Integrin αVβ3 can substitute for collagen‐binding β1‐integrins in vivo to maintain a homeostatic interstitial fluid pressure

New Findings What is the central question of this study? Collagen‐binding β1‐integrins function physiologically in cellular control of dermal interstitial fluid pressure (P IF) in vivo and thereby participate in control of extravascular fluid volume. During anaphylaxis, simulated by injection of compound 48/80, integrin αVβ3 takes over this physiological function. Here we addressed the question whether integrin αVβ3 can replace collagen‐binding β1‐integrin to maintain a long‐term homeostatic P IF. What is the main finding and its importance? Mice lacking the collagen‐binding integrin α11β1 show a complex dermal phenotype with regard to the interstitial physiology apparent in the control of P IF. Notably dermal P IF is not lowered with compound 48/80 in these animals. Our present data imply that integrin αVβ3 is the likely candidate that has taken over the role of collagen‐binding β1‐integrins for maintaining a steady‐state homeostatic P IF. A better understanding of molecular processes involved in control of P IF is instrumental for establishing novel treatment regimens for control of oedema formation in anaphylaxis and septic shock. Abstract Accumulated data indicate that cell‐mediated contraction of reconstituted collagenous gels in vitro can serve as a model for cell‐mediated control of interstitial fluid pressure (P IF) in vivo. A central role for collagen‐binding β1‐integrins in both processes has been established. Furthermore, integrin αVβ3 takes over the role of collagen‐binding β1‐integrins in mediating contraction after perturbations of collagen‐binding β1‐integrins in vitro. Integrin αVβ3 is also instrumental for normalization of dermal P IF that has been lowered due to mast cell degranulation with compound 48/80 (C48/80) in vivo. Here we demonstrate a role of integrin αVβ3 in maintaining a long term homeostatic dermal P IF in mice lacking the collagen‐binding integrin α11β1 (α11−/− mice). Measurements of P IF were performed after circulatory arrest. Furthermore, cell‐mediated integrin αVβ3‐directed contraction of collagenous gels in vitro depends on free access to a collagen site known to bind several extracellular matrix (ECM) proteins that form substrates for αVβ3‐directed cell attachment, such as fibronectin and fibrin. A streptococcal collagen‐binding protein, CNE, specifically binds to and blocks this site on the collagen triple helix. Here we show that whereas CNE perturbed αVβ3‐directed and platelet‐derived growth factor BB‐induced normalization of dermal P IF after C48/80, it did not affect αVβ3‐dependent maintenance of a homeostatic dermal P IF. These data imply that dynamic modification of the ECM structure is needed during acute patho‐physiological modulations of P IF but not for long‐term maintenance of a homeostatic P IF. Our data thus show that collagen‐binding β1‐integrins, integrin αVβ3 and ECM structure are potential targets for novel therapy aimed at modulating oedema formation and hypovolemic shock during anaphylaxis.


INTRODUCTION
Loose connective tissue structures surround all peripheral blood and lymph vessels, nerves and muscles, as well as underlying epithelial sheets forming what is commonly referred to as the interstitium. The interstitium harbours the extracellular fluid, whose volume amounts to some 15% of the total body weight. Interstitial fluid volume is determined by the influx of fluid across the capillary wall and drainage This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.  (Curry & Adamson, 2013;Michel & Curry, 1999). The interstitial volume is the volume resulting from the balance between this influx of fluid and the lymphatic drainage. Finally, the interstitial fluid pressure (P IF ) is a function of the interstitial fluid volume and the Experimental Physiology. 2018;103:629-634. wileyonlinelibrary.com/journal/eph

