**3. Discussion**

Vascular remodeling and angiogenesis are important for maintaining tissue perfusion upon ischemia [2]. We previously demonstrated that the tyrosine phosphatase SHP-2 drives hypoxia mediated HIF-1α upregulation, resulting in revascularization of wounds in vivo [12]. Here, we show that this is achieved by SHP-2 dependent inhibition of 26S proteolytic activity via Src kinase/p38 MAPK signalling.

In a former study, we demonstrated that SHP-2 promotes HIF-1α stabilization and activity in endothelial cells during hypoxia as well as revascularization of hypoxic wounds in vivo by increasing HIF-1α activity, resulting in higher expression of VEGF, MMP-2 and PDGF [12]. Moreover, we showed SHP-2 to negatively influence the proteasomal degradational pathway, as the impaired HIF-1α expression seen upon SHP-2 inactivation was rescued by inactivation of the proteasome as well as the PHD [12]. While these results indicate that SHP-2 influences the prolylhydroxylation of HIF-1α and in this way its proteasomal degradation, we further investigated here the mechanisms of SHP-2 and its regulation of proteasomal activity. The data obtained in this study confirm our previous findings and additionally reveal a second mechanism of promoting HIF-1α upregulation by directly affecting 26S proteasomal activity. Inactivation of SHP-2 prevented the cellular accumulation of HIF1-ODD-Luc, which inversely correlated with 26S proteasomal activity, in endothelial cells in vitro. The in vivo relevance of SHP-2 mediated proteasome regulation was confirmed in hypoxic wounds of intact animals using co-transduction of HIF1-ODD-Luc and SHP-2 constructs. Of note, expression of constitutively active SHP-2 further enhanced HIF1-ODD-Luc levels, demonstrating an increased inhibition of the 26S proteasome. This correlates well with the impaired 26S CT-L proteolytic activity measured in cells with constitutive SHP-2 activation. Thus, we conclude that SHP-2 activity negatively affects proteasomal activity during wound healing under hypoxic conditions in vivo.

As we previously found SHP-2 dependent Src kinase activation to be involved in HIF-1α upregulation during hypoxia [12], we now investigated if it is also important for proteasomal activity. Indeed, Src integrity was crucial for 26S proteasomal activity and in addition, was important for activation of the p38 MAPK upon hypoxia. p38 MAPK, in turn, positively affected HIF-1α upregulation while negatively influencing 26S proteasomal activity, a mechanism which was promoted by SHP-2. These results are not only in accordance with but additionally extend the study from Lee et al. performed in HeLa cells, demonstrating that p38 MAPK negatively a ffects 26S proteolytic activity also in endothelial cells [20]. This inhibition was further shown to be due to the phosphorylation of Thr-273 on the subunit Rpn2 in the 19S regulatory particle [20]. Whether this is the case during hypoxia, remains to be investigated. Posttranslational phosphorylation of the 26S proteasome has repeatedly been shown to regulate its enzymatic activities [15]. Several stimuli, such as stress, metabolic changes, and growth factor signalling have been demonstrated to induce its phosphorylation [15]. Here, we identified hypoxia as a novel stimulus of 26S proteasomal inactivation, particularly the CT-L activity. In addition, we found the caspase-like (C-L) activity to be deprived upon hypoxia (Figure S4). Intriguingly, we measured an increased trypsin-like (T-L) activity during hypoxia, suggesting that the di fferent proteolytic activities of the 26S proteasome may be di fferentially regulated. One may hypothesize that the di fferential regulation of activities is involved in substrate specificity of proteasomal degradation. However, more investigations are needed to elucidate the mechanisms behind this. The observed reduction in CT-L and C-L activity, however, is supported by the fact that the 26S proteasome relies on the activity of ATPases in the 19S regulatory particle for substrate de-ubiquitinylation, unfolding and 20S gate opening and activation [15], which do not function upon ATP deprivation, as is the case during hypoxia. Moreover, we observed that inhibition of Src kinase/p38 MAPK signalling rescued proteasomal activation. Additionally, an inhibition of 26S proteasomal activity via direct dephosphorylation of subunits by SHP-2 was not investigated here and has to be the focus of further studies. However, this is per se possible, as Zong et al. demonstrated the phosphatase PP2A to negatively a ffect 20S proteolytic activity via dephosphorylation [21]. Nevertheless, this is to our knowledge the first study to report a regulation of 26S proteolytic activity by SHP-2.

