Advertisement
Lnk controls mouse hematopoietic stem cell self-renewal and quiescence through direct interactions with JAK2
Alexey Bersenev,1 Chao Wu,1 Joanna Balcerek,1 and Wei Tong1,2
1Division of Hematology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA. 2Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA.
Address correspondence to: Wei Tong, Abramson Building, Suite 316A, Children’s Hospital of Philadelphia, 3615 Civic Center Boulevard, Philadelphia, Pennsylvania 19104-4318, USA. Phone: (267) 426-0930; Fax: (267) 426-5476; E-mail: tongw@email.chop.edu.
Find articles by Bersenev, A. in: JCI | PubMed | Google Scholar
1Division of Hematology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA. 2Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA.
Address correspondence to: Wei Tong, Abramson Building, Suite 316A, Children’s Hospital of Philadelphia, 3615 Civic Center Boulevard, Philadelphia, Pennsylvania 19104-4318, USA. Phone: (267) 426-0930; Fax: (267) 426-5476; E-mail: tongw@email.chop.edu.
Find articles by Wu, C. in: JCI | PubMed | Google Scholar
1Division of Hematology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA. 2Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA.
Address correspondence to: Wei Tong, Abramson Building, Suite 316A, Children’s Hospital of Philadelphia, 3615 Civic Center Boulevard, Philadelphia, Pennsylvania 19104-4318, USA. Phone: (267) 426-0930; Fax: (267) 426-5476; E-mail: tongw@email.chop.edu.
Find articles by Balcerek, J. in: JCI | PubMed | Google Scholar
1Division of Hematology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA. 2Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA.
Address correspondence to: Wei Tong, Abramson Building, Suite 316A, Children’s Hospital of Philadelphia, 3615 Civic Center Boulevard, Philadelphia, Pennsylvania 19104-4318, USA. Phone: (267) 426-0930; Fax: (267) 426-5476; E-mail: tongw@email.chop.edu.
Find articles by Tong, W. in: JCI | PubMed | Google Scholar
Published July 10, 2008 - More info
J Clin Invest. 2008;118(8):2832–2844. https://doi.org/10.1172/JCI35808.
© 2008 The American Society for Clinical Investigation
Received: March 31, 2008; Accepted: May 27, 2008
-
Results
Lnk deficiency rescues the HSC frequency defect in Mpl–/– BM. To study the genetic interactions between Mpl and Lnk in regulating the stem cell compartment, we examined HSC frequencies in the double nullizygous mice for Mpl and Lnk, using limiting dilution competitive BM transplant (cBMT) assay (Figure 1; for detailed experimental parameters, see Supplemental Table 1; supplemental material available online with this article; doi: 10.1172/JCI35808DS1). Lnk–/– mice had over a 10-fold increase in HSC frequency compared with WT mice (1 in 1,848 in Lnk–/– mice versus 1 in 18,570 in WT mice) (Figure 1A), in agreement with previous reports (30). In contrast, the HSC frequency in Mpl–/– BM was 1 in 109,380, a 5-fold decrease compared with WT mice (Figure 1A). Importantly, the HSC concentration of Mpl–/–Lnk–/– mice was about 1 in 29,970, slightly compromised compared with WT HSCs (Figure 1A).
