Zinc-finger protein ZFP318 is essential for expressionof IgD, the alternatively spliced Igh product madeby mature B lymphocytesAnselm Endersa,1, Alanna Shorta, Lisa A. Miosgea, Hannes Bergmanna, Yovina Sontania, Edward M. Bertrama,b,Belinda Whittleb, Bhavani Balakishnanb, Kaoru Yoshidac, Geoff Sjollemab, Matthew A. Fielda, T. Daniel Andrewsa,b,Hiromi Hagiwarac, and Christopher C. Goodnowa,1

aDepartment of Immunology and bAustralian Phenomics Facility, John Curtin School of Medical Research, Australian National University, Canberra, ACT 2601,Australia; and cDepartment of Biomedical Engineering, Toin University of Yokohama, Aoba-ku, Yokohama 225-8503, Japan

Contributed by Christopher C. Goodnow, February 13, 2014 (sent for review January 14, 2014)

IgD and IgM are produced by alternative splicing of long primaryRNA transcripts from the Ig heavy chain (Igh) locus and serve asthe receptors for antigen on naïve mature B lymphocytes. IgM ismade selectively in immature B cells, whereas IgD is coexpressedwith IgM when the cells mature into follicular or marginal zone Bcells, but the transacting factors responsible for this regulatedchange in splicing have remained elusive. Here, we use a geneticscreen in mice to identify ZFP318, a nuclear protein with two U1-typezinc fingers found in RNA-binding proteins and no known role inthe immune system, as a critical factor for IgD expression. A pointmutation in an evolutionarily conserved lysine-rich domainencoded by the alternatively spliced Zfp318 exon 10 abolishedIgD expression on marginal zone B cells, decreased IgD on follic-ular B cells, and increased IgM, but only slightly decreased thepercentage of B cells and did not decrease expression of othermaturation markers CD21, CD23, or CD62L. A targeted Zfp318null allele extinguished IgD expression on mature B cells andincreased IgM. Zfp318 mRNA is developmentally regulated inparallel with IgD, with little in pro-B cells, moderate amountsin immature B cells, and high levels selectively in mature follicu-lar B cells. These findings identify ZFP318 as a crucial factor reg-ulating the expression of the two major antibody isotypes on thesurface of most mature B cells.

IgHm | IgHd | immunoglobulin isotype | ENU mutation

Ig isotypes with different heavy (H)-chain constant regions aremade by B lymphocytes in a developmentally regulated series

(1). The different antibody isotypes serve as cell surface markersof B-cell maturation, as functionally distinct receptors for B-cellactivation by antigens and as secreted mediators of different anti-body effector functions (2). All B cells begin as immature B cells inbone marrow or fetal liver that express only the IgM isotype on theircell surface (3), comprised of H chains with an N-terminal variabledomain and C-terminal constant region domains, transmem-brane segment, and cytoplasmic tail, paired with Ig light chains.Maturation into follicular B cells, which recirculate among thespleen, lymph nodes, and other secondary lymphoid tissues, ismarked by coexpression of a second isotype, IgD. Each maturefollicular B cells displays a mixture of cell surface B-cell receptors(BCRs) comprising the same variable domain joined to either IgDor IgM constant regions, with greater levels of IgD than IgM (4, 5).B cells undergo isotype switching after activation by microbialantigens and helper T cells: They irreversibly lose IgM and IgD andswitch to expressing the same variable domain linked to IgG, IgA,or IgE constant region domains. Although the process of isotypeswitching to IgG, IgA, and IgE is well understood, the mechanismfor developmentally regulated IgD expression remains obscure.The developmental order of antibody isotype expression is

reflected in the layout of the Ig heavy chain locus, Igh. In surfaceIgM+ immature B cells, transcription begins with two variableexons (LH and VDJH) formed by intrachromosomal recombination

of separate LHV, D, and J elements in pre-B cells. Downstreamfrom the VDJH exon are six Ighm constant region exons encodingthe extracellular and transmembrane segments of membrane IgM,then five Ighd constant region exons encoding the correspondingsegments of IgD, and finally similar sets of Ighg, Ighe, or Igha exonsencoding the constant regions of IgG, IgE, and IgA. Isotypeswitching results from further DNA recombination within the locusthat deletes the Ighm and Ighd exons and brings either the Ighg,Ighe, or Igha exons immediately 3′ to the VDJH exon, so that thelatter is spliced to IgG, IgE, or IgA constant region exons in theresulting mRNA (6–8). IgD is the exception, however, because mostB cells do not express IgD by DNA recombination but insteadvia a reversible, developmentally regulated process of alterna-tive mRNA splicing of the VDJH exon to the Ighm and Ighd exons(5, 9, 10). This unique arrangement for coexpression of IgM andIgD mRNA by alternative splicing is conserved in bony fish,amphibians, reptiles, monotremes, and mammals (11), yet it is notknown how IgD mRNA is selectively produced in mature B cells.Pre-B cells and immature B cells express very little IgD

mRNA and express only IgM, despite transcribing the Ighd exonsat levels that are often only two- to threefold lower than the Ighmexons and not differing between IgD+ and IgD− IgM+ B cells,when measured by RNA-polymerase run-on experiments inisolated nuclei (12–16). These results have led to the hypothesisthat 25-kb-long Igh pre-mRNA transcripts traverse from theVDJH exon through the Ighd exons in both immature and mature

Significance

Mammalian B lymphocytes make antibodies of five differentheavy chain isotypes, IgM, IgD, IgG, IgE, and IgA. The differentisotypes are produced at discrete stages in B-cell developmentfrom a single immunoglobulin heavy chain (Igh) gene, either byirreversible rearrangement of the gene to make IgG, IgE, or IgAor by alternative splicing of the RNA transcribed from the Ighgene to coexpress IgM and IgD. Developmentally regulatedtrans-acting factors have been hypothesized to control IgMand IgD expression from large Igh RNAs, but these factors haveremained elusive for several decades. Here, using a genome-wide mutation screen in mice, we identify an obscure gene,Zfp318, as encoding a specific and essential factor promotingIgD expression in mature B cells.