New Findings
• What is the central question of this study?
Collagen-binding 1 -integrins function physiologically in cellular control of dermal interstitial fluid pressure (P IF ) in vivo and thereby participate in control of extravascular fluid volume. During anaphylaxis, simulated by injection of compound 48/80, integrin V 3 takes over this physiological function. Here we addressed the question whether integrin V 3 can replace collagen-binding 1 -integrin to maintain a long-term homeostatic P IF .
• What is the main finding and its importance?
Mice lacking the collagen-binding integrin 11 1 show a complex dermal phenotype with regard to the interstitial physiology apparent in the control of P IF . Notably dermal P IF is not lowered with compound 48/80 in these animals. Our present data imply that integrin V 3 is the likely candidate that has taken over the role of collagen-binding 1 -integrins for maintaining a steady-state homeostatic P IF . A better understanding of molecular processes involved in control of P IF is instrumental for establishing novel treatment regimens for control of oedema formation in anaphylaxis and septic shock.
interstitial compliance, but as we have shown, it is also actively controlled by connective tissue cells. In skin P IF is normally slightly below ambient pressure, i.e. around −1 mmHg compared at a capillary hydostatic pressure of around 10 mmHg and a net capillary pressure, i.e. the net pressure that creates filtration across the capillaries, of 0.5-1 mmHg (Reed, Liden, & Rubin, 2010). P IF normally acts to maintain a constant interstitial volume while in particular conditions like inflammation a lowered P IF transiently becomes the main driving force for the rapid initial fluid movement out of the microvasculature during early innate immunity responses (Reed et al., 2010). A lowering of P IF by even a few mmHg will represent an important part of the driving force for capillary filtration together with increased capillary hydrostatic pressure and increased capillary permeability since the lowering of P IF must be compared with a net capillary filtration pressure of a 0.5-1 mmHg (Reed et al., 2010). Once oedema has formed, P IF will reach positive values and further maintenance of filtration and oedema relies on increased capillary hydrostatic pressure and increased capillary permeability.
Under steady-state conditions connective tissue cells balance the slightly negative P IF by exerting tensional forces that maintain the proteoglycan/hyaluronan ground substance of the extracellular matrix (ECM) in an underhydrated state (Reed et al., 2010). The necessary force is generated by the cytoskeletal machinery that connects to ECM fibres via integrins (Berg, Rubin, & Reed, 2001;Reed, Rubin, Wiig, & Rodt, 1992). At homeostasis 1 -integrins are operative in rat and mouse dermis whereas during inflammatory reactions, in which P IF is lowered, e.g. during anaphylaxis, there is a shift in integrin usage such that the V 3 -integrin, and not 1 -integrins, connects the cellular contractile apparatus to ECM fibres (Liden, Berg, Nedrebø, Reed, & Rubin, 2006;Svendsen, Liden, Nedrebø, Rubin, & Reed, 2008).
Available data suggest that the collagen-binding integrins 2 1 (Rodt, Ahlen, Berg, Rubin, & Reed, 1996) and 11 1 (Svendsen et al., 2009) are operative to maintain a homeostatic P IF in rat and mice dermis, respectively. In 11 1 -deficient mice blockage of 1 -integrins does not lower P IF whereas such blockage lowers P IF in wild-type mice (Reed et al., 1992;Svendsen et al., 2009). Local administration of plateletderived growth factor (PDGF)-BB normalizes P IF in mouse and rat dermis in which P IF has been lowered by mast cell degranulation (Liden et al., 2006;Rodt et al., 1996). This effect of PDGF-BB requires functional integrin V 3 (Liden et al., 2006). Furthermore, dermal P IF is not significantly lowered in 11 1 -deficient mice, but readily lowered in wild-type mice during compound 48/80 (C48/80)-induced anaphylaxis (Svendsen et al., 2009).
The traits for integrin usage in cellular control of P IF in vivo are paralleled by cell-mediated contraction of three-dimensional reconstituted collagen gels in vitro. Thus, collagen-binding 1 integrins mediate, when present, the cell-collagen contacts that are necessary for contraction (Gullberg et al., 1990); in their absence integrin V 3 becomes operative (Grundström Grundström, Mosher, Sakai, & Rubin, 2003). Integrin V 3 -directed contraction by myoblasts requires that the cells synthesize fibronectin, a synthesis that in these cells is stimulated by PDGF-BB van Wieringen et al., 2010). Available data suggest that fibronectin forms a bridge between the collagen fibres and integrin V 3 thereby enabling collagen gel contraction van Wieringen et al., 2010).
Here we investigated the role of integrin V 3 -integrin in maintaining P IF in the dermis of mice with a constitutively perturbed function of collagen-binding 1 -integrins, such as in 11 1 -deficient mice (Svendsen et al., 2009). Furthermore, we investigated the potential role of collagen-binding proteins that may bridge the collagen fibres to cellular V 3 by investigating potential effects of the streptococcal protein CNE on cellular control of P IF .

