In most cases, both alleles of autosomal genes are co-expressed. In recent years random monoallelic expression (RME) has emerged as an important exception that may apply to ~0.5–10% of expressed genes in a given tissue and has been characterized as the autosomal analog of X-inactivation (Gendrel et al., 2016). In RME, different cells of a given type express only one allele, both or neither, and this expression pattern is mitotically stable. Notably, RME genes do not share an overarching unifying feature or function (Deng et al., 2014; Eckersley-Maslin and Spector, 2014; Gendrel et al., 2016; Gimelbrant et al., 2007; Reinius et al., 2016; Xu et al., 2017), and the biological role for RME is in most cases not known. RME is distinct from X-inactivation, genomic imprinting, and allelic exclusion of antigen receptor and odorant receptor genes, in that biallelic expression occurs at an appreciable frequency, and expression is largely stochastic rather than imposed by strict feedback regulatory mechanisms (Gendrel et al., 2016).

The molecular determinants of RME are poorly understood, in part because of difficulties analyzing single primary cells (Reinius and Sandberg, 2015). Chromatin analysis of RME alleles in primary populations has not been possible due to the difficulty of isolating pure cell populations ex vivo with defined RME expression patterns (Gendrel et al., 2016; Xu et al., 2017). As a result, previous analyses have been limited to clonal cell lines derived from F₁ hybrids, where allelic expression is known and clonally stable (Eckersley-Maslin et al., 2014; Gendrel et al., 2014; Levin-Klein et al., 2017; Xu et al., 2017). These analyses revealed that enhancers of RME genes are constitutively accessible irrespective of gene or allelic expression status, whereas promoters are accessible only at active alleles (Levin-Klein et al., 2017; Xu et al., 2017). Therefore, promoter accessibility, rather than enhancer opening and activation, might be the ‘gatekeeper’ of RME, whereas enhancers, being constitutively open, were proposed to permit rather than impose expression of RME alleles (Xu et al., 2017).

Enhancers may play more than a permissive role in RME, however, in light of evidence that enhancers primarily influence the probability of mitotically stable expression, rather than the amount of expression per cell (Walters et al., 1995; Weintraub, 1988). In fact, deletions of enhancers resulted in mitotically stable gene variegation in both cell lines and normal tissues—notably Igh in B cell hybridomas and Cd8a in primary thymocytes, among others (Ellmeier et al., 2002; Garefalaki et al., 2002; Ronai et al., 1999; Sleckman et al., 1997; Walters et al., 1995; Weintraub, 1988; Xu et al., 1996). Collectively, these data support the binary or ‘on/off’ model of enhancer action (Blackwood and Kadonaga, 1998; Fiering et al., 2000), where an increase in enhancer activity at a genetic locus results in an increase in the probability of gene expression, rather than an increase in expression per cell. Conversely, weak or reduced enhancer activity results in a lower likelihood of expression, but the cells that express the gene express a similar amount of gene product.

We reasoned that the binary action of enhancers—when limiting—might be a driving principle of RME, and sought to test this in examples of RME with a clear biological purpose: the Klra genes (which encode the Ly49 family receptors) and the Klrc1 gene (which encodes the NKG2A receptor) (Chess, 2012; Eckersley-Maslin and Spector, 2014; Gendrel et al., 2016). These genes, clustered in a ~ 1 Mb stretch of the NK gene complex (NKC) on chromosome 6 in mice, encode cell surface receptors expressed by NK cells that bind MHC I molecules. They are expressed in a variegated (Raulet et al., 1997; Yokoyama et al., 1990), monoallelic (Held et al., 1995), stochastic and largely mitotically stable fashion (Raulet et al., 2001), resulting in subpopulations of NK cells that express random combinations of the receptors and consequently exhibit distinct reactivities for cells expressing different MHC I molecules. Regulation of each gene allele is independent, and expression of one Klra gene has minimal effects on expression of others (Tanamachi et al., 2001). While a clear biological purpose for RME at many genes is lacking, RME at Klra genes underlies the basis of the ‘missing self’ mode of NK cell target recognition (Kärre et al., 1986). Furthermore, the system represents a powerful in vivo genetic model of RME, where allelic expression states can be easily assessed at the population level in primary cells using allele-specific antibodies that we and others previously generated (Tanamachi et al., 2001; Vance et al., 2002), circumventing previous technical limitations to studying RME in single primary cells.

Importantly, competition between Klra genes for interaction with a shared enhancer or locus control region is not required for variegation of Klra genes, as a Klra1 (encodes Ly49A) genomic transgene ectopically integrated in different genomic sites was usually expressed with a frequency similar to that of the native Klra1 gene (~17% of NK cells) (Tanamachi et al., 2004). We previously identified a key DNase I hypersensitive element, Klra1_Hss1, ~ 5 kb upstream of the Klra1 gene that is conserved in other Klra genes and required for expression of the Klra1 transgene (Tanamachi et al., 2004).

Our central hypothesis is that enhancers, rather than simply being permissive for RME, both limit and directly control the probability of expression of Klra genes—and RME alleles generally—in a stochastic and binary fashion. Binary enhancer action, when limiting, may represent a causal mechanism of RME, explaining the pervasiveness of RME across genes and cell types. We have carried out genetic dissection and population analyses to demonstrate that enhancers control the probability of allelic expression and have provided a more general model of the role of enhancers in RME as well as in other developmentally regulated genes.

The Klra1_Hss1 element was previously reported to be a ‘switch’ element active only in immature, Ly49-negative NK cells (Saleh et al., 2004). Our extensive analysis herein demonstrated that Hss1 displays properties of enhancers in mature cells, consistent with the conclusions of others based on reporter analysis (Gays et al., 2015). Furthermore, the loss of Ly49G2 expression after deletion of Klra7_Hss1 that we have documented in mature Ly49G2+ cells is inconsistent with a solely developmental role of Hss1, and further supports its enhancer identity in mature NK cells. Finally, our results showed that variegation arises and is modulated by enhancer deletion (including Klra7_Hss5 and Klrk1_5′E) rather than introduction of variegating switch elements.

The main significance of our results is to link RME with the binary model of enhancer action, placing previous observations of enhancer deletion-associated variegation in the context of the pervasive and naturally-occurring RME phenomenon. Remarkably, deletion of an enhancer upstream of the Klrk1 gene imparted an RME expression pattern that fully recapitulated the stochasticity, mitotic stability and promoter accessibility features of naturally variegated NK receptor genes. In this instance, enhancer-like elements downstream of Klrk1 may suffice to impart the lower frequency of expression. The striking commonalities of enhancer deletion variegation and natural variegation of NK receptor genes and other RME genes argues that RME is an extreme manifestation of the inherent probablistic nature of stable gene activation rather than a specialized mechanism to impose a variegated expression pattern.

The data also reveal the quantitative impact of enhancer strength on allelic expression frequencies. Deletion of Klra7_Hss5, a relatively minor and constitutively accessible enhancer, reduced the frequency of expression of Klra7, a natural RME gene, directly tying the enhancer deletion-associated variegation phenomenon to RME. This result powerfully argues that enhancers are not simply permissive for expression at RME genes, but are also instructive regarding expression probability. We hypothesize that the broad range of frequencies with which different Klra genes are naturally expressed (~5–60%) in large part reflects differences in enhancer strength. Probabilistic enhancer action has also been documented in Drosophila, where regulation of the gap genes by multiple enhancers has been suggested to ensure a high probability of gene expression, and removal of one of multiple enhancers was shown to increase gene ‘failure rate’ (Perry et al., 2011). We propose that the binary decision to express a gene is regulated by quantitatively varying enhancer activity, which is comprised of both the number of enhancers acting upon a gene and the strength of individual enhancers within that set. Our results are consistent with recent findings that enhancers are probabilistic regulators of transcription burst frequency rather than burst size (Bartman et al., 2016; Larsson et al., 2019). How enhancer control of the probability of stable gene expression interfaces with the control of transcription burst frequency is an exciting area for future investigation.

Our data showing RME of the lineage-defining Klrk1, Ptprc, Cd8a, and Thy1 genes support the generality of probabilistic gene expression, and suggest that RME is even more prevalent than the previous estimates (Eckersley-Maslin et al., 2014; Gendrel et al., 2014; Nag et al., 2013; Reinius et al., 2016). Although RME has been associated previously with poorly expressed genes (Gendrel et al., 2014; Reinius et al., 2016), our results extend the phenomenon to relatively highly expressed genes. We propose that genes lie along a spectrum of allelic failure rates that are largely controlled by enhancer strength, with documented RME genes on the highest end of that spectrum. RME applies to genes encoding cell surface receptors (Klra, Klrc1, Klrk1 etc.) and to a gene encoding a transcription factor, Bc11b (Ng et al., 2018). It applies as well to CpG poor promoters (the aforementioned receptor genes) and to genes that harbor promoter proximal CpG islands (Thy1 and the gene encoding Bcl11b Ng et al., 2018). High-resolution genome-wide approaches will eventually provide a comprehensive picture of the full extent of RME.

We predict that the mechanism of mitotic stability of active and silent RME alleles is likely related to maintenance of gene expression states broadly, and may involve bookmarking of promoters (Xu et al., 2017), rather than repressive chromatin at silent alleles. Indeed, we found that expressed and silent Klra7 alleles are distinguished only by the presence of active marks at the promoters and gene bodies of active alleles. Enhancers are constitutively accessible and activated on expressed and silent alleles alike, consistent with the decoupling of enhancer and promoter accessibility seen at RME loci generally. The absence of a repressive chromatin state on silent NK receptor alleles suggests that the RME ‘off state’ can reflect a stably inactive, as opposed to repressed, chromatin state. Numerous lineage non-specific genes that are also silent in NK cells (e.g. Cd19, Cd3e) also lacked traditional repressive chromatin modifications, yet are generally not subject to subsequent activation after an initial failure to be activated. In the case of the NK receptor genes and RME genes broadly it appears that the inactive state is maintained in mature cells in spite of continued enhancer activation, suggesting silent promoters are no longer competent for activation—perhaps due to lack of critical promoter-activating pioneer factor activity or reduced nucleosome remodeling or ‘promoter opening’ capacity that is only present at sufficient levels for developmentally regulated gene activation during differentiation. The relevant biochemical activity is likely a property of general rather than gene-specific factors, given the apparent pervasiveness of RME even among lineage-defining genes.

