a similar manner as does the full LCR, whereas insertion of hypersensitive site 3 alone of the LCR does not result in a changed expression pattern. This analysis revealed a dynamic interaction between the hCD2-LCR and the endogenous regulatory elements of the CD8 genes.

The human CD2 (hCD2) locus control region (LCR) inserted in the mouse CD8 gene complex activates expression of the CD8 genes in T cell subsets in which the CD8 locus is normally silenced (e.g., CD4 + single-positive T cells). In this article, we show that, in conditional mCD8/hCD2-LCR (CD8/LCR) knock-in mice, the continuous presence of the hCD2-LCR is required for this effect. Deletion of the inserted hCD2-LCR in a developmental stage and cell lineage-specific manner revealed that the temporary presence of the LCR during early development does not permanently alter the expression pattern of the CD8 genes. As a result, cells that have been affected by the insertion of the LCR can convert to their destined phenotype once the LCR is removed. DNaseI hypersensitive sites 1 and 2 of the hCD2-LCR influence the expression of the CD8 genes in a similar manner as does the full LCR, whereas insertion of hypersensitive site 3 alone of the LCR does not result in a changed expression pattern. This analysis revealed a dynamic interaction between the hCD2-LCR and the endogenous regulatory elements of the CD8 genes. The Journal of Immunology, 2011, 187: 3712-3720. L ocus control regions (LCRs) are cis-acting generegulatory elements functionally defined by their ability to direct expression of a linked transgene in a positionindependent, copy number-dependent, and tissue-specific manner (1). The human CD2 (hCD2)-LCR directs T cell-specific expression of linked transgenes and has been used to generate a number of useful transgenic mouse lines (2,3). Within the hCD2 gene locus, the LCR is located in the 39 flanking region of the hCD2 gene and is characterized by three DNaseI hypersensitive sites (HS1-3) (4,5). Whereas HS1 of the hCD2-LCR was shown to contain classic enhancer activity in transient-transfection assays, HS3 was shown to be essential for the protection from position effects in transgenesis (4)(5)(6). Transgenic mice carrying the hCD2 minigene linked to the full LCR achieved position-independent expression in all thymocytes and T cells (4). In contrast, transgenic mouse lines that lack HS3 of the LCR exhibited variegated CD2 expression when the transgene integrated in centromeric positions (4). However, these experiments could not answer the question whether HS3 by itself is sufficient to remodel a lymphoid locus and force its expression in all lymphoid cells, as the full hCD2-LCR does.
Random integration and frequent multiple copy integration of transgenes are two aspects of transgenesis that are difficult to control and, as a consequence, complicate direct comparison of transgenic mouse lines. Fluorescence in situ hybridization to identify the chromosome composition and chromosomal position of the transgene-integration site has been used to address this problem. Other approaches take advantage of the Cre/lox sitespecific recombination systems to reduce transgene copy numbers, as well as to allow direct comparison of transgenic lines with the same integration site but different copy numbers (7,8). This system also allows the generation of conditionally targeted mouse lines in which deletion of gene sequences is controlled in a cell lineage-specific and developmental stage-specific manner through the expression of Cre recombinase.
Three distinct developmental stages of T cell differentiation in the thymus can be defined based on the expression pattern of the coreceptors CD4 and CD8. Immature thymocytes are negative for the expression of both coreceptors and are known as doublenegative (DN) thymocytes. Commitment to the abTCR T cell lineage prompts these immature thymocytes to proceed to the double-positive (DP) stage, simultaneously expressing both CD4 and CD8 coreceptors. Subsequent commitment to either the CD4 single-positive (SP) or CD8 SP cell fate results in two mature T cell lineages in which the expression of the CD4 and CD8 coreceptors is mutually exclusive. Several cis-regulatory elements responsible for the differential turning on and off of these genes during thymocyte development have been identified. For the CD4 coreceptor, regulatory elements have been identified that are involved in both positive and negative regulation of CD4 gene expression (reviewed in Ref. 9).
The CD8 coreceptor is expressed as CD8a and CD8b heterodimers in most thymus-derived T cells. However, some cells, such as intraepithelial lymphocytes from the gut or CD8 + dendritic cells, express only CD8aa homodimers (10)(11)(12). The two CD8 chains are encoded by two distantly related, but closely linked, genes, CD8a and CD8b, which, together with regulatory elements, are required for the appropriate developmental stage-and lineage-specific expression of the CD8 receptor (13)(14)(15)(16)(17)(18)(19)(20). Their expression patterns suggest that the CD8a and CD8b genes are regulated both coordinately and independently. Recent analysis of spatial arrangement of the CD8 regulatory elements, as well as their relationship to each other, at different developmental stages, suggests relocation and clustering of these elements in CD8expressing cells, confirming the dynamic state of their chromatin during development (21).
