**2. Results**

#### *2.1. SHP-2 Inactivation Leads to Increased Proteasome Dependent HIF-1*α *Degradation during Hypoxia*

As HIF-1 α has been shown to be degraded by the proteasome as well as calpain [10], we first investigated whether this is also true in endothelial cells upon hypoxia. Treatment with the specific proteasome inhibitor epoxomicin or the calpain inhibitor MG101, respectively, resulted in an increase in HIF-1 α protein accumulation (Figure 1A). We previously observed that SHP-2 inactivation impaired HIF-1 α accumulation [12], which was rescued by treatment with proteasome inhibitors. We now additionally investigated the involvement of calpain. As seen in Figure 1B, overexpression of a dominant negative SHP-2 (SHP-2 CS) impaired HIF-1 α accumulation under hypoxic conditions compared to cells overexpressing SHP-2 wildtype (WT). Treatment with the calpain inhibitor MG101 could not rescue this e ffect. Having observed a proteasome dependent [12] but calpain independent degradation of HIF-1 α upon SHP-2 inactivation, we next investigated 26S proteasome activity in endothelial cells under hypoxia. For this, we induced the lentiviral expression of a construct containing the oxygen-dependent degradation (ODD) domain of HIF-1 α [16], which guides its proteasomal degradation upon ubiquitinylation [3], fused to a luciferase gene (HIF1-ODD-Luc) with simultaneous expression of mCherry, in endothelial cells. The expression of HIF1-ODD-Luc thus

inversely correlates with 26S proteasome activity and has been used as a measure of proteasome activity before [16]. First, to test its function in endothelial cells, luciferase activity upon treatment with proteasome inhibitors or hypoxia was measured 72 h after transduction with HIF1-ODD-Luc. As seen in Figure 1C, treatment with the proteasome inhibitors Bortezomib and MG132 as well as hypoxic exposure (4 h) significantly increased the accumulation of HIF1-ODD-Luc, reflecting the inhibition of the 26S proteasome. The transduction of endothelial cells with a control reporter construct lacking the HIF-1α ODD (Ctrl-Luc) showed a strong constitutive expression of luciferase, as expected, which did not differ between normoxic and hypoxic conditions (Figure S1). Expression of dominant negative SHP-2 (CS) impaired HIF1-ODD-Luc accumulation compared to SHP-2 WT expressing cells upon hypoxia, thus demonstrating an increase in 26S proteasome activity (Figure 1D).

**Figure 1.** SHP-2 inactivation enhances proteasome dependent hypoxia inducible factor 1α (HIF-1α) degradation in endothelial cells during hypoxia. (**A**) HIF-1α protein levels were increased during hypoxia upon inhibition of the 26S proteasome (Epoxomicin, 10 μM) as well as calpain (MG101, 5 μM) (*n* = 3). Graph underneath blot shows the protein band densities normalized to β-actin. (**B**) Expression of dominant negative SHP-2 (CS) prevents hypoxic HIF-1α protein upregulation, which could not be rescued by calpain inhibition (MG101, 5 μM; \* *p* < 0.05; *n* = 3). Graph underneath blot shows the protein band densities normalized to β-actin. (**C**) The reporter construct HIF1-ODD-Luc accumulated upon inhibition of the proteasome (Bortezomib 64 nM; MG132 10 μM) during normoxia as well as under only hypoxia (\* *p* < 0.05; *n* = 16–25), confirming the specificity of the reporter constructs. (**D**) Expression of dominant negative SHP-2 (CS) increased HIF-1α degradation by the proteasomal pathway, as detected by lower expression of HIF1-ODD-Luc (\* *p* < 0.05, *n* = 17).