Author contributions: A.E., A.S., L.A.M., H.B., Y.S., E.M.B., B.W., B.B., G.S., M.A.F., T.D.A.,and C.C.G. designed research; A.E., A.S., L.A.M., H.B., Y.S., B.W., B.B., G.S., M.A.F., T.D.A.,and C.C.G. performed research; K.Y. and H.H. contributed new reagents/analytic tools;A.E., A.S., L.A.M., H.B., Y.S., E.M.B., B.W., B.B., G.S., M.A.F., T.D.A., and C.C.G. analyzeddata; A.E. and C.C.G. wrote the paper.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.1To whom correspondence may be addressed. E-mail: chris.goodnow@anu.edu.au oranselm.enders@anu.edu.au.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1402739111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1402739111 PNAS | March 25, 2014 | vol. 111 | no. 12 | 4513–4518

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

28, 2

020

B cells, but an unknown transacting factor alters splicing eitherby: (i) promoting RNA cleavage at Ighm polyadenylation sites inimmature B cells to preclude VDJH splicing to Ighd; or (ii) si-lencing Ighm cleavage and polyadenylation sites in mature Bcells to allow splicing to Ighd (13). In some immature B cells,failure to express IgD also appears to reflect unloading ofRNA Pol II at an attenuation region 3′ to Ighm and 5′ to Ighd,but when this region is removed, there is still little splicing toIgD (16, 17). In contrast, in terminally differentiated plasmacells, transcription termination occurs upstream of Ighd, re-sulting in very low expression of Ighd mRNA.Although differential expression of IgM and IgD was one of the

first examples of developmentally regulated alternative mRNAsplicing, progress to understand its basis has stalled because it hasnot been possible to identify the nature of the transacting factors.Here, we use a phenotype-driven genetic screen in mice to identifya gene that fulfils the criteria for encoding the elusive trans-activating factor promoting IgD expression.

ResultsIdentification of a Missense Mutation in Zfp318 Causing DecreasedIgD and Increased IgM. In a peripheral blood screen of mice inher-iting ethylnitrosourea (ENU)-induced point mutations, we identi-fied a pedigree with a Mendelian recessive mutation characterized

by decreased IgD and increased IgM on mature B cells (Fig. 1 Aand B). Homozygotes had one-third as much IgD as wild-typelittermates and unrelated controls, whereas heterozygotes had an∼25% decrease. The frequency of B cells was slightly reduced(Table 1). Exome sequencing of an affected animal followed bygenotyping of candidate mutations in a large cohort of siblingsand offspring revealed that the low IgD trait was completely cor-related with inheritance of a point mutation in the gene encodingzinc-finger protein (ZFP) 318.ZFP318 is also called testicular zinc-finger protein (TZF) and

has been implicated in transcriptional regulation in testes withgenetic deficiency causing infertility in mice (18–21) but has noknown function in the immune system. Microarray comparisonsof gene expression in B-cell subsets have identified Zfp318 asa member of a set of mRNAs that increases during maturation ofimmature B cells into mature follicular B cells (22–24). By ana-lyzing flow-sorted B-cell subsets, we confirmed expression ofZfp318 mRNA closely parallels IgD heavy chain (Ighd) mRNAduring B-cell development (Fig. 1C). There was very little Zfp318in pro-B cells in the bone marrow, moderate amounts in imma-ture (CD93+, CD62L−) B cells in the spleen, and high amounts inmature follicular (CD93−, CD62L+) B cells. The close correlationbetween IgD and Zfp318 expression is reinforced in the IMM-GEN dataset (22, 23), which shows highest Zfp318 expression in

wild-type homozygote

CD19

CD

3

Lymphocytes

IgM

IgD

0

10

20

30

0

2

4

6

8

IgM

MFI

x 1

03

IgD

MFI

x 1

03

wt het hom wt het hom

pro-B

012345

0

2

4

6

Ighd

(a.u

. x 1

03 )

