In the United States, sepsis is the 6th and 7th leading cause of death in neonates and infants, respectively (Heron, 2015). While the mortality rate associated with sepsis in children has fallen over the years, it still remains unacceptably high at 5–25% (Hartman et al., 2013; Ruth et al., 2014; Weiss et al., 2015). Severe sepsis is defined as sepsis with cardiovascular distress and multiple organ dysfunction syndrome (MODS), which increases morbidity and mortality in this understudied patient population (Goldstein et al., 2005; Weiss et al., 2015).

The heart and kidneys are two organs commonly affected by sepsis. The development of both acute cardiac dysfunction and acute kidney injury during severe sepsis is categorized as type 5 CRS (McCullough et al., 2013). The etiology of sepsis-induced CRS in infants is poorly understood (Jefferies and Goldstein, 2013) but appears to be different than that in adults (Aneja and Carcillo, 2011; Deep et al., 2013; Ranjit et al., 2014; Zaky et al., 2014). For example, adults with septic shock typically present in a hyperdynamic state with high cardiac output and low systemic vascular resistance (warm shock) whereas infants more often exhibit low cardiac output and high systemic vascular resistance (cold shock) (Aneja and Carcillo, 2011). Treatment for sepsis-induced CRS focuses on interventions for the underlying systemic cause while considering functional support for the accompanying cardiac and renal dysfunctions (Brophy, 2008; de Oliveira et al., 2008; de Oliveira, 2010). Nevertheless, mortality rates for these critically ill children with CRS can still approach 50% (Romanovsky et al., 2013). The reason for poor outcomes is related to the fact that treatments are not specifically targeted and what supportive care is available is usually begun only after symptoms are present and after organ injury may have already been initiated (Simmons et al., 2012; Fortenberry et al., 2013).

Phosphodiesterase catalyzes the hydrolysis of cAMP and cGMP. In the cardiovascular system, increased levels of cAMP or cGMP can decrease vascular tone, strengthen endothelial tight junctions, and increase cardiac contractility. Tissue distribution of the large PDE superfamily is generally broad; however, differences in their regulation and selectivity (or not) for cAMP and cGMP can be exploited pharmacologically using selective PDE inhibitors (Maurice et al., 2014). For example, in the volume-resuscitated septic infant with acute cardiac dysfunction, PDE3 inhibitors increase levels of both cAMP and cGMP to improve cardiac contractility (Barton et al., 1996; Irazuzta et al., 2007; Meyer et al., 2011) and increase survival (de Oliveira et al., 2008; Brierley et al., 2009). PDE4 is selective for cAMP (Boswell-Smith et al., 2006; Maurice et al., 2014) and in adult animal model of sepsis, inhibitors of PDE4 reduce systemic vascular resistance (Carcillo et al., 1996), improve cardiac contractility (Thomas et al., 2003) and improve renal function (Carcillo et al., 1996; Tanahashi et al., 1999; Holthoff et al., 2013). PDE4 inhibitors also have potent anti-inflammatory activity and reduce microvascular leakage (Sanz et al., 2007; Schick et al., 2012), all of which could have added benefits in infants with severe sepsis.

The pathophysiology of sepsis-induced-CRS in infants is poorly understood due to limited studies in this patient population and the paucity of studies that have examined sepsis-induced MODS in age-appropriate animal models. To address this lack of understanding and the need for an appropriate animal model in which to conduct relevant preclinical studies, we developed a model of infant sepsis-induced cold shock in rat pups subjected to CLP (Seely et al., 2011). Here, we describe the development of type 5 CRS in this model of infant sepsis and then use of this model to evaluate the therapeutic potential of the PDE 4 inhibitor, rolipram.

