TITLE: Direct sum of injective modules is injective
QUESTION [6 upvotes]: By the Bass-Papp Theorem, for a unital ring $R$, any direct sum of injective left $R$-modules is injective if and only if $R$ is left Noetherian.  I would like to restrict my consideration to an arbitrary abelian subcategory $\mathcal{C}$ of the category $R\text{-mod}$ of unitary left $R$-modules.  

We say that an abelian subcategory $\mathcal{C}$ of $R\text{-mod}$ is injectively closed if it satisfies the property that, for an arbitrary family $\left(I_\alpha\right)_{\alpha\in A}$ of injective objects in $\mathcal{C}$ such that $I:=\bigoplus\limits_{\alpha \in A}\,I_\alpha$ is an object in $\mathcal{C}$, $I$ is an injective object in $\mathcal{C}$.  Is it true that if $R$ is left Noetherian, then any abelian subcategory of $R\text{-mod}$ is injectively closed?  If not, can you please provide a counterexample?  Is there a sufficient condition for $\mathcal{C}$ to be injectively closed?  References are greatly appreciated.

For a nontrivial example, let $\mathfrak{g}$ be a finite-dimensional semisimple Lie algebra over an algebraically closed field of characteristic $0$ with a triangular decomposition $$\mathfrak{g}=\mathfrak{n}^-\oplus \mathfrak{h}\oplus \mathfrak{n}^+\,.$$
Denote by $\bar{\mathcal{O}}$ the full subcategory of the category of $\mathfrak{U}(\mathfrak{g})$-modules (where $\mathfrak{U}(\mathfrak{g})$ is the enveloping algebra of $\mathfrak{g}$) consisting of $\mathfrak{U}(\mathfrak{g})$-modules $M$ with the following properties:

$M$ is a weight module with respect to the Cartan subalgebra $\mathfrak{h}$,
each weight space of $M$ is finite dimensional, and
$M$ is locally $\mathfrak{n}^+$-finite (that is, $\mathfrak{U}\left(\mathfrak{n}^+\right)\cdot v$ is a finite-dimensional vector subspace of $M$ for any $v\in M$).

Then, $\bar{\mathcal{O}}$ is injectively closed.  (In this example, note that $\mathfrak{U}(\mathfrak{g})$ is both left and right Noetherian.)
P.S. I:  The Bass-Papp Theorem can be found, for example, in Theorem 3.39 on Page 123 of An Introduction to Homological Algebra by Joseph Rotman.
P.S. II:  I posted this question here as well, but didn't receive any answer.  I figured that MathOverflow users may be able to help.  I don't know how to move my Math.StackExchange post to MathOverflow.

REPLY [3 votes]: EDIT Thanks to Jeremy Rickard for several corrective insights in the comments!
In the direction of positive conditions, I'm not sure whether the following conditions are reasonable for your purposes:
Proposition: Let $\mathcal{C}$ be an abelian category which is 

locally finitely presentable (i.e. it has a generator of finitely presentable objects and is complete, or equivalently cocomplete) and
"strongly Noetherian" in the sense that any subobject of a finitely-presentable object is finitely-presentable, and similarly for higher presentability degrees.

Then $\mathcal{C}$ is injectively closed.
Proof sketch: One shows that in such a category, an object is injective if and only if it lifts against monomorphisms between finitely-presentable objects. This uses the "strong Noetherian" property. This implies, in conjunction with the Noetherianity property again, that injective objects are closed under filtered colimits; since they are also closed under finite direct sums, they are thus closed under arbitrary direct sums.
Corollary: Let $R$ be a "strongly Noetherian ring" in the sense that $R$-$\mathrm{Mod}$ is "strongly Noetherian", and suppose that $\mathcal{C}$ is a full abelian subcategory of $R$-$\mathrm{Mod}$ which is 

closed under kernels and colimits in $R$-$\mathrm{Mod}$, and 
generated under colimits by finitely-presentable $R$-modules.

Then $\mathcal{C}$ is injectively closed.
Proof sketch: Check that $\mathcal{C}$ satisfies the hypotheses of the proposition.
Notes:

As Jeremy Rickard points out, for countable $R$, $R$-$\mathrm{Mod}$ is "strongly Noetherian" iff it is Noetherian. So it seems that it is not that much stronger a condition than simply being Noetherian.
In Jeremy Rickard's example, where $\mathcal{C}$ is the subcategory of $R$-$\mathrm{Mod}$ given by the image of $S$-$\mathrm{Mod}$, it is the case that $\mathcal{C}$ is closed under kernels and colimits, and is itself locally finitely presentable. But it doesn't satisfy the hypotheses of the corollary because $\mathcal{C}$ is not generated under colimits by modules which are finitely-presentable as $R$-modules. Another way to say this is that the embedding $S$-$\mathrm{Mod} \to R$-$\mathrm{Mod}$ doesn't preserve finite presentability: the module $S$ itself is finitely-presentable as an $S$-module but not as an $R$-module.