The results, in concert with previous studies, provide strong evidence that the principles underlying enhancer deletion-associated variegation and RME are broadly applicable to the regulation of many if not all genes. A possible mechanistic explanation of the probabilistic nature of gene expression in RME is that the initially inactive promoters present an energetic threshold that must be overcome to stably activate gene expression, and that enhancer strength determines the probability of overcoming this threshold. Once overcome, the active state is largely stable even in the absense of the initial signal or enzymatic activity, which may be restricted to a cellular differentiation window. The energy barrier may be due to the initially chromatinized state of the promoter and the energy required for nucleosome remodeling of promoters. It may be this property that is the measure of ‘enhancer strength’, that is, the capacity of the enhancer to overcome the energy threshold presented by the promoter within a limited window during differentiation. The number of enhancer elements in a gene, the availability of relevant transcription factors that bind the enhancers and the affinity with which the enhancer binds the factors are all likely relevant in determining enhancer strength. It remains possible that in some cases of RME, additional modifications of the promoter may contribute to the energy barrier as well.

Differential DNA methylation of alleles is not broadly associated with all or even most RME genes (da Rocha and Gendrel, 2019; Eckersley-Maslin et al., 2014), nor does it appear to be responsible for maintenance of silent alleles in the majority of tested RME genes (Eckersley-Maslin and Spector, 2014; Gupta et al., 2022; Marion-Poll et al., 2021). Silent Klra and Klrc1 alleles in mouse NK cells were reported to be hypermethylated relative to active alleles (Rouhi et al., 2007; Rouhi et al., 2006; Rouhi et al., 2009), and our data show that silent Klrk1 alleles in 5′EΔ/5′EΔ NK cells are also hypermethylated. It is doubtful that DNA methylation is sufficient to maintain the silent state, however, since inhibitors of DNA methyltransferases failed to derepress inactive Klra or Klrc1 genes in cell lines (Rouhi et al., 2007; Rouhi et al., 2006; Rouhi et al., 2009), or other silent RME genes that were hypermethylated (Eckersley-Maslin and Spector, 2014; Gupta et al., 2022; Marion-Poll et al., 2021). In light of the collective data on DNA methylation and silent alleles in RME we suspect it is not generally causative in maintaining RME. Furthermore, the Klra, Klrc1, and Klrk1 genes are all hypermethylated on both alleles in progenitor hematopoietic stem cells, suggesting that demethylation is associated with gene activation as opposed to a model where alleles are randomly silenced by de novo DNA methylation. The possibility remains that for these genes, preexisting DNA methylation contributes to the energy barrier that must be overcome by enhancer action, and that demethylation accompanies de novo gene activation. DNA methylation is not a requisite feature of RME promoters, however, as many RME genes are hypomethylated on both alleles.

The RME of the NK receptor genes resembles the monoallelic expression pattern of cytokine genes including Il2, Il4, Il5, Il10, and Il13 (Bix and Locksley, 1998; Gendrel et al., 2016; Kelly and Locksley, 2000; Rivière et al., 1998). The cytokine genes are inducible in response to TCR stimulation and therefore expression is inherently unstable, but impressive stability over several mitotic divisions was observed for Il4 (Bix and Locksley, 1998), and the probability of Il4 allelic activation and biallelic expression correlated with the strength of the inducing signal (Rivière et al., 1998). Intriguingly, the Il4 and Il13 genes are closely linked and are co-regulated by an enhancer, CNS1, that was found to be constitutively acetylated at histone H3, and thus permissive for expression (Guo et al., 2005). Expression of the Il4 and Il13 gene alleles was independent, however, in a manner strikingly similar to the NK receptor genes. It is probable that the general principles uncovered by us and others apply to genes that are induced via stimulation as well as genes whose expression is acquired during differentiation. Importantly, enhancer deletions have been observed to reduce the stochastic expression probabilities of the monoallelically expressed odorant receptor genes (Khan et al., 2011) and trace amine-associated receptors of the olfactory epithelium (Fei et al., 2021). It therefore seems likely that the probabilistic nature of gene activation and binary model of enhancer-promotor communication accounts for the initial stochastic step across many or even all types, even if they might differ with respect to feedback and allelic exclusion mechanisms.

Apparently, RME often occurs at such a low rate that it is both beneath ready detection and presumably irrelevant for the function of a cell lineage. In the case of NK receptors, we propose that evolution has exploited RME to generate a complex combinatorial repertoire of NK cell specificities. More speculatively, by regulating expression of fate-determining mediators, RME may underlie stochastic cell fate decisions in some instances of cellular development (Ng et al., 2018). From an evolutionary perspective, appreciable RME of a gene could arise by mutation of strong enhancers of a precursor gene, by providing a new gene with a weak enhancer, or by diminishing the concentration of relevant enhancer-binding transcription factors in a given lineage of cells (as has been shown for several relevant TFs for the Klra genes) (Ioannidis et al., 2003; Ohno et al., 2008).

Finally, our results suggest that allelic failure rates may in some cases dwarf the rates of null alleles generated by somatic mutation. As a novel mechanism of genetic haploinsufficiency at the cellular level, RME might have broad implications in genetic disease etiology and penetrance of disease phenotypes in heterozygous individuals.

Sequencing data have been deposited in GEO under the accession code GSE181197.

1. 1. Muljo SA
   2. Raulet DH

   (2021) NCBI Gene Expression Omnibus

   ID GSE181197. Binary outcomes of enhancer activity underlie stable random monoallelic expression.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE181197

1. 1. Lara-Astiaso D
   2. Weiner A
   3. Lorenzo-Vivas E
   4. Zaretsky I
   5. Adhemar Jaitin D
   6. David E
   7. Keren-Shaul H
   8. Mildner A
   9. Winter D
   10. Jung S
   11. Friedman N
   12. Amit I

   (2014) NCBI Gene Expression Omnibus

   ID GSE59992. Chromatin state dynamics during blood formation (ATAC-seq).

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE59992
2. 1. Lara-Astiaso D
   2. Weiner A
   3. Lorenzo-Vivas E
   4. Zaretsky I
   5. Adhemar Jaitin D
   6. David E
   7. Keren-Shaul H
   8. Mildner A
   9. Winter D
   10. Jung S
   11. Friedman N
   12. Amit I

   (2014) NCBI Gene Expression Omnibus

   ID GSE60103. Chromatin state dynamics during blood formation.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE60103
3. 1. Levanon D
   2. Negreanu V
   3. Lotem J
   4. Bone KR
   5. Brenner O
   6. Leshkowitz D
   7. Groner Y

   (2014) NCBI Gene Expression Omnibus

   ID GSE52625. Runx3 Regulates Interleukin-15-Dependent Natural Killer Cell Activation.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE52625
4. 1. Shih HY
   2. Sciume G
   3. Mikami Y
   4. Guo L
   5. Sun HW
   6. Brooks SR
   7. Urban JF
   8. Davis FP
   9. Kanno Y
   10. O'Shea JJ

   (2016) NCBI Gene Expression Omnibus

   ID GSE77695. Developmental Acquisition of Regulomes Underlies Innate Lymphoid Cell Functionality.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE77695
5. 1. Shih HY
   2. Sciume G
   3. Mikami Y
   4. Nagashima H
   5. Sun HW
   6. Brooks SR
   7. Jankovic D
   8. Yao C
   9. Villarino A
   10. Davis FP
   11. Kanno Y
   12. O'Shea JJ

   (2020) NCBI Gene Expression Omnibus

   ID GSE145299. Rapid remodeling of poised chromatin landscapes and transcription factor repurposing facilitate gene induction in natural killer cells.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE145299
6. 1. Zhou J
   2. Liu D

   (2021) NCBI Gene Expression Omnibus

   ID GSE167237. DNA methylation regulates haematopoietic development.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE167237

1. 1. Bartman CR
   2. Hsu SC
   3. Hsiung CC-S
   4. Raj A
   5. Blobel GA

   (2016) Enhancer Regulation of Transcriptional Bursting Parameters Revealed by Forced Chromatin Looping

   Molecular Cell 62:237–247.

   https://doi.org/10.1016/j.molcel.2016.03.007

   • PubMed
   • Google Scholar
2. 1. Bix M
   2. Locksley RM

   (1998) Independent and epigenetic regulation of the interleukin-4 alleles in CD4+ T cells

   Science 281:1352–1354.

   https://doi.org/10.1126/science.281.5381.1352

   • PubMed
   • Google Scholar
3. 1. Blackwood EM
   2. Kadonaga JT

   (1998) Going the distance: a current view of enhancer action

   Science 281:60–63.

   https://doi.org/10.1126/science.281.5373.60

   • PubMed
   • Google Scholar
4. 1. Buenrostro JD
   2. Giresi PG
   3. Zaba LC
   4. Chang HY
   5. Greenleaf WJ

   (2013) Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position

   Nature Methods 10:1213–1218.

   https://doi.org/10.1038/nmeth.2688

   • PubMed
   • Google Scholar
5. 1. Calo E
   2. Wysocka J

   (2013) Modification of enhancer chromatin: what, how, and why

   Molecular Cell 49:825–837.

   https://doi.org/10.1016/j.molcel.2013.01.038

   • PubMed
   • Google Scholar
6. 1. Chess A

   (2012) Mechanisms and consequences of widespread random monoallelic expression

   Nature Reviews. Genetics 13:421–428.

   https://doi.org/10.1038/nrg3239

   • PubMed
   • Google Scholar
7. 1. Coles MC
   2. McMahon CW
   3. Takizawa H
   4. Raulet DH

   (2000) Memory CD8 T lymphocytes express inhibitory MHC-specific Ly49 receptors

   European Journal of Immunology 30:236–244.

   https://doi.org/10.1002/1521-4141(200001)30:1<236::AID-IMMU236>3.0.CO;2-X

   • PubMed
   • Google Scholar
8. 1. da Rocha ST
   2. Gendrel A-V

   (2019) The influence of DNA methylation on monoallelic expression

   Essays in Biochemistry 63:663–676.

   https://doi.org/10.1042/EBC20190034

   • PubMed
   • Google Scholar
9. 1. Deng Q
   2. Ramsköld D
   3. Reinius B
   4. Sandberg R

   (2014) Single-cell RNA-seq reveals dynamic, random monoallelic gene expression in mammalian cells

   Science (New York, N.Y.) 343:193–196.