Combining the lymphoid lineage-specific hCD2-LCR with the highly developmentally regulated murine CD8 (mCD8) gene locus allows analysis of LCR function in the context of interactions with endogenous genes within a dynamic lineage-specific environment. In this article, we describe the conditional insertion of the hCD2-LCR in the mCD8 gene locus to study the requirement for temporary or continuous presence within the genome of an LCR to effect its action. Furthermore, we compare knock-in mouse lines that differ in the composition of inserted hCD2-LCR sequences in the mCD8 gene complex. We report that the continuous presence of the hCD2-LCR is required for its effect on the expression of the CD8 genes and that cells, influenced by the LCR during development, can convert to their destined phenotype once the LCR is removed. Insertion of HS1,2 of the hCD2-LCR in the mCD8 gene locus have similar effects as the full LCR. In contrast, knockin mice with only HS3 of the LCR inserted display no overt effect on the expression pattern of the CD8 coreceptor.

Materials and Methods
Generation of mCD8-(hCD2-LCR)-targeting constructs (CD8/LCR1,2, CD8/LCR3, CD8/lxLCRlx) To generate the targeting constructs for insertion of DNaseI HS1,2 or HS3 of the hCD2-LCR, a similar approach was used to that previously described (22). A 1.3-kb HindIII-XmnI fragment (HS1,2) or a 0.8-kb XmnI-HindIII fragment (HS3) of the LCR was ligated into a HindIII/EcoRV linearized pBluescript SK plasmid containing the neomycin gene flanked with loxP sites (23). The fragments containing lx-neo-lx and hCD2-LCR sequences were lifted from their vector by digestion with SalI-NotI, blunted, and ligated into the HpaI site of a subcloned 6.3-kb BamHI fragment of the mCD8 gene complex. This genomic CD8 fragment in pBluescript SK contains DNaseI HS1 of cluster III and downstream sequences of the CD8 gene complex and provided a 3.9-kb 59 homology and a 2.4-kb 39 homology for the respective targeting constructs. A 2.7-kb BamHI-HindIII fragment containing the HSV-TK gene for negative selection was lifted from the pPNT vector (24), blunted, and ligated into the SmaI site of each particular targeting vector. Both constructs were linearized with NotI and used for embryonic stem (ES) cell transfection.
To insert the hCD2-LCR flanked by loxP sites into the mCD8 gene complex, a loxP site was ligated to the 39 HindIII site of the hCD2-LCR in a pBluescript SK plasmid. The hCD2-LCR-loxP fragment was lifted from its vector and inserted 39 of the neomycin gene flanked with loxP sites contained in a pBluescript SK plasmid (23). The fragment containing sequences of lx-neo-lx-(hCD2-LCR)-lx was lifted from its vector and inserted into the HpaI site of a subcloned 6.3-kb BamHI fragment of the mCD8 gene complex. For negative selection, the HSV-TK gene was lifted from the pPNT vector (24) and ligated into the SmaI site of the targeting vector. The vector was linearized by NotI digestion before being used for ES cell transfection.

Transfection and selection of ES cells
ES cells (cre recombinase + PC3 cells) were maintained in their undifferentiated state by growth on a feeder layer of gamma-irradiated primary embryonic fibroblasts, as described previously (25). Electroporation of 0.8 3 10 7 ES cells with 50 mg/ml linearized targeting constructs and selection of homologous recombined clones with G418 (300 mg/ml) and ganciclovir (2 mM) were carried out as previously described (24). Neoresistant ES cell clones for each targeting event were subjected to Southern blot analysis and hybridized with probes outside their homology region (22). Clones that had undergone correct homologous recombination were expanded and analyzed further using additional probes to confirm the integrity of the targeted CD8 gene complex. After karyotype analysis, to verify their euploid karyotype, two clones for each targeting event were used for blastocyst injection to generate CD8/lxLCRlx, CD8/LCR1,2, or CD8/ LCR3 chimeric mice.