Zfp3

18 (a

.u. x

102 )

immatu

re

mature

pro-B

immatu

re

mature

1 2 3 4 5 6 7 89

1 2 3 4 5 6 7 11

10

Ile1347ThrCCDS 28827.2 Zfp318 long

CCDS 28826.2 Zfp318 short

U1-type C2H2 zinc finger

Lysine-rich IgD-promoting domain

Proline-rich domain

2237 aa

1154 aa

0102 105104103

0102

105

104

103

0102 105104103

0102

105

104

103

wild-type homozygote B cells

87.2 76.3

57.2

30.9

64

20.4

A B

C

D

E

Fig. 1. Decreased IgD and increased IgM on circulating B cells from mice with a point mutation in the long isoform of Zfp318. (A) Representative flowcytometry of peripheral blood lymphocytes from a homozygous Zfp318 mutant mouse and a wild-type littermate showing the frequency of CD3+ T cells andCD19+ B cells among lymphocytes (Upper), and the frequency of IgMlow IgDhi mature B cells among CD19+ B cells (Lower). (B) Geometric mean fluorescentintensity (MFI) of IgM and IgD on blood CD19+ B cells in Zfp318 homozygous mutant, heterozygous, and wild-type littermates. (C) Relative abundance ofZfp318 and Ighd mRNA measured in arbitrary units in sorted bone marrow pro-B cells and splenic immature (CD93+ IgMhi) and mature (IgMlo CD93−) B cells.(D) Schematic of the two isoforms of Zfp318 generated by alternative splicing of the numbered exons and location of ENU-induced mutation. (E) Evolutionaryconservation of the LRID. Highly conserved residues are in bold and the mutated Ile1347 residue in red. Accession nos.: Homo sapiens (human),XP_005249038.1; Pan troglodytes (chimpanzee), XP_518490.3; Mus musculus (mouse), AAI50731.1;Monodelphis domestica (opossum), XP_001365292.1; Falcocherrug (falcon), XP_005439139.1; Gallus gallus (chicken), XP_419507.4; Chelonia mydas (sea turtle), EMP24462.1; Ophiophagus hannah (king cobra),ETE68519.1; Alligator sinensis (alligator), XP_006034767.1; Xenopus tropicalis (frog), XP_004914880.1; Latimeria chalumnae (coelacanth), XP_006013622.1;Metriaclima zebra (cichlid fish), XP_004575369.1; Danio rerio (zebrafish), XP_002664136.3.

4514 | www.pnas.org/cgi/doi/10.1073/pnas.1402739111 Enders et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

28, 2

020

IgDhigh follicular B cells and T3 transitional B cells, lower Zfp318in IgDlow marginal zone B cells and T1-T2 transitional B cells,and very low expression in IgD-negative IgM+ immature B cells inthe bone marrow.Two alternatively spliced isoforms of Zfp318 mRNA are an-

notated in the genome and described in the literature (18) (Fig.1D). The long isoform (CCDS 28827.2) includes exons 1–10 andencodes a 2,237-aa protein containing two C2H2 zinc-fingerdomains of the U1 ribonucleoprotein type (25). A 44-aa linkerseparates the two zinc fingers, similar to the 34- to 44-aa linkersbetween the dsRNA-binding U1-type zinc fingers in the JAZ andZFa proteins and unlike the 6- to 8-aa linkers typically presentbetween C2H2 zinc fingers in DNA binding proteins (26). Theshort isoform (CCDS 28826.2) skips exons 8, 9, and 10 and,

consequently, encodes a protein of 1,154 aa lacking the secondzinc finger and the polyproline domain. The IgD-lowering mu-tation was a nonsynonymous T > C transition in the differentiallyspliced exon 10. Because this mutation only alters the mRNA andprotein sequence of the long form, it demonstrates that the longZfp318 isoform promotes normal IgD expression. The mutationchanged codon 1347 in the long isoform from a highly conservedhydrophobic isoleucine into a polar threonine. Ile1347 lies in anunannotated domain between the second zinc finger and thepolyproline domain, containing 16 lysine residues that are highlyconserved in mammals, birds, reptiles, and bony fish (Fig. 1E).We refer to this domain as the lysine-rich IgD-promoting domain(LRID). It is notable that the conserved Ile residue is substitutedto Glu in Maylandia zebra and other Cichlid fish species,

Table 1. Frequency of B-cell subpopulations in bone marrow and spleen in Zfp318 point mutant mice and in the blood of Zfp318−/−

Genotype Organ B cells Pro-B cells Pre-B cellsImmatureB cells

MatureB cells

TransitionalB cells

FollicularB cells MZ B cells No. of animals

WT BM 21.1 ± 2.7 9.0 ± 1.5 22.1 ± 0.6 15.1 ± 1.0 36.5 ± 0.4 — — — 5Het BM 20.4 ± 2.1 8.4 ± 0.9 18.7 ± 3.2 14.2 ± 0.9 39.6 ± 3.2 — — — 6Hom BM 21.1 ± 2.0 7.4 ± 2.4 17.2 ± 3.2 12.6 ± 2.5 45.2 ± 7.6 — — — 4WT Spleen 57.9 ± 3.9 — — — — 25.5 ± 7.0 56.0 ± 4.7 10.9 ± 3.4 9Het Spleen 56.1 ± 4.6 — — — — 25.4 ± 6.8 56.9 ± 4.7 10.3 ± 2.1 10Hom Spleen 51.0 ± 3.0 — — — — 21.9 ± 7.2 59.6 ± 5.2 11.4 ± 2.8 7Zfp318+/+ Blood 62.0 ± 4.8 — — 21.4 ± 5.1 71.7 ± 6.7 — — — 4Zfp318+/− Blood 57.9 ± 5.1 — — 17.5 ± 4.2 74.6 ± 5.0 — — — 22Zfp318−/− Blood 53.5 ± 4.2 — — 19.1 ± 6.1 69.8 ± 7.4 — — — 15

