GILMAN, J., delivered the opinion of the court in which SUHRHEINRICH, J., joined. McKEAGUE, J. (pp. 450-55), delivered a separate dissenting opinion.
OPINION
RONALD LEE GILMAN, Circuit Judge.This case arises out of the drinking-water crisis in Flint, Michigan. The Plaintiffs are residents of the City of Flint who represent themselves and seek to represent a class of similarly situated individuals. They allege that they have been harmed since April 2014 by the toxic condition of the Flint water supply. The Plaintiffs filed suit against several City and State officials in the Genesee County Circuit Court, asserting various state-law tort claims.
Complete diversity of citizenship is lacking, and no federal question is presented on the face of the complaint. Nevertheless, four of the State officials who are present or former employees of the Michigan Department of Environmental Quality (the MDEQ Defendants) removed the action from the state court to federal court on two grounds. They first invoked the “federal-officer removal” provision under 28 U.S.C. § 1442(a)(1), contending that all of their conduct in question was performed under the supervision and direction of the United States Environmental Protection Agency (the EPA). Second, the MDEQ Defendants contend that the Plaintiffs’ claims necessarily implicate a substantial federal issue that merits federal-question jurisdiction under 28 U.S.C. § 1441.
The Plaintiffs objected to removal. They filed a motion seeking to have the district court remand the case back to the state court, which the district court granted. On appeal, the MDEQ Defendants ask us to reverse the remand order. For the reasons set forth below, we instead AFFIRM the judgment of the district court.
I. BACKGROUND
A. Factual background
Prior to April 2014, Flint had obtained its drinking water under contract with the City of Detroit. That month Flint switched its source of drinking water to the Flint River in order to save money.
In January 2016, several of the Plaintiffs filed a class-action lawsuit in the Gen-esee County Circuit Court. The complaint alleged state-law claims of gross negligence, fraud, assault and battery, and intentional infliction of emotional distress. According to the complaint, the MDEQ Defendants committed these tortious actions by allowing Flint to switch its water supply without using an anti-corrosive agent, despite knowledge that the water was “highly corrosive and unsafe.” This knowledge allegedly came from a 2011 report commissioned by the City of Flint, which concluded that the Flint River water could not be safely used for drinking unless it was treated with an anti-corrosive agent. Such an agent would be necessary to prevent lead and other chemicals from leaching into the water. The MDEQ *441received a copy of this report in 2013. But the MDEQ Defendants nevertheless allegedly failed to introduce corrosion-control treatments in a timely manner.
Just days after the switch in the water supply, the Plaintiffs allege that the City of Flint began receiving complaints “that the water was cloudy and discolored in appearance and foul in taste and odor.” Water users also began reporting physical symptoms that included hair loss, rashes, and vomiting -within weeks of the switch. Similar complaints were continually made to both the City of Flint and MDEQ officials for the next eight months. Numerous children in Flint were found to have elevated blood-lead levels by late 2014, and the MDEQ Defendants allegedly knew of this problem by early 2015. But the MDEQ Defendants, according to the Plaintiffs, did not reveal this problem to the public until after an August 2015 report was publicly disclosed.
The complaint further contends that the MDEQ Defendants deliberately ignored evidence that the water was unsafe and “falsely reassure[d] [the public] and insisted] that the water was safe even though they knew that the foul taste, odor and appearance was attributable to the highly corrosive Flint River water, untreated with the proper anti-corrosive agents.” It also alleges, among other things, that the MDEQ Defendants refused an opportunity to reconnect with Detroit’s safe drinking-water supply, falsely told the EPA in February 2015 that the Flint River water was receiving corrosion-control treatments, failed to maintain proper records needed to identify which water users had lead pipes, and failed to carry out proper water-quality tests.