   https://doi.org/10.1126/science.1245316

   • PubMed
   • Google Scholar
10. 1. Dreos R
    2. Ambrosini G
    3. Groux R
    4. Cavin Périer R
    5. Bucher P

    (2017) The eukaryotic promoter database in its 30th year: focus on non-vertebrate organisms

    Nucleic Acids Research 45:D51–D55.

    https://doi.org/10.1093/nar/gkw1069

    • PubMed
    • Google Scholar
11. 1. Eckersley-Maslin MA
    2. Spector DL

    (2014) Random monoallelic expression: regulating gene expression one allele at a time

    Trends in Genetics 30:237–244.

    https://doi.org/10.1016/j.tig.2014.03.003

    • PubMed
    • Google Scholar
12. 1. Eckersley-Maslin MA
    2. Thybert D
    3. Bergmann JH
    4. Marioni JC
    5. Flicek P
    6. Spector DL

    (2014) Random monoallelic gene expression increases upon embryonic stem cell differentiation

    Developmental Cell 28:351–365.

    https://doi.org/10.1016/j.devcel.2014.01.017

    • PubMed
    • Google Scholar
13. 1. Ellmeier W
    2. Sunshine MJ
    3. Maschek R
    4. Littman DR

    (2002) Combined deletion of CD8 locus cis-regulatory elements affects initiation but not maintenance of CD8 expression

    Immunity 16:623–634.

    https://doi.org/10.1016/s1074-7613(02)00309-6

    • PubMed
    • Google Scholar
14. 1. Ernst J
    2. Kellis M

    (2012) ChromHMM: automating chromatin-state discovery and characterization

    Nature Methods 9:215–216.

    https://doi.org/10.1038/nmeth.1906

    • PubMed
    • Google Scholar
15. 1. Fei A
    2. Wu W
    3. Tan L
    4. Tang C
    5. Xu Z
    6. Huo X
    7. Bao H
    8. Kong Y
    9. Johnson M
    10. Hartmann G
    11. Talay M
    12. Yang C
    13. Riegler C
    14. Herrera KJ
    15. Engert F
    16. Xie XS
    17. Barnea G
    18. Liberles SD
    19. Yang H
    20. Li Q

    (2021) Coordination of two enhancers drives expression of olfactory trace amine-associated receptors

    Nature Communications 12:3798.

    https://doi.org/10.1038/s41467-021-23823-4

    • PubMed
    • Google Scholar
16. 1. Fiering S
    2. Whitelaw E
    3. Martin DI

    (2000) To be or not to be active: the stochastic nature of enhancer action

    BioEssays 22:381–387.

    https://doi.org/10.1002/(SICI)1521-1878(200004)22:4<381::AID-BIES8>3.0.CO;2-E

    • PubMed
    • Google Scholar
17. 1. Garefalaki A
    2. Coles M
    3. Hirschberg S
    4. Mavria G
    5. Norton T
    6. Hostert A
    7. Kioussis D

    (2002) Variegated expression of CD8 alpha resulting from in situ deletion of regulatory sequences

    Immunity 16:635–647.

    https://doi.org/10.1016/s1074-7613(02)00308-4

    • PubMed
    • Google Scholar
18. 1. Gays F
    2. Koh ASC
    3. Mickiewicz KM
    4. Aust JG
    5. Brooks CG

    (2011) Comprehensive analysis of transcript start sites in ly49 genes reveals an unexpected relationship with gene function and a lack of upstream promoters

    PLOS ONE 6:e18475.

    https://doi.org/10.1371/journal.pone.0018475

    • PubMed
    • Google Scholar
19. 1. Gays F.
    2. Taha S
    3. Brooks CG

    (2015) The distal upstream promoter in Ly49 genes, Pro1, is active in mature NK cells and T cells, does not require TATA boxes, and displays enhancer activity

    Journal of Immunology (Baltimore, Md 194:6068–6081.

    https://doi.org/10.4049/jimmunol.1401450

    • PubMed
    • Google Scholar
20. 1. Gendrel AV
    2. Attia M
    3. Chen CJ
    4. Diabangouaya P
    5. Servant N
    6. Barillot E
    7. Heard E

    (2014) Developmental dynamics and disease potential of random monoallelic gene expression

    Developmental Cell 28:366–380.

    https://doi.org/10.1016/j.devcel.2014.01.016

    • PubMed
    • Google Scholar
21. 1. Gendrel AV
    2. Marion-Poll L
    3. Katoh K
    4. Heard E

    (2016) Random monoallelic expression of genes on autosomes: Parallels with X-chromosome inactivation

    Seminars in Cell & Developmental Biology 56:100–110.

    https://doi.org/10.1016/j.semcdb.2016.04.007

    • PubMed
    • Google Scholar
22. 1. Gimelbrant A
    2. Hutchinson JN
    3. Thompson BR
    4. Chess A

    (2007) Widespread monoallelic expression on human autosomes

    Science (New York, N.Y.) 318:1136–1140.

    https://doi.org/10.1126/science.1148910

    • PubMed
    • Google Scholar
23. 1. Guerra N
    2. Tan YX
    3. Joncker NT
    4. Choy A
    5. Gallardo F
    6. Xiong N
    7. Knoblaugh S
    8. Cado D
    9. Greenberg NM
    10. Greenberg NR
    11. Raulet DH

    (2008) NKG2D-deficient mice are defective in tumor surveillance in models of spontaneous malignancy

    Immunity 28:571–580.

    https://doi.org/10.1016/j.immuni.2008.02.016

    • PubMed
    • Google Scholar
24. 1. Guo L
    2. Hu-Li J
    3. Paul WE

    (2005) Probabilistic regulation in TH2 cells accounts for monoallelic expression of IL-4 and IL-13

    Immunity 23:89–99.

    https://doi.org/10.1016/j.immuni.2005.05.008

    • PubMed
    • Google Scholar
25. 1. Gupta S
    2. Lafontaine DL
    3. Vigneau S
    4. Mendelevich A
    5. Vinogradova S
    6. Igarashi KJ
    7. Bortvin A
    8. Alves-Pereira CF
    9. Nag A
    10. Gimelbrant AA

    (2022) RNA sequencing-based screen for reactivation of silenced alleles of autosomal genes

    G3 (Bethesda, Md.) 12:jkab428.

    https://doi.org/10.1093/g3journal/jkab428

    • PubMed
    • Google Scholar
26. 1. Held W
    2. Roland J
    3. Raulet DH

    (1995) Allelic exclusion of Ly49-family genes encoding class I MHC-specific receptors on NK cells

    Nature 376:355–358.

    https://doi.org/10.1038/376355a0

    • PubMed
    • Google Scholar
27. 1. Ioannidis V
    2. Kunz B
    3. Tanamachi DM
    4. Scarpellino L
    5. Held W

    (2003) Initiation and limitation of Ly-49A NK cell receptor acquisition by T cell factor-1

    Journal of Immunology (Baltimore, Md 171:769–775.

    https://doi.org/10.4049/jimmunol.171.2.769

    • PubMed
    • Google Scholar
28. 1. Joncker NT
    2. Fernandez NC
    3. Treiner E
    4. Vivier E
    5. Raulet DH

    (2009) NK cell responsiveness is tuned commensurate with the number of inhibitory receptors for self-MHC class I: the rheostat model

    Journal of Immunology (Baltimore, Md 182:4572–4580.

    https://doi.org/10.4049/jimmunol.0803900

    • PubMed
    • Google Scholar
29. 1. Kärre K
    2. Ljunggren HG
    3. Piontek G
    4. Kiessling R

    (1986) Selective rejection of H-2-deficient lymphoma variants suggests alternative immune defence strategy

    Nature 319:675–678.

    https://doi.org/10.1038/319675a0

    • PubMed
    • Google Scholar
30. 1. Kelly BL
    2. Locksley RM

    (2000) Coordinate regulation of the IL-4, IL-13, and IL-5 cytokine cluster in Th2 clones revealed by allelic expression patterns

    Journal of Immunology (Baltimore, Md 165:2982–2986.

    https://doi.org/10.4049/jimmunol.165.6.2982

    • PubMed
    • Google Scholar
31. 1. Khan M
    2. Vaes E
    3. Mombaerts P

    (2011) Regulation of the probability of mouse odorant receptor gene choice

    Cell 147:907–921.

    https://doi.org/10.1016/j.cell.2011.09.049

    • PubMed
    • Google Scholar
32. 1. Krueger F
    2. Andrews SR

    (2016) SNPsplit: Allele-specific splitting of alignments between genomes with known SNP genotypes

    F1000Research 5:1479.

    https://doi.org/10.12688/f1000research.9037.2

    • PubMed
    • Google Scholar
33. 1. Kubota A
    2. Kubota S
    3. Lohwasser S
    4. Mager DL
    5. Takei F

    (1999)

    Diversity of NK cell receptor repertoire in adult and neonatal mice

    Journal of Immunology (Baltimore, Md 163:212–216.