Generation of mCD8-(hCD2-LCR) knock-in mice
All animal husbandry and experimental procedures were carried out according to U.K. Home Office regulations and Institute guidelines. The generated chimeric males were bred to C57BL/6 wild-type females for germline transmission, and resulting offspring were screened for neodeletion by Southern blot analysis of tail DNA. For line CD8/lxLCRlx, offspring were screened for recombination events, which deleted the floxed neomycin gene but left behind the inserted hCD2-LCR flanked by loxP sites in the CD8 gene complex. For lines CD8/LCR1,2 and CD8/LCR3, expression of the Cre recombinase in the male germline resulted in the deletion of the floxed neomycin gene, leaving behind a single loxP site next to either 1.3-or 0.8-kb hCD2-LCR sequences, respectively, in the CD8 gene complex. The Cre recombinase transgene, derived from the PC3 ES cell background, was bred out in the process of backcrossing. Comparison of mice, derived from the two independent ES clones for each CD8/LCRtargeting event, revealed similar phenotypes; therefore, backcrosses to C57BL/6 and homozygous CD8/LCR knock-in mouse lines were generated from one clone only for each line. All mice used for these studies descend from mice backcrossed for a minimum of 10 times to C57BL/6 mice. Data described below result from knock-in mice homozygous for the integration of the full or partial hCD2-LCR or from knock-in mice heterozygous for the integration of the floxed hCD2-LCR and positive or negative for a developmental stage-specific Cre transgene.

Southern blot analysis
Genomic DNA was extracted from sorted ES cell clones, tail biopsies, T cell-enriched spleen cells, or total thymus cells by phenol/chloroform extraction and ethanol precipitation. DNA samples were digested overnight with the recommended concentration of the restriction enzyme EcoRV (Roche) and resolved on 1% agarose gels. Southern blots were set up to immobilize the samples on Hybond XL nylon membranes (GE Healthcare) and hybridized with probes labeled with [ 32 P]dCTP (GE Healthcare). DNA probes specific for the CD8 locus were used to identify the targeted and wild-type or deleted CD8 allele (22).
For lineage identification of peripheral cells or bone marrow cells and CD8 expression, the following Abs were used: (allophycocyanin) anti-

Cell sorting
For the isolation of distinct T cell populations, single-cell suspensions were prepared from spleen and lymph node pooled from up to 20 mice from line CD8/lxLCRlx (cII-III). Cells were stained for anti-TCRb, CD4, CD8, and CD44 and sorted using a MoFlo cell sorter (Beckman Coulter).

Generation of floxed mCD8-(hCD2-LCR) knock-in mice
It was postulated that LCRs effect their action either by setting up active chromatin structures, but may not be required subsequently, or by obligatory continuous interactions with the regulatory elements and promoters in their proximity. To distinguish between these two possibilities, we targeted the hCD2-LCR flanked by loxP sites into the intergenic region of the endogenous mouse CD8 gene complex between DNaseI hypersensitive clusters II and III. The generated knock-in mouse line was designated CD8/lxLCRlx (cII-III) (Fig. 1A). Deletion of the inserted LCR at specific stages during hematopoietic development was achieved by crossing CD8/lxLCRlx (cII-III) knock-in mice to transgenic mice with lineage-and stage-specific expression of Cre (2, 26).

Developmental stage-specific deletion of the inserted hCD2-LCR by Cre recombination
To see the effect of deletion in hematopoietic stem cells, knock-in mice were crossed to vav-iCre transgenic mice (2). For LCR deletion in early thymocytes, knock-in mice were crossed to hCD2-iCre transgenic mice; for deletion in mature T cells, they were crossed to dLck-iCre transgenic mice (2,26).
Resulting mice, positive or negative for the Cre recombinase transgene and heterozygous for the insertion of the hCD2-LCR, were analyzed by FACS ( Fig. 1B-E, Supplemental Fig. 1). CD8/ lxLCRlx knock-in mice negative for a Cre transgene deregulate the  expression of the CD8 coreceptor, as seen previously in knock-in mice with constitutive insertion of the hCD2-LCR (Fig. 1B-E) (22).
Deletion of the floxed hCD2-LCR in hematopoietic stem cells or immature lymphocytes by expression of a vav-iCre or a hCD2-iCre transgene, respectively, results in mice in which the developmentalexpression pattern of the CD8 genes is not perturbed compared with wild-type. This indicates that the specific subsequent expression pattern of the CD8 coreceptor is not affected by the presence of the hCD2-LCR during embryonic development (Fig. 1B, Supplemental  Fig. 1A) or at early stages during thymocyte development (Fig. 1C,  Supplemental Fig. 1B).