In October 2015, the Governor of Michigan ordered Flint to reconnect with the Detroit water system. The Plaintiffs allege that, even after the reconnection, the MDEQ Defendants were grossly negligent in failing to properly monitor water quality and to protect residents from exposure when they knew that the water remained unsafe,
B. Procedural background
In April 2016, the MDEQ Defendants filed a notice of removal in the United States District Court for the Eastern District of Michigan. The MDEQ Defendants sought removal under two statutory provisions: (1) 28 U.S.C. § 1442, the federal-officer removal statute, and (2) 28 U.S.C. § 1441, which allows removal for state-law causes of action that raise substantial federal questions.
According to the notice of removal, federal-officer removal is appropriate because the MDEQ Defendants are being sued for actions that they took while acting under the direction of the EPA. The MDEQ Defendants assert that they were acting under the EPA because the EPA delegated primary enforcement authority to the MDEQ to implement the federal Safe Drinking Water Act (SDWA) in Michigan. Among the SDWA’s requirements is the submission to the EPA of quarterly and annual reports detailing the MDEQ’s compliance with the EPA’s Lead and Copper Rule (LCR). See 40 C.F.R. § 142.15. Under this primary enforcement scheme, the MDEQ Defendants contend that the EPA retains “tremendous oversight authority” over the MDEQ, including the ability to intervene or to withdraw primary enforcement authority in the event that the State fails to meet regulatory requirements.
The MDEQ Defendants also assert that they are being sued for actions that were “guided by repeated written and verbal dialogue with a number of EPA officers who advised and oversaw” the MDEQ’s regulation of the Flint water system. According to the notice of removal, the fact *442that the MDEQ Defendants were “acting under” the EPA “is most clearly demonstrated by” an emergency order that the EPA issued on January 21, 2016. The emergency order stated that the MDEQ and the City of Flint had failed to adequately respond to the drinking-water crisis, and it directed the MDEQ to take certain actions deemed necessary by the EPA. In addition, the EPA announced that it would begin conducting its own water-quality tests in the City of Flint and publishing the results on its website.
To support their alternative ground for removal, the MDEQ Defendants assert that the Plaintiffs’ claims will raise a substantial federal question: whether the MDEQ Defendants complied with the SDWA and the LCR. The MDEQ Defendants argue that the applicability of the SDWA and the LCR to the present case is ambiguous and that a need for uniform interpretation of those laws is a substantial federal interest supporting removal.
In response to the MDEQ Defendants’ notice of removal, the Plaintiffs filed a motion to remand the case back to the Genesee County Circuit Court. The district court granted the Plaintiffs’ motion and remanded the case. This timely appeal of the remand order is authorized by 28 U.S.C. § 1447(d) because the MDEQ Defendants removed the case under 28 U.S.C. § 1442. Our jurisdiction to review the remand order also encompasses review of the district court’s decision on the alternative ground for removal under 28 U.S.C. § 1441. See Lu Junhong v. Boeing Co., 792 F.3d 805, 811-13 (7th Cir. 2015).
II. ANALYSIS
A. Standard of review
We review de novo the district court’s determination that it lacked subject-matter jurisdiction and its consequent decision to issue a remand order. Smith v. Nationwide Prop. & Cas. Ins. Co., 505 F.3d 401, 404 (6th Cir. 2007). As the party seeking removal, the MDEQ Defendants bear the burden of establishing federal court jurisdiction. Eastman v. Marine Mech. Corp., 438 F.3d 544, 549 (6th Cir. 2006). Where, as here, the district court treats the motion to remand as a facial attack on the court’s jurisdiction, we look only to the pleadings—the complaint and the notice of removal—for the relevant facts. Gentek Bldg. Prods., Inc. v. Sherwin-Williams Co., 491 F.3d 320, 330 (6th Cir. 2007). This includes the consideration of exhibits attached to the pleadings “so long as they are referred to in the [pleadings] and central to the claims contained therein.” Rondigo, L.L.C. v. Twp. of Richmond, 641 F.3d 673, 680-81 (6th Cir. 2011) (quoting Bassett v. Nat’l Collegiate Athletic Ass’n, 528 F.3d 426, 430 (2008)). As this court has previously stated, removal statutes are to be strictly construed, and “all doubts should be resolved against removal.” Harnden v. Jayco, Inc., 496 F.3d 579, 581 (6th Cir. 2007); see also Eastman, 438 F.3d at 550.