#### *2.2. SHP-2 Regulates Proteasomal Degradation of HIF-1*α *in Hypoxic Wounds In Vivo*

As we previously found SHP-2 inactivation to prevent HIF-1 α accumulation and activity in endothelial cells upon hypoxia, resulting in impaired wound healing angiogenesis in vivo [12] and, as we now observed that SHP-2 inactivation increases 26S proteasomal activity under hypoxia in endothelial cells in vitro, we investigated the proteasomal activity in vivo. For this, HIF1-ODD-Luc or Ctrl-Luc were expressed in wounds of the dorsal skin of mice by localized magnetic nanoparticles-assisted lentiviral transduction (Figure S3). By using lentiviruses (LV) coupled to magnetic nanoparticles (MNP) and the application of an external magnetic field, the simultaneous transduction of three individual wounds in the same animal can be achieved [12]. As seen in Figure 2A and Figure S2B, HIF1-ODD-Luc only accumulated in the malperfused wound and not after transduction of healthy tissue, confirming that the wound is hypoxic and that proteasome activity is higher in normoxic tissues. As a positive control, wounds were transduced with Ctrl-Luc, which causes a continuous strong expression of luciferase, as this construct does not contain the HIF-1 α ODD domain. Next, we performed co-transductions of individual wounds in the same animal with HIF1-ODD-Luc and the di fferent SHP-2 constructs, to investigate the influence of SHP-2 on proteasome activity in vivo. Whereas the expression of inactive SHP-2 CS in hypoxic wounds significantly inhibited HIF1-ODD-Luc accumulation via increased proteasome activity, introduction of the constitutively active SHP-2 E76A (Glu76 to Ala76) enhanced the HIF1-ODD-Luc protein accumulation compared to SHP-2 WT expressing wounds (Figure 2B and Figure S2C). This indicates that SHP-2 regulates HIF-1 α stabilization and accumulation in hypoxic wounds by decreasing 26S proteasome activity.

**Figure 2.** SHP-2 inactivation induces HIF-1α degradation via the proteasome pathway in hypoxic wounds in vivo. ( **A**) Wounds in the same dorsal skin fold chamber in mice were simultaneously transduced with HIF1-ODD-Luc or Ctrl-Luc lacking HIF1-ODD using site directed lentiviral magnetic

targeting [12]. HIF1-ODD-Luc was expressed in wounds (1) but was degraded by the proteasome in healthy tissue (2), demonstrating the specificity of the lentiviral constructs and that the wounds are hypoxic (\* *p* < 0.05; *n* = three animals). (3) Ctrl-Luc lacking the HIF-ODD domain was therefore constitutively expressed in the wound (\* *p* < 0.05; *n* = three animals). (**B**) Wounds in the same dorsal skin fold chamber in mice were simultaneously co-transduced with HIF1-ODD-Luc and different SHP-2 constructs. While HIF1-ODD-Luc accumulated in hypoxic wounds upon transduction with SHP-2 WT, expression of dominant negative SHP-2 (CS) impaired this, demonstrating an increased 26S proteasomal activity (\* *p* < 0.05; *n* = 3–4 animals). Expression of constitutively active SHP-2 (E76A) further enhanced HIF1-ODD-Luc accumulation, demonstrating enhanced inhibition of 26S proteasome activity (\* *p* < 0.05; *n* = 3–4 animals). Wounding was performed the day after implantation of the dorsal skin fold chamber. Transduction of wounds was performed 24h after wounding and measurements of luciferase activity were performed eight days after transduction (see also Figure S2A).