Figure 1Mpl–/–Lnk–/– HSCs show decreased self-renewal ability. (A) We quantified HSC frequencies of WT, Mpl–/–, Lnk–/–, and Mpl–/–Lnk–/– mice using limiting dilution cBMT. Results are pooled from 3–5 independent experiments. (B) Top: BM cells from WT, Mpl–/–, Lnk–/–, and Mpl–/–Lnk–/– mice were mixed at 1:1 ratio with CD45.1 competitor cells and transplanted into irradiated recipient mice. Chimerisms of transplanted mice were measured 1, 4, and 8 months after transplant (mean ± SD). n = 10. Middle panel: Transplant results from WT, Mpl–/–, and Mpl–/–Lnk–/– mice when mixed at 9:1 ratio with the competitors. Bottom: Transplant results from Lnk–/– BM cells when mixed at different ratios with the competitors. n = 5. *P < 0.005, WT compared with Mpl–/–Lnk–/– at the same time points, 2-tailed unequal variance t test; #P < 0.005, compared with same strain at 4 months, 2-tailed paired t test; **P < 0.05, compared with same strain at 4 months, 2-tailed paired t test. (C and D) Primary transplanted mice were sacrificed at 8 months and total BM cells were transplanted into secondary recipients. (C) Transplant results from WT, Mpl–/–, and Mpl–/–Lnk–/– mice when mixed at 9:1 ratio and Lnk–/– mice when mixed at 1:19 ratio with the competitors. (D) Transplant results from WT, Mpl–/–, and Mpl–/–Lnk–/– mice when mixed at 3:1 ratio and Lnk–/– mice when mixed at 1:99 ratio with the competitors. Chimerisms of the primary transplanted mice (mean ± SD) are shown in the left panels and those of the secondary transplant (mean ± SEM) are shown in the right panels. **P < 0.05; †P = not significant, n = 5.
We next performed BM-mixing cBMT experiments to follow stem cell repopulating activity and maintenance. This competitive repopulation assay allows a direct and sensitive comparison of reconstitution potentials of stem cells from various genotypes with normal competitor stem cells. Total BM cells (CD45.2) from WT, Mpl–/–, Lnk–/–, and Mpl–/–Lnk–/– mice were mixed at different ratios with CD45.1 competitor cells and transplanted into lethally irradiated CD45.1 recipient mice. Short-term and long-term repopulations were quantified at 1, 4, and 8 months after transplant by flow cytometry analysis of the percentage of chimerism in peripheral blood for donor CD45.2 cells. Close to the expected values, WT mice showed 47% chimerism at 4 months with a 1:1 test to competitor ratio (Figure 1B, top panel), and 85% at 4 months with a 9:1 ratio (Figure 1B, middle panel). Mpl–/–Lnk–/– BM cells had no significant difference in short-term reconstitution as judged by chimerism 1 month after transplant (Figure 1B, top and middle panels) but did exhibit a slight but significant reduction in long-term HSC activity at 4 and 8 months compared with WT (P < 0.005) at both 1:1 and 9:1 ratios (Figure 1B, top and middle panels). In sharp contrast, Mpl–/– BM cells showed diminished stem cell activity with time at both test to competitor ratios (Figure 1B, top and middle panels). On the other hand, Lnk–/– HSCs showed vastly superior repopulation capability, with greater than 90% reconstitution at a 1:1 ratio (Figure 1B, top panel), greater than 80% reconstitution at a 1:19 input ratio (Figure 1B, bottom panel), and 50% reconstitution at a 1:99 input ratio (Figure 1B, bottom panel). Therefore, consistent with data in Figure 1A, Lnk–/– mice have an expanded HSC compartment, and Mpl–/–Lnk–/– mice have slightly compromised long-term HSC activities in primary BM transplant (BMT) assay.
In addition, it is noted that Lnk–/– BM reconstitutions were significantly increased from 4 to 8 months after transplant (Figure 1B, bottom panel). In contrast, WT and Mpl–/–Lnk–/– BM reconstitutions were maintained from 4 to 8 months after transplant (Figure 1B, top and middle panels), while Mpl–/– BM reconstitution was diminished (Figure 1B, top panel). These data indicate that HSC frequency and maintenance defects in Mpl–/– mice are partially rescued by loss of Lnk.
Lnk affects HSC self-renewal largely through TPO/Mpl. Donor-derived HSCs are thought to be subjected to significant stress during BMT (33). These normally quiescent cells must self-renew, proliferate, and differentiate to support the increased number of lineage-restricted progenitors required to reconstitute myeloid and lymphoid populations after ablation of the recipient’s BM. HSC concentrations and their repopulating capacity drop significantly after each round of transplant (34–36). Thus, we performed serial BMT to analyze stem cell self-renewal in Mpl–/–Lnk–/– mice.