In CD8/lxLCRlx mice positive for the transgene dLck-iCre, the deletion of the LCR is effected at a late (post-DP) stage of lymphocyte development (26). In such mice, the LCR establishes an open chromatin configuration in DN3-DN4 and CD4 + thymocytes, and these cells are positive for CD8 expression. However, upon deletion of the LCR after the positive-selection stage, a large number of the CD4 + (CD8 + ) mature thymocytes convert to a CD4 + CD8 2 phenotype (Supplemental Fig. 1C). In dLck-iCre + knock-in mice, 38% of all TCR high thymocytes are CD4 SP compared with 62% in wild-type mice (Supplemental Fig. 1C).
Peripheral CD4 + T cells from spleen of dLck-iCre + knock-in mice exhibit an intermediary phenotype for CD8 coreceptor expression between Cre negative knock-in mice and wild-type mice, which indicates a gradual ceasing of CD8 coreceptor expression in CD4 + cells (Fig. 1D, 1E). In Cre + CD4 + splenocytes 18% of cells express the CD8 coreceptor as CD8ab heterodimers compared with 77% in Cre 2 knock-in mice (Fig. 1E). This indicates that cells that have been affected by the hCD2-LCR can convert to their destined phenotype once the LCR is removed through Cremediated recombination.
The high expression level for the CD8 coreceptor in CD4 + (CD8 + ) T cells is reflected in the intense band for the targeted knock-in allele. In CD4 + (CD8 med ) T cells, the lower level of CD8 coreceptor expression is associated with a faint band for the floxed knock-in allele. This indicates that within the CD4 + (CD8 med ) population, some cells have not deleted the inserted LCR. However, a proportion of these cells seem to have lost the LCR but continue to express the CD8 coreceptor at low levels on the surface, suggesting a possible delay between deletion of the LCR and extinction of CD8 cell surface expression. In CD4 + CD8 2 cells, the absence of a band for the targeted knock-in allele corresponds to the termination of CD8 coreceptor expression in these cells (Fig. 2). The deletion of the LCR appears complete in CD8 + SP cells, since no band for the floxed knock-in allele is detectable, indicating that these cells convert to wild-type expression levels for the CD8 coreceptor, as shown by mean fluorescence intensities (Supplemental Fig. 2A).
To confirm tissue and developmental stage-specific deletion of the hCD2-LCR by the activity of the Cre recombinase, total thymocytes, T cell-enriched spleen cells, or tail tissue from CD8/ lxLCRlx knock-in mice negative or positive for the dLck-iCre transgene were analyzed (Supplemental Fig. 2B). Southern blot analysis of total thymocytes (∼90% DP cells) from dLck-iCre + knock-in mice confirmed that the hCD2-LCR is still present in immature thymocytes, indicating that the Cre recombinase has not yet been active. A weak band for the targeted allele in T cellenriched spleen cells and the strong band in tail tissue from dLck-iCre + knock-in mice prove the specificity of the Cre recombinase activity.

Influence of age on CD8 expression on CD4 cells in CD8/ lxLCRlx knock-in mice
FACS analysis of CD8/lxLCRlx mice positive for the dLck-iCre transgene identified a proportion of CD4 + cells, which maintain expression of the CD8 coreceptor, despite migration to peripheral lymphoid organs (Fig. 1D, 1E). Southern blot analysis of these CD4 + (CD8 + ) cells confirmed that these cells have not deleted the inserted hCD2-LCR (Fig. 2). To assess whether the complete deletion of the LCR and the loss of CD8 expression in the CD4 lineage are a function of time, CD4 and CD8 coreceptor expression profiles of young and old mice were compared. Such analysis revealed a decrease in the percentage of peripheral CD4 + (CD8 + ) cells with increasing age. Whereas in young (1.5-mo-old) knockin mice, ∼17% of all CD4 + cells express the CD8 coreceptor, in old (12-mo-old) knock-in mice the proportion decreases to ∼6% (Fig. 3). The persistence of CD4 + (CD8 + ) cells throughout the lifespan of dLck-iCre + CD8/lxLCRlx knock-in mice indicates that the deletion of the inserted hCD2-LCR by Cre recombinase is not complete. This is consistent with reports that recombination driven by the dLck-iCre transgene is less efficient within CD4 + lymphocytes than in CD8 + lymphocytes (26). However, peripheral CD4 + (CD8 + ) cells could represent recent emigrants from the thymus, in which deletion of the inserted LCR by Cre activity is delayed. The continuous presence, but decreasing number, of CD4 + (CD8 + ) cells with age could be attributed to the reduced thymic output due to involution of the thymus. In CD8/LCR (cII-III) knock-in mice with constitutive insertion of the hCD2-LCR, a reduction in CD4 + (CD8 + ) cells with age has been detected. Characterization of those mice indicated a link between the at-tenuated influence of the hCD2-LCR on CD8 coreceptor and the proliferation history of a cell (22). However, in dLck-iCre + CD8/ lxLCRlx knock-in mice, the decrease in CD4 + (CD8 + ) cells is more pronounced as a result of the deletion of the LCR (Fig. 3B).