B. Removal by the MDEQ Defendants under the federal-officer removal statute was properly denied.
The federal-ofScer removal statute allows removal of actions against “[t]he United States or any agency thereof or ány officer (or any person acting under that officer) of the United States or of any agency thereof, in an official or individual capacity, for or relating to any act under color of such office.” 28 U.S.C. § 1442(a)(1) (emphasis added). Persons like the MDEQ Defendants, who are not federal officers, must satisfy three requirements in order to invoke the federal-officer removal statute: (1) the defendants must establish that they acted under a federal officer, (2) those actions must have been performed *443under color of federal office, and (3) the defendants must raise a colorable federal defense. Bennett v. MIS Corp., 607 F.3d 1076, 1085 (6th Cir. 2010).
1. History of the federal-officer removal statute
The Supreme Court discussed the federal-officer removal statute most recently in Watson v. Philip Morris Cos., 551 U.S. 142, 127 S.Ct. 2301, 168 L.Ed.2d 42 (2007). A discussion of the history of the federal-officer removal statute begins the Court’s analysis. The Court noted that Congress enacted the original federal-officer removal statute during the War of 1812. Id. at 147, 127 S.Ct. 2301. That war was especially unpopular in New England, where many state court actions had been filed against federal customs officials whose duties included enforcing a trade embargo on England. Id. The Court explained that the “initial removal statute was ‘obviously . .■. an attempt to protect federal officers from interference by hostile state courts.’ ” Id. at 148, 127 S.Ct. 2301 (quoting Willingham v. Morgan, 395 U.S. 402, 405, 89 S.Ct. 1813, 23 L.Ed.2d 396 (1969)).
Originally, the federal-officer removal statute covered only federal customs officials and “any other person aiding or assisting” those officials. Id. (emphasis removed) (quoting Customs Act of 1815, ch. 31, s 8, 3 Stat. 198). Over time, Congress gradually expanded the scope of the statute, first to include persons assisting federal revenue officials, and later to include all federal officials and persons acting under them. Id. at 148-49, 127 S.Ct. 2301. But as the Court noted in Watson, these changes simply provided that more types of federal officials could take advantage of removal, with no indication that Congress intended to expand the scope of the words “acting under.” Id. at 149, 127 S.Ct. 2301.
Early uses of the somewhat broadened version of the federal-officer removal statute involved cases where people were killed when federal officers raided illegal distilleries. Id. at 149-50, 127 S.Ct. 2301. When murder charges were brought against the federal officers in state court, the federal officers were allowed to remove the cases to federal court. Id. In one of those cases, the Supreme Court reasoned that a private person acting as a chauffeur to the federal officers “had ‘the same right to the benefit of the removal provision as did the federal agents,’ ” even though the Court denied removal for other reasons. Id. at 150, 127 S.Ct. 2301 (quoting Maryland v. Soper (No. 1), 270 U.S. 9, 30, 46 S.Ct. 185, 70 L.Ed. 449 (1926)).
The Watson Court explained that these early cases “illustrate that the removal statute’s ‘basic’ purpose is to protect the Federal Government from the interference with its ‘operations’ ” that would occur if a federal officer could be tried in state court for a state offense related to the operation. Id. Concerns about “ ‘local prejudice’ against unpopular federal laws or federal officials” thus underlie the federal-officer removal statute. Id. As the Court explained, states that were antagonistic toward federal government operations might use state court proceedings to thwart the enforcement of federal law. Id.
2. The Supreme Court’s application of the federal-officer removal statute in Watson
Watson involved a cigarette manufacturer, Philip Morris, which filed a notice of removal based on the federal-officer removal statute. The plaintiff-smokers in Watson alleged that Philip Morris had engaged in deceptive business practices by using testing methods for its cigarettes that underreported the levels of tar and nicotine contained in the cigarettes. Concluding that the complaint challenged Phil*444ip Morris’s use of the federal government’s method for testing cigarettes, the district court held that the federal-officer removal statute applied, and the Eighth Circuit affirmed that decision. The Supreme Court reversed, holding that “the fact that a federal regulatory agency directs, supervises, and monitors a company’s activities in considerable detail” was not enough to establish that Philip Morris was “acting under” an officer of the federal government. Id. at 145,127 S.Ct. 2301.