The frequencies are expressed as mean values ± SD out of all live cells (B cells) or as a percentage of all B cells for all other populations. The last columnshows the number of animals analyzed per genotype analyzed in one (BM and blood) or two (spleen) experiments. Statistical analysis was done comparing allgroups within one organ by one-way ANOVA followed by Bonferroni post hoc test to compare each population in the heterozygous and homozygous mice tothe relevant WT control. Statistically significant differences are marked in bold. B cells, B220+ in live lymphocytes gate for BM and in lymphocyte gate forspleen. Immature B cells, IgM+CD93+; mature B cell, CD93−; pre-B cell, CD24hiCD43−CD93+IgD−IgM−; pro-B cell, CD24intCD43+CD93+IgD−IgM−; in the bonemarrow. Follicular, CD93−CD23hiCD21med; marginal zone B cells, CD93−CD23−CD21hi; transitional B cells, CD93+; in the spleen and total B cells (CD19+),immature (CD93+), and mature (CD93−) B cells in the blood of Zfp318−/− mice as well as Zfp318+/− and Zfp318+/+ littermate controls.

A

B Immature B cells Mature B cells

IgM

MFI

x 1

02

IgM

MFI

x 1

02

WT

Hom

CD93-CD19+ B cells

0 102 105104103

0102

105

104

103

0 102 105104103

IgM

IgD

** ##

* ##

ns ***#

ns nsns

IgD

MFI

x 1

03

IgD

MFI

x 1

02

WT Het Hom0

1

2

3

4

WT Het Hom0

20

40

60

80

WT Het Hom0

5

10

15

20

25

WT Het Hom0

20

40

60

Live cells

FSC

B220 0 200k100k

0102

105

104

103

25.1

23.7

B220+ cells

CD93

CD19 0 102 105104103

0102

105

104

103

35.9 52.4

38.9 47.6

CD93+ B cells

0 102 105104103

0102

105

104

103

5 0 102 105104103

IgM

IgD

64.4

64.7

precursor B cells

0 102 105104103

0102

105

104

103

0 102 105104103

CD43

CD24

64.7 26.5

3.6

64.7 27.2

4.3

Fig. 2. B-cell development in bone marrow ofZfp318 point mutant mice. (A) Flow cytometry ofbone marrow lymphocytes from Zfp318 mutant mice(Hom) and wild-type littermate controls (WT). B cellswere gated as B220+ cells. Expression of IgM and IgDwas analyzed on CD93+ immature and CD93− matureB cells. (B) MFI of IgM and IgD on immature andmature B cells in the bone marrow gated as in A.Statistical analysis by one-way ANOVA followed byBonferroni post hoc test: ns, P > 0.05; *P < 0.05; **P <0.01; ***P < 0.001; #P < 0.0001.

Enders et al. PNAS | March 25, 2014 | vol. 111 | no. 12 | 4515

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

28, 2

020

although we are unaware of any data on IgD expression inthese species.

Effect of Zfp318I1347 Mutation on B-Cell Developmental Subsets inBone Marrow and Spleen. Analysis of B-cell development in thebone marrow of Zfp318 mutant mice and littermate controlsshowed no significant difference in either percentage of B cellsor subset distribution of developing B cells but a small increasein the proportion of mature recirculating B cells (Fig. 2A andTable 1). Interestingly, the subset of CD93+, IgM+ immature Bcells in the bone marrow that starts to express IgD at very lowlevels already showed a reduced expression of IgD comparedwith either wild-type (P < 0.0001) or heterozygous (P < 0.001)littermate controls without a statistically significant change ofIgM expression (Fig. 2B). In contrast, the mature, recirculating Bcells in the bone marrow had reduced expression of IgD andincreased expression of IgM (Fig. 2B).Splenic B cells in homozygous mutants showed a similarly de-

creased IgD and increased IgM, but normal expression of othermature B-cell surface markers with the exception of slightly in-creased CD23 (Fig. 3 A–C). In normal B cells, CD93 expression isextinguished, whereas CD62L, CD23, CD21, and BAFF-Receptorare induced when immature B cells in the spleen mature into re-circulating follicular B cells, and these developmentally regulated

events were comparable in Zfp318I1347T homozygous mutantand wild-type mice. The two main subsets of mature B cells in thespleen, CD23+ CD21med follicular B cells and CD23− CD21hi

marginal-zone B cells, were both present in normal frequencies(Fig. 3A and Table 1). The lower level of IgD expressed onwild-type marginal zone B cells was completely extinguishedin Zfp318I1347T homozygotes, representing a decrease of ∼10-foldon marginal zone B cells compared with only ∼threefold decreasedIgD on follicular B cells (Fig. 3B).