    • PubMed
    • Google Scholar
34. 1. Lara-Astiaso D
    2. Weiner A
    3. Lorenzo-Vivas E
    4. Zaretsky I
    5. Jaitin DA
    6. David E
    7. Keren-Shaul H
    8. Mildner A
    9. Winter D
    10. Jung S
    11. Friedman N
    12. Amit I

    (2014) Immunogenetics: Chromatin state dynamics during blood formation

    Science (New York, N.Y.) 345:943–949.

    https://doi.org/10.1126/science.1256271

    • PubMed
    • Google Scholar
35. 1. Larsson AJM
    2. Johnsson P
    3. Hagemann-Jensen M
    4. Hartmanis L
    5. Faridani OR
    6. Reinius B
    7. Segerstolpe Å
    8. Rivera CM
    9. Ren B
    10. Sandberg R

    (2019) Genomic encoding of transcriptional burst kinetics

    Nature 565:251–254.

    https://doi.org/10.1038/s41586-018-0836-1

    • PubMed
    • Google Scholar
36. 1. Levanon D
    2. Negreanu V
    3. Lotem J
    4. Bone KR
    5. Brenner O
    6. Leshkowitz D
    7. Groner Y

    (2014) Transcription factor Runx3 regulates interleukin-15-dependent natural killer cell activation

    Molecular and Cellular Biology 34:1158–1169.

    https://doi.org/10.1128/MCB.01202-13

    • PubMed
    • Google Scholar
37. 1. Levin-Klein R
    2. Fraenkel S
    3. Lichtenstein M
    4. Matheson LS
    5. Bartok O
    6. Nevo Y
    7. Kadener S
    8. Corcoran AE
    9. Cedar H
    10. Bergman Y

    (2017) Clonally stable Vkappa allelic choice instructs Igkappa repertoire

    Nature Communications 8:15575.

    https://doi.org/10.1038/ncomms15575

    • PubMed
    • Google Scholar
38. 1. Li X
    2. Liu D
    3. Zhang L
    4. Wang H
    5. Li Y
    6. Li Z
    7. He A
    8. Liu B
    9. Zhou J
    10. Tang F
    11. Lan Y

    (2021) The comprehensive DNA methylation landscape of hematopoietic stem cell development

    Cell Discovery 7:86.

    https://doi.org/10.1038/s41421-021-00298-7

    • PubMed
    • Google Scholar
39. 1. Marion-Poll L
    2. Forêt B
    3. Zielinski D
    4. Massip F
    5. Attia M
    6. Carter AC
    7. Syx L
    8. Chang HY
    9. Gendrel A-V
    10. Heard E

    (2021) Locus specific epigenetic modalities of random allelic expression imbalance

    Nature Communications 12:5330.

    https://doi.org/10.1038/s41467-021-25630-3

    • PubMed
    • Google Scholar
40. 1. McCullen MV
    2. Li H
    3. Cam M
    4. Sen SK
    5. McVicar DW
    6. Anderson SK

    (2016) Analysis of Ly49 gene transcripts in mature NK cells supports a role for the Pro1 element in gene activation, not gene expression

    Genes and Immunity 17:349–357.

    https://doi.org/10.1038/gene.2016.31

    • PubMed
    • Google Scholar
41. 1. Modzelewski AJ
    2. Chen S
    3. Willis BJ
    4. Lloyd KCK
    5. Wood JA
    6. He L

    (2018) Efficient mouse genome engineering by CRISPR-EZ technology

    Nature Protocols 13:1253–1274.

    https://doi.org/10.1038/nprot.2018.012

    • PubMed
    • Google Scholar
42. 1. Nag A
    2. Savova V
    3. Fung HL
    4. Miron A
    5. Yuan GC
    6. Zhang K
    7. Gimelbrant AA

    (2013) Chromatin signature of widespread monoallelic expression

    eLife 2:e01256.

    https://doi.org/10.7554/eLife.01256

    • PubMed
    • Google Scholar
43. 1. Ng KK
    2. Yui MA
    3. Mehta A
    4. Siu S
    5. Irwin B
    6. Pease S
    7. Hirose S
    8. Elowitz MB
    9. Rothenberg EV
    10. Kueh HY

    (2018) A stochastic epigenetic switch controls the dynamics of T-cell lineage commitment

    eLife 7:e37851.

    https://doi.org/10.7554/eLife.37851

    • PubMed
    • Google Scholar
44. 1. Ohno S
    2. Sato T
    3. Kohu K
    4. Takeda K
    5. Okumura K
    6. Satake M
    7. Habu S

    (2008) Runx proteins are involved in regulation of CD122, Ly49 family and IFN-gamma expression during NK cell differentiation

    International Immunology 20:71–79.

    https://doi.org/10.1093/intimm/dxm120

    • PubMed
    • Google Scholar
45. 1. Perry MW
    2. Boettiger AN
    3. Levine M

    (2011) Multiple enhancers ensure precision of gap gene-expression patterns in the Drosophila embryo

    PNAS 108:13570–13575.

    https://doi.org/10.1073/pnas.1109873108

    • PubMed
    • Google Scholar
46. 1. Raulet DH
    2. Held W
    3. Correa I
    4. Dorfman JR
    5. Wu MF
    6. Corral L

    (1997) Specificity, tolerance and developmental regulation of natural killer cells defined by expression of class I-specific Ly49 receptors

    Immunological Reviews 155:41–52.

    https://doi.org/10.1111/j.1600-065x.1997.tb00938.x

    • PubMed
    • Google Scholar
47. 1. Raulet DH
    2. Vance RE
    3. McMahon CW

    (2001) Regulation of the natural killer cell receptor repertoire

    Annual Review of Immunology 19:291–330.

    https://doi.org/10.1146/annurev.immunol.19.1.291

    • PubMed
    • Google Scholar
48. 1. Reinius B.
    2. Sandberg R

    (2015) Random monoallelic expression of autosomal genes: stochastic transcription and allele-level regulation

    Nature Reviews. Genetics 16:653–664.

    https://doi.org/10.1038/nrg3888

    • PubMed
    • Google Scholar
49. 1. Reinius B
    2. Mold JE
    3. Ramsköld D
    4. Deng Q
    5. Johnsson P
    6. Michaëlsson J
    7. Frisén J
    8. Sandberg R

    (2016) Analysis of allelic expression patterns in clonal somatic cells by single-cell RNA-seq

    Nature Genetics 48:1430–1435.

    https://doi.org/10.1038/ng.3678

    • PubMed
    • Google Scholar
50. 1. Rivière I
    2. Sunshine MJ
    3. Littman DR

    (1998) Regulation of IL-4 expression by activation of individual alleles

    Immunity 9:217–228.

    https://doi.org/10.1016/s1074-7613(00)80604-4

    • PubMed
    • Google Scholar
51. 1. Rogers SL
    2. Rouhi A
    3. Takei F
    4. Mager DL

    (2006) A role for DNA hypomethylation and histone acetylation in maintaining allele-specific expression of mouse NKG2A in developing and mature NK cells

    Journal of Immunology (Baltimore, Md 177:414–421.

    https://doi.org/10.4049/jimmunol.177.1.414

    • PubMed
    • Google Scholar
52. 1. Roland J
    2. Cazenave PA

    (1992) Ly-49 antigen defines an alpha beta TCR population in i-IEL with an extrathymic maturation

    International Immunology 4:699–706.

    https://doi.org/10.1093/intimm/4.6.699

    • PubMed
    • Google Scholar
53. 1. Ronai D
    2. Berru M
    3. Shulman MJ

    (1999) Variegated expression of the endogenous immunoglobulin heavy-chain gene in the absence of the intronic locus control region

    Molecular and Cellular Biology 19:7031–7040.

    https://doi.org/10.1128/MCB.19.10.7031

    • PubMed
    • Google Scholar
54. 1. Roth TL
    2. Puig-Saus C
    3. Yu R
    4. Shifrut E
    5. Carnevale J
    6. Li PJ
    7. Hiatt J
    8. Saco J
    9. Krystofinski P
    10. Li H
    11. Tobin V
    12. Nguyen DN
    13. Lee MR
    14. Putnam AL
    15. Ferris AL
    16. Chen JW
    17. Schickel J-N
    18. Pellerin L
    19. Carmody D
    20. Alkorta-Aranburu G
    21. Del Gaudio D
    22. Matsumoto H
    23. Morell M
    24. Mao Y
    25. Cho M
    26. Quadros RM
    27. Gurumurthy CB
    28. Smith B
    29. Haugwitz M
    30. Hughes SH
    31. Weissman JS
    32. Schumann K
    33. Esensten JH
    34. May AP
    35. Ashworth A
    36. Kupfer GM
    37. Greeley SAW
    38. Bacchetta R
    39. Meffre E
    40. Roncarolo MG
    41. Romberg N
    42. Herold KC
    43. Ribas A
    44. Leonetti MD
    45. Marson A

    (2018) Reprogramming human T cell function and specificity with non-viral genome targeting

    Nature 559:405–409.

    https://doi.org/10.1038/s41586-018-0326-5

    • PubMed
    • Google Scholar
55. 1. Rouhi A
    2. Gagnier L
    3. Takei F
    4. Mager DL

    (2006) Evidence for epigenetic maintenance of Ly49a monoallelic gene expression

    Journal of Immunology (Baltimore, Md 176:2991–2999.

    https://doi.org/10.4049/jimmunol.176.5.2991

    • PubMed
    • Google Scholar
56. 1. Rouhi A
    2. Brooks CG
    3. Takei F
    4. Mager DL

    (2007) Plasticity of Ly49g expression is due to epigenetics

    Molecular Immunology 44:821–826.

    https://doi.org/10.1016/j.molimm.2006.04.006

    • PubMed
    • Google Scholar
57. 1. Rouhi A
    2. Lai CB
    3. Cheng TP
    4. Takei F
    5. Yokoyama WM
    6. Mager DL

    (2009) Evidence for high bi-allelic expression of activating Ly49 receptors

    Nucleic Acids Research 37:5331–5342.

    https://doi.org/10.1093/nar/gkp592

    • PubMed
    • Google Scholar
58. 1. Saleh A
    2. Makrigiannis AP
    3. Hodge DL
    4. Anderson SK

    (2002) Identification of a novel Ly49 promoter that is active in bone marrow and fetal thymus

    Journal of Immunology (Baltimore, Md 168:5163–5169.

    https://doi.org/10.4049/jimmunol.168.10.5163

    • PubMed
    • Google Scholar
59. 1. Saleh A
    2. Davies GE
    3. Pascal V
    4. Wright PW
    5. Hodge DL
    6. Cho EH
    7. Lockett SJ
    8. Abshari M
    9. Anderson SK

    (2004) Identification of probabilistic transcriptional switches in the Ly49 gene cluster: a eukaryotic mechanism for selective gene activation

    Immunity 21:55–66.

    https://doi.org/10.1016/j.immuni.2004.06.005

    • PubMed
    • Google Scholar
60. 1. Sciumè G
    2. Mikami Y
    3. Jankovic D
    4. Nagashima H
    5. Villarino AV
    6. Morrison T
    7. Yao C
    8. Signorella S
    9. Sun H-W
    10. Brooks SR
    11. Fang D
    12. Sartorelli V
    13. Nakayamada S
    14. Hirahara K
    15. Zitti B
    16. Davis FP
    17. Kanno Y
    18. O’Shea JJ
    19. Shih H-Y

    (2020) Rapid Enhancer Remodeling and Transcription Factor Repurposing Enable High Magnitude Gene Induction upon Acute Activation of NK Cells

    Immunity 53:745–758.

    https://doi.org/10.1016/j.immuni.2020.09.008

    • PubMed
    • Google Scholar
61. 1. Shih H-Y
    2. Sciumè G
    3. Mikami Y
    4. Guo L
    5. Sun H-W
    6. Brooks SR
    7. Urban JF Jr
    8. Davis FP
    9. Kanno Y
    10. O’Shea JJ