Primary transplanted mice with 9:1 (Figure 1C) and 3:1 (Figure 1D) test to competitor ratios from WT, Mpl–/–, and Mpl–/–Lnk–/– mice were sacrificed at 8 months, and total BM cells were transplanted into lethally irradiated secondary recipients. Since Lnk–/– BM cells showed an expansion in the HSC compartment, we used the primary transplants with 1:19 (Figure 1C) and 1:99 (Figure 1D) test to competitor ratios for secondary BMT. Four months after secondary transplant, WT BM cells still showed expected reconstitution levels, 90% at a 9:1 ratio (Figure 1C, right panel) and 70% at a 3:1 test to competitor ratio (Figure 1D, right panel). Lnk–/– BM cells also sustained high levels of reconstitutions (Figure 1, C and D). Strikingly, Mpl–/–Lnk–/– BM cells had dramatically reduced reconstitution levels compared with WT controls (Figure 1, C and D). In fact, similarly to Mpl–/– BM cells, Mpl–/–Lnk–/– cells barely had any donor-derived repopulating cells (below 5%) after secondary BMT (Figure 1, C and D). Thus, both Mpl–/– and Mpl–/–Lnk–/– HSCs showed markedly decreased self-renew activity, indicating that Lnk affects HSC self-renewal largely through the TPO/Mpl pathway.
Loss of Lnk results in increased resistance to repeated 5-fluorouracil treatments in vivo. The massive expansion of Lnk–/– HSC compartment prompted us to examine the cell cycle kinetics upon loss of Lnk. Intriguingly, purified “side population” CD45+Sca-1+ HSCs from Lnk–/– mice did not display increased S/M/G2 populations (our unpublished observations).
To functionally determine HSC numbers undergoing self-renewing division in vivo, we performed limiting dilution cBMT experiments before and after treatment with the cytoablative agent 5-fluorouracil (5-FU) (Table 1 and see Supplemental Table 2 for detailed experimental parameters). Untreated Lnk–/– mice had an approximately 14-fold increase in HSC number per BM compared with WT mice (Table 1). However, after 5-FU treatment the difference between Lnk–/– and WT HSC numbers per BM was decreased to 10-fold (Table 1). Since the read-out of this experiment was the functional transplant assay, our results indicate that there are more HSCs undergoing self-renewing division in Lnk–/– BM than in WT BM. Interestingly, Lnk–/– mice still possess 10-fold more HSC numbers than WT mice after a single 5-FU treatment (Table 1), indicating a subset of Lnk–/– HSCs is resistant to 5-FU or in quiescence. These data led us to speculate that Lnk deficiency keeps HSC in quiescence, that once they enter the cell cycle, they are more likely to undergo symmetric or asymmetric self-renewing division than asymmetric non-renewing division.
Table 1
HSC frequencies and absolute numbers of WT and Lnk–/– mice untreated or treated once or twice with 5-FU
It is predicted that if there is a larger quiescent population in Lnk–/– HSCs compared with that of WT after a single 5-FU treatment, then these cells will be more resistant to a second 5-FU treatment. To test this hypothesis, two 5-FU treatments were administered 4 days apart, and cBMT assays were used to quantify HSC numbers remaining in WT and Lnk–/– mice (Table 1). Remarkably, Lnk–/– mice exhibited over 30-fold more HSCs following consecutive 5-FU treatments than did WT mice (Table 1). This further indicates that a larger quiescent HSC fraction exists in Lnk–/– mice compared with WT mice. Indeed, Lnk–/– mice survived significantly longer compared with WT mice after repeated weekly 5-FU treatments (Supplemental Figure 1). Importantly, the effect of Lnk deficiency on survival was abolished in an Mpl–/– background, as both Mpl–/– and Mpl–/–Lnk–/– showed reduced survival rate compared with WT mice (Supplemental Figure 1).