Watson did not formulate a clear test for when the acting-under requirement is satisfied. But the Court set forth several considerations that are helpful in analyzing the present case. After reviewing the history of the federal-officer removal statute, as discussed above, the Court concluded that “the word ‘under’ must refer to what has been described as a relationship that involves ‘acting in a certain capacity, considered in relation to one holding a superi- or position or office.’ ” Id. at 151, 127 S.Ct. 2301 (quoting 18 Oxford English Dictionary 948 (2d ed. 1989)). The acting-under relationship “typically involves-‘subjection, guidance, or control.’ ” Id. (citing dictionary definitions of “under” as “subordinate or subservient to” or “subject to”). In addition, the Court explained that a private entity must be assisting the federal government in carrying out the government’s own tasks in order to invoke federal-officer removal. Id. at 152, 127 S.Ct. 2301. Simply complying with a regulation is insufficient, even if the regulatory scheme is “highly detailed” and the defendant’s “activities are highly supervised and monitored.” Id. at 153,127 S.Ct. 2301.
3. Watson supports the conclusion that the MDEQ Defendants were not “acting under” the EPA for puiposes of the federal-officer removal statute.
In the present case, the MDEQ Defendants emphasize that the EPA would have to enforce the SDWA in Michigan if the MDEQ did not have primary enforcement authority to do so. Watson indicates that this is a factor supporting removal. 551 U.S. at 153-54, 127 S.Ct. 2301. If this were the only factor for us to consider, then the MDEQ Defendants might well be entitled to remove the case under the federal-officer removal statute.
But we read Watson as requiring more. Specifically, Watson interprets “acting under” to mean that the defendant seeking removal must be in a relationship with the federal government where the government is functioning as the defendant’s superior. Id. at 151, 127 S.Ct. 2301. The Court went on to explain that the federal-officer removal statute was inapplicable to the facts in Watson because there was “no evidence of any delegation of legal authority from the FTC to the industry association to undertake testing on the Government agency’s behalf. Nor is there evidence of any contract, any payment, any employer/employee relationship, or any principal/agent arrangement.” Id. at 156, 127 S.Ct. 2301.
In the present case, the MDEQ Defendants argue that they were acting under the EPA because the EPA delegated to the MDEQ primary enforcement authority over the SDWA in Michigan. The district court disagreed, concluding that there is no contract, no employer/employee relationship, nor any other indication of a principal/agent arrangement between the MDEQ and the EPA.
We acknowledge at the outset that the MDEQ does receive funds from the EPA. But we conclude that the receipt of federal funding alone cannot establish a delegation of legal authority because finding such a delegation on that basis is way beyond the reasoning of Watson and would allow myr*445iad state agencies to invoke federal-officer removal.
We now turn to the question of whether the EPA’s delegation to the MDEQ of primary enforcement authority warrants federal-officer removal under Watson. Our research has revealed no caselaw from the Supreme Court, this court, or our sister circuits addressing the issue of whether state officers working in a joint federal-state regulatory system can invoke the federal-officer removal statute. As the Watson Court noted, federal-officer removal has been found to be applicable to certain government contractors. Id. at 158-54, 127 S.Ct. 2801. The Court suggested that contracts could delegate legal authority as required for federal-officer removal where a government contractor is helping to fulfill the federal government’s tasks. Id.
We further note that a government contractor entitled to removal would presumably be contractually required to follow the federal government’s specifications in making products or providing services. See Isaacson v. Dow Chem. Co., 517 F.3d 129, 134, 137-38 (2d Cir. 2008) (holding that a government contractor that manufactured Agent Orange for the government was entitled to federal-officer removal in a state-law tort action for injuries caused by the product, noting that the government had contractually required that the product be made to meet certain specifications). And a government contractor would not ordinarily have any authority to take actions beyond those specified in the contract. In these cases, the federal government can properly be viewed as acting in a superior relationship to the private entity.