There is growing appreciation that the adult animal is not the appropriate model for studying sepsis in the infant population (Solhaug et al., 2004; Seely et al., 2011; Wheeler et al., 2011; Deep et al., 2013; Fink and Warren, 2014; Efron et al., 2015). While no animal model can fully replicate the complexities of human sepsis, the CLP model in the rat pup displays key clinically relevant characteristics of human infant sepsis-induced MODS including increased lactate, microvascular hypoperfusion, cold shock with reduced cardiac output, and the rapid development of cardiac and renal dysfunction (type 5 CRS), all of which are associated with poor outcomes in this understudied patient population (Ceneviva et al., 1998; Aneja and Carcillo, 2011; Wheeler et al., 2011; De Backer et al., 2013; Casserly et al., 2015).

The clinical relevance of this model directed us to use it in preclinical studies to evaluate the therapeutic potential of the PDE4 inhibitor, rolipram. Our previous study in the adult aged mouse made septic by CLP showed that rolipram was effective in reversing renal injury (Holthoff et al., 2013). However, the adult mouse model is associated with profound hypotension, which is absent in the pup model (Seely et al., 2011). The differences in these rodent models are clinically relevant because approximately 60% of septic infants present with cold shock while approximately 90% of septic adults present with warm shock (hypotension and increased cardiac output) (Parker et al., 1987; Ceneviva et al., 1998). Hypotension in septic infants occurs later in the course of sepsis because infants compensate initially by vasoconstriction to maintain blood pressure (Ceneviva et al., 1998). This appears to also occur in septic pups, which receive fluids after surgery and again at 6 h to help maintain blood pressure. However, we have not measured blood pressure in septic pups beyond 18 h (Seely et al., 2011), a time when mortality is 50%. Inhibitors of PDE4 can reduce vascular resistance and protect the microcirculation (Carcillo et al., 1996; Holthoff et al., 2013), which is why preservation of renal peritubular capillary perfusion was used as the end-point for the dose-response study. An important finding was that the lowest most efficacious dose of rolipram in the septic adult mouse was lethal in the septic rat pup. Because inhibitors of PDE4 can decrease blood pressure (Thomas et al., 2001), higher doses could hinder cardiovascular compensation in septic pups and increase mortality. While additional studies are required to understand whether high-dose toxicity was due to hypotension, species differences, or pharmacodynamic/pharmacokinetic differences (Krause and Kuhne, 1988), the results of our dose-response study emphasized the importance of evaluating potential therapies in the appropriate age model.

Kidney injury developed rapidly following CLP. The decrease in renal peritubular capillary perfusion and GFR were profound as early as 6 h, a time when lactate levels were increased as well as the inflammatory cytokines TNF-α and IL-1β (Seely et al., 2011). Since therapy for the septic infant is begun only after the onset of symptoms, we evaluated rolipram using a delayed dosing paradigm begun at 6 h post-CLP. Rolipram not only restored renal capillary perfusion at 18 h, it restored renal perfusion velocity and prevented microvascular leakage. However, rolipram did not prevent mitochondrial oxidant generation by the tubular epithelium and did not restore GFR. This finding suggests that improving renal microvascular perfusion alone is not enough to restore GFR when the tubular epithelium is under oxidative stress.

We observed that impairment of cardiac function, as measured by cardiac catheterization, developed rapidly following CLP. Cardiac catheterization is the gold standard to assess systolic and diastolic cardiac function in pediatric patients. At 6 h LV dP/dt max, an indicator of systolic function was slightly but significantly decreased. This was not observed by echocardiography, which showed no significant changes in shortening fraction or ejection fraction at 6 h. Several studies demonstrate limitations of standard echocardiographic measures like M-Mode fractional shortening and ejection fraction in assessment of LV systolic dysfunction in septic children. Ranjit et al. (2014) showed that the majority of pediatric patients (68%) with septic shock had only mild to moderate impairment of LV function measured by echocardiogram. These findings are also consistent with several adult studies (Poelaert et al., 1997; Dalla et al., 2015). Although pediatric patients with septic shock often have compromised LV function, non-invasive assessment of ventricular dysfunction by echocardiogram might fail to detect the early changes before ejection fraction and cardiac output decrease. The decrease in stroke volume and cardiac output observed by echocardiography at 6 h in septic pups could be also related to diastolic dysfunction and limited filling related to low ventricular compliance. At 18 h, echocardiography showed slight but significant decreases in fractional shortening and ejection fraction, and a sustained decrease in stroke volume. These data suggest a worsening of systolic cardiac function over time.