    (2016) Developmental Acquisition of Regulomes Underlies Innate Lymphoid Cell Functionality

    Cell 165:1120–1133.

    https://doi.org/10.1016/j.cell.2016.04.029

    • PubMed
    • Google Scholar
62. 1. Skene PJ
    2. Henikoff JG
    3. Henikoff S

    (2018) Targeted in situ genome-wide profiling with high efficiency for low cell numbers

    Nature Protocols 13:1006–1019.

    https://doi.org/10.1038/nprot.2018.015

    • PubMed
    • Google Scholar
63. 1. Sleckman BP
    2. Bardon CG
    3. Ferrini R
    4. Davidson L
    5. Alt FW

    (1997) Function of the TCR alpha enhancer in alphabeta and gammadelta T cells

    Immunity 7:505–515.

    https://doi.org/10.1016/s1074-7613(00)80372-6

    • PubMed
    • Google Scholar
64. 1. Stempor P
    2. Ahringer J

    (2016) SeqPlots - Interactive software for exploratory data analyses, pattern discovery and visualization in genomics

    Wellcome Open Research 1:14.

    https://doi.org/10.12688/wellcomeopenres.10004.1

    • PubMed
    • Google Scholar
65. 1. Tanamachi DM
    2. Hanke T
    3. Takizawa H
    4. Jamieson AM
    5. Raulet DR

    (2001) Expression of natural killer receptor alleles at different Ly49 loci occurs independently and is regulated by major histocompatibility complex class I molecules

    The Journal of Experimental Medicine 193:307–315.

    https://doi.org/10.1084/jem.193.3.307

    • PubMed
    • Google Scholar
66. 1. Tanamachi DM
    2. Moniot DC
    3. Cado D
    4. Liu SD
    5. Hsia JK
    6. Raulet DH

    (2004) Genomic Ly49A transgenes: basis of variegated Ly49A gene expression and identification of a critical regulatory element

    Journal of Immunology (Baltimore, Md 172:1074–1082.

    https://doi.org/10.4049/jimmunol.172.2.1074

    • PubMed
    • Google Scholar
67. 1. Thorvaldsdóttir H
    2. Robinson JT
    3. Mesirov JP

    (2013) Integrative Genomics Viewer (IGV): high-performance genomics data visualization and exploration

    Briefings in Bioinformatics 14:178–192.

    https://doi.org/10.1093/bib/bbs017

    • PubMed
    • Google Scholar
68. 1. Vance RE
    2. Jamieson AM
    3. Cado D
    4. Raulet DH

    (2002) Implications of CD94 deficiency and monoallelic NKG2A expression for natural killer cell development and repertoire formation

    PNAS 99:868–873.

    https://doi.org/10.1073/pnas.022500599

    • PubMed
    • Google Scholar
69. 1. Walters MC
    2. Fiering S
    3. Eidemiller J
    4. Magis W
    5. Groudine M
    6. Martin DI

    (1995) Enhancers increase the probability but not the level of gene expression

    PNAS 92:7125–7129.

    https://doi.org/10.1073/pnas.92.15.7125

    • PubMed
    • Google Scholar
70. 1. Weintraub H

    (1988) Formation of stable transcription complexes as assayed by analysis of individual templates

    PNAS 85:5819–5823.

    https://doi.org/10.1073/pnas.85.16.5819

    • PubMed
    • Google Scholar
71. 1. Wensveen FM
    2. Jelenčić V
    3. Polić B

    (2018) NKG2D: A Master Regulator of Immune Cell Responsiveness

    Frontiers in Immunology 9:441.

    https://doi.org/10.3389/fimmu.2018.00441

    • PubMed
    • Google Scholar
72. 1. Xu Y
    2. Davidson L
    3. Alt FW
    4. Baltimore D

    (1996) Deletion of the Ig kappa light chain intronic enhancer/matrix attachment region impairs but does not abolish V kappa J kappa rearrangement

    Immunity 4:377–385.

    https://doi.org/10.1016/s1074-7613(00)80251-4

    • PubMed
    • Google Scholar
73. 1. Xu J
    2. Carter AC
    3. Gendrel AV
    4. Attia M
    5. Loftus J
    6. Greenleaf WJ
    7. Tibshirani R
    8. Heard E
    9. Chang HY

    (2017) Landscape of monoallelic DNA accessibility in mouse embryonic stem cells and neural progenitor cells

    Nature Genetics 49:377–386.

    https://doi.org/10.1038/ng.3769

    • PubMed
    • Google Scholar
74. 1. Yokoyama WM
    2. Kehn PJ
    3. Cohen DI
    4. Shevach EM

    (1990)

    Chromosomal location of the Ly-49 (A1, YE1/48) multigene family: Genetic association with the NK 1.1 antigen

    Journal of Immunology (Baltimore, Md 145:2353–2358.

    • PubMed
    • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "Binary outcomes of enhancer activity underlie stable random monoallelic expression" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Tadatsugu Taniguchi as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Anne-Valerie Gendrel (Reviewer #1); Alvaro Rada-Iglesias (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Overall, the reviewers found the manuscript to be relevant, the data clearly presented and the main conclusions well supported by the data. They appreciated the usage of an impressive battery of mouse models, and recognized novel insights brought by this study, into how RME can be regulated and how enhancers can work in a binary manner controlling the probability rather than the level of gene expression. Two slight critics may be (1) the limited mechanistic insights the manuscript provides onto how enhances contribute to RME, and (2) whether the enhancer-RME process described here applies to other types of genes and other cell types, i.e. for genes that are rather CpG-rich than CpG-poor, that are not expressed in the hematopoietic lineage and that do not encode surface markers.

Please refer to the two points further expanded below to prepare your revisions, and please consult the detailed points raised by the reviewers, for further rephrasing of the manuscript and discussion in a point-by-point rebuttal.

1 – Provide additional analyses to investigate the nature of the promoters studied here, before making the claim that RME may be more widespread than originally thought. Using published data, it should be feasible to study these peculiar promoters for their CpG content, and motifs for transcription factor binding sites. In comparison, are genes with CpG-rich promoters also subject to RME and do enhancers influence RME at CpG-rich genes? Finally, although the authors suggest in the discussion that DNA methylation is unlikely to play a major role in RME, it is definitely worth investigating, as an attempt to explain the lack of enhancer-gene communication and also to link with a potential influence on RME extent. Promoters of active and inactive genes should be compared for their DNA methylation states, either using public datasets (if they exist) or using targeted PCR-based DNA methylation analysis.

2 – Moderate the tone regarding the claims that HSSs identified by ATAC-seq are definitely enhancers and not promoters (solely based on epigenomic profiling), provided that this topics is quite controversial in the field.

Reviewer #1 (Recommendations for the authors):

Here are my specific comments and questions for the authors to improve the manuscript.

1. Throughout the manuscript, mature NK, immature NK, primary NK, IL-2 expanded NK cells are mentioned in different instances. It is unclear whether this matters or not? Is this important for the expression of the different genes analysed? This should be clarified, especially for non-specialists in these lineages.

What is the stimulation of NK cells with IL-2 important for; is this for proliferation, for activity? Why are cells sometimes expanded with IL-2 and sometimes not?

2. How were the % of expression frequencies on Figure 1—figure supplement 1A calculated?

3. There is no ATAC-seq peak visible for the promoter of the Ly49a gene (Figure 1A, Figure 3A, Figure 4A), unlike for the other genes of the cluster. Could the authors please comment on this? Doesn´t this argue against the claim that HSS1 is an enhancer and not a promoter?

4. On the FACS plot on Figure 2A, it is difficult to distinguish the population of cells expressing the B6 allele only from both alleles. It would help to draw a circle around each cell populations that were selected for the sorting and further analysed for ATAC-seq. Moreover, was there a control of populations after sorting?

5. On Figure 2E, for the control mouse IgG2ak (cIgG) antibody in IL-2 expanded NK cells, there is no ATAC-seq peak at the enhancer (nor CUT&RUN signal). Shouldn't an ATAC-seq peak at the enhancer expected in this population too? Isn't it similar to cells expressing neither alleles?

6. On Figure 3—figure supplement 2C and F. Isn't there a slight increase in the % of cells expressing the BALB only allele in heterozygous F1 mice for both Ly49a and Nkg2a?

7. On lines 344-345, the authors say: "the corresponding region of the highly related Ly49a gene, which is expressed by only ~17% of NK cells, is much less accessible and presumably less active (Figure 4A)". How is the level of accessibility seen/observed? This is not clear to me.

8. Figure 4E – similar to comment #4, isn't the % of cells expressing the BALB allele only increased? …which would be indicative of a slight dosage compensation mechanism?

9. The possible role of the HSS5 enhancer in increasing the expression probability/frequency of the Ly49g gene in NK cells is appealing. Given the similarity between the Ly49a and Ly49g loci regarding the HSS1 enhancer, would it be possible to test this by introducing the HSS5 element upstream of the Ly49a gene to see whether this would increase the probability/frequency of Ly49a expression? (maybe using a transgene construct, not necessarily a knockin).

10. Line 442; there is no introduction for the knockout model of Nkg2d. What is the KO allele exactly? What is deleted: promoter, exons, whole locus?

11. Line 464; The authors mention the 3'enhancer for the Nkg2d gene. Could they speculate on its function?

12. Line 573-575; the following sentence is not clear to me: "Importantly, NKG2D- cells in both Nkg2d5'ED/5'ED and Nkg2d+/- mice were as likely to express NKG2A, Ly49G2 or Ly49I as NKG2D+ cells, suggesting stochastic expression of the WT Nkg2d allele (Figure 7—figure supplement 1, A-C)." Could the authors please clarify?

13. The following two important points remain unanswered in this study and could be discussed further in the discussion part of the manuscript:

a – What define the strength of an enhancer for the Ly49 and Nkg2 genes. What is the difference between a weak and a strong enhancer? Related to the binding of specific transcription/chromatin factors? Level of enrichment of specific histone marks? 3D conformation?

b – Could this model apply to chromatin or nuclear factors, expressed in other cell lineages than the hematopoietic lineage?