The MDEQ Defendants insist that the government-contractor cases apply here. They heavily rely on a few such district court cases to support their argument, but these cases are distinguishable. In Clio Convalescent Center v. Michigan Department of Consumer & Industry Services, 66 F.Supp.2d 875, 876 (E.D. Mich. 1999), for example, the district court allowed a state agency to invoke the federal-officer removal statute because the agency had entered into an agreement with the Secretary of Health and Human Services to ensure compliance with Medicaid’s nursing-home certification requirements. That agreement explicitly provided that the state agency would act on the Secretary’s behalf. Id. The district court found that the Secretary was the real party in interest and that the state agency was acting as the Secretary’s agent, evidenced in part by the fact that the Secretary moved to intervene in the case. Id. at 877. A similar outcome was reached in Thompson v. Community Insurance Co., No. C39831999 U.S. Dist. LEXIS 21725, *12-13 (S.D. Ohio 1999), because that case involved a contractual agreement with an insurance company to act as a “fiscal intermediary” of the federal government in administering a Medicare program.
In the present case, there is no comparable contract or other delegation of legal authority to the MDEQ to act as the EPA’s agent or on the EPA’s behalf. Furthermore, this court has recently indicated, albeit in an unpublished opinion, that the absence of language allowing a private entity to act on the federal government’s behalf weighs against allowing federal-officer removal. See Ohio State Chiropractic Ass’n v. Humana Health Plan Inc., 647 Fed.Appx. 619, 623-24 (6th Cir. 2016) (explaining that a Medicare Advantage Organization could not invoke the federal-officer removal statute because the relevant regulations did not permit the organization to act on the federal government’s behalf, the organization had the freedom to use some innovative private-market techniques, the government was not controlling *446the entity’s daily operations, and the government would not have to perform the exact same tasks as the organization in the absence of the contractual arrangement).
Another case relied on by the MDEQ Defendants is City of St. Louis v. Velsicol Chemical Corp., 708 F.Supp.2d 632 (E.D. Mich. 2010). In that case, the district court allowed federal-officer removal by a defendant trust that had been created pursuant to a settlement under a federal environmental statute. Id. at 661-62. The trust’s objective was to resolve claims for cleanup costs and to manage the money allocated for the cleanup. Id. at 644-47. As in Clio, the federal government intervened in the action, arguing that removal was proper on the basis of the EPA’s control of the trusts. Id. at 661-62, 667-68. We conclude that Velsicol is distinguishable from the present case because it involved a trust—a relationship that required the defendant trustee to act “solely in a fiduciary capacity consistent with, and in furtherance of, the purposes” of the trust and the settlement agreements. Id. at 647 (quoting trust agreements). As previously explained, there is no comparable legal relationship here.
Instead, in order for the MDEQ to obtain primary enforcement authority, the state of Michigan chose to enact its own state safe-drinking-water laws. See 42 U.S.C. § 300g-2(a)(l) (providing that a state can obtain primary enforcement responsibility if, among other things, it “has adopted drinking water regulations that are no less stringent than the national primary drinking water regulations”). A-though Michigan’s laws had to comply with the standards set forth in the SDWA and its accompanying regulations in order for the MDEQ to obtain primary enforcement authority, Michigan was free to enact more stringent requirements for the protection of its residents. See id. That Michigan decided to enact a regulatory scheme that essentially mirrors the federal one does not mean that Michigan was required to do so.
Once the MDEQ received primary enforcement authority, it had to periodically submit reports to the EPA detailing compliance with regulations that had been adopted into state law. See 40 C.F.R. § 142.15. The MDEQ points to these reports as evidence of the EPA’s control over it, but Watson indicates that compliance reporting, even if detailed, is insufficient by itself to merit federal-officer removal. See Watson, 551 U.S. at 156, 127 S.Ct. 2301.
