A nonsynonymous LNK polymorphism associated with idiopathic erythrocytosis

Abstract

Idiopathic erythrocytosis (IE) comprises a heterogeneous group of disorders characterized by hyperplasia of the erythroid lineage; however, in many cases, the molecular basis remains undetermined. Serum erythropoietin (EPO) levels can be raised, normal, or reduced, suggesting that there are at least two underlying etiologies involving either the control of EPO production or modulation of EPO-induced signaling. EPO production is regulated by the oxygen-sensing pathway via the hypoxia inducible transcription factor (HIF) complex. Proteasomal turnover of HIF is controlled by interactions with the von Hippel Lindau (VHL) and prolyl hydroxylase domain 2 (PHD2) proteins. Erythrocytosis-associated mutations have been detected in the oxygen sensing pathway indicating that EPO is regulated by the HIF-2alpha-PHD2-VHL axis (reviewed by McMullin [1]). Aberrant EPO-induced signaling in IE patients with subnormal serum EPO levels can arise from mutations in the EPO receptor (EpoR) gene which result in the receptor being hypersensitive to EPO with prolonged activation of the EPO-dependent signaling pathways (reviewed by Percy [2]).

Recent reports have uncovered several mutations in the lymphocyte-specific adaptor protein (LNK) in a variety of myeloproliferative neoplasm (MPNs), including JAK2V617F-negative erythrocytosis, primary myelofibrosis (PMF), and chronic or blast-phase MPNs [3,4].

LNK is a plasma membrane-bound protein whose functions include inhibition of both wildtype and mutant JAK2 activity [5]. LNK is mainly expressed in hematopoietic tissues and contains pleckstrain homology (PH) and SH2 interaction domains. It modulates thrombopoietin and EPO signaling by interacting with JAK2, inhibiting downstream Signal transducer and activators of transcription (STAT) activation [6–8]. Thus, it is plausible that aberrant LNK function may allow dysregulation of EPO-induced signaling resulting in hypersensitivity to EPO and consequently erythrocytosis.

The first known disease-associated LNK mutations were somatically acquired. These include a small internal deletion that leads to a premature termination of protein translation and a functionally relevant mis-sense mutation in the PH domain (p.Glu208Gln) in patients with PMF and ET, respectively [4]. In a study of 341 MPN patients, 11 LNK mutations were found in cases of ET, PMF, and chronic myelomonocytic leukemia but not erythrocytosis [9]. Notably, in two instances the mutations were found in the germline of the patients. In an independent study of a cohort of 172 patients with chronic phase and blast-phase MPN, eight additional mutations in the LNK PH domain were described. These reports suggest that LNK mutations target an exon 2 “hot spot” in the PH domain spanning residues Glu208-Asp234 [10,11].

Finally, in a group of eight patients of IE with subnormal serum EPO levels and no alterations in JAK2, two cases were found to harbor mutations in the PH domain of LNK [3]. The above results support LNK mutations a possible cause of MPNs. This includes otherwise unexplained cases of erythrocytosis such as those with normal or low EPO levels. To understand the potential role of LNK in the development of IE, we investigated LNK in a group of 23 patients with subnormal serum EPO levels, with wild-type EPOR, and JAK2. The clinical parameters of the patient cohort are displayed in Table I.

TABLE I.

Clinical Parameters of 23 Idiopathic Erythrocytosis Patients with Subnormal Serum EPO Levels

	Male	Female	Overall
Sex	15	8	23
Age at presentation	39 (20–60)	35 (19–57)	36 (19–60)
Hb (g/dl)	20.8 (18.1–23.2)	18.6 (17.3–19.9)	19.5 (17.3–23.2
WCC (×109/L)	8.1 (2.4–14.0)	9.1 (6.3–12.2)	8.5 (2.4–13)
Platelets (×109/L)	199 (166–312)	256 (160–400)	221 (160–400)
Erythropoietin (mU/L)	3.1 (0–5)	4.4 (0–9.3)	3.4 (0–9.3)

Our analysis of LNK coding regions did not detect any LNK PH domain mutations in any of the 23 erythrocytosis patients. Instead, we identified, in two cases in the heterozygous state, a nonsynonymous change in the SH2 domain of LNK, p.Glu400Lys (or p.E400K; rs72650673; Fig. 1A). This alteration corresponds to a known single nucleotide polymorphism (SNP) but its frequency in the general population is unknown based on searches of public SNP database (NCBI dbSNP data base located at website address: http://www.ncbi.nlm.nih.gov/snp?term=rs72650673). However, a group of 200 normal control samples were screened by allele refractory mutation specific (ARMS)-PCR but none were positive for the p.Glu400Lys SNP (Table II). In contrast, we found that three additional patients with variable EPO levels were found to be heterozygous for this SNP (Table II). Thus, five out of 96 (5%) of the total IE patients possess this SNP, in comparison to zero out of 200 (0%) control subjects (P = 0.0033, two-tailed Fisher’s exact test, Table II). Taken together, our data suggest an association of the LNK p.Glu400Lys polymorphism with erythrocytosis.

Figure 1.