Reviewer #2 (Recommendations for the authors):

Overall, I found the manuscript highly relevant and the main conclusions are well supported by the presented data. However, there are some aspects that could be modified and/or expanded in order to improve the overall relevance of the work:

My main concern is that, as presented, the work strongly suggest that enhancers play an important and instructive work in RME. However, there are limited insights into the molecular mechanisms behind such an instructive role. Given that the promoter but not the enhancer chromatin state correlates with RME, I think the authors should explore more how enhancer-gene communication might be altered in expressing and non-expressing alleles. Since the investigated enhancers are quite proximal to the target gene promoters, investigating topological aspects of enhancer-gene communication can be technically quite challenging. Moreover, it is highly intriguing that in alleles subject to RME the promoters stably lose their responsiveness to proximal active enhancers. In this regard, I have a few suggestions:

– Although H3K27ac is strongly associated with active enhancers, it can be uncoupled from expression of the target genes (Pachano et al., 2021), while the production of eRNAs seems tope strongly coupled with mRNA expression (Kim et al., 2010; Hirabayashi et al., 2019). Therefore, I think it could be quite relevant to investigate eRNA levels at the investigated enhancers in the context of expressing and non-expressing alleles.

– As the authors clearly show, the promoters of the NK receptor genes do not acquire histone modifications associated with gene repression. This is not surprising given that, as the authors mention in the discussion, these promoters are CpG-poor. Inactive CpG-poor promoters do not typically show enrichment for these histone marks but display high DNA methylation levels instead, in contrast with CpG-rich promoters, which are typically hypomethylated regardless of whether they are active or not. Therefore, although the authors mention in the Discussion that DNA methylation is unlikely to play a major role in RME, I still think that it would be quite insightful to compare the DNA methylation levels of active and inactive promoters for example by traditional locus-specific bisulfite sequencing. Moreover, the authors could investigate whether their findings still apply to genes with CpG-rich promoters (are CpG-rich genes subject to RME? Do enhancers influence RME at CpG-rich genes?) and/ or, alternatively, introduce a CpG island into one of the NK receptor gene promoters.

– If the authors find clear differences in the DNA methylation status of active and inactive promoters, one possibility is that epigenetic heterogeneity at gene promoters might occur before enhancer activation and thus influence, together with enhancer strength, the extent of RME. The authors could investigate DNA methylation at promoter regions at earlier cellular states, which could represent a critical temporal window at which DNA demethylating agents could have an impact on RME.

Reviewer #3 (Recommendations for the authors):

1 – In the first section of the results, the authors test the idea that Ly49Hss and NKg25'E regulatory elements are enhancers rather than promoters in NK cells as previously suggested. A full demonstration of this is not provided since the data presented is essentially epigenetic features and ATAC-seq. The provided evidence does not for example demonstrate that Ly49Hss can not behave as promoters in immature cells as proposed before since no RNA data is shown in the article for NKR analyses.

2 – In the same section, based on epigenetic profiles and the relative amount of H3K4me1/3 and other predictive methods (ATAC, K27ac) the authors formally conclude on 'enhancer identity' (lines 186-187) and 'these findings provide definitive support for the conclusion that Ly49Hss1 and Nkg25'E elements represent enhancers' (lines 196-197). It is important to keep in mind that epigenomic features never demonstrate enhancer identity but rather suggest of putative enhancer features. Thus, the phrasing used here should be toned down.

3 – For all ATAC-seq and ChIP-seq data shown, including examples the units should appear on the graphs as normalized equivalently to the read counts (for example in Figure 1B, Figure 1s2C and all the following). A table indicating all published and original data sets used with GSE numbers should be provided, including replicates used and sequencing depth. The presented novel data does not seem to have been submitted to GEO database. This should be done.

4 – If the model proposed by the authors is correct, it implies a level of stochasticity in promoter-enhancer dependent activation. Based on sequence elements (for example presence or absence of CpG islands), do the promoters of the NKR locus tend to be less constitutive? More generally how do the authors envision the critical parameters limiting promoter's action and their strong dependency to enhancer stochastic activation? These points could be discussed in the Discussion section.

1 – Provide additional analyses to investigate the nature of the promoters studied here, before making the claim that RME may be more widespread than originally thought. Using published data, it should be feasible to study these peculiar promoters for their CpG content, and motifs for transcription factor binding sites. In comparison, are genes with CpG-rich promoters also subject to RME and do enhancers influence RME at CpG-rich genes? Finally, although the authors suggest in the discussion that DNA methylation is unlikely to play a major role in RME, it is definitely worth investigating, as an attempt to explain the lack of enhancer-gene communication and also to link with a potential influence on RME extent. Promoters of active and inactive genes should be compared for their DNA methylation states, either using public datasets (if they exist) or using targeted PCR-based DNA methylation analysis.

We address these concerns in detail in the response to the reviewers’ comments below. In summary, RME can be found at genes with both high and low CpG content (Eckersley-Maslin et al., Dev. Cell 2014 PMID: 24576421). As one example from our work, in Figure 7, we show random monoallelic expression of the Thy1 gene, which is expressed by T cells and is widely regarded as a T cell “marker”. As documented in the UCSC genome browser screenshot depicted in reviewer data, Figure 2 (embedded in the response to reviewer 2) the Thy1 gene includes a CpG island at its promoter. Furthermore, the Bcl11b gene, which encodes a transcription factor that commits thymocytes to the T cell fate, contains a promoter CpG-island and was recently shown to be an RME gene and whose expression probability is further reduced by enhancer-deletion (Ng et al., eLife 2018 PMID: 30457103). Therefore, we do not believe there is reason to believe that our results are unique to a class of genes with respect to CpG density; rather, the RME phenomenon (as well as the enhancer deletion-associated variegation phenomenon) appears to apply across cell types and gene classes.

Concerning DNA methylation, we have extensively revised the manuscript to address this issue and will discuss this below in response to reviewer 2.

2 – Moderate the tone regarding the claims that HSSs identified by ATAC-seq are definitely enhancers and not promoters (solely based on epigenomic profiling), provided that this topics is quite controversial in the field.

We softened the language, but note that our data provide evidence beyond epigenomic profiling that these elements are enhancers in mature NK cells: specifically, we showed that deletion of Hss1 in mature NK cells extinguishes expression, which is known to initiate at promoters well downstream of the enhancers in mature NK cells.

Reviewer #1 (Recommendations for the authors):

Here are my specific comments and questions for the authors to improve the manuscript.

1. Throughout the manuscript, mature NK, immature NK, primary NK, IL-2 expanded NK cells are mentioned in different instances. It is unclear whether this matters or not? Is this important for the expression of the different genes analysed? This should be clarified, especially for non-specialists in these lineages.

Where relevant, we have attempted to explain the characteristics of each type of NK cell. Briefly, immature NK cells are not yet terminally differentiated and importantly do not yet express Ly49 genes, expression of which is associated with mature, or fully differentiated NK cells. This is now indicated on page 10.

What is the stimulation of NK cells with IL-2 important for; is this for proliferation, for activity? Why are cells sometimes expanded with IL-2 and sometimes not?

Generally, we use IL-2 to expand NK cells when it is necessary obtain sufficient cells for analysis, which we now indicate in the text on page 15. To elaborate, NK cells are fairly rare among spleen cells (3%) and only 3 % of those co-express both alleles of Ly49G2 in (B6 x BALB/c)F₁ hybrid mice, for a frequency of 0.09% of splenocytes. Generally, CUT&RUN experiments utilized IL-2 expanded cells, while ATAC-seq experiments utilized fresh cells. These data appear to cross-validate each other well (Figure 2). Furthermore, Ly49 expression states (both on and off) are stable in NK cells expanded in IL-2. For these reasons, we used IL-2 expanded NK cells when we needed larger numbers of primary cells for chromatin assays.

2. How were the % of expression frequencies on Figure 1—figure supplement 1A calculated?

The percentages are based on previously published historical data generated by both our group and others. We have added citations in the Figure 1—figure supplement 1 legend in the revised manuscript. The numbers depict approximations only.

3. There is no ATAC-seq peak visible for the promoter of the Ly49a gene (Figure 1A, Figure 3A, Figure 4A), unlike for the other genes of the cluster. Could the authors please comment on this? Doesn´t this argue against the claim that HSS1 is an enhancer and not a promoter?

The transcription start sites (TSSs) of the Ly49 genes were previously documented by 5’ oligo-capping RACE (Gays et al., PLos One 2011 PMID: 21483805), and for the Ly49 genes these range from ~5-6kb downstream of HSS1, at sites called “Pro2” and “Pro3,” hereafter referred to as the TSSs. The citation is now included in the Figure 2 legend. mRNAs in mature, Ly49+ NK cells do NOT initiate at the HSS1 elements, and this is true across the gene family. Thus, HSS1 is not the promoter for Ly49a or any of the Ly49 genes in mature NK cells, nor has any published report made the claim that it is. While the signals from the documented TSSs for Ly49a are weak, they are present (see Author response image 1). The weakness of these ATAC sites is likely due to (1) Multiple TSSs such that the ATAC-seq accessibility is diffused to several sites, and (2) the fact that less than 20% of cells express the gene. The signal would therefore be ~5 times higher (or more) if the cells were sorted to be Ly49A+.

Author response image 1

Download asset Open asset

4. On the FACS plot on Figure 2A, it is difficult to distinguish the population of cells expressing the B6 allele only from both alleles. It would help to draw a circle around each cell populations that were selected for the sorting and further analysed for ATAC-seq. Moreover, was there a control of populations after sorting?

The NKG2A^B6-specific mAb (16a11) was first generated in our lab (Vance et al. PNAS 2002 PMID 11782535), and it was subsequently shown by Dixie Mager’s lab that the use of this clone in combination with the NKG2A^B6+BALB/c reactive clone 20d5 enables the sorting the populations expressing each separately, which was confirmed at the mRNA level (Rogers et al., J. Immunol. 2006 PMID: 16785537) and is now cited on p. 11. We added a new figure (Figure 2 figure supplement 1) that shows the gating and post sort analysis of the NKG2A-sorted cells and the Ly49G2-sorted cells used to generate ATAC-seq data and the CUT&RUN data presented in Figure 2.

5. On Figure 2E, for the control mouse IgG2ak (cIgG) antibody in IL-2 expanded NK cells, there is no ATAC-seq peak at the enhancer (nor CUT&RUN signal). Shouldn't an ATAC-seq peak at the enhancer expected in this population too? Isn't it similar to cells expressing neither alleles?