Similarly, the MDEQ Defendants vaguely allege that they were guided by “repeated written and verbal dialogue” with the EPA in taking the actions (and inac-tions) that are challenged in this case. We note, however, that counsel for the MDEQ Defendants conceded at oral argument that the EPA was not involved in the key action underlying the Plaintiffs complaint—approval of the decision to switch Flint’s water supply to the Flint River. And the notice of removal does not identify any specific actions or inactions alleged in the complaint that the EPA required the MDEQ Defendants to take or refrain from taking. Nor do the MDEQ Defendants explain how the SDWA required any of their actions or inactions, and they fail to identify any specific EPA officials who allegedly directed their conduct.
The MDEQ Defendants cite several communications with the EPA in their brief. But because the MDEQ Defendants did not cite these exhibits in the notice of removal, the communications are not among the pleadings that we can review on an appeal of the district court’s determination that it facially lacked jurisdiction. See Casias v. Wal-Mart Stores, Inc., 695 F.3d 428, 433 (6th Cir. 2012) (“When deciding a motion to remand, ... [t]he court may *447look to material outside the pleadings for the limited purpose of determining whether there are ‘undisputed facts that negate the claim’ ” (quoting Walker v. Philip Morris USA, Inc., 443 Fed.Appx. 946, 955-56 (6th Cir. 2011))). And even if we could review these communications, they simply suggest that the EPA and the MDEQ discussed issues related to the Flint water crisis, but not that the EPA ordered the MDEQ to take any specific action.
We also find telling the MDEQ Defendants’ statement in their notice of removal that the clearest indicator that they were acting under the EPA is the emergency order issued on January 21, 2016. That order, however, was issued two days after January 19, 2016, the date on which the Plaintiffs filed their complaint. So any actions that the MDEQ Defendants took pursuant to that order were obviously unrelated to the actions that the Plaintiffs challenge.
True enough, the EPA retains the ability to intervene when a state with primary enforcement authority fails to meet the requirements to maintain such authority. See 40 C.F.R. § 142.17(a)(2). But we disagree with the MDEQ Defendants’ argument that this ability to intervene supports them invocation of federal-officer removal. The fact that the EPA can intervene if a state fails to properly exercise its primary enforcement authority suggests that, absent any such failure by the state, the state retains the freedom to enforce its own safe-drinking-water laws and regulations. Michigan was so governing itself when the alleged actions and inactions giving rise to the Plaintiffs’ claims occurred.
4. Allowing the MDEQ Defendants to invoke the federal-officer removal statute would not serve the purpose of the statute.
According to the MDEQ Defendants, the MDEQ and the EPA are working together in a “joint Federal-State system” that purportedly supports federal-officer removal. But a joint federal-state system suggests to us that the MDEQ was working alongside the EPA, not under it.
We agree with the Plaintiffs that the MDEQ-EPA relationship is a model of cooperative federalism, not an agency relationship. See Philip J. Weiser, Towards a Constitutional Architecture for Cooperative Federalism, 79 N.C. L. Rev. 663, 671 (2001) (discussing the popularity of cooperative federalism, of which the environmental regulations of the 1960s and 1970s are a notable example, and explaining that “the real authority under such regimes often rest with the states which ultimately exercise considerable discretion,” even if the federal government sets forth the regulatory standards). The facts that the MDEQ Defendants allege in their notice of removal—such as the communications with the EPA and the MDEQ’s attempts to follow guidance documents issued by the EPA— suggest that a joint relationship exists, not that the MDEQ is controlled by the EPA.
We are also persuaded by the Plaintiffs’ argument that allowing removal in this case would “stand cooperative federalism on its head.” Our research has not found any caselaw from the Supreme Court or our sister circuits allowing officers of a state agency to invoke the federal-officer removal statute. If the MDEQ Defendants were given the benefit of that statute, then the benefit could be extended to many other state and municipal employees whose agencies work with federal agencies or receive federal funding. This would likely open a floodgate of federal litigation of state-law tort claims based on actions undertaken by state officers.