Loss of IgD Expression in Mice with a Zfp318 Null Mutation. Theabove finding that IgD expression was extinguished on marginalzone B cells but persisted at moderate levels on homozygousmutant follicular B cells had two alternative explanations: (i)another gene might also contribute to IgD expression in follic-ular B cells; or (ii) the Zfp318I1347T point mutation may onlypartially compromise ZFP318 function. To resolve these alter-natives, we analyzed circulating B cells in mice carrying a nullZfp318 allele generated by a targeted insertion in exon 2. Homo-zygous Zfp318 null mice had a normal percentage of circulatingB cells and a normal fraction of these B cells that had matured tothe CD93-negative stage (Fig. 4A), but IgD expression was almostcompletely abolished on immature and mature B cells (Fig. 4B and C). IgM was increased threefold on mature B cells from

WT

HetHom W

THet

Hom WT

HetHom

0

5

10

15

Immature Follicular Marginal zone

** ##

# ##

# ##

IgD

MFI

x 1

03

WT

HetHom W

THet

Hom WT

HetHom

0

5

10

15

20

25

ns ##

* ##

ns ##

IgM

MFI

x 1

03

Immature Follicular Marginal zone

B

WT

Hom

0 102 105104103

0102

105

104

103

0 102 10104103

IgM

IgD

CD93+ B cellsA

0 102 105104103

0102

105

104

103

0 102 104103

B220

TCRβ

Lymphocytes

35.2

55.8

42.3

47.1

0 102 105104103

0102

105

104

103

0 102 1104103

CD93

CD62L

B220+ B cells

64.5 31.1

62.6 32.4

0 102 105104103

0102

105

104

103

0 102 1104103

CD23

CD21

CD93- B cells

78.514.1

79.314.5

20

BA

FF-R

MFI

x 1

02

WT Het Hom0

5

10

15 ns nsns

WT Het Hom0

1

2

3

4 ns nsns

CD

21 M

FI x

103

WT Het Hom0

5

10

15ns ns

ns

BA

FF-R

MFI

x 1

02

WT Het Hom0

5

10

15 ns nsns

WT Het Hom0

0.5

1

ns nsns

CD

62L

MFI

x 1

03

WT Het Hom0

2

4

ns nsns

CD

62L

MFI

x 1

03

WT Het Hom0

1

2

3 ns ****

CD

23 M

FI x

103

CD

21 M

FI x

103

WT Het Hom0

1

2

3 ns nsns

CD

23 M

FI x

102

Follicular B cells

MZ B cells

C

Fig. 3. B-cell subsets in spleen of Zfp318 point mutant mice. (A) Representative flow cytometry of splenic lymphocytes from Zfp318 mutant mice (Hom) andWT littermate controls showing the frequency of TCRβ+ T cells and B220+ B cells among lymphocytes (Left), frequency of CD93+ immature and CD93-CD62L+

mature B cells (Center Left), frequency of CD23+ follicular and CD21+ marginal zone B cells among CD93− mature B cells as well as straining for IgM and IgD onmature CD93− B cells (Center Right and Right). (B) MFI of IgM and IgD on immature, follicular, and marginal zone B cells in the spleen gated as in A. (C) MFI ofantibody staining for BAFF-receptor, CD23, CD21, and CD62L on mature follicular B cells (Upper) and marginal zone B cells (Lower) in the spleen. Statisticalanalysis by one-way ANOVA followed by Bonferroni post hoc test: ns, P > 0.05; *P < 0.05; **P < 0.01; #P < 0.0001.

4516 | www.pnas.org/cgi/doi/10.1073/pnas.1402739111 Enders et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

28, 2

020

homozygous Zfp318 null mice. Thus, ZFP318 is absolutelyrequired for IgD expression and the point mutation onlypartially inactivates its function.

DiscussionThe findings above identify ZFP318 as a long-sought transactingfactor governing differential expression of IgM and IgD isotypesduring B-cell maturation. Earlier work demonstrated that 25-kb-long heavy chain pre-mRNAs are transcribed from the VDJHexon through the Ighm and Ighd exons in both immature andmature B cells (12–16). Consequently, it was hypothesized thatthe alternative splice acceptor sites on the first Ighm and Ighdconstant region exons must compete for the single splice donorsequence in the VDJH exon, with unknown transacting factorseither suppressing RNA splicing to the Ighd exons in immature Bcells or promoting Ighd splicing in mature B cells at the expenseof Ighm. The results here support the latter hypothesis: Zfp318mRNA increases during B-cell maturation in parallel with in-creasing IgD, and Zfp318 mutations increase IgM and extinguishIgD expression.Zfp318 appears to be specifically required for balancing IgD

and IgM output from Igh, and not for the overall program ofB-cell maturation. Zfp318 mutation had no effect on the accu-mulation of mature B-cell subsets nor did it diminish the ex-pression of CD21, CD62L, or CD23 on mature B cells, despitethese markers being transcriptionally regulated in parallel with IgDas part of the B-cell maturation program. CD23 and CD21 areinduced during B-cell maturation by BAFF-receptor signaling toNF-κB transcription factors, whereas IgD expression does not de-pend on this transcriptional regulatory system (27, 28). It will beinteresting to see whether there are other Zfp318-dependent genesthrough global gene expression analyses in the mutant B cells.