In summary, we were able to further characterize the mechanism behind the regulation of hypoxia mediated HIF-1 α upregulation by SHP-2, which is essential for revascularisation of malperfused wounds. Moreover, we demonstrate for the first time that the 26S proteasomal activity is regulated by SHP-2 during hypoxia in vitro and is functionally relevant in vivo. We show that SHP-2 not only inhibits the proteasomal degradation pathway by influencing HIF-1 α prolylhydroxylation [12] but in addition by directly inhibiting 26S proteolytic activity via p38 MAPK (for a summary of our findings, see Figure 5). We thus believe that SHP-2 is important for HIF-1 α upregulation in vitro and in wounds in vivo by inhibition of the proteasomal pathway via activation of Src kinase/p38 MAPK signalling. Together with our previous results, this regulation may be achieved by redundant and or additive pathways, involving external regulators of the proteasome (PHD and pVHL activation), and directly via the CT-L activity of the 26S proteasome. Finally, our findings confirm SHP-2 to be essential for hypoxic HIF-1 α upregulation in vivo. SHP-2 may therefore constitute a novel therapeutic target in ischemic vascular disease, aiming for the revascularization of ischemic tissues.

**Figure 5.** Illustration of 26S proteasome regulation and HIF-1α accumulation by SHP-2 in hypoxia. HIF-1α is constitutively expressed but degraded by the proteasomal pathway during hypoxia (left). Upon hypoxia (right), the degradation of HIF-1α is inhibited. In our previous study, we found SHP-2 to influence the prolyl hydroxylation of HIF-1<sup>α</sup>, which targets it for degradation [12]. In this study, we found a second mechanism of SHP-2 mediated HIF-1α upregulation: SHP-2 positively affects Src kinase and p38 MAPK activity during hypoxia, which in turn negatively influences the activity of the 26S proteasome. As a consequence, HIF-1α is stabilized and accumulates in the cell, promoting hypoxic angiogenesis and vascular remodeling.

### **4. Materials and Methods**

### *4.1. Antibodies and Chemicals*

Rabbit phospho-p38 MAPK (Thr180/Tyr182) (D3F9) XP™ (#4511) and β-Actin (13E5) antibodies were from Cell Signaling Technology, Frankfurt am Main, Germany. HIF-1α clone H1α67, HIF-1α clone EP1215Y, anti-mouse and -rabbit horseradish peroxidase-conjugated secondary antibodies were from Merck Millipore, Darmstadt, Germany. Src-Inhibitor PP2 was purchased from Sigma-Aldrich (#P0042), Darmstadt, Germany. p38 MAPK inhibitor (SB203580), MG132 and MG101 were from Tocris, Wiesbaden-Nordenstadt, Germany. Bortezomib and Epoxomicin were from Calbiochem. All other chemicals were from Sigma-Aldrich, Darmstadt, Germany.

### *4.2. Human Microvascular Endothelial Cell (HMEC) Culture*

Human dermal microvascular endothelial cells (HMEC) [22] were cultivated as described earlier [23]. In detail, HMEC were cultivated in DMEM (Sigma-Aldrich, Darmstadt, Germany) containing 10% fetal calf serum (Biochrom, Berlin, Germany) and 1% Penicillin-Streptomycin (Sigma-Aldrich, Darmstadt, Germany) and kept in an incubator with 5% CO2 at 37 ◦C.

### *4.3. Lentiviral Constructs and Transductions*

Wild type (WT) SHP-2 and catalytically inactive mutant SHP-2 CS (Cys459 to Ser459) plasmid vectors were kind gifts from the Bennett laboratory [24]. The constitutively active SHP-2 E76A (Glu76 to Ala) was generated as previously described [12]. The lentiviral constructs containing the above mentioned cDNAs were generated as described earlier [12] and lentiviral particles were produced as described elsewhere [25]. Flow cytometry of transduced HEK293T cells was used to determine the biological titer as previously described [25]. The FUW-ODD-Luc-mCherry lentiviral plasmid (referred to as HIF1-ODD-Luc in this manuscript) and the control vector FUW-Luc-mCherry (referred to as Ctrl-Luc in this manuscript) were kindly provided by Kimbrel et al. [16] and packaged into lentiviral particles as previously described [25]. The ODD-Luc insert encodes a reporter fusion protein consisting of the oxygen-dependent domain (ODD) from hypoxia-inducible factor 1a (HIF-1a) and firefly luciferase, inversely reflecting proteasomal activity, with simultaneous expression of mCherry after a self-cleavage 2A site. Lentiviral transduction of HMECs was carried out using a multiplicity of infection (MOI) of five. Lentiviral particles were diluted in Hank's solution and applied onto subconfluent HMEC. After incubation for 4–6 h at 37 ◦C, culture medium was added. The next day the medium was changed, and cells were left 72 h before assaying.