Loss of Lnk results in an increased quiescent HSC fraction and decelerated cell cycle kinetics. Because our 5-FU experiment indicated that there is a higher quiescence fraction in Lnk–/– HSCs, we next directly analyzed the HSC turnover rate by measuring BrdU incorporation using purified stem cells. Mice were fed with BrdU in drinking water for 7 or 12 days. Lin–Sca-1+Kit+ (LSK) HSCs were purified using flow cytometric sorting (Supplemental Figure 2) and subsequently stained with antibodies specific for BrdU and propidium iodide. Seven days after BrdU labeling, Lnk–/– LSK HSCs had a marked increase in the BrdU– population (Figure 2) compared with WT HSCs (23.3% in Lnk–/– versus 6.9% in WT; P < 0.05). In contrast, Mpl–/– LSK HSCs had only a 1% BrdU– population (P < 0.01 vs. WT), while the BrdU– population in Mpl–/–Lnk–/– HSCs was intermediate between those in WT and Mpl–/– HSCs (Figure 2). Lin–Sca-1–Kit+ (LK) progenitor cells were also sorted at the same time, and the cell cycle kinetics were similar among all strains of mice (Figure 2). These data suggest that Lnk and Mpl specifically control HSC cell cycle progression.
Figure 2Lnk–/– LSK HSCs show decelerated cell cycle kinetics. WT, Mpl–/–, Lnk–/–, and Mpl–/–Lnk–/– mice were fed with water containing BrdU for 7 days. Left panels show representative FACS plots of Lin–Kithi BM cells gated for LSK and LK HSC progenitors. Numbers indicate percentage of cells within each boxed area. Bold numbers indicate that Lnk–/– and Mpl–/– mice were significantly different from WT mice. BrdU incorporation analysis in purified LSK (middle panels) and LK (right panels) cells were determined by FACS. Representative pseudocolor plots are shown. BrdU– 2N, BrdU+ 2N, and BrdU+ populations with over 2N DNA content were quantified (mean ± SD). n = 4.
Since LSK cells contain primitive progenitors in addition to stem cells, we next to set out to distinguish cell cycle characteristics using further purified stem cell populations, Flk2–LSK long-term HSCs and Flk2+LSK short-term HSCs. After 12 days, BrdU labeling, both Flk2– and Flk2+ LSK HSCs from Lnk–/– mice showed a 3-fold increase in BrdU– population compared with those from WT mice (P < 0.01) (Figure 3A). Mpl–/– HSCs had decreased BrdU– populations (P < 0.05 vs. WT), and cell cycle kinetics in Mpl–/–Lnk–/– HSCs were intermediate between those of WT and Mpl–/– HSCs (Figure 3A). These data suggest that loss of Lnk results in decelerated cell cycle progression and that Lnk controls HSC cell cycle kinetics at least in part through the Mpl pathway. It should be noted that the S/M/G2 fractions determined in this assay could not differentiate self-renewing and non–self-renewing divisions.
Figure 3Lnk–/– HSCs have an increased quiescent HSC population compared with WT HSCs. (A) WT, Mpl–/–, Lnk–/–, and Mpl–/–Lnk–/– mice were fed with water containing BrdU for 12 days. Left panels show representative FACS plots of LSK cells gated for Flk2 surface expression. The solid line indicates the reduction staining control, which contains all fluorochromes except Flk2-PE. Shaded areas indicate Flk2 antibody staining. Numbers indicate the percentage of cells represented by the solid line and shaded area. Bold numbers indicate that WT and Mpl–/– mice were significantly different from Lnk–/– mice. BrdU incorporation analysis in purified Flk2–LSK (middle panels) and Flk2+LSK (right panels) HSCs were determined by FACS. Representative pseudocolor plots are shown (mean ± SD). n = 4. For middle and right columns, percentages indicate the percentage of cells gated in the respective quadrants. Bold numbers indicate that Lnk–/– and Mpl–/– mice were significantly different from WT mice. (B) Sorted Flk2–LSK HSCs were stained with Py and Ho, and representative contour plots are shown. The quiescent G0 populations defined as Py–Ho– are indicated (mean ± SEM). n = 4.
The increased BrdU– populations observed in Lnk–/– HSCs after long-term BrdU labeling, implies there is a higher quiescent fraction in Lnk–/– HSCs than in WT HSCs. To directly measure the quiescent HSC population, we stained HSCs with an RNA-specific dye, Pyronin Y (Py), in conjunction with a DNA-specific dye, Hoechst 33342 (Ho), to differentiate G0 from G1 populations. Py–Ho– represents low RNA content and 2N DNA content, thereby defining the G0 fraction (37). Consistent with our BrdU experiment, Lnk–/– HSCs exhibited a much higher G0 population than those of WT HSCs (P < 0.05) (Figure 3B). In contrast, Mpl–/– and Mpl–/–Lnk–/– HSCs had much reduced quiescent populations (P < 0.01) (Figure 3B). Together, our data suggest that Lnk controls HSC quiescence and cell cycle kinetics predominantly through Mpl.