Impaired LV diastolic function is a frequent occurrence in septic infants (Raj et al., 2014; Sankar et al., 2014). In agreement with these clinical findings, pups showed significant decreases in LV dP/dt min and increases in Tau at 6 h following CLP, suggesting severe diastolic dysfunction due to a direct effect of sepsis on cardiac muscle. Several studies have shown that Tau is a preload-independent measure of isovolumic relaxation of LV. It is independent of end-diastolic pressure, peak LV pressure, stroke volume and heart rate. Although Tau might be slightly influenced by very high or very low heart rate (Raff and Glantz, 1981), we did not observe significant changes in heart rate among groups during the catheterization measurements.

The decrease in LV function was also associated with elevated cTnl, a biomarker for cardiac injury. This finding is consistent with what has been reported in pediatric septic patients. Fenton et al. (2004) showed that elevated serum levels of troponin are associated with cardiac dysfunction in pediatric patients with septic shock and correlates with the severity of illness. Moreover, levels of cTnI are highest early in the course of sepsis (Fenton et al., 2004). Interestingly, LV function is often reversible in septic children and did recover by 18 h in septic pups. However, the nature of these experiments required that different groups of animals be used for measurements of cardiac function at 6 and 18 h. Moreover, measurements at 18 h were made in survivors, which were only approximately 50% of septic pups at that time point. Consequently, the data cannot suggest a true rate of recovery.

The ability of rolipram to prevent LV dysfunction at 6 h was impressive. To determine whether rolipram might be clinically useful in improving cardiac contractility in septic infants as are PDE3 inhibitors (Barton et al., 1996; Irazuzta et al., 2007; Meyer et al., 2011), rolipram was administered at the time of CLP and cardiac catheterization was performed at 6 h. All the measurements of systolic and diastolic function by cardiac catheterization were improved with no change in the mean arterial pressure. These findings were confirmed by echocardiography using clinically relevant delayed administration of rolipram. For these experiments, rolipram was given at 6 h and echocardiography was performed at 18 h. Even with delayed therapy, rolipram reversed the decline in stroke volume and cardiac output and prevented the decreases in fractional shortening and ejection fraction. These improvements in cardiac function suggest direct effects of rolipram on cardiomyocyte contractility and relaxation; however, additional studies are required to confirm this.

There are relatively few detailed studies describing the development of CRS in infants with sepsis (Jefferies and Goldstein, 2013) and, to our knowledge, none in a relevant animal model of infant sepsis. The rapid development of cardiac and renal dysfunction in this rat pup model is consistent with type 5 CRS. In both animal models of sepsis and postmortem septic human adult tissue samples, the generally mild changes in kidney and heart morphology observed by light microscopy do not explain the severity of kidney or heart dysfunction (Smeding et al., 2012; Takasu et al., 2013). In other forms of CRS, crosstalk between the heart and kidneys, which can worsen organ injury are better understood (Doi and Rabb, 2016). During sepsis, heart and kidney dysfunction occur essentially simultaneously so it is difficult to experimentally uncover interdependence. For example, the ability of rolipram to improve renal microvascular perfusion may have been, at least partially, due to an improvement in cardiac output. Studies from other types of CRS suggest that the sustained decline in cardiac and kidney function may be influenced by heart–kidney crosstalk through tissue release of cytokines and vasoactive hormones (Doi and Rabb, 2016). Additional studies are required to address this directly in this model.