Generation of CD8/LCR knock-in mouse lines with partial LCR sequences
To identify the ability of each of the DNaseI hypersensitive regions within the hCD2-LCR to overcome heterochromatinization, constructs containing either hypersensitive regions HS1,2 or only hypersensitive region HS3 of the hCD2-LCR were inserted in the mCD8 locus, and the resultant mice were compared with mice containing the full LCR (Fig. 4) (22). The knock-in lines generated were designated CD8/LCR1,2 and CD8/LCR3, respectively.

Expression of CD8 genes on cells from CD8/LCR knock-in mouse lines
Insertion of HS1,2 of the hCD2-LCR results in the expression of CD8a and CD8b in CD4 + cells, similar to mice with a full LCR knocked in the CD8 locus (Fig. 5A, 5B) (22). In both knock-in lines, $75% of CD4 + cells express the CD8 coreceptor as CD8ab heterodimers, whereas $9% of CD4 + cells express the CD8 coreceptor as CD8aa homodimers (Fig. 5C).
Interestingly, in both of these knock-in lines, some lymphocytes are CD4 SP, indicating a resistance to the activating effect of the LCR sequences on the CD8 gene complex. These CD4 SP cells seem to follow their normal developmental program. HS1,2 of the hCD2-LCR, which can be subject to position effects in transgenesis (4), seem sufficient to deregulate the CD8 gene complex in the majority of cells within a lineage, which is normally programmed to turn off CD8 expression.
In contrast, HS3 sequences of the hCD2-LCR seem insufficient for overcoming the endogenous regulation of the CD8 gene complex, because no deregulation of CD8 coreceptor expression was observed in peripheral lymphocytes from CD8/LCR3 knock-in mice. This suggests that HS3 of the LCR, which is essential for the protection of position effects in transgenic mice (4), is unable, by itself, to influence the CD8 gene complex.
Enhanced expression of the CD8a and CD8b-chain, as measured by mean fluorescence intensity, was observed in CD8 SP cells in peripheral lymphocytes from knock-in lines CD8/LCR (cII-III) or CD8/LCR1,2, but not from knock-in line CD8/LCR3. No change in ratio between CD4 + cells and CD8 SP cells was observed, indicating that the insertion of hCD2-LCR sequences does not influence the cell lineage decision of immature lymphocytes. Mature thymocytes, identified by high abTCR expression, from each CD8/ LCR knock-in line show a similar CD8 expression profile as observed in peripheral T cells (data not shown).
Expression of transgenes under the regulation of the hCD2-LCR was shown from the early stages of thymocyte development onward (2). By the DN3 stage, almost all cells from knock-in lines CD8/LCR (cII-III) and CD8/LCR1,2 express the CD8a-chain. However, in cells from knock-in line CD8/LCR3 no expression of the CD8 coreceptor is detected. In thymocytes at the DN4 stage almost 100% from knock-in lines CD8/LCR (cII-III) and CD8/LCR1,2 express the CD8a coreceptor (Fig. 6A) (mainly as CD8ab heterodimers; data not shown). CD8 expression detected in thymocytes from wild-type and CD8/LCR3 at the DN4 stage is attributed to immature CD8 SP cells most likely included in the cell preparation due to the staining protocol. This analysis of immature thymocytes revealed that HS1,2 of the LCR is sufficient to overrule the regulation of the CD8 coreceptor during early T cell development. However, HS3 of the LCR alone is unable to initiate expression of the CD8 coreceptor (Fig. 6A).