Ethical approval
The animal experiments were conducted according to the European Measurements of interstitial fluid pressure (P IF ) were performed on the dorsal side of the hind paw with the mouse lying on its back.
After a control measurement with intact circulation, the remaining measurements (90 min) were performed after circulatory arrest and the animal was killed with cervical dislocation in Group A (see below).
In Groups B and C (see below) the animals were killed with intravenous saturated KCl. Furthermore, the duration of anaesthesia in all three groups was no more than 5-10 min including measurement of control P IF and I.V. injections in any of the groups. Before and during the experiments sufficient depth of anaesthesia was confirmed by lack of response to hindlimb toe pinch.

Effects of the anti-integrin 3 IgG on dermal interstitial fluid pressure
After measurement of control P IF with intact circulation, circulatory arrest was induced by dislocation of the neck. Thereafter 1 l of antiintegrin 3 IgG (1 g l −1 ) was injected intradermally and P IF was measured for the next 90 min. Measurements were performed in wildtype C57BL/6 mice and in littermate mice deficient in 11 1 .

Effects of compound 48/80 and subsequent injection of PDGF-BB alone or with CNE
After a control measurement of P IF the mice were injected intravenously with 200 g C48/80 in 100 l phosphate-buffered saline.
C48/80 induces a generalized mast cell degranulation that as part of the clinical picture is associated with a lowering of P IF within 30 min.
Also, the effect is seen as increased respiratory rate and lowering of blood pressure. Circulatory arrest was induced by I.V. injection of saturated KCl 2 min after injection of C48/80. Measurement of P IF was started and continued for the next 90 min. Mice that did not demonstrate a lowering of P IF of at least 0.5 mmHg were excluded from the study since a lack of response to C48/80 means that PDGF-BB will not have a lowered P IF to act on. One microlitre of PDGF-BB (0.7 g ml −1 ) was injected intra-dermally after 30 min either alone or combined with CNE at 0.7 mg ml −1 .

Effects of CNE in wild-type and 11 1 -deficient mice
After control measurement of P IF with intact circulation, the animals were given saturated KCl intravenously to induce circulatory arrest.
One microlitre of CNE at 0.7 g ml −1 was injected subcutaneously and measurement of P IF continued for 90 min.

Statistical methods
Data are presented as means ± SD unless specified otherwise.
Repeated measurements ANOVA and post-hoc test (Sidak's multiple comparison test) correcting for multiple corrects were used.

Measurements of P IF were compared using one-and two-tailed
Student's t test as specified in Results. P < 0.05 was considered statistically significant.

Effects of anti-integrin 3 IgG on dermal interstitial fluid pressure
In accordance with previously reported findings showing that 1integrin and not V 3 is operative in maintaining P IF at homeostasis, local intradermal injection of anti-integrin 3 IgG in wild-type naïve C57BL/6 mice had no effect on P IF (Figure 1). In contrast, intradermal but not from its own control (P = 0.72) measured prior to injection of C48/80 (in both cases using paired comparison and two-tailed testing).

Effects of CNE in wild-type and 11 1 -deficient mice
Injection of 1 l 0.7 mg ml −1 CNE in wild-type and 11 1 -deficient ( 11 −/− ) mice did not change P IF compared to the respective controls ( Figure 3). P IF in both wild-type and 11 1 -deficient mice was unaffected by injection of CNE (P > 0.05 using one-way repeated ANOVA). P IF in the 11 1 -deficient mice was lower in this experimental series than in wild-type. The control P IF values did not, however, differ between wild-type and 11 1 -deficient mice in the experimental series shown in Figure 1, nor in those reported by Svendsen et al. (2009).