Where, as here, a state agency is sued based on state-law tort claims for actions *448taken while enforcing the State’s own statutes and regulations, we see no need for federal-officer removal. As discussed above, the purpose of the federal-officer removal statute is to protect federal officers from hostility toward the federal government or the enforcement of federal laws. The MDEQ Defendants might have been guided by the SDWA and its accompanying regulations in the present ease, but they were ultimately enforcing Michigan law, which was designed to mirror federal regulations.
More importantly, any bias that the MDEQ Defendants will face in state court differs from the historic biases that the federal-officer removal statute was designed to protect against. Unlike prior cases involving local resentment of the customs laws during the War of 1812 or of the actions of federal prohibition officers, this is not a case where a state court has any reason to be hostile to the federal safe-drinking-water laws. We acknowledge that the MDEQ Defendants are likely to be unpopular figures in Genesee County. But any bias that the MDEQ Defendants might face in state court would almost certainly come from the perception that their actions and inactions created the Flint water crisis by failing to guarantee that the Flint River water was safe to drink. We do not see how the MDEQ’s relationship with the EPA or the relationship between Michigan’s safe-drinking-water laws and the SDWA could be viewed as a source of bias.
Finally, we acknowledge the Watson Court’s statement that “[t]he words ‘acting under’ are broad.” Watson, 551 U.S. at 147, 127 S.Ct. 2301. But the Court also went on to explain that “[b]road language is not limitless. And a liberal construction can nonetheless find limits in a text’s language, context, history, and purposes.” Id. The Court then explained that the history of the federal-officer removal statute indicated that its purpose is to protect federal officers from state court hostility to the federal government, as discussed above. Id. at 148,127 S.Ct. 2301.
All three of the cases cited by Watson for the point that the statute’s scope is broad involve situations where the undisputed facts showed that the defendant was a federal officer or an employee of a federal officer. In Arizona v. Manypenny, 451 U.S. 232, 234-36, 101 S.Ct. 1657, 68 L.Ed.2d 58 (1981), for example, the defendant was a border-patrol agent employed by the Immigration and Naturalization Service who was indicted in state court for assault with a deadly weapon after shooting a man fleeing toward the border after the defendant told him to stop. There was no dispute that the defendant had satisfied the acting-under requirement. See id. at 236, 101 S.Ct. 1657. The Court noted that the statute was to be construed broadly in the context of discussing the fact that Arizona law would be applied in the federal proceeding. Id. at 242,101 S.Ct. 1657.
Similarly, in Willingham v. Morgan, 395 U.S. 402, 403, 89 S.Ct. 1813, 23 L.Ed.2d 396 (1969), the defendants were federal-prison employees who were sued under state tort law. The court remarked on the federal-officer removal statute’s broad scope in the context of concluding that its “color of office” language covered acts taken while the defendants were allegedly “on a frolic of their own” instead of being engaged in official duties. Id. at 404-10, 89 S.Ct. 1813 (quoting motion for remand). And Colorado v. Symes, 286 U.S. 510, 514-16, 52 S.Ct. 635, 76 L.Ed. 1253 (1932), involved the classic case of a prohibition agent who faced a murder prosecution in state court, which arose out of an encounter with a suspect who was openly drinking wine at a restaurant. Removal in these cases thus clearly served the historical *449purpose of the federal-officer removal statute. The present case is markedly different because the MDEQ Defendants are state officers who have been sued for actions and inactions related to their enforcement of state law.
We therefore conclude that the MDEQ Defendants were not “acting under” the EPA and thus are not eligible for federal-officer removal. Because we conclude that the acting-under requirement is not satisfied, we need not decide whether the MDEQ Defendants are being sued for actions taken under color of federal law or whether the MDEQ Defendants have raised a colorable federal defense.
C. This case does not present a substantial federal question that merits removal under 28 U.S.C. § 1441.
The MDEQ Defendants argue in the alternative that they are entitled to removal under 28 U.S.C. § 1441. Removal pursuant to § 1441 is proper where a state-law claim involves a federal issue that is “(1) necessarily raised, (2) actually disputed, (3) substantial, and (4) capable of resolution in federal court without disrupting the federal-state balance approved by Congress,” “which a federal forum may entertain without disturbing a eongressionally approved balance of federal and state judicial responsibilities.” Gunn v. Minton, 568 U.S. 251, 258, 133 S.Ct. 1059, 185 L.Ed.2d 72 (2013). As the Supreme Court has cautioned, however, removal based on the theory that a state-law claim raises a substantial federal question is appropriate only in “a special and small category of cases.” Id.