To assess how T cells, which do not normally open the chromatin of the CD8 gene complex during their development, are affected by the inserted hCD2-LCR sequences, gdTCR T cells from thymus and spleen of the CD8/LCR knock-in mice were analyzed (Fig.  6B, Supplemental Fig. 3A).
In two knock-in lines, lines CD8/LCR (cII-III) and CD8/ LCR1,2, a large proportion of gdTCR + cells from thymus ($91%) and spleen ($74%) express the CD8 coreceptor either as CD8aa homodimers or CD8ab heterodimers. Interestingly, both knock-in lines have a significant population of gdTCR + CD8 2 cells in thymus and spleen, suggesting that some cells are able to elude the influence of the inserted hCD2-LCR sequences (Supplemental Fig. 3A). In contrast, gdTCR + cells from knock-in line CD8/LCR3 show no increase in CD8 coreceptor-expressing cells containing only a small population of gdTCR + CD8 + cells in thymus and spleen, similar to the ones observed in wild-type mice. These data suggested the HS3 sequences of the hCD2-LCR are unable to initiate CD8 coreceptor expression in cells, which normally do not express CD8. However, HS1,2 sequences of the LCR or the complete hCD2-LCR can initiate expression of the CD8 coreceptor in a large proportion of gdTCR T cells, which are not programmed to express the CD8 genes.
Analysis of non-CD8 coreceptor-expressing lymphoid and myeloid cell lineages by flow cytometry revealed the inserted hCD2-LCR sequences cannot activate CD8 expression in these lineages (data not shown). This indicated that the effect of the LCR is probably restricted to T lineage lymphoid cells.

Age-related loss of hCD2-LCR-dependent CD8 expression on CD4 cells
To determine whether individual HS regions of the hCD2-LCR are responsible for the loss of LCR influence on the CD8 gene complex, FIGURE 5. CD8 coreceptor expression in peripheral lymphocytes from CD8/LCR knockin mouse lines. A and B, CD8 coreceptor expression on abTCR + spleen cells from CD8/ LCR (cII-III), CD8/LCR1,2, and CD8/LCR3 knock-in mice analyzed by plotting CD4 against CD8a or CD8b, respectively. C, The CD8a and CD8b coreceptor profile on CD4 + T cells.
cohorts of young and old knock-in mice from line CD8/LCR1,2 and CD8/LCR3 were compared with CD8/LCR (cII-III) mice (Supplemental Fig. 3B). These experiments showed a similar increase of CD4 + CD8 2 lymphocytes with age, as previously shown for mice with the full hCD2-LCR inserted in the CD8 locus (22).
Knock-in mice with only HS3 sequences of the LCR inserted (CD8/LCR3) show no change in their CD4 and CD8 expression profile with age. We conclude from this analysis that the information contained within the HS1,2 sequences of the LCR is sufficient to exert the full effect of the hCD2-LCR on CD8 coreceptor expression.
To follow the loss of CD8 expression on CD4 cells more precisely, we sorted peripheral CD4 + (CD8 + ) T cells from either knock-in line CD8/LCR (cII-III) or CD8/LCR1,2 and transferred them into Rag2 2/2 mice (Supplemental Fig. 4). The transferred cells are induced into rapid proliferative expansion within the lymphopenic environment of the host. As described previously, a loss of CD8 expression was observed in donor-derived CD4 + T cells (22); however, no significant difference between the knock-in lines with full or HS1,2 hCD2-LCR sequences was detected. This suggests that, when inserted within the CD8 gene complex, HS1,2 sequences implement a similar effect to the full length hCD2-LCR (Supplemental Fig. 4).

Discussion
Lineage-and developmental stage-specific deletion of the inserted hCD2-LCR Targeted insertion of an LCR as a single entity into a defined genomic region has been shown to influence the expression of endogenous genes (22,27). Using a Cre recombinase-based approach, we addressed the question of whether the continuous presence of an LCR is required to exert this effect.
No effect on the CD8 coreceptor-expression profile was observed in CD8/lxLCRlx knock-in mice when the LCR was deleted during early development by the expression of a vav-iCre or hCD2-iCre transgene, respectively (Fig. 1B, 1C). This suggests that the temporary presence of the LCR during early hematopoietic development (stem cell stage and early lymphoid development, respectively) does not influence the subsequent developmental stage-specific expression pattern of the CD8 gene.