### *4.4. Hypoxia Treatment*

For hypoxia treatment, HMECs were incubated in cultivation media in a hypoxia chamber (Cell Systems, Troisdorf, Germany) at pO2 8 ± 2 mmHg equivalent to a O2 concentration of 1 ± 0.2% for 4 h as previously described [26]. To reach these experimental hypoxic conditions, the chamber was flooded for four minutes with 15–20 L/min with an anoxic gas mixture (5% CO2, 95% N2). After hypoxia, the media was discarded, and cells quickly washed once with phosphate bu ffered saline and lysed for further processing.

### *4.5. In Vitro 26S Proteasome Activity*

The 26S CT-L proteasome activity was measured as previously described [19]. In detail, HMEC were subjected to hypoxia (4 h), rinsed with cold phosphate bu ffered saline supplemented with calcium (PBS+) and lysed at 4 ◦C with lysis bu ffer (1mM DTT, 1× Roche PhosphoStop Tablet in homogenizing bu ffer containing 20 mM HEPES, 1 mM MgCl2, 150 mM NaCl, and 0.5 mM EDTA). 10 μg protein in assay bu ffer containing 1mM DTT, 50 μM ATP and 100 μM Suc-LLVY-AMC (R&D systems, Wiesbaden, Germany) in homogenizing bu ffer was measured at 37 ◦C with excitation wavelength 339 nm and emission wavelength 439 nm for 2 h.

### *4.6. In Vitro Luciferase Assay*

HMEC were seeded in 24-well plates and transduced the following day. Indicated inhibitor treatments and hypoxia was performed 72 h post transduction and cells were lysed on ice for 30 min (250 mM Tris base pH7.8, 0.1% Triton-X). Then, 10 μL of cell lysate were transferred to a black 96-well plate in triplicates and 100 μL luciferin assay bu ffer (60 mM DTT, 10 mM MgSO4, 1 mM ATP, 25 mM Glycil-Glycin, 0.3 mM D-Luciferin) were added to each well. Bioluminescence was detected using the Spectrafluor (Tecan, Männedorf, Switzerland) with a one-second integration time and normalized to protein concentrations.

### *4.7. In Vivo Transduction and Luciferase Imaging*

Animal studies were conducted in accordance with the German animal protection law and approved by the district governmen<sup>t</sup> of upper Bavaria (Regierung von Oberbayern, approval reference number AZ55.2-1-54-2532-172-13). The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). 16–20-week old male and female C57 BL6/J mice (Charles River) were anesthetized

(5 mg/kg midazolam, 0.5 mg/kg medetomidin, and 0.05 mg/kg fentanyl) and the dorsal skinfold chamber was implanted as described before [27]. Wounds were introduced using a hot probe as previously described 24 h after dorsal skinfold chamber implantation [12]. Proteasomal activity in avascular wounds in the dorsal skin of mice was detected by localized magnetic nanoparticles-assisted transduction of individual wounds in the same animal with HIF1-ODD-Luc (FUW-ODD-Luc-mCherry) and Ctrl-Luc (FUW-Luc-mCherry) lentiviral vectors as well as co-transductions of with the ODD-Luc vector and SHP-2 WT, CS and E76A lentiviral vectors, respectively, as previously described [12]. Luciferase activity was detected eight days after transduction by application of 3 mg/mL D-Luciferin directly to the imaging window of the dorsal skinfold chamber. Bioluminescence was imaged using the IVIS® spectrum in vivo imaging system from PerkinElmer at medium binning and exposure times between five and 10 minutes. Signal intensities were quantified using the Fiji software.