Identification and functional analysis of LNK p.Glu400Lys mutation. (a) LNK p.Glu400Lys (or p.E400K) mutation in patients with idiopathic erythrocytosis. G to A change at base g.41,559 (genomic sequence NT_009775, NCBI) in exon 5 of LNK, which results in exchange of glutamic acid at amino acid 400 with lysine (p.E400K). (b) The p.E400K mutant does not compromise LNK growth inhibitory function. 32D/EpoR cells were infected with retroviruses encoding either wildtype or mutant human LNK in MSCV-IRES-GFP (MIG) bicistronic vectors. The percentage of GFP⁺ cells relative to the initial infection rate is plotted as a measurement of effects in cell growth (details in Design and Methods). Representative of three independent experiments are shown. (c) The W262R mutant does not compromise LNK growth inhibitory function. 32D/EpoR cells were infected with retroviruses encoding either wildtype or mutant human LNK in MIG vectors. The growth inhibition was measures as percentage of GFP⁺ populations decline. W262 was mutated to R, as well as to P that is the equivalent residue in mouse Lnk. As a control, the W217A mutation in LNK, which is equivalent to the PH domain null mutation in mouse Lnk W191A, was generated and tested. Representative of two independent experiments are shown.

TABLE II.

Frequency of Glu400Lys SNP Detected by ARMS-PCR in IE Patients and Normal Control Subjects

Sample	EPO level	Number screened	Number (percentage) positive for Glu400Lys	Overall percentage	P Value
Normal control	N/A	200	0 (0%)	0 %	0.0033*
IE	Low	23	2 (8.6%)	5.2 %
IE	Normal	46	2 (4.3%)
IE	Elevated	27	1 (3.7%)

^aTwo-tailed P value calculated with Fisher’s exact test.

We failed to detect another previously reported polymorphism in the LNK PH domain, p.Trp262Arg (or p.W262R; rs3184504), which is associated with increased proliferation of peripheral blood monocytes in diabetic patients [12]. However, Trp262 is not conserved between human and mouse, raising questions as to its functional significance. Nevertheless, because of previous work suggesting a role of this residue in altered LNK function in diverse human diseases [13–17], we examined its function in this context.

To examine the functional consequences of this polymorphism, we stably expressed E400K and, as controls, a previously established SH2-null mutant, R392E, and wild-type LNK in hematopoietic 32D/EpoR cells (a hematopoietic progenitor cell line stably expressing EpoR). In parallel, we included in our analysis the W262R version. As previously reported, wild-type LNK impaired cell growth while the R392E version was inactive [7,18,19]. Neither the E400K nor the W262R mutation measurably impaired LNK function in this assay (Fig. 1B). We also examined a W262P conversion as this corresponds to the equivalent residue in mouse Lnk. As an additional control, we generated the W217A mutation, which is equivalent to the PH domain null mutation previously reported in mouse Lnk [6,7]. While the W262P version displayed growth inhibitory activity similar to wild-type LNK, the W217A mutation significantly impaired LNK function as expected (Fig. 1C). Together, our data suggest that neither p.Glu400Lys nor p.Trp262Arg SNPs produce any obvious defects in LNK based on this cell system. One caveat of this study is that although the 32D cell assay has been a reliable tool in measuring LNK function, it remains possible that subtle defects are only revealed in the context of whole animal studies.

In summary, we have identified a nonsynonymous polymorphism in LNK (p.Glu400Lys) which is associated with IE. Although functional assays using a cell line model do not indicate that p.Glu400Lys impairs the ability of LNK to inhibit JAK2 activation of STAT5, it does not negate the possibility of a subtle loss of function. Consequently, impaired negative regulation of EPOinduced signaling, albeit minor, may support the erythrocytosis phenotype independent of the patient’s serum EPO level.

Design and Methods

Patients

A group of 181 patients with a raised red cell mass and who did not fulfil the polycythemia vera diagnostic criteria proposed by the British Committee for Standards in Haematology [20] have been referred to Belfast City Hospital from clinics throughout the United Kingdom and Ireland. All patients gave informed written consent on entering the study. The study was approved by the office for Research Ethics Committee, Northern Ireland (06/ NIRO1/57).

Mutation screening

Polymerase chain reaction (PCR)-direct sequencing of exons 2–6 of LNK was performed using standard protocols. A group of 200 normal control samples (Human Random Control DNA panels, ECACC, Salisbury, UK) was screened for the p.Glu400Lys base change by ARMS-PCR. Primers were designed using the primer design program devised by Ye et al. [21].

Functional assays

Interleukin (IL)-3-dependent 32D hematopoietic cells were used to establish a stable cell line expressing the EpoR, designated as 32D/EpoR. While the parental cells did not respond to EPO, 32D/ EpoR cells proliferated in a dose-dependent manner [7]. The effect of LNK in EPO-dependent 32D cell growth was examined by overexpression of wild-type LNK using the Murine Stem Cell Virus (MSCV)-internal ribosomal entry site (IRES)-green fluorescent protein (GFP) (MIG) vector. MIG is a bicistronic vector containing GFP downstream of an IRES. As GFP expression is tightly correlated with the expression of the gene cloned upstream of the IRES, we were able to identify cells expressing LNK by analyzing GFP fluorescence. We introduced either MIG vector alone or LNK into 32D/EpoR cells and determined the fraction of GFP⁺ infected cells 2 days later. We then measured the GFP⁺ fraction every 3 days, as the cells divide, relative to the level 2 days after infection.