Lnk–/– HSCs potentiate JAK2 activation in response to TPO. The absence of Lnk enhances 3 major Mpl-mediated signaling pathways, Stat3/5, Akt, and p44/42 MAPK in CD41+ megakaryocytes (29). These findings prompted us to investigate the relationship of Lnk and JAK2 activation in HSCs.
Purified LSK HSCs from WT, Mpl–/–, Lnk–/–, and Mpl–/–Lnk–/– animals (Supplemental Figure 2), were stimulated with TPO for 10 minutes after serum starvation. FACS analysis with an antibody specific for activated phosphorylated JAK2 demonstrated that TPO induced a 2-fold increase in JAK2 phosphorylation in WT HSCs, whereas Lnk–/– HSCs showed a 4-fold increase (Figure 4, A and B). In contrast, JAK2 was not activated in response to TPO in either Mpl–/– or Mpl–/–Lnk–/– HSCs (Figure 4, A and B). Mpl surface expression was comparable between WT and Lnk–/– LSK HSCs (Figure 4C), indicating the potentiated JAK2 activation in Lnk–/– HSCs is due to their intrinsic ability to respond to TPO rather than elevated Mpl expression. Since many type I cytokines, such as G-CSF and IL-3, activate JAK2, we next investigated whether Lnk regulates JAK2 activation induced by cytokines other than TPO. Interestingly, G-CSF and IL-3 induced a similar extent of JAK2 activation in WT and Lnk–/– HSCs (Figure 4, D–F), while SCF did not activate JAK2 as expected (data not shown). These results indicate that Lnk negatively regulates TPO-mediated JAK2 activation in HSCs and that loss of Lnk potentiates JAK2 activation, specifically in response to TPO.
Figure 4Lnk–/– HSCs showed potentiated activation of JAK2 in response to TPO, but not to IL-3 or G-CSF. Purified LSK HSCs from mice of various genotypes were starved and stimulated with TPO (A and B) or IL-3 or G-CSF (D–F) for 10 minutes. Representative histograms of the fluorescence intensity of phosphorylated JAK2 (p-JAK2; pY1007/Y1008) in response to TPO (A), IL-3 (D), or G-CSF (E) are shown. Green lines represent second antibody–only controls, blue solid lines represent lack of TPO stimulation, and shaded areas represent TPO stimulation. Quantifications of the fold inductions (mean ± SEM) of phosphorylated JAK2 TPO (B) or IL-3 or G-CSF (F) are shown (n = 3–8). (C) Representative histograms of Mpl surface expression in LSK (left) and LK (right) cells are shown. Green lines represent Mpl expression in Mpl–/– mice, blue lines represent WT mice, and shaded areas represent Lnk–/– mice.
The Lnk SH2 domain associates with kinase-active JAK2 in response to TPO. Because our in vivo studies suggested Mpl and Lnk interact functionally, leading to potentiated JAK2 activation, we next tested whether they interact physically to probe the mechanisms of this genetic interaction. Lnk potently inhibits TPO-dependent 32D/Mpl cell growth (29), and therefore coexpression of the leukemic oncogene, Bcr/Abl, was necessary to counteract the growth-inhibitory effect of Lnk, allowing stable Lnk expression in a hematopoietic progenitor cell line, 32D cells. This 32D-Bcr/Abl cell line, referred to here as 32D-B/A cells, was used for all subsequent cell culture studies unless otherwise indicated.