#### *2.3. The Proteasomal Degradation of HIF-1*α *is Dependent on Src Kinase and p38 MAPK Activation*

In a former study, we could show that SHP-2 induces HIF-1α expression via a Src kinase dependent mechanism in endothelial cells upon hypoxia [12]. We thus hypothesized that the observed effect of SHP-2 on 26S proteasome activity and HIF-1α stabilization in this study may be mediated by Src as well. To test this, endothelial cells were transduced with the HIF1-ODD-Luc reporter construct and treated with the pharmacological Src inhibitor PP2 upon hypoxia. Whereas hypoxia induced the stabilization, and thus accumulation, of HIF1-ODD-Luc, representing a decrease in proteasome activity, Src inhibition significantly impaired this response (Figure 3A). Src kinases have been demonstrated to induce the activation of p38 MAPK during hypoxia [17]. Thus, we next explored whether this was the case in endothelial cells. The hypoxia induced HIF-1α accumulation was prevented upon inhibition of p38 MAPK (Figure 3B). Moreover, hypoxia induced the phosphorylation and thus activation of p38 MAPK and this was abrogated when treating cells with the Src kinase inhibitor PP2 (Figure 3C). Moreover, treatment with the p38 MAPK inhibitor SB203580 increased proteasome activity, reflected by reduced HIF1-ODD-Luc accumulation (Figure 3D). Finally, the phosphorylation of p38 MAPK was impaired in endothelial cells expressing SHP-2 CS and enhanced in cells expressing the constitutively active SHP-2 E76A compared to the expression of SHP-2 WT upon hypoxia (Figure 3E).

#### *2.4. SHP-2 Activity Inhibits the Chymotrypsin-Like Activity of the 26S Proteasome upon Hypoxia*

As previously published data from our group showed that treatment with epoxomicin and MG132, which are both inhibitors of the CT-L activity of the 26S proteasome [18], rescued the low HIF-1α protein level caused by SHP-2 inactivation [12], we next detected the 26S CT-L proteolytic activity in endothelial cells under hypoxia. We used experimental conditions optimized to investigate proteolytic protease activity of the 26S proteasome as previously described [19]. For this, cells expressing SHP-2 WT were exposed to hypoxia and the chymotrypsin-like (CT-L) activity of the 26S proteasome was assessed by a specific fluorogenic proteasome substrate (Suc-LLVY-AMC). Hypoxia significantly reduced CT-L activity (Figure 4A and Figure S4), and endothelial cells expressing constitutively active SHP-2 E76A exhibited an even lower 26S CT-L activity upon hypoxia compared to SHP-2 WT expressing cells (Figure 4B), indicating a downregulation of the CT-L proteolytic activity of the 26S proteasome.

**Figure 3.** The proteasomal HIF-1α degradation is dependent on Src kinase and p38 mitogen-activated protein kinase (MAPK) signaling. (**A**) Whereas hypoxia inhibited 26S proteasome activity in endothelial cells, as seen by increased expression of HIF1-ODD-Luc, inhibition of Src kinase (PP2, 100 nM) reversed this (\* *p*<0.05; *<sup>n</sup>*=9). (**B**) Inhibition of p38 MAPK (SB203580, 10 μM) in endothelial cells impaired hypoxia induced HIF-1α expression (*n* = 6). (**C**) Src kinase inhibition (PP2, 100 nM) reduced hypoxia induced p38 MAPK activation (*n* = 2). (**D**) p38 MAPK inhibition (SB203580, 10 μM) increased 26S proteasome activity, as measured by a lower level of HIF1-ODD-Luc reporter expression (\* *p* < 0.05; *n* = 4). (**E**) Expression of dominant negative SHP-2 (CS) impaired hypoxia induced p38 MAPK phosphorylation, whereas expression of constitutively active SHP-2 (E76A) enhanced this compared to SHP-2 WT (\* *p* < 0.05; *n* = 4). Graphs underneath blots show the protein band densities normalized to β-actin.

**Figure 4.** SHP-2 inhibits 26S chymotrypsin-like (CT-L) proteasomal activity during hypoxia. (**A**) The CT-L activity of the 26S proteasome was decreased upon hypoxia (\* *p* < 0.05, *n* = 11), as measured by the fluorogenic substrate Suc-LLVY-AMC. Treatment with epoxomicin (10 μM) was used as a positive control and effectively inhibited the CT-L proteolytic activity of the 26S proteasome (\* *p* < 0.05; *n* = 3). (**B**) Expression of constitutively active SHP-2 (E76A) also significantly impaired the CT-L activity of the 26S proteasome compared to SHP-2 WT during hypoxia (\* *p* < 0.05, *n* = 9), whereas the expression of dominant negative SHP-2 (CS) showed a tendency towards increased CT-L activity (*n* = 9).