DISCUSSION
Here we show that the integrin V 3 functions physiologically to maintain the homeostatic P IF in mouse dermis lacking the integrin 11 1 . During acute inflammatory reactions collagen-binding 1integrins decouple and their role in controlling P IF is taken over by the V 3 integrin (Liden et al., 2006;Svendsen et al., 2008). Our present data are a further elaboration on how P IF can be modulated by cellular and molecular pathways and show that the V 3 integrin can Values are means ± SD participate in P IF control also in the absence of inflammation. The data also expand on previous findings on a potential role of the 11 1 integrin in control of dermal P IF in mice (Svendsen et al., 2009). Taken together with the data presented here, it is possible to conclude that the collagen-binding 1 -integrin 11 1 is a key operator in maintaining a homeostatic P IF in normal dermis. In mouse dermis lacking 11 1 ( 11 −/− mice), P IF was only marginally lowered after induction of anaphylaxis by the mast cell degranulator C48/80 (Svendsen et al., 2009) suggesting that V 3 integrin-operated P IF control works also during anaphylaxis, which is in line with previously published reports (Liden et al., 2006;Svendsen et al., 2008).
In a previous publication, we presented data on a role of the collagen-binding 1 -integrin 2 1 in controlling P IF in rat dermis (Rodt et al., 1996). This conclusion was based on experiments in which the anti-rat 2 1 monoclonal antibody Ha1/29 (Mendrick & Kelly, 1993) lowered P IF in naïve rat dermis. It is thus possible that mice and rats differ as to preferred usage of collagen-binding 1 -integrin to control dermal P IF . Alternatively, both integrins are required and perturbation of any of them distorts dermal P IF -control. It can furthermore not be excluded that the Ha1/29 antibody inhibits both 2 1 and 11 1 .
It is not clear whether collagen-binding 1 -integrins bind directly to collagen molecules in the ECM fibres in vivo or only via accessory proteins as has been suggested to be the case for chondrocyte binding to cartilage collagenous fibres (Woltersdorf et al., 2017). Our present data do not discriminate between these two possibilities but together with previously reported data show that integrins play an important physiological role in controlling P IF .
To further delineate V 3 integrin-operated P IF control in mouse dermis deficient in the 11 1 integrin ( 11 −/− mice), we took advantage of the streptococcal protein CNE. CNE binds to and blocks a collagen site that is necessary for binding of several proteins that can function as a bridge between cellular V 3 and the collagen fibres, such as fibrin and fibronectin. Integrin V 3 -mediated contraction of collagen gels in vitro relies on these interactions and is inhibited by CNE (Reyhani et al., 2014;van Wieringen et al., 2010). Our present data demonstrate an in vivo effect of CNE, namely that it inhibited PDGF BB-induced and integrin V 3 -mediated normalization of P IF that has been lowered by induction of anaphylaxis in naïve mouse dermis using the mast cell degranulator C48/80. This implies, first, that the ECM is altered during early innate immune responses. Second, that a collagen-binding site needs to be available in order for the cellular binding to ECM fibres via V 3 to occur, a defined site known to bind several proteins that can associate with collagen fibres (Farndale et al., 2008;Fields, 2014;Howes et al., 2014;Kalamajski et al., 2016;Manka et al., 2012;Reyhani et al., 2014;van Wieringen et al., 2010). Based on our present finding that CNE had no effect on P IF in naïve mouse dermis lacking 11 1 ( 11 −/− mice) or in wild-type dermis it can be concluded that integrin V 3 -directed processes that are operative in P IF control during homeostasis differ from the dynamic changes resulting from acute inflammatory reactions. Based on the induction of an acute inflammation in mouse dermis lacking 11 1 ( 11 −/− mice) not resulting in a lowering of P IF , the present findings with CNE suggest the need for a change of ECM build-up in order for the tissue to be able to respond to inflammatory insults by forming oedema.
In conclusion, the present data show that integrin V 3 can fully substitute for loss of collagen-binding 1 -integrins with regard to maintaining a homeostatic dermal P IF. Taken together with results presented by Svendsen et al. (2009), the data also imply that V 3 integrin-operated P IF control does not respond to acute inflammatory challenges and thereby does not enable oedema formation during innate immunity responses. Furthermore, our data show that whereas in normal dermis V 3 integrin-operated P IF control requires changes of the ECM build-up, they are not needed in dermis in which impaired collagen-binding 1 -integrin activity is a constitutive property.

ACKNOWLEDGEMENT
Dr Donald Gullberg (University of Bergen, Norway) is gratefully acknowledged for supplying the integrin 11 −/− mice and for critical comments.

COMPETING INTERESTS
None declared.

AUTHOR CONTRIBUTIONS
The experiments were performed in the laboratory space of the