We established 32D-B/A cells stably expressing Mpl and Flag-tagged Lnk and infected them with retroviruses encoding either myc-tagged WT JAK2 or kinase-inactive JAK2 mutant (JAK2 Y1007F/Y1008F, or Y1007F/Y1008F) (38). Cells were starved and stimulated with TPO for 10 minutes prior to lysis. Flag-tagged Lnk was able to coimmunoprecipitate myc-tagged WT JAK2, but not kinase-inactive Y1007F/Y1008F, in response to TPO (Figure 5A, left panel). There was also some basal-level interaction between Lnk and JAK2 in the absence of TPO (Figure 5A, left panel). The blot was next probed with phosphotyrosine-specific antibodies (4G10) (Figure 5A, left panel), demonstrating that TPO induces robust Lnk phosphorylation in the presence of WT myc-JAK2. In contrast, in the presence of JAK2 Y1007F/Y1008F, Lnk exhibited a small increase in phosphorylation after TPO induction, probably due to endogenous JAK2. The right panel of Figure 5A demonstrates that myc-tagged WT JAK2, but not Y1007F/Y1008F, was able to coimmunoprecipitate with phosphorylated Lnk in response to TPO. These results suggest that Lnk is phosphorylated by JAK2 in response to TPO and robustly binds to kinase-active JAK2.
Figure 5The Lnk SH2 domain associates with kinase-active JAK2. (A) Stable 32D-B/A cells expressing Mpl and Flag-tagged Lnk, in conjunction with either myc-tagged WT JAK2, or kinase-inactive JAK2 mutant Y1007F/Y1008F, were starved and stimulated with TPO for 10 minutes. Cell lysates were then precipitated with Flag-specific antibodies (left panel) and blotted with antibodies specific for myc (top), phosphotyrosine 4G10 (middle), or Lnk (bottom). The right panels show precipitation with myc-specific antibodies and blots with 4G10-specific (top) or myc-tagged antibodies (bottom). (B) We established stable 32D-B/A cell lines expressing Mpl along with Flag-tagged WT Lnk or various Lnk mutants with individual functional domains (R364E, W191A, and Y536F) abolished. Cells were starved and stimulated with TPO. Cell lysates were precipitated with Flag-specific antibodies (left panel) and blotted with antibodies specific for JAK2 (top) or Lnk (bottom). The right panels show precipitation with JAK2-specific antibodies and blots with antibodies specific for 4G10 (top) and JAK2 (bottom). C, control parental 32D-B/A cells; IgG(H), IgG heavy chain.
We next studied whether Lnk binds to endogenous JAK2 and which domain of Lnk are important for its interaction with JAK2. We established stable 32D-B/A cell lines expressing Mpl along with Flag-tagged WT Lnk or various Lnk mutants with individual domains abolished: R364E, W191A, and Y536F (29). Ten minutes after TPO stimulation, WT Lnk was able to coimmunoprecipitate with endogenous JAK2 (Figure 5B, left panel). The Lnk SH2 domain mutant, R364E, failed to pulldown JAK2 (Figure 5B, left panel). In contrast, the PH domain mutant W191A and mutation of the conserved tyrosine at the C terminus, Y536F, did not affect the interaction between Lnk and JAK2 (Figure 5B, left panel). We next performed reciprocal IP studies (Figure 5B, right panel). WT Lnk and Lnk PH domain mutant W191A coimmunoprecipitated with phosphorylated JAK2 (Figure 5B, right panel). The Lnk SH2 domain mutant R364E failed to coimmunoprecipitate with JAK2, and Lnk Y536F mutant showed reduced phosphorylation (Figure 5B, right panel). These results indicate that the Lnk SH2 domain associates with phosphorylated JAK2, and JAK2 in turn phosphorylates Lnk. Since the Lnk SH2 domain is critical to Lnk’s function in inhibiting cytokine receptor signaling (28, 29), these results implicate its interaction with JAK2 as an important transducer of Lnk inhibitory function.