The similarities and differences between the Zfp318 pointmutant and null mutant have several implications. In mice withthe point mutation, follicular B cells retained substantial IgDexpression, whereas IgD was almost fully extinguished by the nullmutation. Because the point mutation is within the alternativelyspliced exon 10 and only alters the mRNA and protein sequenceof the ZFP318 long isoform, the residual IgD-promoting activitymay reflect action of the short isoform. However, the fact thatthe point mutation fully extinguished IgD expression on marginalzone B cells and decreased IgD on follicular B cells shows thatthe long isoform has the major role in promoting normal IgD ex-pression. In cell transfection studies, the long and short ZFP318isoforms have been shown to have opposite stimulatory and in-hibitory effects, respectively, on transcription induced by the an-drogen receptor (20). Hence, it is possible that only the ZFP318long form promotes IgD expression and the point mutation crip-ples, but does not abolish, its activity, so that residual IgD ex-pression occurs only in follicular B cells with the highest Zfp318mRNA levels. The point mutation changes an isoleucine in thelong isoform that is conserved from fish to humans within anunannotated domain marked by 16 lysine residues that are alsohighly conserved. Based on the effects of the point mutation, wepropose to call this conserved region the LRID domain. Asoccurs in well-characterized RNA-binding domains, the chargedlysine residues in the LRID domain may bind RNA and co-operate with RNA binding by the two U1-type zinc finger domainsin the ZFP318 long isoform.Although it is possible that ZFP318 controls IgD and IgM

expression at the level of protein trafficking or by inhibiting ex-pression or activity of mRNA polyadenylation cleavage enzymes,several lines of evidence favor a simpler hypothesis that ZFP318directly regulates alternative RNA splicing of Ighm and Ighdconstant region exons. First, differential expression of IgD dur-ing B-cell maturation has been shown to occur at the level ofmRNA production (12, 13, 29–31) Second, the ZFP318 zincfingers are of the U1 type defined by the RNA-binding zincfinger in the spliceosomal U1C protein (25, 32) and of thezf-C2H2_JAZ superfamily that bind double-stranded RNA orRNA-DNA hybrids (33). Third, both the long and short isoformsof ZFP318 accumulate selectively in the nucleus, with the shortisoform localized to subnuclear speckles that contain histonedeacetylase 2 (HDAC2) and are adjacent to nuclear specklescontaining serine/arginine-rich splicing factor 2 (SRSF2, SC-35;refs. 18–21). ZFP318 binding to HDAC2 (21) is potentiallysimilar to the recently demonstrated interaction between theRNA-binding protein HuR and HDAC2 to influence alternativemRNA splicing (34). A large amount of alternative splicingoccurs cotranscriptionally when pre-mRNAs are still chromatinassociated, where it is governed by two-way cooperation betweenRNA-binding splicing factors such as HuR or SRSF proteins, theextent of histone acetylation along intragenic chromatin, and thespeed of RNA polymerase II (Pol II) transcript elongation alongchromatin (34–38).The evidence above, combined with earlier models for Ighd

expression (12, 13, 31), leads us to propose the following simplehypothesis for ZFP318-dependent IgD expression (Fig. S1). (i)The rate of VDJH exon splicing to Ighd competes with the rate ofIgh pre-mRNA cleavage at the Ighm polyadenylation site. Be-cause the latter is located 5′ to Ighd on the pre-mRNA, ifcleavage occurs first, it precludes VDJH splicing to Ighd. (ii) Inthe absence of ZFP318, Pol II elongates the Ighm-Ighd pre-mRNA at a slower rate than polyadenylation site cleavage, sothat most pre-mRNAs are cleaved at the Ighm polyadenylationsite before Pol II has transcribed the Ighd exons. (iii) In mature Bcells, ZFP318 is recruited to the Igh pre-mRNA via its U1-typezinc fingers and its LRID domain and associates with and inhibitsHDAC2, thereby promoting hyperacetylation of Igh chromatinand more rapid Pol II elongation of the Ighm-Ighd pre-mRNA.When the rate of Igh pre-mRNA elongation exceeds the rateof polyadenylation site cleavage, a substantial fraction of Ighdexons are now spliced to VDJH. Variations to this hypothesis

0 10 101010

010

10

10

10

0 10 101010

010

10

10

10

CD19

CD

3

CD93

CD

19

+/+

-/-

Lymphocytes B cells

25.7

66.3

76 21.9

30.4

61.7

78.2 20

Zfp318-/- T cells Zfp318-/- B cells Zfp318+/+ B cells

IgM

CD19+ B cells

IgD

0 10 101010

0 10 101010

A B

C

0

5

10

15

IgM

MFI

x 1

0

+/+ -/-+/-0

5

10

15Ig

D M

FI x

10

+/+ -/-+/-

mature B cellsimmature B cells

IgM

MFI

x 1

0

0

5

10

+/+ -/-+/-

IgD

MFI

x 1

0

0

0.5

1.0

1.5

2.0

2.5

+/+ -/-+/-

Fig. 4. Loss of IgD expression on circulating B cells of mice with a Zfp318null mutation. (A) Representative flow cytometry of peripheral blood lym-phocytes from a Zfp318 homozygous null (−/−) and wild-type littermatecontrol (+/+) showing the frequency of CD3+ T cells and CD19+ B cells amonglymphocytes (Left), and the frequency of CD93− mature and CD93+ imma-ture B cells among CD19+ lymphocytes (Right). (B) Expression of IgM (Upper)and IgD (Lower) on B cells from Zfp318 homozygous null (−/−, black line) andwild-type (+/+, shaded). As a negative control, staining on T cells fromZfp318−/− mice is also shown (dashed line). (C) MFI of IgM and IgD on CD93+

immature and CD93− mature B cells from Zfp318−/− mice compared withheterozygous and wild-type littermate controls.