Cell lineage-specific deletion of the inserted hCD2-LCR at a later lymphoid-specific stage, after the effects of the LCR have been manifested, causes a conversion of the deregulated CD8 coreceptor expression to its wild-type expression pattern. The expression of the CD8 coreceptor in CD4 + cells was shown to be directly linked to the presence of the hCD2-LCR (Fig. 2). Taken together, these data indicated that the inserted hCD2-LCR does not "indelibly mark" the CD8 gene complex and that the continuous presence of the LCR is required for the deregulation of the normal pattern of CD8 expression. Cells whose expression of the CD8 coreceptor is influenced by the inserted hCD2-LCR can convert to their destined phenotype after deletion of the LCR.
A declining influence of the LCR on the expression of the CD8 coreceptor, which correlates with the proliferation history of a cell, was reported for CD8/LCR knock-in mice with a constitutively inserted LCR (22). Such a loss in LCR activity was also observed in CD8/lxLCRlx +/2 /cre 2 knock-in mice. Although it cannot be excluded that some CD4 + CD8 2 lymphocytes from CD8/lxLCRlx +/2 / dLck cre + knock-in mice no longer express the CD8 coreceptor due to such an inactivation of the LCR, the majority of cells lose CD8 coreceptor expression as a result of LCR deletion (Fig. 3).
Targeted deletion of the endogenous mouse b-globin LCR to address the role of the LCR in its native locus revealed a contributory, rather than exclusive, role in transcriptional regulation. The dominant role of the b-globin LCR was described as conferring high-level transcription of the globin genes, whereas its role in establishing and maintaining an open chromatin configuration was dispensable (28). In contrast, in gb-thalassemia patients, the deletion of the b-globin LCR resulted in an inactive chromatin configuration of the b-globin genes in vivo, which prevented expression of the globin genes (29). In the context of ectopic integration of a single LCR, the continuous presence of the LCR seems essential for the influence on endogenous genes, as shown in the generated CD8/lxLCRlx knock-in mice.
When used to drive expression of a linked transgene, the LCR acts in concert with the promoter of the transgene to achieve tissuespecific and position-independent expression (30). However, expression of endogenous genes adjacent to the integration site of a transgenic construct can also be influenced by the inserted LCR. In CD8/LCR knock-in mice, the inserted hCD2-LCR interacts with the promoters of the endogenous CD8 genes and influences the developmentally regulated expression of these genes. As a consequence of the presence of the LCR, some cell lineages (e.g., CD4 + cells) that are normally programmed to silence the CD8 genes at a certain stage of development are prevented from doing so. However, as shown previously, the endogenous CD8 generepressive regulatory processes in these cells appear to continuously contest the influence of the inserted LCR, which can result in the eventual inactivation of the LCR (Fig. 3) (22). Therefore, deletion of the inserted LCR by Cre-mediated recombination facilitates the conversion of these cells to their destined phenotype. This apparent reversibility of the LCR influence indicates that the endogenous CD8 genes are not permanently marked by the temporary presence of the LCR and that the continuous presence of the LCR is essential for the deregulation of the endogenous CD8 genes. Influence of partial hCD2-LCR sequences on the mCD8 gene complex In addition, to further functionally characterize the specific regions of the hCD2-LCR, we generated knock-in mice with partial LCR sequences inserted in the mCD8 locus. The three defined DNaseI HSs (HS1-3) that make up the hCD2-LCR have been extensively characterized in transgenic mice (4,6). The generation of CD8/ LCR knock-in mice with partial hCD2-LCR sequences allowed further characterization of the DNaseI HSs of the LCR when inserted in a defined locus, making possible the direct comparison of their influence on the endogenous CD8 genes, unlike in transgenic mouse models where random construct integration often obfuscates analysis (4,8).
Whereas lineage-specific and position-independent transgene expression requires the full LCR (4), HS1,2 of the hCD2-LCR seem to be sufficient for this effect in CD8/LCR knock-in mice. In both knock-in lines, CD8/LCR (cII-III) and CD8/LCR 1,2, peripheral CD4 + cells are prevented from closing the CD8 locus, expression of the CD8 coreceptor is initiated in immature thymocytes, and the majority of gdTCR + cells are forced to express the CD8 coreceptor (Figs. 5, 6, Supplemental Fig. 3A). Furthermore, in CD8/LCR 1,2 mice, a similar resistance to the inserted LCR sequences was observed as in CD8/LCR (cII-III) mice, which is shown in the decline of CD4 + (CD8 + ) cells with age and after adoptive transfer (Supplemental Figs. 3, 4). Interestingly, the insertion of HS3 alone of the LCR does not result in a changed expression of the endogenous CD8 genes, indicating that, as an individual region, HS3 is unable to influence the regulation of adjacent genes under the circumstances described in this article. It is possible that heterochromatinization at centro-or telomeric sites is different from the one occurring in T cells that close-down the CD8 gene complex. Whereas HS3 is essential to avoid the former type of heterochromatinization, it may be dispensable in a more favorable environment, such as the nucleus of T lymphocytes.