The ability of rolipram to improve cardiac function and the renal microcirculation, even without improving GFR, is an important finding because it has been shown that organ failures have a cumulative effect on mortality and that reducing even one organ dysfunction can improve the prognosis of the septic patient (Martin et al., 2003). Still, the ultimate test of a new therapeutic for the septic patient is its effect on mortality. A single dose of rolipram administered after the onset of symptoms of CRS significantly improved 4-days survival. While additional studies are required to establish whether additional doses could have improved renal function and survival even further, the therapeutic potential of rolipram in this model of infant sepsis is encouraging.

There are several important limitations of our studies. One limitation is fluid status. All pups received fluids at the end of surgery (25 ml/kg of saline) and again at 6 h (40 ml/kg of saline). Both systolic and diastolic dysfunctions are dependent on load and fluid status (Vasan and Levy, 2000). Intravascular volume depletion reduces LV filling and could therefore result in what appears to be diastolic dysfunction. We could confirm that mean arterial pressure was unchanged in septic pups as we reported previously (Seely et al., 2011), and that hematocrit was unchanged; however, actual intravascular volume status was unknown. Regardless, rolipram clearly improved LV function and ultimately cardiac output. A second limitation is the use of short axis M-mode echocardiography to assess cardiac output in rat pups. While short axis M-mode measurements have been used previously in mice and children with some correlation with invasive measurements of cardiac output (Walther et al., 1986; Wallerson et al., 1990; Tournoux et al., 2011), these data must be interpreted cautiously. A third limitation is that we did not specifically address the mechanism of action of rolipram. PDE4 isoforms are present in rat and human cardiomyocytes (Johnson et al., 2012; Soler et al., 2015); however, the potential beneficial effect of PDE4 inhibitors in the human heart during sepsis remains to be determined (Eschenhagen, 2013). PDE4 is also expressed in various cell types in the kidney (Cheng and Grande, 2007). We know rolipram reached in the kidney even with delayed administration despite reduced GFR because renal capillary leakage was prevented. However, we do not know if rolipram reached pharmacologically active concentrations in the heart or whether the levels of cyclic nucleotides were increased. These studies are currently underway. While we cannot rule out the possibility that the anti-inflammatory actions of rolipram contributed to the clinically relevant improvements in organ function and survival, they cannot account for all its effects because inflammatory cytokines were already elevated at the time rolipram was administered (Seely et al., 2011).

In summary, our studies are the first to characterize cardiac and renal dysfunction in an animal model of infant sepsis-induced CRS. Our findings with rolipram suggest that PDE4 inhibitors, by targeting microvascular leakage and the inflammatory process in addition to cardiac contractility and vascular tone (Sanz et al., 2007; Schick et al., 2012), may offer therapeutic advantages over PDE3 inhibitors, such as milrinone, which is currently recommended for pediatric “cold shock” in the current American College of Critical Care Medicine Guidelines for Septic Shock Resuscitation (Barton et al., 1996; Irazuzta et al., 2007; Brierley et al., 2009; Meyer et al., 2011). These additional therapeutic targets would have added benefits in infants with sepsis.

While early goal-directed therapy intended to maintain systemic hemodynamics with the intent of preserving organ perfusion has been shown to reduce mortality in septic infants (Brophy, 2008; Brierley et al., 2009), the lack of targeted therapies remains a limitation of current management of the septic infant. Clinical trials for sepsis therapy have often resulted in a worsening of patient outcomes because preclinical studies often do not use age-appropriate models nor evaluate therapy for multiple organ dysfunction (Fink and Warren, 2014; Cohen et al., 2015; Sims et al., 2016). Using our model of type 5 CRS in the rat pup, we present preclinical pharmacological evidence that the PDE4 inhibitor rolipram may offer a more targeted approach to treating the septic infant.

CS, SM, SS, TN, and PM were involved in the study design and conception of the manuscript. CS, SM, and SS conducted the animal experiments and were responsible for data collection and analysis along with DZ, TN, NG, and PM. All authors contributed to the writing of the manuscript and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer YZ and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.