In the present case, the Plaintiffs have alleged state-law claims of gross negligence, fraud, assault and battery, and intentional infliction of emotional distress. The MDEQ Defendants argue that this case nevertheless meets all of the requirements for removal under 28 U.S.C. § 1441 because the complaint alleges that the MDEQ Defendants breached duties to the Plaintiffs that were based on the SDWA and the LCR. According to the MDEQ Defendants, the interpretation of the LCR is “inherently tied to” the determination of whether the MDEQ Defendants were negligent. They contend that there is a sufficient federal interest to merit removal because of the need to avoid inconsistent interpretations of the LCR. The Plaintiffs counter that the district court correctly concluded that this case does not raise a substantial federal issue.
We agree with the Plaintiffs and the district court. As the removing party, the MDEQ Defendants bear the burden of demonstrating that a basis for federal jurisdiction exists. The MDEQ Defendants have vaguely alleged that the Plaintiffs’ gross-negligence claims are inextricably related to the interpretation of the LCR because those regulations are “ambiguous and susceptible to multiple interpretations as applied to Flint.”
But the Supreme Court has held that, when a plaintiffs complaint raises garden-variety tort claims, “the presence of a claimed violation of [a federal] statute as an element of a state cause of action is insufficiently ‘substantial’ to confer federal-question jurisdiction.” Merrell Dow Pharm. Inc. v. Thompson, 478 U.S. 804, 814, 106 S.Ct. 3229, 92 L.Ed.2d 650 (1986) (concluding that a negligence per se claim based on alleged violations of a federal statute did not warrant federal jurisdiction); see also Grable & Sons Metal Prods., Inc. v. Darue Eng’g & Mfg., 545 U.S. 308, 318-19, 125 S.Ct. 2363, 162 L.Ed.2d 257 (2005) (reaffirming that Merrell Dow's reasoning applies in cases presenting garden-variety tort claims, but allowing removal on the basis that the issue before the Court in Grable was a relatively rare state *450quiet-title claim based directly on the effect of an IRS notice of seizure to satisfy a federal tax lien). Both Grable and Merrell Dow noted that allowing federal jurisdiction over typical negligence claims that implicated issues of federal law could dramatically increase the volume of federal litigation over state-law claims. See Merrell Dow, 478 U.S. at 811-12, 106 S.Ct. 3229; Grable, 545 U.S. at 319, 125 S.Ct. 2363. This court applied the holding of Merrell Dow post Grable in 2012, reiterating that state-law negligence claims arising out of violations of a federal statute do not merit federal jurisdiction. See Hampton v. R.J. Gorman R.R. Switching Co., 683 F.3d 708, 712-13 (6th Cir. 2012).
Furthermore, where Congress has not created a private cause of action for violations of a federal statute, there is less likely to be a substantial interest in favor of removal. Grable, 545 U.S. at 318, 125 S.Ct. 2363. And there is no private right of action for damages arising from a violation of the SDWA or the LCR. Harding-Wright v. D.C. Water & Sewer Auth., 350 F.Supp.2d 102, 106-07 (D.D.C. 2005) (explaining that, although private parties may bring lawsuits against violators of the SDWA under 42 U.S.C. § 300j-8(a)-(e), there is no private right of action for compensatory damages and, under Merrell Dow, there is no federal jurisdiction over state-law claims based on the SDWA).
In light of the aforementioned precedents, the MDEQ Defendants have not met their burden of demonstrating that federal jurisdiction exists. We therefore conclude that removal under 28 U.S.C. § 1441 is inappropriate in the present case.
III. CONCLUSION
For all of the reasons set forth above, we AFFIRM the judgment of the district court.
DISSENT