The analysis of CD8/LCR knock-in mice complements the characterization of LCRs using transient and transgenic assays. It was reported that LCRs contain a core sequence with enhancer activity in transient assays but that does not function as an LCR in transgenic assays. However, when additional flanking elements, without measurable activity on their own, are included with the core sequence, LCR activity is observed (reviewed in Ref. 31).
For the hCD2-LCR, HS1,2 contain the core sequence with enhancer function, and HS3 is the flanking sequence, which is required for complete LCR function in transgenic mice. In CD8/ LCR knock-in mice, the core sequence of the hCD2-LCR influences the regulation of the CD8 genes in the same manner as does the full-length LCR, indicating that inclusion of HS3 has no additional/synergistic effect. This suggests that, when inserted as an isolated unit, the core sequence of the LCR contains all functional elements needed to influence the regulation of the endogenous CD8 genes. However, to confirm this suggestion, other endogenous gene loci adjacent to inserted LCR sequences need to be analyzed.
In the described CD8/LCR knock-in mice, interaction of the inserted LCR sequences with the adjacent gene-regulatory regions of the endogenous CD8 genes was assessed by examining the expression pattern of the CD8 coreceptor. The activity of the hCD2-LCR coincides with the establishment of hypersensitivity and accessibility of the endogenous CD8 locus before the onset of CD8 transcription (2,32). Therefore, the initiation of CD8 coreceptor expression in immature thymocytes at the DN3 stage in CD8/LCR knock-in mice coincides with this preparative stage of the chromatin environment of the CD8 gene locus. Thus, the LCR initiates expression in DN thymocytes at the same time that the CD8 locus is being prepared for normal CD8 expression (32). This preparative stage of the CD8 locus chromatin region may explain the readiness of immature thymocytes to express the CD8 receptor in the presence of specific LCR sequences (Fig. 6A). Chromatin accessibility may also explain why the partial LCR sequences HS1,2 effect the same phenotype as does the full-length LCR.
Bivalent chromatin as the basic mechanism that influences gene regulation at key points in thymocyte differentiation-and lineagefate decisions has been suggested by Harker et al. (32) and others (33)(34)(35). Remodeling of the CD8 promoter was reported at the DN3 stage (32), and this coincides with the completion of lineage divergence of the ab and gd T cell lineages (reviewed in Ref. 36). This overlap of lineage-commitment decision and poised chromatin accessibility of the CD8 gene complex may explain the increase in CD8 coreceptor-expressing gdTCR + thymocytes observed in knock-in lines CD8/LCR (cII-III) and CD8/LCR1,2 as opposed to what occurs in wild-type mice (37,38). Interestingly, expression of the CD8 coreceptor in gdTCR + thymocytes from CD8/LCR knock-in mice can occur in the CD8ab heterodimer or CD8aa homodimer form, which is also observed in the peripheral gd T cell lineages of these mice. Some cell lineages (e.g., gdTCR intraepithelial lymphocytes and dendritic cells) express only the CD8a gene; it will be interesting to study the influence of the hCD2-LCR sequences on the expression of the CD8b gene in these cells.
The absence of a measurable influence of the inserted hCD2-LCR sequences in non-CD8-expressing hematopoietic cells indicates that the interaction of the LCR with gene-regulatory regions of the endogenous CD8 gene complex is restricted to cells descending from lymphoid lineage progenitors.
Taken together, our results showed that cells that have been affected by the inserted hCD2-LCR can convert to their destined fate upon deletion of the LCR. However, CD8 coreceptorexpression analyses in hematopoietic lineages suggest that the influence of inserted LCR sequences is largely dependent on chromatin accessibility of the CD8 gene complex in lymphoid cells. During early developmental stages (e.g., immature thymocytes) and during lineage commitment, when remodeling of the CD8 gene complex occurs, interference by the inserted LCR seems to be unrestricted; however, its influence seems to weaken throughout the life span of a cell.