The control IgG “cIgG” refers to the antibody used as a control for the CUT&RUN itself, not as a control for sorting; we do not expect to observe a signal with an IgG that should not interact with the chromatin. We regret the omission of a clear explanation. The cells used for the cIgG control are a mixture of NK cells that express either the B6 or BALB/c allele, in approximately equal amounts. In order to clarify this, we changed the coloring of the cIgG label to black to distinguish it from the labeling of the allele expressed. We further explain in the revised legend for Figure 2 the nature of this control experiment and specify which cells were used to generate the cIgG track. The protein A in the pA-Mnase fusion binds mIgG2ak with high affinity, which is the reason for the selection of this control. In general, CnR produces remarkably low noise, and in all cIgG CnR datasets we have produced we have not seen appreciable signal enrichment at ATAC-seq accessible sites.

6. On Figure 3—figure supplement 2C and F. Isn't there a slight increase in the % of cells expressing the BALB only allele in heterozygous F1 mice for both Ly49a and Nkg2a?

In each of the referenced panels, we expect to see an increase in frequency of BALB/c only cells without invoking a compensation mechanism. This is not due to an increased likelihood of expression of the BALB/c allele, but rather because the cells that would have expressed both alleles are no longer able to express the B6 allele as they lack the critical enhancer on the B6 chromosome. In other words, while the proportion of cells expressing only the BALB/c allele increases, the total proportion of cells expressing the BALB/c allele remains unchanged.

If the reviewer is instead referring to the second sub-panel in Figure 3—figure supplement C (where there is a slight increase in the observed BALB/c-only cells over expected) this is a separate point. We expect an increase in the BALB/c only cells for reasons entirely independent of a compensatory effect, as described above. However, this slight increase in observed over expected may imply some slight compensatory mechanism. This increase is slight, and the overall result is that the total number of cells expressing the BALB/c allele is not drastically increased as we might expect if there were a trans-acting transcriptional “counting” mechanism to ensure a particular percentage of cells express NKG2A.

7. On lines 344-345, the authors say: "the corresponding region of the highly related Ly49a gene, which is expressed by only ~17% of NK cells, is much less accessible and presumably less active (Figure 4A)". How is the level of accessibility seen/observed? This is not clear to me.

Figure 4A shows that the region of Ly49a that aligns with Hss5 of Ly49g has only a very weak ATAC signal. This was true for all ATAC-seq datasets generated in NK cells that we have examined. The ATAC-seq data in Reviewer Data Figure 1 also shows a weak Hss5 signal that is barely distinguishable above background, and certainly much less detectable than Ly49g-Hss5 (Figure 2D; Figure 4A).

8. Figure 4E – similar to comment #4, isn't the % of cells expressing the BALB allele only increased? …which would be indicative of a slight dosage compensation mechanism?

See comment 6. Cells that would otherwise express both alleles in a WT mouse that fail to express the B6 allele in the mutant F1 are now identified as BALB only cells, resulting in an increased proportion without any compensation mechanism.

If, again, the reviewer is referring to the slight increase in observed BALB/c-only cells as compared to expected, the effect is even more slight here and is much more consistent with stochastic and independent regulation of alleles rather than a compensatory mechanism that acts in trans.

9. The possible role of the HSS5 enhancer in increasing the expression probability/frequency of the Ly49g gene in NK cells is appealing. Given the similarity between the Ly49a and Ly49g loci regarding the HSS1 enhancer, would it be possible to test this by introducing the HSS5 element upstream of the Ly49a gene to see whether this would increase the probability/frequency of Ly49a expression? (maybe using a transgene construct, not necessarily a knockin).

This is potentially a revealing experiment that we would love to do and plan to do. In order to do it properly, it should be a germline knockin. Transgenes suffer from various technical problems including variable copy number insertions and position effects. We previously successfully used a genomic transgene to establish the proximal regulation of Ly49a (Tanamachi et al., J. Immunol. 2004 PMID: 14707081). The data, however, were variable for different founders, to an extent that we believe would compromise this approach for the question at hand. Knockin approaches for these genes have proven to be extremely challenging in our hands likely due to the homology between the targeting sequence and multiple related Ly49 genes which appear to diminish the frequency of replacement of the desired gene. For these reasons we decided this experiment, while potentially revealing, will be subject of future efforts and, if successful, included in a future publication.

10. Line 442; there is no introduction for the knockout model of Nkg2d. What is the KO allele exactly? What is deleted: promoter, exons, whole locus?

The Nkg2d knockout model was generated and characterized by our group (Guerra et al., Immunity 2008 PMID: 18394936). The gene is comprised by 7 exons, of which exons 2-6 are deleted in the knockout model. We added a statement to this effect and a reference on p. 27 of the revised manuscript.

11. Line 464; The authors mention the 3'enhancer for the Nkg2d gene. Could they speculate on its function?

We added a sentence on p 28 to note the possibility that the 3’ element constitutes the residual enhancer that drives expression after 5’E deletion, and which acts in combination with 5’E to drive a very high probability of Nkg2d expression in NK cells. However, we have no direct evidence at present that the 3’ ATAC-seq peak is actually an enhancer of Nkg2d.

12. Line 573-575; the following sentence is not clear to me: "Importantly, NKG2D- cells in both Nkg2d5'ED/5'ED and Nkg2d+/- mice were as likely to express NKG2A, Ly49G2 or Ly49I as NKG2D+ cells, suggesting stochastic expression of the WT Nkg2d allele (Figure 7—figure supplement 1, A-C)." Could the authors please clarify?

We revised this sentence (p. 33) to lend more clarity. It now states:

“Importantly, in both Nkg2d^{5′ED/5′ED} and Nkg2d^+/- mice, NKG2D-negative cells were as likely as NKG2D+ cells to express NKG2A, Ly49G2 or Ly49I (Figure 7—figure supplement 1, A-C). These data suggest that in both knockout and WT mice, NKG2D is randomly expressed with respect to the other RME receptor genes.”

13. The following two important points remain unanswered in this study and could be discussed further in the discussion part of the manuscript:

a – What define the strength of an enhancer for the Ly49 and Nkg2 genes. What is the difference between a weak and a strong enhancer? Related to the binding of specific transcription/chromatin factors? Level of enrichment of specific histone marks? 3D conformation?

We provide a speculative answer in our response to the public reviews and in the revised manuscript on p. 43. On p 43 of the revised manuscript, after proposing that the randomness of RME may reflect the probability of overcoming the energy barrier required for nucleosome remodeling of initially chromatinized promoters, we added the following passage:

“It may be this property that is the measure of “enhancer strength”, i.e., the capacity of the enhancer to overcome the energy threshold presented by the promoter within a limited window during differentiation. The number of enhancer elements in a gene, the availability of relevant transcription factors that bind the enhancers and the affinity with which the enhancer binds the factors are all likely relevant in determining enhancer strength.”

b – Could this model apply to chromatin or nuclear factors, expressed in other cell lineages than the hematopoietic lineage?

We believe the answer is yes. As already mentioned, the gene encoding the transcription factor Bcl11b is expressed in an RME fashion during thymocyte development, and the probability of expression of this gene is further altered by enhancer deletion variegation (Ng et al., eLife 2018 PMID: 30457103). We add a reference to this in this discussion on page 41, indicating that this supports the generality of our model to genes not encoding cell surface receptors. With respect to non-hematopoietic lineages, deletion of enhancers regulating the odorant receptor genes (Khan et al., Cell 2011 PMID: 22078886) as well as trace amine-associated and trace-amine associated receptors (Fei et al., Nat. Commun. 2021 PMID: 34145235) in the olfactory epithelium reduced the expression probability of the target genes, demonstrating that RME rooted in binary enhancer action applies to non-hematopoietic lineage. On page 44 in the revised manuscript, we add statements and citations to these papers in order to clarify that the mechanisms we describe are likely not restricted to hematopoietic cells.

Reviewer #2 (Recommendations for the authors):

Overall, I found the manuscript highly relevant and the main conclusions are well supported by the presented data. However, there are some aspects that could be modified and/or expanded in order to improve the overall relevance of the work:

My main concern is that, as presented, the work strongly suggest that enhancers play an important and instructive work in RME. However, there are limited insights into the molecular mechanisms behind such an instructive role. Given that the promoter but not the enhancer chromatin state correlates with RME, I think the authors should explore more how enhancer-gene communication might be altered in expressing and non-expressing alleles. Since the investigated enhancers are quite proximal to the target gene promoters, investigating topological aspects of enhancer-gene communication can be technically quite challenging. Moreover, it is highly intriguing that in alleles subject to RME the promoters stably lose their responsiveness to proximal active enhancers. In this regard, I have a few suggestions:

– Although H3K27ac is strongly associated with active enhancers, it can be uncoupled from expression of the target genes (Pachano et al., 2021), while the production of eRNAs seems tope strongly coupled with mRNA expression (Kim et al., 2010; Hirabayashi et al., 2019). Therefore, I think it could be quite relevant to investigate eRNA levels at the investigated enhancers in the context of expressing and non-expressing alleles.

We appreciate the close relationship between eRNA and target mRNA production. While we have not directly assayed eRNAs ourselves, they have been analyzed extensively by RT-PCR in both primary NK cells and NK cell related cell lines (Gays et al., J. Immunol. 2015 PMID: 25926675) (cited on p. 9 of revised manuscript). This study showed that bidirectional enhancer-derived transcripts are present in cells that do not express the corresponding receptor gene. Therefore, even when enhancer-derived transcripts are used to measure enhancer activity, Hss1 activity can be decoupled from target gene expression status, corroborating our results. Taken together, these results suggest that the relationship between the enhancer activity acting on a locus is fundamentally probabilistic rather than deterministic (i.e., not 1:1). This decoupling likely takes place infrequently, when overall enhancer activity is below the threshold to activate the target gene in the large majority of cells. In many and perhaps most cases, enhancer activity as measured by eRNA production will correlate very strongly with mRNA production, especially when measured in bulk populations rather than single cells. We added a sentence on p. 14 addressing the relationship between eRNA production and target Ly49 gene expression status.