Enders et al. PNAS | March 25, 2014 | vol. 111 | no. 12 | 4517

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

28, 2

020

should nevertheless also be considered, including the possibilitythat ZFP318 binds particular sites in Igh pre-mRNA to suppressrecognition of Ighm splice acceptors or polyadenylation sites or toenhance spliceosomal recognition of Ighd splice acceptors.Testing these hypotheses in the future can be facilitated by the

Zfp318 mouse mutants described here, for example by ChipSeqexperiments to define ZFP318 binding sites and measurement ofhistone acetylation and pol II elongation rates in the Igh locusof mutant and wild-type B cells. Overall, the findings herereveal the function of a regulator of B-cell maturation andopen up avenues to understand the regulation of alternativemRNA splicing.

Materials and MethodsMouse Strains and Procedures. The Zfp318 point mutant strain was identifiedby flow cytometry screening of peripheral blood lymphocytes in third-gen-eration offspring from ENU-treated C57BL/6 mice as described (39). Identi-fication of the causal mutation was done by whole exome sequencing asdescribed (40, 41). The Zfp318−/− mice were generated by H.H. by insertinga GFP-Neomycin cassette into exon 2 of the Zfp318 gene. Knock-out mice ona C57BL/6 background were obtained from the RIKEN BioResource Center(RBRC01768). All animals were housed under specific pathogen-freeconditions at the Australian Phenomics Facility. All animal experiments

were approved by the Australian National University Animal Ethics andExperimentation Committee.

FACS. Lymphocytes from blood, spleen, and bone marrow were prepared,stained, and analyzed by flow cytometry according to published methods(42). Samples were analyzed by using a BD LSR II or LSRFortessa flowcytometer.

Microarray Analysis. IgD−IgM−CD43intCD24int Pro-B cells were sorted fromthe bone marrow, and CD93−IgMlow mature and CD93highIgMhigh immatureB cells were sorted from the spleen of naive wild-type C57BL/6 mice by flowcytometry. Isolated cells were pelleted and snap frozen in liquid nitrogenbefore shipment to Miltenyi Biotech Genomic Services (Bergisch Gladbach)for RNA extraction and global gene expression analysis by Agilent singlecolor 8 × 60K Whole Mouse Genome Microarray.

ACKNOWLEDGMENTS. We thank the staff the Australian Phenomics Facilityfor excellent animal husbandry and the staff of the John Curtin School ofMedical Research Microscopy and Cytometry Resource Facility. We alsothank Nadine Barthel for excellent technical assistance. This work wassupported by National Institutes of Health Grant U19 AI100627, a RamaciottiFoundation grant (to A.E. and C.C.G.), and National Health and MedicalResearch Council of Australia Grants 585490, 1016953 (to C.G.G.), and1035858 (to A.E.).

1. Hirano M, Das S, Guo P, Cooper MD (2011) The evolution of adaptive immunity invertebrates. Adv Immunol 109:125–157.

2. Hardy RR, Kincade PW, Dorshkind K (2007) The protean nature of cells in the Blymphocyte lineage. Immunity 26(6):703–714.

3. Lawton AR, 3rd, Asofsky R, Hylton MB, Cooper MD (1972) Suppression of immuno-globulin class synthesis in mice. I. Effects of treatment with antibody to -chain. J ExpMed 135(2):277–297.

4. Vitetta ES, et al. (1975) Cell surface immunoglobulin. XI. The appearance of an IgD-like molecule on murine lymphoid cells during ontogeny. J Exp Med 141(1):206–215.

5. Blattner FR, Tucker PW (1984) The molecular biology of immunoglobulin D. Nature307(5950):417–422.

6. Honjo T, Kataoka T (1978) Organization of immunoglobulin heavy chain genes andallelic deletion model. Proc Natl Acad Sci USA 75(5):2140–2144.

7. Cory S, Jackson J, Adams JM (1980) Deletions in the constant region locus can accountfor switches in immunoglobulin heavy chain expression. Nature 285(5765):450–456.

8. Muramatsu M, et al. (2000) Class switch recombination and hypermutation requireactivation-induced cytidine deaminase (AID), a potential RNA editing enzyme. Cell102(5):553–563.

9. Moore KW, et al. (1981) Expression of IgD may use both DNA rearrangement and RNAsplicing mechanisms. Proc Natl Acad Sci USA 78(3):1800–1804.

10. Maki R, et al. (1981) The role of DNA rearrangement and alternative RNA processingin the expression of immunoglobulin delta genes. Cell 24(2):353–365.

11. Ohta Y, Flajnik M (2006) IgD, like IgM, is a primordial immunoglobulin class perpet-uated in most jawed vertebrates. Proc Natl Acad Sci USA 103(28):10723–10728.

12. Yuan D, Tucker PW (1984) Transcriptional regulation of the mu-delta heavy chainlocus in normal murine B lymphocytes. J Exp Med 160(2):564–583.

13. Mather EL, Nelson KJ, Haimovich J, Perry RP (1984) Mode of regulation of immuno-globulin mu- and delta-chain expression varies during B-lymphocyte maturation. Cell36(2):329–338.

14. Kelley DE, Perry RP (1986) Transcriptional and posttranscriptional control of immu-noglobulin mRNA production during B lymphocyte development. Nucleic Acids Res14(13):5431–5447.

15. Tisch R, Watanabe M, Letarte M, Hozumi N (1987) Assessment of antigen-specificreceptor function of surface immunoglobulin M and D with identical hapten speci-ficity. Proc Natl Acad Sci USA 84(11):3831–3835.