Lnk constitutively associates with Mpl. As we observed a weak but significant interaction between Lnk and JAK2 in the absence of TPO (Figure 5), it is possible that Lnk indirectly interacts with JAK2 through Mpl at the basal state, since the Mpl/JAK2 complexes are preassembled before they reach cell surface (39). Stable 32D-B/A cell lines expressing Flag-tagged Lnk along with HA-tagged WT Mpl or various Mpl mutants were established. TPO induced Lnk-JAK2 binding equivalently in cells expressing WT Mpl, and the Mpl mutant with all 5 cytoplasmic tyrosine residues mutated to phenylalanines (Y5F) (Figure 6A, left panel). The Y5F mutant was no longer phosphorylated in response to TPO, which abolished the Mpl binding sites for Stats or Shc/MAPK downstream signaling molecules but retained JAK2 activation (Figure 6A, right panel). Lnk was associated with WT and Y5F Mpl at similar levels and phosphorylated in response to TPO in cells expressing either WT or Y5F Mpl (Figure 6A, left panel). Reciprocal coimmunoprecipitation showed that Mpl-specific antibodies were able to pulldown similar levels of Lnk in the absence and presence of TPO, regardless of the phosphorylation status of Mpl (Figure 6A, right panel). These data indicate that Lnk constitutively associates with Mpl.
Figure 6Lnk constitutively associates Mpl. We established stable 32D-B/A cell lines expressing Flag-Lnk along with WT Mpl or various Mpl mutants. Cells were starved and then stimulated with TPO. (A) Left: Lysates from cells expressing either WT Mpl or the Y5F mutant were precipitated with Flag-specific antibodies and sequentially probed with antibodies specific for JAK2, Mpl, 4G10, and Lnk. Right: Cell lysates were also reciprocally precipitated with Mpl-specific antibodies and sequentially probed with antibodies specific for Flag, 4G10, and HA. (B) Left: Lysates from cells expressing WT Mpl or the Mpl mutants, Box1 or LW, were precipitated with Flag-specific antibodies and sequentially probed with antibodies specific for JAK2, Mpl, and Lnk. Right: Cell lysates were precipitated with Mpl-specific or JAK2-specific antibodies and probed with 4G10-specific antibodies, followed by antibodies specific for total Mpl and JAK2.
We next generated 2 Mpl mutations to abolish its capacity to activate JAK2: Mpl mutant with the Box1 sequence P521SLP mutated to AAAA [abbreviated as Mpl(Box1)] and Mpl with hydrophobic motif L519W mutated to AA [abbreviated as Mpl(LW)] (40, 41). Cells expressing WT Mpl showed strong tyrosine phosphorylation of Mpl and JAK2 after TPO stimulation, but cells expressing the Box1 and LW mutants did not (Figure 6B, right panel). TPO induced Lnk-JAK2 binding only when WT Mpl, but not Box1 or LW mutant, was expressed (Figure 6B, left panel). In contrast, Lnk coimmunoprecipitated equivalently with WT and mutant Mpl proteins (Figure 6B, left panel). Our data further suggest that active JAK2 is required for binding to Lnk, but Lnk associates with Mpl at the basal level, independently of JAK2 phosphorylation.
Lnk specifically associates with pY613 and pY813 residues on JAK2. After finding that the Lnk SH2 domain associates with phosphorylated JAK2 in response to TPO, we next set out to pinpoint this Lnk-JAK2 interaction. There are 49 tyrosine residues in JAK2, and thus we made peptides containing both nonphosphorylated (Y) and phosphorylated (pY) tyrosines to each tyrosine motif in JAK2. Recombinant Lnk SH2 domain was used to pulldown various JAK2 Y and pY peptides, revealing pY813 as a major site and pY613 as a minor site for Lnk SH2 domain association (data not shown).
We next made point mutations to mutate tyrosine residues to phenylalanines, and 32D-B/A cell lines were established expressing both Mpl and Flag-Lnk, in conjunction with either vector alone, or WT or various myc-JAK2 mutants (Figure 7A). myc-JAK2 Y613F and myc-JAK2 Y813F showed reduced binding to Lnk in response to TPO (Figure 7A, lanes 4–7). Importantly, the double Y613F/Y813F mutation in JAK2 abolished its binding to JAK2 in 32D cells (Figure 7A, lanes 8 and 9). Mutation of the non-relevant residue JAK2 Y1099F displayed normal Lnk-JAK2 interaction (Figure 7A, lanes 10 and 11), while kinase-inactive JAK2 mutant Y1007F/Y1008F failed to interact with Lnk (Figure 7A, lanes 12 and 13). We also probed the blots with JAK2-specific antibodies that identified both exogenous myc-JAK2 and endogenous JAK2 and found that the binding between Lnk and endogenous JAK2 remained in WT JAK2 and various JAK2 mutants, while Y613F/Y813F abolished the binding between Lnk and exogenous myc-JAK2 (Figure 7A). Y613F or Y813F did not significantly affect JAK2 overall level of phosphorylation, and similar levels of total Lnk and JAK2 proteins were expressed in all cell lines (Figure 7A).