– As the authors clearly show, the promoters of the NK receptor genes do not acquire histone modifications associated with gene repression. This is not surprising given that, as the authors mention in the discussion, these promoters are CpG-poor. Inactive CpG-poor promoters do not typically show enrichment for these histone marks but display high DNA methylation levels instead, in contrast with CpG-rich promoters, which are typically hypomethylated regardless of whether they are active or not. Therefore, although the authors mention in the Discussion that DNA methylation is unlikely to play a major role in RME, I still think that it would be quite insightful to compare the DNA methylation levels of active and inactive promoters for example by traditional locus-specific bisulfite sequencing. Moreover, the authors could investigate whether their findings still apply to genes with CpG-rich promoters (are CpG-rich genes subject to RME? Do enhancers influence RME at CpG-rich genes?) and/ or, alternatively, introduce a CpG island into one of the NK receptor gene promoters.

We regret that we did not dedicate more space in the first version of the paper to the discussion of DNA methylation. We have added data and much discussion on the topic, including a section in the results beginning on p 28 and supplementary figure (Figure 5—figure supplement 2) showing differential methylation of the Nkg2d promoter on active and silent alleles in Nkg2d5’E deletion mice. We further revised the discussion to address the role of DNA methylation in RME beginning on p44. Briefly, many RME genes do not display differential promoter methylation (da Rocha, Gendrel Essays Biochem. PMID: 31782494). Recent studies have further indicated that only a small minority of RME genes are subject to derepression of silent alleles through pharmacological inhibition of DNA methylation (Gupta et al., G3 2021 https://doi.org/10.1093/g3journal/jkab428 ; Marion-Poll et al., Nat. Commun. 2021 PMID: 34504093). Furthermore, inhibition of DNA methylation did not derepress silent Igh alleles in a B cell hybridoma model of enhancer-deletion variegation (Ronai et al., J. Immunol. 2002 PMID: 12471125). The role of DNA methylation in RME was addressed in a recent review (da Rocha, Gendrel Essays Biochem. PMID: 31782494), and it appears to be accepted in the RME field that DNA methylation is unlikely to underly all or most cases of RME (Eckersley-Maslin, Spector Trends Genet. 2014 PMID: 24780084). As we mention in the discussion (page 43), there remains the possibility that DNA methylation may play a role in determining the probability of promoter activation for some RME genes, including the ones we studied in the paper.

Regarding the issue of CpG density: Ly49 promoters are indeed CpG-poor, but it is the case that CpG-rich genes are also subject to RME (Eckersley-Maslin et al., Dev. Cell 2014 PMID: 24576421). Indeed, the Thy1 gene, which we show is an RME gene despite being considered a marker of all T cells, harbors a CpG island annotated on the UCSC genome browser (Author response image 2). The RME Bcl11b (Ng et al., eLife 2018 PMID: 30457103) also harbors a CpG island (Author response image 3). We add a statement in the discussion beginning (p. 42) to clarify that RME (and enhancer-controlled RME) apply to genes with CpG-dense and CpG poor promoter regions.

Author response image 2

Download asset Open asset

Author response image 3

Download asset Open asset

– If the authors find clear differences in the DNA methylation status of active and inactive promoters, one possibility is that epigenetic heterogeneity at gene promoters might occur before enhancer activation and thus influence, together with enhancer strength, the extent of RME. The authors could investigate DNA methylation at promoter regions at earlier cellular states, which could represent a critical temporal window at which DNA demethylating agents could have an impact on RME.

With respect to the methylation of CpG sites in the Ly49 and Nkg2a promoters themselves, several studies have been published by Dixie Mager’s group demonstrating a clear correlation between DNA methylation and silent alleles (Rouhi et al., J. Immunol. 2006 PMID: 16493057; Rouhi et al., Mol. Immunol. 2007 PMID: 16750269; Rogers et al., J. Immunol. 2006 PMID: 16785537). Inhibition of DNA methylation on its own appears insufficient to derepress silent alleles, however. In the case of Nkg2a, the promoter region is heavily methylated both in immature cells that do not yet express NKG2A, and in the case of silent alleles in mature cells, suggesting that methylation is not specifically deposited on silent alleles, but rather that active alleles are de-methylated. This is mentioned in the revised discussion (page 43). Furthermore, the Ly49a promoter region (Pro2) was shown to be methylated in non-hematopoietic cells. All of these data support the notion that DNA methylation of the NK receptor gene promoters is associated with the default off state, and that demethylation takes place specifically on activated alleles, likely concomitant with activation. These considerations make it less likely that there exists an immature developmental state in which the two alleles might be differentially methylated, preparatory of subsequent RME. Another important point is that the enhancer deletion variegation (and associated allelic “failure”) was observed in Drosophila (Perry et al., PNAS 2011 PMID: 21825127), where DNA methylation is not thought to be present at detectable levels and thus does not play a major role in gene regulation. This consideration confirms that DNA methylation is not generally required for the binary enhancer-driven RME phenomenon.

As discussed in the revised Discussion section (p 42), we favor a mechanism in which enhancer action must overcome the energy barrier associated with the promoters being initially in an inactive state, and success occurs with a random probability, albeit dependent on enhancer strength. In our thinking, the inactive state that must be overcome could simply be the closed chromatin state. However, we also propose the possibility that promoter DNA methylation could contribute to the energy barrier in some cases.

Reviewer #3 (Recommendations for the authors):

1 – In the first section of the results, the authors test the idea that Ly49Hss and NKg25'E regulatory elements are enhancers rather than promoters in NK cells as previously suggested. A full demonstration of this is not provided since the data presented is essentially epigenetic features and ATAC-seq. The provided evidence does not for example demonstrate that Ly49Hss can not behave as promoters in immature cells as proposed before since no RNA data is shown in the article for NKR analyses.

As requested, we toned down the statements claiming the chromatin modification data provided “definitive” support for enhancer as opposed to promoter identity (p. 10). Nevertheless, we feel that the evidence is strong. Our argument depends not only on the epigenetic marks and ATACseq results, but also on our finding that the elements are necessary for maintaining gene expression in mature NK cells, where it is known that the mRNAs are initiated well downstream of Hss1 (Figure 1—figure supplement 2). The revised text on p. 10-11 emphasizes this point (see our response to reviewer 1, comment 3 for further details). The reviewer points out that we did not examine RNA, but others have, and we do not contest their findings that transcripts initiating at Hss1 and transcribing through the gene have been identified in immature NK cells. The difficulty is that transcripts (eRNAs) are known to initiate from enhancers, so the RNA data does not resolve the issue. Another group has made a good argument that the observed transcripts initiating at Hss1 are indeed eRNAs (Gays et al., J. Immunol. 2015 PMID: 25926675). We note that the existence of eRNAs in other genes has not led to theories that such RNAs are part of a developmental switch that turns those genes on and off. An additional argument in favor of our conclusions is that deletion of the 5’ enhancer of Nkg2d resulted in an expression pattern indistinguishable from that of Ly49 or Nkg2a genes. Since it is unlikely that enhancer deletion would CREATE a bidirectional promoter that drives variegated expression, the simplest hypothesis is that variegation in Ly49 and enhancer deletion variegation alike is the consequence of weak enhancer activity as opposed to the action of a bidirectional upstream promoter.

2 – In the same section, based on epigenetic profiles and the relative amount of H3K4me1/3 and other predictive methods (ATAC, K27ac) the authors formally conclude on 'enhancer identity' (lines 186-187) and 'these findings provide definitive support for the conclusion that Ly49Hss1 and Nkg25'E elements represent enhancers' (lines 196-197). It is important to keep in mind that epigenomic features never demonstrate enhancer identity but rather suggest of putative enhancer features. Thus, the phrasing used here should be toned down.

See our response to the previous comment, where we describe the modifications we made and their basis.

3 – For all ATAC-seq and ChIP-seq data shown, including examples the units should appear on the graphs as normalized equivalently to the read counts (for example in Figure 1B, Figure 1s2C and all the following). A table indicating all published and original data sets used with GSE numbers should be provided, including replicates used and sequencing depth. The presented novel data does not seem to have been submitted to GEO database. This should be done.

We have added the units on the graphs as requested. We have submitted to GEO and provided the accession number: GSE181197. The secure token avcpeismhpehlgd is required to access the series prior to publication. Furthermore, we have provided the metadata for the GEO series, with an added column indicating sequencing depth for each file as the file “220302_GEO submission metadata.”

4 – If the model proposed by the authors is correct, it implies a level of stochasticity in promoter-enhancer dependent activation. Based on sequence elements (for example presence or absence of CpG islands), do the promoters of the NKR locus tend to be less constitutive? More generally how do the authors envision the critical parameters limiting promoter's action and their strong dependency to enhancer stochastic activation? These points could be discussed in the Discussion section.

As noted in our response to Reviewer 2’s comments, we have now expanded the Discussion section to include discussion of our conception of the critical parameters that determine stochasticity. Some of these (enhancer strength) come from our work and others, such as the abundance of relevant transcription factors, come from the work of others, which we cite.

In conclusion, we have made substantial changes to the manuscript to address the comments of all the reviewers. We believe the manuscript is improved as a result.

Author details

1. Djem U Kissiov

   Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Resources, Software, Validation, Visualization, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-6279-342X

2. Alexander Ethell

   Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

   Contribution

   Data curation, Formal analysis, Methodology, Validation, Visualization

   Competing interests

   No competing interests declared

3. Sean Chen

   Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

   Contribution

   Investigation, Methodology, Resources

   Competing interests

   No competing interests declared

4. Natalie K Wolf

   Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

   Contribution

   Data curation, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

5. Chenyu Zhang

   Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

   Contribution

   Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

6. Susanna M Dang

   Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

7. Yeara Jo

   Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

8. Katrine N Madsen

   Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

   Contribution

   Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

9. Ishan Paranjpe

   Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

10. Angus Y Lee

    Cancer Research Laboratory, University of California, Berkeley, Berkeley, United States

    Contribution

    Methodology, Resources, Writing - review and editing

    Competing interests

    No competing interests declared

11. Bryan Chim

    Laboratory of Immune System Biology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, United States

    Contribution

    Formal analysis, Investigation, Methodology, Writing - review and editing

    Competing interests

    No competing interests declared

12. Stefan A Muljo

    Laboratory of Immune System Biology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, United States

    Contribution

    Methodology, Project administration, Resources, Writing - review and editing

    Competing interests

    No competing interests declared

13. David H Raulet

    Division of Immunology and Pathogenesis, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States

    Contribution

    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing - original draft, Writing - review and editing

    For correspondence

    raulet@berkeley.edu

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-1257-8649

• 2,710

  Page views

• 124

  Downloads

• 0

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.