16. Yuan D, Witte PL, Tan J, Hawley J, Dang T (1996) Regulation of IgM and IgD heavychain gene expression: Effect of abrogation of intergenic transcriptional termination.J Immunol 157(5):2073–2081.

17. Yuan D, Witte PL (1988) Transcriptional regulation of mu and delta gene expressionin bone marrow pre-B and B lymphocytes. J Immunol 140(8):2808–2814.

18. Ishizuka M, et al. (2003) Molecular cloning and characteristics of a novel zinc fingerprotein and its splice variant whose transcripts are expressed during spermatogenesis.Biochem Biophys Res Commun 301(4):1079–1085.

19. Ishizuka M, et al. (2005) A zinc finger protein TZF is a novel corepressor of androgenreceptor. Biochem Biophys Res Commun 331(4):1025–1031.

20. Tao R-H, et al. (2006) Opposite effects of alternative TZF spliced variants on androgenreceptor. Biochem Biophys Res Commun 341(2):515–521.

21. Tao R-H, et al. (2006) Testicular zinc finger protein recruits histone deacetylase 2 andsuppresses the transactivation function and intranuclear foci formation of agonist-boundandrogen receptor competitively with TIF2. Mol Cell Endocrinol 247(1-2):150–165.

22. Debnath I, et al. (2008) Defining a transcriptional fingerprint of murine splenic B-celldevelopment. Genes Immun 9(8):706–720.

23. Heng TSP, Painter MW; Immunological Genome Project Consortium (2008) The Im-munological Genome Project: Networks of gene expression in immune cells. NatImmunol 9(10):1091–1094.

24. Jojic V, et al.; Immunological Genome Project Consortium (2013) Identification oftranscriptional regulators in the mouse immune system. Nat Immunol 14(6):633–643.

25. Muto Y, et al. (2004) The structure and biochemical properties of the human spli-ceosomal protein U1C. J Mol Biol 341(1):185–198.

26. Möller HM, Martinez-Yamout MA, Dyson HJ, Wright PE (2005) Solution structure ofthe N-terminal zinc fingers of the Xenopus laevis double-stranded RNA-bindingprotein ZFa. J Mol Biol 351(4):718–730.

27. Gorelik L, et al. (2004) Cutting edge: BAFF regulates CD21/35 and CD23 expressionindependent of its B cell survival function. J Immunol 172(2):762–766.

28. Grossmann M, et al. (2000) The anti-apoptotic activities of Rel and RelA requiredduring B-cell maturation involve the regulation of Bcl-2 expression. EMBO J 19(23):6351–6360.

29. Yuan D, Tucker PW (1984) Regulation of IgM and IgD synthesis in B lymphocytes. I.Changes in biosynthesis of mRNA for mu- and delta-chains. J Immunol 132(3):1561–1565.

30. Yuan D (1984) Regulation of IgM and IgD synthesis in B lymphocytes. II. Translationaland post-translational events. J Immunol 132(3):1566–1570.

31. Tisch R, Kondo N, Hozumi N (1990) Parameters that govern the regulation of im-munoglobulin delta heavy-chain gene expression. Mol Cell Biol 10(10):5340–5348.

32. Förch P, Puig O, Martínez C, Séraphin B, Valcárcel J (2002) The splicing regulator TIA-1interacts with U1-C to promote U1 snRNP recruitment to 5′ splice sites. EMBO J 21(24):6882–6892.

33. Yang M, May WS, Ito T (1999) JAZ requires the double-stranded RNA-binding zincfinger motifs for nuclear localization. J Biol Chem 274(39):27399–27406.

34. Zhou H-L, et al. (2011) Hu proteins regulate alternative splicing by inducing localizedhistone hyperacetylation in an RNA-dependent manner. Proc Natl Acad Sci USA108(36):E627–E635.

35. de la Mata M, et al.; la Mata de M et al (2003) A slow RNA polymerase II affects al-ternative splicing in vivo. Mol Cell 12(2):525–532.

36. Luco RF, et al. (2010) Regulation of alternative splicing by histone modifications.Science 327(5968):996–1000.

37. Schor IE, et al. (2012) Perturbation of chromatin structure globally affects localizationand recruitment of splicing factors. PLoS ONE 7(11):e48084.

38. Khan DH, et al. (2014) RNA-dependent dynamic histone acetylation regulates MCL1alternative splicing. Nucleic Acids Res 42(3):1656–1670.

39. Nelms KA, Goodnow CC (2001) Genome-wide ENU mutagenesis to reveal immuneregulators. Immunity 15(3):409–418.

40. Bergmann H, et al. (2013) B cell survival, surface BCR and BAFFR expression, CD74metabolism, and CD8- dendritic cells require the intramembrane endopeptidaseSPPL2A. J Exp Med 210(1):31–40.

41. Andrews TD, et al. (2012) Massively parallel sequencing of the mouse exome to ac-curately identify rare, induced mutations: An immediate source for thousands of newmouse models. Open Biol 2(5):120061.

42. Yabas M, et al. (2011) ATP11C is critical for the internalization of phosphatidylserineand differentiation of B lymphocytes. Nat Immunol 12(5):441–449.

4518 | www.pnas.org/cgi/doi/10.1073/pnas.1402739111 Enders et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

28, 2

020