Figure 7Lnk binds to JAK2 Y613 and Y813 residues in response to TPO. (A) We established stable 32D-B/A cell lines expressing Flag-Lnk and Mpl along with myc-tagged WT JAK2 or various JAK2 mutants. Cells were starved and stimulated with TPO. Top: IPs with Flag-specific antibodies and blots sequentially probed with antibodies specific for myc, JAK2, and Lnk. Bottom: IPs with myc-specific antibodies and blots sequentially probed with 4G10-specific and JAK2-specific antibodies. C, control 32D cells expressing Lnk alone. (B) Top: Flag-Lnk and either vector alone or various myc-JAK2 constructs were transiently transfected into 293T cells. Lysates were precipitated with Flag-specific antibodies and sequentially probed with antibodies specific for myc, 4G10, and Lnk. Bottom: Cell lysates were precipitated with myc-specific antibodies and probed with 4G10- or JAK2-specific antibodies. (C) Stable 32D-B/A cell lines expressing Flag-Lnk and Mpl, along with vector alone or myc-tagged WT JAK2 or JAK2 V617F (J2V617F) mutant were starved and stimulated with TPO. Top: IPs with Flag-specific antibodies and blots with antibodies specific for myc, JAK2, and Lnk. Bottom: IPs with myc-specific antibodies and blots with 4G10- and JAK2-specific antibodies.
Flag-Lnk was reproducibly able to pulldown WT myc-JAK2 and JAK2 Y1099F mutant species in transiently transfected 293T cells, since JAK2 was constitutively activated in this transient transfection assay (Figure 7B, lanes 3 and 7). Single tyrosine mutants Y613F or Y813F reduced Lnk-JAK2 interaction, while double tyrosine mutant Y613F/Y813F and kinase-inactive JAK2 Y1007F/Y1008F completely abolished Lnk-JAK2 binding (Figure 7B, lanes 4–6 and 8). Flag precipitates were probed with phosphotyrosine-specific antibodies (4G10), revealing both phosphorylated JAK2 and Lnk with either WT JAK2 or JAK2 Y1099F and, to a lesser extent, with JAK2 Y613F (Figure 7B, lanes 3 and 7). Y813F, Y613F/Y813F, or kinase-inactive Y1007F/Y1008F JAK2 failed to associate or phosphorylate Lnk (Figure 7B, lanes 4–6 and 8).
We next examined the consequences of JAK2 Y813F mutation in conferring cell proliferation using IL-3–dependent parental 32D cells. Parental 32D cells and cells stably expressing Mpl in combination with WT JAK2 or JAK2 mutants were established and cultured in different concentrations of TPO. We found that 32D cells expressing the JAK2 Y813F mutant were more sensitive to TPO in supporting cell growth compared with WT JAK2 (Supplemental Figure 3). We next introduced either vector alone (MIG) or MIG-Lnk into these cells and determined their sensitivity to Lnk growth inhibition by analyzing GFP percentage in comparison with the initial infection rate (Supplemental Figure 3). The results demonstrated that JAK2 Y813F–expressing cells are less sensitive to Lnk growth inhibition compared with WT JAK2 cells when cultured in TPO.
We next asked whether Lnk associates with the JAK2 V617F mutant, which is found at high frequencies in myeloproliferative diseases (MPDs). JAK2 V617F was expressed at considerably lower levels compared with those of WT JAK2 (Figure 7C, lanes 3–6). However, the JAK2 V617F mutant retained its ability to associate with Lnk (Figure 7C).
Copyright © 2025 American Society for Clinical Investigation
ISSN: 0021-9738 (print), 1558-8238 (online)